![2 Wind turbine system design
manufacturers are able to understand the load simulations in accordance with the
requirements of international standards, independently of the WT manufacturer, and
to generate relevant load information for the design process themselves. Although
the parameters of generic WTs are never 100% consistent with the parameters of
the manufacturers and generic loads are always subject to uncertainty compared to
manufacturer loads, generic WTs offer indispensable added value for suppliers and
research. After all, they offer the only way to determine and research the loads and
dynamics of WTs independently of the manufacturer.
Fraunhofer Institute for Wind Energy Systems IWES develops generic WTs
itself and has already published a 7.5 MW design, the onshore WT IWES Wind
Turbine IWT-
7.5-
164 [1]. The 7.5 MW WT has a rotor diameter of approximately
164 m, a hub height of 119.3 m (hybrid tower) and a total mass of 2004 t. The design
combines a large onshore WT with relatively low specific power and a focus on
detailed blade design with a load mitigation control strategy. Therefore, the WT was
developed for coastal locations with high average wind speeds and turbulence inten-
sity (IEC IA). The key figures of the WT can be found in Table 1.1.
An excerpt of previous use cases shows the potential of generic WTs. The IWT-
7.5-
164 was designed in the Smart Blades project [2] and used to investigate various
concepts for passive and active load reduction techniques. The loads generated were
later used for the IWES Highly Accelerated Pitch Bearing Test (HAPT) test bench
[3]. In the Towards an Advanced Design of Large Monopiles (TANDEM) research
project [4], it was used to calculate loads for the design of an XL monopile, in the
InterWiLa research project [5] for load simulations with new types of wind fields,
and for the optimised [6] and automatic [7] upscaling of the IEA Wind Task 30 OC4
Table 1.1 Summary of IWT-
7.5-
164 main properties
Description Value Unit
Rated electrical power 7542 kW
Rotor diameter 163.96 m
Hub radius 2 m
Arc length of the blade 79.98 m
Hub height 119.3 m
Cut-
in wind speed 3 m/s
Rated wind speed 11.7 m/s
Cut-
out wind speed 25 m/s
Rated tip speed 85.53 m/s
Cut-
in rotor speed 5 rpm
Rated rotor speed 10 rpm
Rated tip speed ratio 7.31 –
Blade cone angle 2 degrees
Shaft tilt angle 5 degrees
Mass of rotor-
nacelle assembly 536.78 t
Single blade mass 30.93 t
Specific power per rotor swept area 360 W/m²](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-31-320.jpg)
![Load calculation and load validation 3
semi-
submersible [8] and OC3 spar floater [9], respectively, from a 5 MW to the 7.5
MW WT.
This chapter deals with the requirements from the following standards for deter-
mining the loads of WTs:
•
• IEC 61400-
1 design requirements
•
• IEC 61400-
3 design requirements for fixed offshore WTs
•
• IEC 61400-
13 measurement of mechanical loads
Section 1.1 first introduces the topic of load calculation and then goes into load
determination requirements for onshore and offshore WTs. Topics such as taking
environmental conditions into account when determining the loads by simulation
and the procedure for determining the design loads are considered. At the end of
the section, two use cases for load calculation are presented. Section 1.2 deals with
the validation of the load calculation models. After the general procedure for model
validation is explained, the acquisition of measurement data is presented. Finally,
the specific procedure for validating the load calculation models is outlined.
1.1
Design loads of wind turbines
The design loads are determined in accordance with the requirements of interna-
tional standards. IEC 61400-
1 [10] for onshore WTs and 61400-
3 [11] for offshore
WTs are mentioned here as examples. The standards define requirements regarding
the consideration of environmental conditions and how these are to be considered
when designing specific components of the WT. The aim is to consider all operating
and extreme load conditions that are likely to occur during the service life of the WT
and their worst combinations when determining the design loads.
The standards contain requirements for the simulative determination of the
design loads. These requirements are sometimes very specific and sometimes leave
a lot of room for interpretation. Detailed simulation scenarios are prescribed, the
so-
called design load cases (DLCs), but the requirements for the numerical simula-
tion tools are only outlined schematically. With the so-
called fully coupled simula-
tion tools, the DLCs specified in the standards can be implemented numerically and
converted into time series. The time series contain the dynamic reaction loads and
deformations for a period of typically 10 minutes. Load spectra are extracted from
the time series and converted into component-
specific operating loads for a service
life of 20 or more years as well as extreme loads. Furthermore, the simulation pro-
vides the natural frequencies of the individual components and the coupled natural
frequencies of the overall WT system as well as the specific characteristic curves,
such as power and thrust depending on the wind speed.
The results of the load simulation are checked by measurements on the WT
in the field to show their validity. The load calculation and load validation are two
essential steps in the type certification process, which is essential for the commercial
sale of the WT.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-32-320.jpg)
![Load calculation and load validation 9
Each DLC uses horizontal and vertical oblique flow. Oblique flow is the devia-
tion of the main wind direction from the rotor plane. Horizontal oblique flow is
taken into account differently for operating load cases and standstill load cases.
While the maximum dead range of the yaw control, in which the yaw control is not
yet active (e.g., ±10 degrees), is to be examined in the case of operating load cases,
special attention is paid to the failure of the yaw control in the standstill load cases,
with angles of up to ±180 degrees. Here, DLC 6.2 in particular leads to numerical
instabilities with many load simulation tools at a 90-
degree oblique flow. Vertical
oblique flow takes into account the effect of terrain on the wind direction. It is rec-
ommended to turn the main wind direction downwards by 8 degrees against the
horizon for onshore load cases. Furthermore, the wind speed, which changes with
altitude, must be taken into account. In IEC 61400-
1, an exponential shear model is
specified with an exponent of 0.2 or 0.11 for onshore load cases, depending on the
wind model.
In addition to specifications on wind conditions, the standard contains other
requirements that must be taken into account. The natural frequencies for the tower,
drivetrain and rotor must be determined. If coupled system frequencies are in the
range of up to the sixth harmonic of the speed (6 P), further analysis must be carried
out to check for any resonance points and, if necessary, countermeasures must be
taken. If there is a risk of earthquakes at the potential installation site of the WT,
the load simulations must also be adapted accordingly. The same applies to regions
where ice formation on the rotor blades is to be expected.
1.1.1.2
Waves and current conditions
In the case of offshore WTs, also sea conditions need to be taken into account in
addition to wind conditions. The underlying standard IEC 61400-
3 [11], hence, is
mainly based on IEC 61400-
1 and extends it by offshore-
specific design require-
ments. Thus, in addition to the wind and operating conditions detailed in 0, different
waves, the directionality between wind and waves, the presence and type of sea cur-
rents as well as the water level need to be considered and are included in the offshore
DLC specification.
The waves occurring offshore are irregular and stochastic. They can therefore be
represented by a wave spectrum, which is based on a significant wave height, peak
spectral period and the mean direction of the waves. Similar to the wind conditions,
it is differentiated between normal (normal sea state, NSS), severe (severe sea state,
SSS) and extreme (extreme sea state, ESS) conditions with the corresponding nor-
mal (NWH), severe (SWH) and extreme (EWH) wave heights, and associated peak
spectral periods. While the SSS is only used in combination with NTM to represent
the severe conditions at a wave-
dominated site, the ESS in conjunction with EWM
reflects the extreme environmental condition that has a recurrence period of one year
or 50 years.
Sea currents are only taken into account in DLCs for ultimate strength analyses
and not for fatigue analyses. Depending on the type of environmental and site condi-
tion, it is differentiated between normal (normal current model, NCM) and extreme](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-38-320.jpg)
![12 Wind turbine system design
1.1.2.1
Rotor blade design
During the design process of the IWT-
7.5-
164 rotor blades, a selection of the five
DLCs 1.1, 1.2, 2.3, 6.1 and 6.2 were simulated in order to determine the loads rel-
evant to the design of the rotor blades. This selection of DLCs was then used to eval-
uate different blade design approaches. For this purpose, the DLCs were repeated
several times and the results were compared.
The loads along the complete rotor blade are required for the rotor blade design.
For the sake of clarity, only the results at the blade root are shown here. Fatigue
results in Table 1.4 are presented in terms of damage equivalent loads (DELs) cal-
culated for all force and moment components: Fx
, Fy
, Fz
,
Mx
,
Myand
Mz
, with the
following settings:
•
• 20 years WT lifetime
•
• 107
load cycles
•
• Wöhler slope exponent m of 4, 8, 10 and 14 for blade outputs
Table 1.4 Fatigue loads at the blade root of the IWT-
7.5-
164
m Fx
[kN]
Fy
[kN]
Fz
[kN]
Mx
[kNm]
My
[kNm]
Mz
[kNm]
4 261.5 549.9 557.3 11879.3 10469.2 304.9
8 252.6 420.6 464.5 9229.3 10623.8 409.0
10 261.0 399.4 461.1 8844.9 11193.0 458.9
14 281.2 377.7 471.1 8540.6 12456.6 535.4
Figure 1.2 Coordinate systems at tower and blade root](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-41-320.jpg)
![Load calculation and load validation 13
Ultimate loads are presented in terms of the minimum and maximum load com-
ponents located on the main diagonal of an ultimate load table with the correspond-
ing (same time step) load components listed in the rows. Each ultimate load table
also lists DLC names (with wsp: wind speed in m/s, yaw: yaw misalignment in
degrees, seed: wind seed), at which the ultimate loads occurred, with the corre-
sponding PSFs. For reasons of space, only the moments in the three spatial direc-
tions x
, yand zare shown in Table 1.5, and the forces are left out.
With the help of the load calculation, variables influencing the loads can also
be examined. The IWT-
7.5-
164 reference blade design was modified in two ways to
account for bend-
twist coupling (BTC). BTC couples the blade bending to the blade
torsion and automatically twists the rotor blade out of the wind at high bending.
This reduces the aerodynamic torque with large blade deflection and thus the loads.
Two BTC approaches were investigated: structural (SBTC) and geometric (GBTC)
bend-twist coupling.
Table 1.6 compares the relative change in fatigue loads at the blade root with
the reference design, as an example for S-
N slope
m = 8
. The three different blade
design methods affect the loads differently. With these exemplary numbers, GBTC
seems the most promising for load reduction. Only the torsional loads
Mzincrease
with BTC, which is to be expected.
In addition to evaluating different design approaches for rotor blades, the blade
root loads of the IWT-
7.5-
164 offer other practical uses. For example, they can be
used for the design of WT manufacturer-
independent test stands for which no other
loads are available. The IWT-
7.5-
164 blade root loads were used to design the IWES
Table 1.5 Extreme loads at the blade root of the IWT-
7.5-
164
DLC Mx
[kNm]
My
[kNm]
Mz
[kNm]
PSF
[-]
Mx
max DLC11_wsp25yaw0seed6 16656.7 −9789.1 −7.9 1.35
Mx
min DLC11_wsp25yaw-8seed5 −14939.6 −6242.1 −199.9 1.35
My
max DLC62_wsp50yaw300seed2 −1968.8 30363.4 728.5 1.10
My
min DLC11_wsp11yaw-8seed6 8983.3 −40819.8 −52.5 1.35
Mz
max DLC11_wsp13yaw-8seed2 −21486.5 −756.4 1231.7 1.35
Mz
min DLC62_wsp50yaw180seed4 −15477.1 −1359.5 −1194.0 1.10
Table 1.6
Comparison of fatigue loads for different blade design approaches at
the blade root of the IWT-
7.5-
164
Fx
Fy
Fz
Mx
My
Mz
Reference 100.00% 100.00% 100.00% 100.00% 100.00% 100.00%
GBTC 95.59% 99.95% 99.21% 100.04% 90.45% 104.69%
SBTC 99.58% 101.35% 101.34% 101.89% 97.78% 112.78%](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-42-320.jpg)
![14 Wind turbine system design
HAPT test stand [3], where endurance tests for large bearings are carried out.
The fatigue and extreme loads served as the basis for defining the capacities of
the test bench, with additional safety factors being taken into account so that the
test bench is also suitable for testing future bearing generations. In addition, the
IWT-
7.5-
164 loads were used for the planning and execution of the pitch bearing
test program.
1.1.2.2 Monopile design
The design process of bottom-
fixed offshore foundation structures usually consists
of the following three sequential steps:
1. Provision of representative cutting loads
2. Design of the monopile
3. Optimization of the overall system
In the example described below, the DLCs 1.1, 1.2, 3.1, 4.1, 6.1, 6.2 and 6.3
were simulated to determine the representative cutting loads at the transition piece
of the tower. The environmental conditions were chosen for a representative North
Sea location. The fatigue and extreme loads at the transition piece of the tower
for six degrees of freedom of the three spatial directions x
, yand zas well as the
resultant loads in the x − y – horizontal plane are shown in Table 1.7. The fatigue
loads were calculated for
107cycles in 20 years lifetime and for the Wöhler slope
exponents
mof 4, 5 and 6.
The extreme loads at the transition piece are shown in Table 1.8 for DLC 6.2.
For reasons of space, only the moments in the three spatial directions x
, yand z and
the resulting moment xyare shown, and the forces are left out.
The values of the extreme loads can be found on the main diagonal (shaded
grey in Table 1.8). The absolute largest and smallest values are given as well as
the subcase of the DLC in which the values occurred (with u: wind speed in m/s, y:
yaw misalignment in degrees, ww: wind-
wave misalignment in degrees, s: seed).
In addition to the extreme values, the other loads that occurred in the simulation
at the time of the extreme load are also entered. In addition to the loads, the PSF
is also listed.
Table 1.7
Fatigue loads at the transition piece of the IWT-
7.5-
164 in the tower
coordinate system
m Fx
[kN]
Fy
[kN]
Fxy
[kN]
Fz
[kN]
Mx
[kNm]
My
[kNm]
Mxy
[kNm]
Mz
[kNm]
4 408.5 873.7 583.0 1819.6 23583.4 18758.7 20532.9 20237.1
5 468.6 975.6 600.7 2113.7 29210.8 23378.8 22087.3 22992.4
6 523.5 1086.4 643.3 2353.9 37778.3 28556.1 25083.4 25550.5](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-43-320.jpg)
![Table 1.8 Extreme loads at the transition piece of the IWT-
7.5-
164 in the tower coordinate system
Load case Mx
[Nm]
My
[Nm]
Mxy
[Nm]
Mz
[Nm]
PSF
[-]
Mx
max DLC62_u37y30ww30s4 1.10E+08 1.29E+07 1.11E+08 3.42E+06 1.10
Mx
min DLC62_u37y240ww30s6 −1.10E+08 −2.97E+07 1.14E+08 −2.53E+06 1.10
My
max DLC62_u37y0ww0s1 2.00E+07 7.52E+07 7.78E+07 −5.86E+05 1.10
My
min DLC62_u37y180ww0s2 1.02E+07 −9.53E+07 9.59E+07 1.15E+06 1.10
Mxy
max DLC62_u37y120ww30s3 8.30E+07 −8.25E+07 1.17E+08 2.78E+06 1.10
Mxy
min DLC62_u37y0ww-30s1 −2.90E+04 1.79E+04 3.41E+04 −1.91E+05 1.10
Mz
max DLC62_u37y30ww-30s4 4.24E+07 −1.00E+07 4.36E+07 5.80E+06 1.10
Mz
min DLC62_u37y330ww-30s1 −7.03E+07 4.96E+06 7.05E+07 −6.79E+06 1.10](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-44-320.jpg)
![16 Wind turbine system design
1.2
Design load validation
Within the numerical design process, the design loads and system dynamics are
calculated and form the basis for the manufacturing process of the turbine. Before
being built into a serial product, the results from the numerical design process must
be validated on the prototype in the field. This provides proof that the real loads and
dynamics are within the limits of the numerical design.
Based on the numerical design, a prototype of the WT is built, on which the
type certification measurements are carried out. The type certification measurements
for load validation are specified in the standard IEC 61400-
13 ‘Measurement of
mechanical loads’ [12], as part of the IEC 61400 series, that comprises all IEC stan-
dards relevant for WTs. In Chapter 10, details of further IEC standards relevant to
the certification process of WTs can be found.
The standard of the American Society of Mechanical Engineers (ASME) for
Verification and Validation in Computational Solid Mechanics ASME VV 10
defines validation as the ‘process of determining the degree to which the model is an
accurate representation of corresponding physical experiments from the perspective
of the indented uses of the model’ [13, p. 3]. Accordingly, the case of intended use
for which the model is to be validated must be defined. However, the area for which
the model can be considered valid after successful validation is not defined by the
intended use domain but by the validation experiments that span the validation space
(see Figure 2.3-
2 in ASME VV 10 [13, p. 5]).
A general process for verification and validation is presented in the standard
ASME VV 10 [13] and illustrated in Figure 1.3*
.
This process should be carried out in parallel with the model development.
However, the different steps of verification of aeroelastic models, such as code veri-
fication and calculation verification, are not part of this book. This chapter focuses
on validation. First, the load measurements to receive the experimental outputs from
the physical experimentation branch are described. Then, the simulation results of
the modelling and simulation branch are described.
1.2.1
Standard load measurements
The measurements of mechanical loads in the design validation follow the standard
IEC 61400-
13. The quantities to be measured can be divided into three different
groups:
•
• Load quantities
•
• Meteorological quantities
•
• Operational quantities
*
Figure 3.3-
1 reprinted from ASME VV 10-
2019, by permission of The American Society of Me-
chanical Engineers. All rights reserved.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-45-320.jpg)
![Load calculation and load validation 17
Meteorological quantities, specified in the standard IEC 61400-
12-
1 [14], com-
prise wind speed, wind direction, TI, air density and wind shear for the lower rotor
half as mandatory parameters. For the load quantities, the minimum instrumentation
according to the standard is given in Table 1.9.
Operational parameters comprise electrical power, rotor or generator speed,
yaw misalignment, rotor azimuth angle, pitch position and speed, brake status and
WT status. The status values are usually taken from the SCADA system of the
Figure 1.3 Verification and validation process [13, p. 9]](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-46-320.jpg)
![Load calculation and load validation 21
1.2.2
Data evaluation process
During the measurement campaign, regular plausibility checks are applied to the
data. The end of the measurement campaign is reached with the completion of the
capture matrix. After a final plausibility check, the data evaluation process starts.
The final documentation is implemented by the report, which end is defined by the
IEC 61400-
13. Additionally, special measurement load cases (MLCs) have to be ful-
filled. Some of these MLCs correspond to specific DLCs. An overview of all MLCs
and DLCs is given in Table 1.2.
Before evaluation, certain data are rejected from the data set. This includes, e.g.,
wind directions outside the valid sector, defined by the site evaluation (for details,
see Chapter 10) or spikes in the data sets.
1.2.3
Standard load validation
Aeroelastic models, like all models, are subject to assumptions about the state and
physics they represent. In order to have confidence in the results of the models,
they must be validated. The standard IEC 61400-
1 specifies that load calculations
must be based on validated methods and approved codes. The aeroelastic simulation
model that is used for the specific design calculations must be subsequently vali-
dated by measurements on a dynamically and structurally similar turbine. However,
they may differ in detail, e.g., in alternative tower designs [10, p. 40].
1.2.3.1 General information
To validate models, output variables are generally compared between the model and
the real system to match the response of the model and the real system and to assess
whether the model validly describes the real system. This requires the model input to
represent the system and its excitation as accurately as possible. However, all model
input data are subject to uncertainties. The model input can be divided into system
data and surroundings data (cf. Figure 1.5). System data include geometry, initial
conditions and physical modelling parameters. Surrounding data include boundary
Figure 1.5 Source of uncertainty, adapted from Reference [15]](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-50-320.jpg)
![22 Wind turbine system design
conditions and system excitation. For WTs, the system excitation is derived from the
boundary conditions.
For aeroelastic models, geometry includes the geometry of the individual com-
ponents: blade and tower as well as the drivetrain, which represents the geometric
connection between blade and tower. The blade geometry includes chord length,
thickness, twist and pre-
bend along the blade. Physical modelling parameters
include density, elastic modulus and lift coefficients.
The real prevailing wind cannot be measured at every position at every time.
Often, only the measurements at one to five measuring points of a permanently
installed met mast are available. Therefore, in the validation of aeroelastic models,
especially the input of the system excitation is subject to uncertainty and poses a
great challenge.
1.2.3.2
Procedure for the validation of load calculation models
A procedure to perform the validation of the load model based on measurements
is provided in the informative Annex E.1 of IEC 61400-
13 [12, p. 81]. This pro-
cedure comprises 10 steps and is reprinted here with the kind permission of the
International Electrotechnical Commission (IEC)†
.
“1) Set validation requirements and acceptance criteria. Specify the
conditions for the tests and the acceptance criteria for the results. The
acceptance criteria are the maximum allowable differences between
the
measured loads and the simulated loads for equivalent wind condi-
tions. The acceptance criteria may be different for each load component
or test condition.
2) Specify the measurement setup which is needed to measure the desired
quantities. The starting point is of course the sensors specified in this stand-
ard [(i.e., IEC 61400-
13)] but the specific turbine design could require other
sensors or measurement techniques. Also special attention should be given
to the inflow conditions where additional sensors could be needed to get a
suitable understanding of the wind.
3) Assure the model is representative of the real turbine to be tested (main
structural data, natural frequencies, controller settings, sensor positions
etc.).
4) Rerun model for test turbine based on site specific inflow conditions to
ensure the structural integrity of the turbine during the test campaign to be
done afterwards.
5) Perform the load measurements campaign. Collect data about turbine
performances, loads and inflow conditions during the tests.
†
IEC 61400-
13 ed.1.0 “Copyright © 2015 IEC Geneva, Switzerland. www.iec.ch”](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-51-320.jpg)
![Load calculation and load validation 23
6) It is recommended to make a small comparison between measurements
and simulations early in the campaign to identify possible errors in the
measurements or the model.
7) Create a database for model validation by filtering the measured data.
Normally a capture matrix is already defined by the measurement Institute,
further filtering can be required to: reduce the number [of] data for a certain
wind speed, trending in the wind speed, yaw activity, data quality issues
etc.
8) Reproduce test conditions in simulations, by rerunning the model for
test turbine with the inflow conditions measured during the tests. Synthetic
wind time series can be created to match measured time history at some
grid points.
9) Compare simulated loads and measured loads. This comparison can
be done based on statistical values, frequency spectral density functions,
point by point time series, etc. (see Clause E.2 [in IEC 61400-
14]). Besides
load magnitudes, it is recommended to compare also other magnitudes as
power, rotor speed, pitch activity, control variables, etc.
10) If there are differences between simulations and measurements look
into potential reasons:
a) External conditions not captured in the model: low level jets, wind
shear, veer, TI, etc.
b) Other external conditions; controller settings, location of measured
loads vs. model, etc.
c) Measurement issues: calibrations mistakes, spikes, cross talks, etc.
d) Model issues: profile data, masses, stiffness’s, etc.”
Remarks on the proposed procedure are given below:
Step 1 states that the validation requirements and acceptance criteria should
be defined. Requirements for the acceptance criteria, which describe the maximum
deviation between measured and simulated loads, are not given in this standard.
In step 2, the measurement setup has to be defined. In addition to the study of
the turbine design in documents, a first inspection of the WT and the site is required.
For example, the presence of electric or hydraulic pitch requires different sensors, or
the unavailability of a slip ring or electrical power supply requires solutions for the
data and power transfer from the rotating to the non-
rotation system.
To assure that the model is representative of the real turbine to be tested, as
stated in step 3, the explanations from section 3.1 structural properties of sub-
components in the publication by Huhn and Popko [16, p. 3] could be used.
Step 4 is especially important if installed measuring equipment can lead to mass
and/or aerodynamic imbalance. The latter can be caused, e.g. by pressure sensors
installed outside the blade.
In step 5, the measurement campaign is conducted. The duration of the cam-
paign depends on the environmental conditions. In addition to the different load
cases, a so-
called capture matrix has to be filled during normal power production.
The measured data are categorized into wind speed intervals (wind bins) and TI](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-52-320.jpg)
![24 Wind turbine system design
intervals (TI bins). For each of these bins, a certain number of time intervals of 10
minutes have to be collected to fill the capture matrix and thus complete the mea-
surement campaign.
A comparison between measurement and simulation at an early stage of the val-
idation campaign, as mentioned in step 6, helps to identify possible differences, e.g.,
in the coordinate systems of measurement sensors and sensors from the simulation.
While adjustments and corrections (including renewed simulations) can usually be
carried out without much effort on the simulation side, this is usually not possible
afterwards for the real measurement.
Filtering the measurement data is an important step in validation. In addition
to the aspects mentioned in step 7, it is essential to filter by wind direction. If wind
measurements from a met mast are used, wind directions where the met mast is in
the wake of the turbine must be excluded. Furthermore, depending on the intended
use case, other wind directions should be excluded if necessary. If the focus of the
model validation is on the power curve, e.g. only the measurement sectors accord-
ing to section 6.3.3 and Annex A from IEC 61400-
12-
1 [14] should be considered.
If measured time series of the wind are used as input variables for the creation of
the wind fields, a so-
called narrow sector, as used by Dimitrov et al. [17], is advan-
tageous. The fact that the met mast is located exactly upwind of the WT increases
the significance of the wind measured at the met mast in relation to the wind expe-
rienced by the WT. In addition, special operating conditions that lead to a power
reduction or lower rotor speed must be filtered out. The following are examples:
•
• Noise reduced mode at night
•
• Safe mode of turbine control that could occur in conjunction with warnings
that may be triggered by high wind direction change or temperature warnings,
among other things
•
• Regulation from grid operator
•
• A shutdown can be triggered at certain times and seasons when bats fly near
the WT.
Step 8 mentions the reproduction of the test conditions. As mentioned in section
1.2.3.1, this is a particular challenge and key role in the validation of aeroelastic
WT models. The use of synthetic wind fields that satisfy the measured time series at
some grid points is a helpful element and especially necessary when the evaluation
is performed by point-
by-
point comparison. The tools TurbSim [18] and PyConTurb
[19] are suitable for the generation of these synthetic wind fields, which fulfil the
time series at the measured points. The wind profile can be measured using vertical
profiler light detection and ranging (LiDAR) systems. The mean values (over the
duration of the wind field) of the wind speeds at the different heights can then addi-
tionally be used as input for the creation of the synthetic wind field.
Step 9 deals with the comparison between simulation and measurement results
and mentions statistical values, frequency spectral density functions and point-
by-
point
time series as comparison methods. These are explained in more detail in Annex E.2
of IEC 61400-
13 [12, pp. 82-
85]. It should be considered that, on the one hand, the](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-53-320.jpg)
![Load calculation and load validation 25
comparison utilizing statistical values leads to a certain loss of information because
the simulation and measurement results of 10 minutes are condensed to one single
value. On the other hand, when comparing using the point-
by-
point method, it must
be ensured that the synthetic wind field also reflects the measurements over time (cf.
step 8). Furthermore, it must be taken into account that the point-
by-
point comparison
is made utilizing plots and thus always represents a subjective observation. In addi-
tion, the runtime length between the wind measuring point (e.g. met mast) and the WT
must be taken into account. For each individual 10-
minute measurement, this can be
determined, e.g., using a cross-
correlation function. For this purpose, on the one hand,
the measured wind speed signal at the wind measuring point and, on the other hand, the
signal of the electrical power can be used as an indicator for the wind speed in the rotor
plane (neglecting the rotor inertia). If synthetic wind fields that satisfy the time series at
the measured points are used, each measurement is reproduced by a specific simulation.
Pairs of measurement and simulation are created. In addition to the comparison meth-
ods presented in IEC 61400-
13, a linear regression can be performed from these simu-
lation and measurement pairs and used as an indicator of the model’s quality [20, 21].
Acknowledgements
“The author thanks the International Electrotechnical Commission (IEC) for per-
mission to reproduce Information from its International Standards. All such extracts
are copyright of IEC, Geneva, Switzerland. All rights reserved. Further information
on the IEC is available from www.
iec.
ch. IEC has no responsibility for the placement
and context in which the extracts and contents are reproduced by the author, nor is
IEC in any way responsible for the other content or accuracy therein.”
References
[1] Popko W., Thomas P., Sevinc A, et al. IWES wind turbine IWT-
7.5-
164. rev 4
[online]. 2018. Available from https://doi.org/10.24406/IWES-N-518562
[2] Teßmer J., Daniele E., Balzani C, et al. ‘Schlussbericht SB: smart blades:
01.12.2012-
30.04.2016’. 2016. Available from 10.2314/GBV:871472740
[3] Stammler M. ‘Endurance test strategies for pitch bearings of wind turbines’
[Dissertation]. Fraunhofer IRB-
Verlag, Gottfried Wilhelm Leibniz Universität
Hannover
[4] Leimeister M., Spill S., Dose B, et al. ‘TANDEM Schlussbericht: towards an
advanced design of large monopiles’. [in de] Research Gate. 2019. Available
from 10.2314/KXP:1678117404
[5] Thomas P., Wächter M., Antoniou A. ‘InterWiLa Schlussbericht: berechnung
und validierung von lasten an windenergieanlagen aufgrund intermittenter
windfelder’. [in de] 2018. Available from 10.2314/GBV:1028484186
[6] Leimeister M., Bachynski E.E., Muskulus M., Thomas P. ‘Rational upscal-
ing of a semi-
submersible floating platform supporting a wind turbine’.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-54-320.jpg)
![26 Wind turbine system design
Energy Procedia. 2016, vol. 94(1), pp. 434–442. Available from 10.1016/j.
egypro.2016.09.212
[7] Leimeister M., Kolios A., Collu M., Thomas P. ‘Larger MW-
class floater
designs without upscaling?: a direct optimization approach’. ASME 2019
38th International Conference on Ocean, Offshore and Arctic Engineering;
Glasgow, Scotland, 2016. pp. 06092019.
[8] Robertson A., Jonkman J., Masciola M, et al. Definition of the semisubmers-
ible floating system for phase II of OC4 [online] [technical report NREL/
TP-
5000-
60601]. 2014. Available from https://www.nrel.gov/docs/fy14osti/
60601.pdf [Accessed 2 Nov 2022].
[9] Jonkman J. Definition of the floating system for phase IV of OC3 [online]
[Technical Report NREL/TP-
500-
47535]. 2010. Available from https://www.
nrel.gov/docs/fy10osti/47535.pdf [Accessed 2 Nov 2022].
[10] IEC ‘Wind turbines-
part 1: design requirements’. [IEC 61400-
1] International
Electrotechnical Commission, Geneva. 2019.
[11] IEC ‘Wind turbines-
part 3: design requirements for offshore wind turbines’.
[IEC 61400-3] International Electrotechnical Commission, Geneva. 2009.
[12] IEC ‘Wind turbines-
part 13: measurement of mechanical loads’. [IEC 61400-
13] International Electrotechnical Commission, Geneva. 2015.
[13] ASME Standard for verification and validation in computational solid me-
chanics: ASME. VV 10. New York, NY: American Society of Mechanical
Engineers; 2019.
[14] IEC ‘Wind turbines-
part 12-
1: power performance measurements of electrici-
ty producing wind turbines’. [IEC 61400-12-1] International Electrotechnical
Commission, Geneva, Switzerland. 2017.
[15] Oberkampf W.L., Roy C.J. Verification and validation in scientific computing.
Cambridge: Cambridge University Press; 2010.
[16] Huhn M.L., Popko W. ‘Best practice for verification of wind turbine nu-
merical models’. Journal of Physics: Conference Series. 2020, vol. 1618(5),
p. 052026. Available from 10.1088/17426596/1618/5/052026
[17] Dimitrov N., Borraccino A., Peña A., Natarajan A., Mann J. ‘Wind turbine
load validation using lidar‐based wind retrievals [online]’. Wind Energy.
2019, vol. 22(11), pp. 1512–1533. Available from 10.1002/WE.2385
[18] Jonkman B. ‘TurbSim user’s guide v2.00.00: draft version’. 2016.
[19] Rinker J.M. ‘PyConTurb: an open-
source constrained turbulence generator’.
Journal of Physics: Conference Series. 2018, vol. 1037, p. 062032. Available
from 10.1088/1742-6596/1037/6/06203
[20] Huhn M.L., Gómez-
Mejía A.F. ‘Aeroelastic model validation with 8 MW
field measurements: influence of constrained turbulence with focus on power
performance’. Journal of Physics: Conference Series. 2022, vol. 2265(3), p.
032058. Available from 10.1088/1742-6596/2265/3/032058
[21] Conti D., Dimitrov N., Peña A. ‘Aeroelastic load validation in wake con-
ditions using nacelle-
mounted lidar measurements’. Wind Energy Science.
2020, vol. 5(3), pp. 1129–1154. Available from 10.5194/wes-5-1129-2020](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-55-320.jpg)
![Models and simulation 29
2.1.2
Requirements of standards for model fidelity
Due to the various levels of detail of the models that are capable of representing the
behaviour of WTs, national and international standards and guidelines have been
agreed on. These standards and guidelines provide a reference for the selection of
valid simulation models for a given task. They ensure that simulation results are
within a reasonable uncertainty band, regardless of the WT model or site analysed,
and they ensure comparability between different simulation tools by defining stand-
ardised models for certain tasks.
A crucial step in the design of a WT system is the iterative load simulation
process in order to estimate the extreme and fatigue loads of the individual compo-
nents (cf. Chapter 1). Since these simulation results are of high importance for both
the design and the certification process of the WT as a whole as well as its subsys-
tems and subcomponents, there are detailed requirements specified on simulation
models to be used for load determination and certification, mainly provided in the
International Electrotechnical Commission (IEC) standards. This chapter will focus
mainly on models described in the following references:
•
• IEC 61400-
1 [1]
•
• IEC 61400-
3 [2]
•
• IEC 61400-
4 [3]
•
• DNVGL-ST-0437 [4]
The first two elements, as well as the last one on this list, focus on the model require-
ments for global load analysis of onshore and (bottom-
fixed) offshore WTs. These docu-
ments give advice on how aeroelastic simulation tools should be implemented in order
to estimate the global WT loads used for design (cf. section 2.1.2.1). The third item,
however, focuses on the design of WT gearboxes (cf. section 2.1.2.2), therefore putting
a focus on higher dynamics and more complex mechanical models in comparison to
global aeroelastic simulation models. These gearbox models require a higher level of
detail with respect to mechanical components, so typically other and more specialised
MBS software is used.
2.1.2.1
Requirements for global load simulation
The international standard IEC 61400-
1 [1] defines the requirements for load simulation
of (onshore) WTs and specifies wind conditions and design situations by defining load
cases. Furthermore, for determining WT design loads, the use of dynamic aeroelastic
simulation models is required by the standard, and it is described what is sufficient to
meet these requirements. Suitable simulation codes should be validated subsequently by
measurements and must take into account at least the following types of loads:
•
• gravitational and inertial loads
•
• aerodynamic loads
•
• actuation loads, mainly caused by the WT controller](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-58-320.jpg)
![30 Wind turbine system design
•
• other loads, e.g. wake loads, impact and ice loads as well as vortex-
induced
tower vibrations
More specifically, IEC 61400-
1 lists relevant effects that should be taken into
account by using appropriate models, which will be described in this chapter.
Among the most relevant ones are wind field perturbations due to the WT (e.g. tower
shadow or wake-
induced velocities), three-
dimensional (3D) flow phenomena, such
as tip and hub losses as well as 3D stall, unsteady aerodynamic effects and the realis-
tic representation of the influence of the control system behaviour on the WT. These
requirements lead to the application of the fully coupled aero-
servo-
elastic simula-
tion tools that are typically used for load calculation in the WT design.
With respect to modelling the environmental conditions causing loading of the
considered WT during design load calculations, the IEC 61400-
1 standard mentions
two turbulence models for representation of the wind inflow: the Mann uniform
shear model and the Kaimal spectral and exponential coherence model. Icing of
the rotor blades shall be accounted for by applying profile coefficient modifications
to the airfoil characteristics as well as adding additional mass to the rotor blades.
Therefore, only model input parameter corrections, and not the implementation of
specific models, are required to consider rotor blade icing during load simulations.
Adding up to IEC 61400-
1, the international standard IEC 61400-
3 [2] focuses
on the additional assessment requirements to be considered for (bottom-
fixed) off-
shore WTs. Since the models used for representing the WT and the aerodynamic
wind loads are the same as for onshore WTs, this standard focuses mainly on the
influence of specific offshore loads, such as sea loads and sea ice loads. In particular,
the following additional loads have to be considered and therefore require specific
models in load simulation to capture aero-
hydro-
servo-
elastic dynamics:
•
• wave loads, considering different sea states as well as breaking waves
•
• sea currents
•
• differences in water level (e.g. due to tides)
•
• sea ice
•
• marine growth
•
• seabed movement, including scour
All loads shall be calculated utilising full dynamic simulations and the appropri-
ate modelling approaches. The IEC standard points out the relevance of correct rep-
resentation of various sources of system damping, such as aerodynamic damping,
hydrodynamic damping, structural damping or soil energy dissipation. Further guid-
ance on the calculation of hydrodynamic loads is given by referring to the Morison
equation as a standard method as well as the MacCamy-
Fuchs approach to account
for diffraction effects. The most relevant models for the above-
mentioned loads will
be presented in this chapter.
Besides the IEC standards, Det Norske Veritas (DNV), formerly DNV GL,
provides supplementary standards and recommended practices, such as the stan-
dard DNVGL-ST-0437 Loads and site conditions for wind turbines [4], which is](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-59-320.jpg)
![Models and simulation 31
applicable to onshore and offshore WTs. This standard defines normal and extreme
turbulence models, which differ from the models given in IEC 61400-
1 [2] to
account for offshore wind conditions. Furthermore, DNVGL-
ST-
0437 gives addi-
tional advice on the estimation of environmental conditions. With respect to the
aeroelastic WT simulation model, it points out the importance of non-
linear effects
to be accounted for where important (e.g. for soil–structure interaction), while linear
elastic theory is considered to be applicable for the structural dynamics, e.g. for
blade and tower/support structure elasticity. Some more specific requirements on
model fidelity, such as consideration of bearing friction moments, elastic machinery
mountings (if relevant) or rotor mass eccentricity, are mentioned as well.
2.1.2.2
Requirements for gearbox and drivetrain simulation
Detailed knowledge of the drivetrain dynamics is of major importance for the evalu-
ation of the design loads, and dynamic analysis of the drivetrain forms a mandatory
part of the certification of WTs. The primary aim of drivetrain dynamic models,
apart from component load calculation, is to investigate and avoid the occurrence
of possible resonances in the drivetrain during normal operation [5]. As drivetrain
models in aeroelastic codes are usually simplified with very few degrees of freedom
(DOFs) for the global WT analysis, the dynamic analysis of the drivetrain is mostly
performed in a separate analysis with a more detailed drivetrain model.
While the WT structural analysis methods have been well established over the
past decades, the methods for drivetrain dynamic analysis have still not achieved
the same level of maturity in terms of standardisation. The standard practices for
designing WT gearbox systems follow the Germanischer Lloyd (GL) guidelines
[6, 7], AGMA 6006 [8] and IEC 61400-
4 [3]. For standard practices of modelling
the drivetrain for dynamic analysis, very limited material is available. To tackle
this issue, GL has set the minimum modelling requirements for drivetrain dynamic
analysis, which are summarised in Table 2.1. These requirements emphasise the use
of multi-
DOF models. However, the GL guidelines also offer the possibility to use
the classic torsional drivetrain models in combination with complementary mea-
surements. These measurements can be carried out either at a gearbox test bench or
taken from drivetrain system testing. The reason for including these measurements
is to verify the modelling assumptions and to identify any missing eigenfrequencies
and mode shapes not covered by the simulation model.
2.2
Modelling of environmental conditions
WTs have to face various environmental impacts due to site-
specific wind and soil
conditions as well as, in the case of offshore systems, sea and marine conditions. An
overview of environmental conditions for the example of a bottom-
fixed offshore
WT system is presented in Figure 2.1.
The commonality of these environmental factors is their stochastic nature. Thus,
e.g. wind, waves and currents are never constant but fluctuate. Similarly, the soil
characteristics change over depth and time depending on further impacts, such as](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-60-320.jpg)
![32 Wind turbine system design
installation method, decommissioning or scour. Icing of blades or sea ice impact
only occurs in some climates at certain times of the year and is subject to random-
ness and the characteristics of nature. And also the marine growth on an offshore
WT structure is ever changing due to the living organisms, and its development
depends on the sea conditions and structural characteristics.
Overall, the environmental conditions are dependent on the specific site of the
WT, its surroundings, the elevation or depth, the time and season and further com-
plex correlations. Due to the random and stochastic nature of the environmental
conditions and their dependency factors, there is therefore always a certain degree
of uncertainty prevailing when trying to model environmental impacts and condi-
tions. In the following, different methods for representing environmental conditions
in numerical models are presented, grouped into wind (section 2.2.1), sea (section
2.2.2) and soil (section 2.2.3) conditions.
2.2.1
Modelling of wind conditions
For WT systems, the wind is actually the resource for energy production. Due
to its fluctuating character, the wind can be best described by a wind speed
Table 2.1
Minimum requirements for modelling drivetrain components for
dynamic analysis according to GL guideline [6, 7]
Component Minimum model Minimum degrees of freedom
Hub Rigid body Torsional, axial, bending
Main shaft Two rigid bodies
(flexible recommended)
Torsional, axial, bending
Low-speed shaft
coupling
Rigid body Torsional, axial, bending
Gearbox housing Rigid body
(flexible recommended)
Torsional, axial, bending
Planet carrier Rigid body
(flexible recommended)
Torsional, axial, bending
Gearbox shafts Three rigid bodies
(flexible recommended)
Torsional, axial, bending
Gearbox gears Rigid bodies Translational, axial, bending
Gearbox support Spring-damper element Translational
Brake disc Rigid body Torsional, axial, bending
High-speed shaft
coupling
Three rigid bodies
(flexible recommended)
Torsional, axial, bending
Generator rotor Rigid body Torsional, axial, bending
Generator housing Rigid body Torsional, translational
Generator legs Spring-damper element Translational
Main frame Rigid body
(flexible recommended)
In compliance with model of component
Bearings Spring-damper element Full stiffness matrix recommended](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-61-320.jpg)
![Models and simulation 33
time series. Most commonly, met-
masts and light detection and ranging (Lidar)
systems are used to record such site-
specific wind speed time series at different
heights and over the years. Additionally, the variable direction of the fluctuat-
ing wind speed is captured by measuring the three components (e.g. horizontal
x-component, horizontal y-
component and vertical z-
component) of the wind
speed vector separately. To reflect the characteristics of real wind speed time
series and naturally occurring wind-
related environmental conditions in numeri-
cal models, different aspects need to be included, which are addressed in the
following.
2.2.1.1
Wind turbulence and power spectral densities
The characteristics of a turbulent wind speed time series can be expressed by
the corresponding power spectral density. Traditional wind speed spectra con-
tain most of the energy at very low frequencies. Thus, the shape of a wind speed
spectrum resembles the curve of a power function with a negative exponent.
This is also reflected in the equations of wind spectral models, which commonly
depend on the frequency f
, the wind speed
Vhubat hub height zhuband the turbu-
lence intensity, mostly represented by the standard deviation of the wind speed.
Based on the example of the Kaimal model and considering only the longitudinal
component of the wind velocity (corresponding to index 1), the power spectral
density function
S1
f
is presented in (2.1) [1].
Currents
and tides
Scour
Marine growth
Soil mechanics
(Extreme)
waves
Sea ice impact
Tidal depth variantions
Wake
Icing
Turbulent wind and gusts
Figure 2.1
Environmental conditions at an offshore WT system ©Fraunhofer
IWES](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-62-320.jpg)
![34 Wind turbine system design
S1
f
= 0.052
1
ƒ1
Vhub
2
3
f 5
3
with ƒ1 =
8
:
0.7zhub for zhub 60 m
42 m for zhub 60 m
(2.1)
While the Kaimal model [9] and the Mann model [10, 11] are mentioned in the stan-
dard IEC 61400-
1 [1], tools for generating turbulent wind speed time series, such
as TurbSim by the National Renewable Energy Laboratory (NREL) [12], support,
among others, the Kaimal model and von Kármán model [13]. Beyond this, special
wind spectral models like the Frøya model [14] are recommended and supported for
representing extreme wind conditions, such as hurricane winds [12, 15, 16].
As mentioned before and apparent from (2.1), the turbulence intensity feeds into
the power spectral density function via the corresponding standard deviation. While
the turbulence intensity I
, which is derived from the standard deviation following
(2.2), decreases with the wind speed, the standard deviation of the turbulent wind
speed increases with the wind speed. Depending on the condition considered (distin-
guishing between normal and extreme turbulence models), the turbulence standard
deviation is computed differently.
I1 =
1
Vhub
(2.2)
2.2.1.2
Wind speed distributions in space and time
The wind speed distribution in space is more commonly referred to as wind shear. Due
to the non-
slip condition and in dependence on the roughness of the environment (i.e.
the roughness of the ground or sea surface if offshore), which is influenced by surround-
ing objects (e.g. the landscape, trees, buildings) as well, the wind speed increases from
low altitudes to higher ones and follows a logarithmic profile. This sheared profile can
be expressed by a power law, setting the wind speeds (
V
) and the associated heights (z)
in relation to each other, as given in (2.3). The power law exponent ˛depends on the
roughness as well as on the conditions considered, i.e. normal or extreme. Thus, values
of 0.2 and 0.14 are recommended for normal wind conditions onshore and offshore,
respectively, while a value of 0.11 should be taken for extreme wind conditions [1, 2].
V
z1
=
z1
z2
˛
V
z2
(2.3)
The distribution of the wind speed in time is the probability distribution of dis-
tinct wind speeds or wind speed bins. Commonly, the distribution is derived for a
specific site based on 10 min average wind speeds from measurement data over at
least a year or averaged over several years. If measurement data is not available or
a distribution function should be fitted to the measurements, a Weibull or Rayleigh
distribution is often used.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-63-320.jpg)
![Models and simulation 35
2.2.1.3
Extreme wind and gust models
Apart from the extreme turbulence model, there are other extreme events that need
to be reflected in the numerical models. Among them is the extreme wind speed
model, which aims at representing extreme wind speeds that are expected to occur
with a certain return frequency. Based on the design standards for WTs, return peri-
ods of 1 year and 50 years are considered. However, due to climate change and
more extreme environmental conditions, it is expected that even more extreme cases
would need to be taken into account when designing WT systems. Special consid-
eration must be given as well to regional extreme wind conditions like hurricanes,
typhoons or cyclones [15].
Other extreme wind conditions are, e.g. extreme operating gusts, meaning a
sudden and significant increase in wind speed for a very short period of time, or
extreme direction changes. These events could, of course, also happen simultane-
ously. Furthermore, the wind shear might be extreme as well and have an impact on
the loading over the rotor plane.
Another extreme and special condition can be the icing of blades. This is not
directly related to a wind or turbulence model; however, it needs to be considered
in the corresponding aerodynamic load calculations due to the affected shape of the
blades, as well as in the loads acting on the blades and entire WT due to gravity and
inertia.
2.2.1.4
Tower shadow and wakes
The ambient wind flow gets disturbed by the presence of a WT. The two main
aspects that need to be addressed at this point are the tower shadow and wake
effects.
While the wake of a WT mainly affects other WTs that are located in the lee
behind it, the tower shadow effect has a direct impact on the WT itself. The tower
influences the undisturbed ambient wind flow in such a way that the wind speed is
reduced at that point of time when a blade passes the tower. This affects the power
output and is visible in the 3P oscillations of the power output, considering a com-
mon three-
bladed WT. The tower shadow can be mainly modelled based on the flow
around a cylindrical structure and potential flow theory.
The wake of a WT, on the other hand, plays a key role in the wind farm’s aero-
dynamics. A WT that operates in the wake of another WT no longer experiences
an undisturbed inflow. The wake behind a WT is characterised by increased turbu-
lence intensity and reduced wind speed. These aspects mainly affect the loads on
the WT operating in a wake as well as the performance and power production. Due
to these wake effects, WTs are placed at certain distances to each other in a wind
farm, and the park layout is designed such that the main wind flow directions are
incorporated as well. To take the wake effects, including their amplification when
considering several rows of WTs, into account in the numerical modelling of a wind
farm, different models can be applied, which can be grouped into analytical or semi-
analytical models, such as the N.O. Jensen model [17], the Sten Frandsen model [18,
19] or dynamic wake meandering models [20–22] and CFD-
type models, such as](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-64-320.jpg)
![36 Wind turbine system design
linearised Navier–Stokes, parabolised Navier–Stokes, Reynolds-
averaged Navier–
Stokes (RANS), detached eddy simulations or large eddy simulation (LES).
2.2.2
Modelling of sea conditions
In the case of offshore WTs, additional environmental impacts have to be taken into
account. These may be sourced from waves, currents, sea ice and marine growth.
2.2.2.1 Wave models
Similar to the wind, waves are also stochastic. Thus, to represent the irregular char-
acteristics of the time series for the wave elevation, the power spectral density is uti-
lised. There are two most commonly followed approaches: the Pierson–Moskowitz
spectrum
SPMand the Joint North Sea Wave Project (JONSWAP) spectrum
SJ[2].
While the Pierson–Moskowitz spectrum, as presented in (2.4), represents a fully
developed sea, an extension and modification are established within JONSWAP to
represent a developing sea. Thus, additionally, the normalising factor
C
(2.5) and
the peak enhancement factor
˛
, which both depend on the peak-
shape parameter
(2.6), are comprised in (2.7) for the JONSWAP spectrum*
. This spectrum model,
however, is limited to only a low amount of swell being present and is mainly repre-
sentative of the North Sea and rather shallow waters.
SPM
f
= 0.3125 H2
s f 4
p f5
exp
1.25
fp
f
4
#
(2.4)
C
= 1 0.287 ln (2.5)
=
8̂
ˆ
ˆ
ˆ
ˆ
ˆ
ˆ
ˆ
ˆ
:̂
5 for
Tp
p
Hs
3.6
exp
5.75 1.15
Tp
p
Hs
for 3.6
Tp
p
Hs
5
1 for
Tp
p
Hs
5
(2.6)
SJ
f
= C
SPM
f
˛
with ˛ = exp
f fp
2
22f 2
p
#
with =
8
:
0.07 for f fp
0.09 for f fp
(2.7)
In any of these spectrum models, the irregular wave elevation time series is specified
by the significant wave height Hsand peak spectral period
Tp(or the corresponding
*
The JONSWAP spectrum can be transferred into the Pierson-
Moskowitz spectrum if the peak-
shape
parameter is set equal to 1.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-65-320.jpg)
![Models and simulation 37
peak frequency fp
). Depending on the specific type considered – distinguishing
between normal, severe or extreme sea states – the significant wave height and peak
spectral period are defined, so that, e.g. a 50-
year event can be represented in the
numerical models as well. Specific adjustments and enhancements may be required
to account for and represent other extreme conditions, such as hurricanes and the
associated wave fields generated [15]. Furthermore, the wave direction is essential
for describing the sea state, for which reason a directional wave spectrum, including
some directional spreading function, is often utilised.
Apart from the power spectral density-
based approach, the irregular wave ele-
vation time series can also be seen as the superimposition of a set of regular wave
elevation time series of different amplitudes (i.e. wave heights), frequencies (i.e.
wave frequencies which are related to the wavelengths and wave numbers as well)
and phase angles. Depending on the specific site and wave conditions, i.e. the water
depth, wave height and wave period, different wave theories are applicable for deter-
mining the wave kinematics [23]:
•
• Linear Airy wave theory, whose approximation – namely that the waves are
linear-
harmonic – is valid only for deep and intermediate waters under the con-
dition that the waves are not very steep;
•
• Stokes wave theories, which can be of different orders – depending on the num-
ber of additional harmonic waves added to the basic linear theory to include
corrections for non-
linear effects – but again are only applicable in deep and
intermediate waters, while they can represent steep waves and can, in the case
of deep waters and high-
order Stokes waves, even be applied up to the breaking
wave limit;
•
• Stream function theory of Dean, which has the same application range as Stokes
wave theories and also considers a set of harmonic waves, but simultaneously
rather than sequentially as in Stokes’ theory;
•
• Cnoidal theory, which accounts for non-
linear effects due to the limited depth in
shallow waters and, hence, is applicable in shallow waters as well as to steeper
waves in intermediate waters; or
•
• Solitary wave theory, which describes a soliton and represents in shallow and
also intermediate waters the very steep waves and waves that are about to break.
The maximum wave height for a certain wave period and at a specific water
depth is limited. At that limit, the wave is breaking.
Since wave kinematics just up to the still water level can be obtained following the
linear wave theory, commonly, stretching methods are applied supplementary to con-
sidering the actual water level elevation. Such wave stretching methods may follow an
extrapolation, a vertical stretching or apply the stretching method by Wheeler [24].
2.2.2.2
Current types and models
Sea currents can be classified by the depth range where they exist as well as the
source causing the current.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-66-320.jpg)
![38 Wind turbine system design
While currents that are generated by the wind are just felt to a limited depth
below the water surface and are, hence, called near-
surface currents, currents that
stem from storm surge, tides as well as density, temperature, pressure or salinity gra-
dients are present over the entire water depth and are, hence, called sub-
surface cur-
rents. Towards the seabed, however, the current velocity decreases due to friction.
In the preceding classification according to the respective depth range of occur-
rence, wind, storms, tides or any gradients are already mentioned as sources for cur-
rents. Beyond that, another source that causes a third current type has to be listed:
breaking waves. These wave-
induced surf currents are only present close to the
shore. Additional sources of currents may be specific local topographies or estuaries
as well as Coriolis forces or other causes leading to ocean circulations on a large
scale or eddy currents.
Depending on the current type, the corresponding current velocity may be mod-
elled as a constant, as a uniform flow in the horizontal direction for a certain depth
range below the water surface or as a velocity distribution over the depth following,
e.g. a linear or power function. The overall current impact may then be reflected by
the superimposition of the different single current models.
2.2.2.3
Modelling of sea ice
In cold regions or winter periods, sea ice may be present and, hence, act as another
environmental impact on an offshore WT.As for the other environmental conditions,
the occurrence and characteristics of sea ice highly depend on the site, time and
environment. Thus, it is always best if some real data exists that can be taken as a
basis for modelling sea ice.
One characteristic of sea ice is its thickness
hice
. After the frost season, this can
be approximated as a value in metres according to (2.8), with the 1-
day mean tem-
perature
mean
.
hice = 0.032
p
0.9Kmax 50
with Kmax =
X
days
ˇ
ˇmean
day
ˇ
ˇ for mean 0ı
C
(2.8)
Further characteristics of sea ice are, among others, its crushing strength, bending
strength and velocity of motion. In general, it is recommended to derive sea ice mod-
els and the behaviour of sea ice impacting the offshore structure based on ice model
tests and detailed site assessments [2].
2.2.2.4
Modelling of marine growth
Due to the presence of living organisms in the oceans, these may settle on parts of
the offshore structure, either those that are fully submerged or those that are just
periodically under water or in the splash zone. The amount, type and distribution
of living organisms on the support structure highly depend on the location of the
offshore WT.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-67-320.jpg)
![Models and simulation 39
Marine growth is commonly not represented by means of a separate model
but taken into account by adjusting certain parameters for the depths where marine
growth occurs. This implies larger outer dimensions of the affected structural parts
and – depending on the type of living organisms – modified surface properties, which
both are especially relevant for the hydrodynamic load calculation. Furthermore, the
additional mass due to the marine growth needs to be modelled as distributed mass
over the affected part of the support structure using an appropriate density (depend-
ing on the type of living organisms). In this way, the impact of marine growth on the
total system mass, its frequencies and dynamic response can be reflected.
2.2.3
Modelling of soil conditions
For bottom-
fixed (onshore or offshore) WTs, the soil is of substantial relevance for
the stability of the entire system. The soil properties are highly site-
specific and even
vary across the depth due to different soil layers. Some main soil characteristics,
such as stiffness, damping and shear strength, can be deduced from the complex
geophysics; however, geotechnical investigations and cone penetration tests should
preferably be conducted at the site of interest. While the soil–structure interaction
can simply be modelled as a clamped beam connection of the foundation to the
ground and, hence, represents just a rigid foundation, more detailed and advanced
soil models may be built based on the site-
specific soil parameters. Thus, the infor-
mation on the soil characteristics may be fed into a stiffness matrix to represent
the soil and soil–structure interaction. An apparent fixity method may be utilised to
reflect the rather flexible characteristics of the foundation. Even the p–y approach
can be followed in numerical modelling by the implementation of non-
linear springs
as well as dampers, distributed along the depth of the structure in the soil.
In the case of offshore WT systems, additional wave–soil–structure interactions
have to be taken into account. These include scour, which might occur locally or
also globally, as well as movements of the seabed, such as soil settlement or mud-
slides. Such effects may be reflected by adjustment of the soil characteristics and the
embedded length of the structure, which may be implemented in the numerical soil
models in a time-
dependent manner, allowing for consideration of any scour protec-
tive measures during maintenance and repair work offshore.
Other soil mechanics that may occur both on- and offshore are earthquakes.
The impact of earthquakes on WT structures is mostly represented by the resulting
soil acceleration on the ground. This may be derived from time series or a response
spectrum [1].
2.3
Fully coupled wind turbine modelling
To determine the load assumptions for WT system design, fully coupled simula-
tions of the WT are required by the standards, as described above in section 2.1.2.
In the following section, an introduction to the most commonly used models for
global load calculation will be given. Since WTs are dominated by the aerodynamic](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-68-320.jpg)
![40 Wind turbine system design
behaviour of their rotor blades, emphasis will be put on the description of the aero-
dynamic models.
2.3.1
Aeroelasticity and standard tools
Over the years, several codes have been developed for the purpose of aeroelastic
WT simulation. While some of them originate from other multi-
purpose dynamic
simulation software, most codes commonly used by both industry and academia
have been specifically developed for WT simulations. Among the most commonly
used and mentioned codes are Bladed [25], HAWC2 [26], Flex [27] as well as the
free and open-
source software OpenFAST [28]. These dedicated aero-
servo-
elastic
tools are used in industry, but some WT original equipment manufacturers (OEMs)
have also developed their own tailor-
made in-
house software. Such in-
house devel-
opments allow WT manufacturers to put a specific focus on real turbine challenges
that may arise during the design or validation stages and can be directly adapted in
code development.
These aeroelastic codes couple the calculation of aerodynamics with a struc-
tural solver in order to be able to determine the material loads at various locations
of the WT. The (time-
varying) loads from an aerodynamic model are applied to the
structural model to calculate its dynamic response. The deflections of the structural
components are in turn provided to the aerodynamic (and – in the case of offshore
WTs – also hydrodynamic) models to update the respective loads. For design load
estimation, WTs are modelled as aeroelastic multibody systems.
2.3.2
Aerodynamic models
Modern WTs use aerodynamically tailored rotor blades to transform the kinetic
energy of the wind into (rotational) mechanical energy that can in turn be used to
generate electricity. This section will describe the fundamentals of aerodynamics
and efficiency of horizontal axis WTs. The classical aerodynamic analysis for WTs
was developed at the beginning of the 20th century and is based on momentum
theory as well as blade element theory. Combining these two, creating the so-
called
blade element momentum (BEM) theory, allows the analysis of an idealised rotor.
Several additional correction models have been developed and implemented in order
to account for more realistic simulation results, as computation capabilities and WT
size have increased over the years. This section will focus on an introduction to
the BEM theory, since it is the most common modelling approach to determine the
aerodynamics of a WT rotor for design load calculation.
2.3.2.1 Momentum theory
The power conversion from the wind’s kinetic energy to mechanical energy of a
horizontal axis WT can be described using fundamental physical equations. First,
the WT will be described as an ideal non-
rotating actuator disc. In a second step, this
simplified one-
dimensional (1D) momentum theory will be extended to account for
rotational effects.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-69-320.jpg)
![44 Wind turbine system design
total far wake wind speed deficit of two-
thirds, i.e.
v1 = 1/3 V0
. Furthermore,
the power coefficient has a theoretical maximum of
Cp,max = 16/27 0.593
. This
means that no WT can extract more than approximately 60 per cent of the wind’s
theoretical kinetic energy. This important relationship is known as the Betz limit,
named after the physicist Albert Betz [29]. Although this finding is crucial for
understanding WT aerodynamics in general, the above-mentioned theory does not
consider the WT itself as a mechanical system but rather as an idealised extrac-
tion of power from the wind. To describe the influence and behaviour of the rotor
and its blades in more detail, further considerations need to be taken into account,
as will be explained in section 2.3.2.2.
Momentum theory with rotational effects
The idealised actuator disc model does not expect any rotor rotation and the cor-
responding torque, which acts both on the rotor as well as on the fluid flow. This
torque, as a reaction, will impose an additional rotational movement on the wake
flow, which is also present in the rotor plane, reducing the usable power that can
be extracted from the wind. First of all, the momentum theory is expanded to
describe this phenomenon. Therefore, an angular or tangential induction factor
Figure 2.4
Power coefficient as a function of the axial induction factor
according to momentum theory](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-73-320.jpg)
![46 Wind turbine system design
dM = rv1,td P
m = v1,tv2r2dr (2.23)
Again, with the tangential induction factor, this can be formulated as
dM = 4a0(1 a)V0r3dr (2.24)
2.3.2.2
Blade element momentum theory
Based on the described momentum theory, which focuses on the wind resource
as an idealised fluid stream, an additional theory was developed by Glauert in the
1930s that takes the influence of the rotor blade into account [30] and extends the
1D momentum theory with respect to the local forces that act on the blades. This
theory is best known as the BEM theory and is the classical formulation for WT
aerodynamics.
Figure 2.6 shows the local cross section of a rotor blade located in a stream
tube at radius r
. The blade has the chord length
cas well as local pitch angle ‚,
and the rotor rotates in such a way that the depicted blade cross section moves from
right to left. The blade cross section experiences a lift force FLand a drag force FD
,
combining to the total resulting force R
. This force can be expressed with a normal
component FNand a tangential component FT
, of which the latter one drives the
rotor rotation. All experienced velocities are seen from an observer moving with the
Figure 2.6
Local relative velocities and forces at an airfoil cross section
©Fraunhofer IWES](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-75-320.jpg)
![48 Wind turbine system design
dM = B r dFT (2.32)
Now, these forces, which were derived from the local velocities at the blade, describe
the same total loads on the annular stream tube as (2.22) and (2.24), which were
derived from the momentum theory. With some algebraic manipulations, combin-
ing (2.22) and (2.31), a formulation for the axial induction factor can be established:
a =
1
4 sin2 ˆ
rCn
+ 1
(2.33)
Here,
rdescribes the solidity of the local annular stream tube, which is the fraction
of the annular area that is covered by the rotor blades of chord length
c
:
r =
cB
2r
(2.34)
In the same way, the torque equations, i.e. (2.24) and (2.32), for the stream tube can
be combined to find an equation for the tangential induction factor:
a0
=
1
4 sin ˆ cos ˆ
rCt
1
(2.35)
Equations (2.33) and (2.35) are the classical formulation of the induction factors for
the BEM. With these formulations, the aerodynamics of a WT rotor can be com-
puted. Since the radial discretisation of the stream tubes assumes the stream tubes
to be independent from each other, each tube can be solved individually. Typically,
an algorithm that solves for the induction factors
aand
a0
is developed, as shown
below:
1. Initialise induction factors
aand
a0
, e.g. by taking the values from the previous
iteration step or assuming
a = 1/3and
a0 = 0as an initial guess.
2. Calculate the local angles ˆ and ˛using (2.26) and (2.25).
3. Determine
Cnand
Ctfrom tabulated data for
Cland
Cdfor the calculated ˛.
4. Calculate updated values
aand
a0with (2.33) and (2.35) and compare them with
the initial guess.
This algorithm scheme to solve the simple BEM is iterated until the predefined
tolerance criteria for the induction factors are met. Instead of solving for the induc-
tion factors, an equivalent algorithm can be formulated to solve for
Cland
a
. To
speed up simulation time, the number of iterations can be limited, which, however,
might result in not fully converged solutions in some cases. A modified solution
method for the BEM equations was suggested by Andrew Ning, which reduces the
two-
dimensional to a 1D problem [31]. Instead of iterating for the induction fac-
tors, the problem is re-
formulated to solve for the local inflow angle ˆ
, which can
increase the robustness of the solution approach as well as reduce the computation
time.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-77-320.jpg)
![50 Wind turbine system design
Turbulent wake state
As mentioned above in section 2.3.2.3, momentum theory, and therefore also classical
BEM theory, is only valid for axial induction factors smaller than 0.5, since this is the
limit for the wake flow field to have a positive velocity (cf. (2.17)). In reality, however,
the axial induction factor can exceed this theoretical limit since the flow field can become
more complex compared to what the simplified theory predicts. For such higher induc-
tion factors, momentum theory suggests that the rotor thrust has a maximum for
a = 0.5
(cf. (2.22)) and approaches zero for
a ! 1
, while in reality, the wake can turn into a
turbulent state and the thrust can increase above the theoretical maximum for
a = 0.5
.
To describe this turbulent wake state, a correction was originally proposed by
Glauert, establishing an empirical relationship between the non-
dimensional thrust
coefficient
CTand the axial induction factor
a[30]. The detailed description will not
be shown here, but using this empirical relationship, a correction for the axial induc-
tion factor can be applied if its value exceeds a certain limit
ac
, which is in this case
ac = 1/3
. A solution was given by Hansen [32], which also includes Prandtl’s tip loss
factor F
. For small axial inductions
a ac
, the standard BEM formulation is valid:
a =
1
4F sin2 ˆ
rCn
+ 1
(2.38)
For larger induction factors,
a ac
, the wake state can become turbulent and the
following solution can be applied:
a = 1
2
2 + (1 2ac)K
q
2 + (1 2ac)K
2
+ 4(Ka2
c 1)
(2.39)
where:
K =
4F sin2
ˆ
rCn
(2.40)
Dynamic inflow
The described aerodynamic theory so far assumes quasi-
static behaviour of the flow,
i.e. that the wake is in equilibrium with the aerodynamic loads acting on the rotor. If
there is a change in the rotor loads, e.g. due to a change in blade pitch angle or wind
speed, the wake will not instantaneously be in equilibrium according to momentum
theory. Instead, a dynamic transient to a new steady state will occur since the flow
cannot respond quickly enough. This effect is known as dynamic inflow or dynamic
wake and can be modelled in different ways. A simple and straight-
forward dynamic
inflow model was proposed by Øye [33], which acts as two filters connected in
series for the induced velocities and can be applied for both the axial and tangential
induced velocities. The dynamic inflow model consists of two first-
order differential
equations, taking into account time derivatives of the induced velocities:
Wint + 1
dWint
dt
= Wqs + 0.6 1
dWqs
dt
(2.41)](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-79-320.jpg)
![Models and simulation 51
W + 2
dW
dt
= Wint
(2.42)
In this model,
Wqsis the quasi-
static value of the induced velocities (both axial and
tangential) coming from the BEM algorithm,
Wintis an intermediate value for the
filter and
W is the final value, including the dynamic inflow correction. The time
constants
1and
2are fitted to:
1 =
1.1
1 1.3a
R
V0
(2.43)
2 =
0.39 0.26
r
R
2
1
(2.44)
where
ais the axial induction factor, Ris the rotor radius,
V0is the undisturbed
inflow velocity and ris the local radius of the considered rotor annulus. When cal-
culating
1
, the induction factor shall be limited to
a 0.5
.
Dynamic stall
Another aerodynamic effect occurring at WT blades, which is not considered in the
static BEM theory, is the dynamic separation and re-
attachment of the flow at the
airfoil, known as dynamic stall. The local angle of attack at an airfoil varies continu-
ously, e.g. due to wind turbulence and shear, tower shadow or yaw misalignment.
The static airfoil polar curves for
Cland
Cdconsider stalling of the blade at a fixed
static angle of attack. In reality, the flow separation is a dynamic phenomenon that
does not happen immediately but has a time delay, comparable with the dynamic
inflow effect. The timescale for the phenomenon of flow separation, however, is
usually faster than for dynamic inflow since it can be estimated with the order of
the time it takes the local wind speed to pass the blade chord. Dynamic stall occurs
when the angle of attack changes rapidly and delays the onset of stall as well as the
re-
attachment of the flow.
This trailing edge flow separation can be modelled with a separation function, as
suggested by Øye [34]. The model interpolates linearly between the lift coefficients
for the attached and the fully separated flows, such that the steady lift coefficient is
continuously restored. This interpolated lift coefficient can be described as
Cl = fs Cl,inv(˛) + (1 fs) Cl,sep (2.45)
where
Cl,inv(˛)is the (idealised) lift coefficient for inviscid flow without separa-
tion,
Cl,sepis the lift coefficient for fully separated flow and fsis the factor which
describes the degree of stall from 0 (fully separated) to 1 (no stall), which is given in
dfs
dt
=
f st
s fs
st
(2.46)
Here, fst
s is a value of the stall factor chosen in such a way that the static airfoil
behaviour is obtained when inserted into (2.45), while
st is a time constant. A
description of how to estimate the parameters in these equations can be found, e.g.,
in References 35, 36 and is not given here. The application of a dynamic stall model
in aeroelastic simulations has been observed to be important for the stability of the](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-80-320.jpg)
![52 Wind turbine system design
turbine, since otherwise blade vibrations might be calculated that do not occur in
reality [34].
Another important dynamic stall model often used in WT simulation is the
so-
called Beddoes-
Leishman model [35, 37]. This model includes more physical
effects of the dynamic stall phenomenon that are neglected by the Øye dynamic stall
model. It is, however, beyond the scope of this chapter to explain this model in more
detail here.
Further blade element momentum theory correction models
Besides the models described above, there are further correction models which
account for specific effects that are not represented in the classical BEM theory,
typically with respect to unsteady aerodynamics [38]. One example is a correction
for the case that the WT rotor is not aligned perfectly with the main wind direc-
tion. In such a yawed or tilted skewed inflow operating point, the induced velocities
vary with the azimuthal rotor angle since the blade tip will point either upstream or
downstream during one rotation, so that the blade will be deeper or less deep into the
wake. An overview of different implementations of such a skewed inflow correction
can be found in Reference 39.
Another correction with respect to the aerodynamics of a WT is the consider-
ation of the tower passage effect when a blade passes the tower. The reduced wind
velocity can be modelled based on potential flow theory, as mentioned in section
2.2.1.4. Furthermore, a BEM correction model has been suggested to account for
radial flow [40].
2.3.2.4
More complex aerodynamic models
The focus of the previous section was the introduction of the BEM theory and its
most important correction models, which is a comparably fast and robust low-
fidelity method to solve the aerodynamics of a WT. Since several hundred or even up
to thousands of single simulations need to be conducted for WT design load calcula-
tions, such efficient engineering models are widely used and therefore explained in
detail above. Besides the BEM theory, the more general formulation of the so-
called
generalised dynamic wake (GDW) theory is utilised in some simulation codes. It is
based on the potential flow theory and accounts for unsteady and 3D effects while
avoiding the requirement for an iterative solution. More information on the model-
ling of GDW theory can be found in Reference 41. There are, however, several other
ways which enable a more accurate estimation of WT rotor aerodynamics. Given the
current trend of ever-
increasing WT and rotor sizes, longer and more flexible rotor
blades might require improved simulation methods in the future.
Aerodynamic models that exceed BEM capabilities can usually be assigned to
either the category of vortex wake and actuator type models or to CFD models.
Compared to BEM models, vortex wake and actuator type models resolve the full
wake instead of just the rotor plane and are thus more computationally expensive
[42]. For certain applications, however, the increasing computational power of mod-
ern computers allows the utilisation of vortex wake models even for WT design load](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-81-320.jpg)
![Models and simulation 53
calculations, which can lead to significantly reduced rotor fatigue loads compared to
classical BEM codes [43]. Vortex wake and actuator type models can be seen as a
mid-
fidelity approach to modelling rotor aerodynamics.
Most details in aerodynamic simulations can be expected from high-
fidelity
CFD codes. These codes typically solve the 3D Navier–Stokes equations and pro-
vide a very detailed solution of the flow field around the blades as well as in the WT
wake. A challenge is the modelling of turbulence effects, which is either extremely
computationally expensive in direct numerical simulations or relies on turbulence
model assumptions to reduce the computational cost. In so-
called LES, only larger
scale motions are considered, while small eddies rely on specific models. A different
approach is the RANS equations, where time-
averaged Navier–Stokes equations are
solved, while different models have been proposed to capture the turbulence effects.
The simulation of WT aerodynamics with CFD codes of all types requires large
computational resources and is therefore not suitable for design load estimation.
2.3.3 Hydrodynamic models
The most traditional method for determining hydrodynamic loads on a structure
is the Morison equation [44]. Based on a semi-
empirical approach, the horizon-
tal component of the hydrodynamic load @Fon a section of a vertical cylindrical
structure (with length
@land diameter D
) is determined based on an inertial and
a drag component with corresponding coefficients
CMand
CD
, respectively, and
depending on the horizontal orbital velocity component
Uand corresponding water
particle acceleration P
Uas well as the water density
water
, as given in (2.47) [45].
The values for the inertial and drag coefficients need to be determined empirically
and depend on the flow and surface conditions. Thus, the drag coefficient may range
from 0.6 for large Reynolds numbers (105
) to about 1.2 for smaller Reynolds num-
bers (105
), while a value of 2.0 may be considered for the inertial coefficient for
small Keulegan–Carpenter numbers (10) and a value of 1.5 for larger Keulegan–
Carpenter numbers (
10) [46].
@F = waterCM
D2
4
P
U@l + 1
2
waterCDDU
ˇ
ˇU
ˇ
ˇ @l (2.47)
The Morison equation, however, assumes that the structure is hydrodynamically
transparent, meaning that the structure experiences a wave impact, whereas the
wave itself is not altered due to the presence of the structure. This assumption is
valid for non-
moving structures that exhibit a very small diameter in relation to the
wave length.
With a more advanced approach based on the potential flow theory [46], some
of these limitations can be overcome. According to this theory, the total velocity
potential is a summation of the potentials of the incident waves, the diffracted waves
as well as the radiated waves. The hydrodynamic load on an offshore structure is
then derived from the pressure distribution, which is itself determined based on the
velocity potential. In numerical modelling, the potential flow theory may be realised
by using the boundary element method. An alternative method, which is also based
on the potential flow theory, is the MacCamy–Fuchs approach [47]. This is based on](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-82-320.jpg)
![54 Wind turbine system design
the Morison equation and extends it: The inertial coefficient is determined in depen-
dence on the structural diameter and the wave number, and the phase shift between
the incident wave and the resulting force is taken into account. Thus, it is no longer
assumed that the wave is not affected by the structure, but radiation and diffraction
effects are now accounted for.
While waves feed into both inertial and drag forces, currents – if assumed to be
constant – only lead to an additional drag force resulting from the current velocity.
Additional hydrodynamic impacts might be impulse loads due to horizontal or ver-
tical wave slamming and slapping, breaking waves close to or at the structure and
wave run-
up. Tailored approaches are required to capture these complex hydrody-
namic phenomena [2, 15, 16].
Another alternative is CFD, by means of which a much higher level of fidelity in
modelling the hydrodynamic effects is achieved. CFD approaches are based on the
Navier–Stokes equations and can represent even complex non-
linear flow phenom-
ena. The underlying equations are the momentum equation (2.48) and the continuum
equation (2.49) for incompressible flows and Newtonian fluids, with the flow veloc-
ity vector
U
, the Nabla operator r
, the time
t
, the pressure p
, the Laplace operator
, the dynamic viscosity
and the volume and gravitational force vector F. These
Navier–Stokes equations are numerically approximated by means of CFD methods.
water
@U
@t
+
U r
U
= rp + U + F
(2.48)
r U = 0 (2.49)
2.3.4
Modelling of structural components
Depending on the purpose of the analysis, different modelling approaches might
be used for representing the structural parts of a WT. In general, the mathematical
model should be as simple as possible and only as complex as necessary to account
for the relevant physical effects while being computationally efficient. Depending
on the purpose of the model, a simple 1D mass spring system might be sufficient to
compute the elastic response, while for other purposes, an in-
depth stress analysis
of, e.g. the blade root trailing edge might be necessary, requiring a detailed descrip-
tion of the local geometry and material distribution as well as an exact knowledge
of the aerodynamic loading scenario. Due to the subject area of this book on WT
system design and therefore the calculation of design loads, this section will focus
on the description of models that are used for aeroelastic simulation. More complex
finite element method (FEM) models are often used in structural analysis of blades,
tower and other mechanical components (see also section 2.4.1.1), but it is beyond
the scope of this section to explain this method here in more detail.
In fully coupled aeroelastic WT simulations for design load calculation, the
drivetrain is most often modelled as a torsional spring–damper system with addi-
tional rotational inertia for the generator, low-
speed shaft (LSS) and high-
speed
shaft (HSS). Some implementations additionally consider bending of the main
shaft as well. For direct-
drive WTs, it is also common to disable any torsional](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-83-320.jpg)
![Models and simulation 55
drivetrain dynamics and just connect a rotational generator inertia rigidly to the
rotor hub. Different modelling approaches of varying complexity for WT drivetrains
are described in more detail in section 2.4. In the following, however, this section
focuses on suitable simulation models for rotor blades and tower since these are the
most important structural components of a WT.
2.3.4.1
Blade and tower modelling as beams
The main structural components that make up a WT are the rotor blades and the
tower. These components not only mainly define the optical appearance of a WT,
but they are also the most important structural components with respect to creating
a model of an (onshore) WT. WT blades and towers are both tall and slender struc-
tures that have one dimension significantly larger than the other two, so physically
it is a good assumption to be considered as beams. While the tower is mostly a sym-
metrical steel tube, the blade shape and, therefore, the blades’ bending behaviour
are more complex compared to a tower, but the beam-
like shape is comparable. To
combine the structural WT component models for time domain simulations, typi-
cally a multibody approach is chosen, where different bodies, which might be rigid
or flexible, are connected to each other with certain kinematic joints or links (see
also section 2.4.1.2).
During operation of a WT, the rotor blades experience deflections in flapwise
(out-
of-
plane) as well as in edgewise (in-
plane) direction. Furthermore, especially
the large and flexible blades of modern WTs experience considerable elastic tor-
sional deflections as well as rigid body blade pitch rotations around their main axis.
Similarly, WT towers experience large fore–aft bending moments as well as side–
side deflections and torsion due to elastic yaw rotations of the rotor-
nacelle assem-
bly. To allow for these types of loading in simulation models, WT blades and towers
are usually represented as a 1D beam model with beam elements. For more detailed
design simulations, 3D FEM shell element models are often used, which allow for
local stress determination. However, this section will focus on the beam modelling
approach in the following, since this simplified approximation is computationally
faster and therefore suitable for design load calculation.
Commonly used beam models are the Euler–Bernoulli beam model, as
described, e.g. in References 48, 49, and the Timoshenko beam model [50]. Both
models account for axial and bending loads as well as torsion of the beam. While the
former neglects shear deformation and is therefore not suitable for shorter beams,
the Timoshenko beam model includes the deformation of the local cross section
due to shear in its mathematical formulation. This, however, makes it slightly more
computationally expensive, so that the simpler Euler–Bernoulli beam model has
been the standard beam model in aeroelastic load simulation in the past [51]. This
section will focus on blade modelling since the blades usually require more detailed
models compared to the tower. However, the same model may be used to model the
tower as well.
To apply the beam models, a discretisation method is required. Most often, one
of the following methods is used [52]:](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-84-320.jpg)
![56 Wind turbine system design
•
• Modal reduction method: In a preprocessing step, the flexible eigenmode shapes
of the beam are calculated based on a static FEM implementation. During the
dynamic simulation runtime, the beam deflection is approximated by a linear
combination of the calculated mode shapes. This method solves computation-
ally efficiently but may be less accurate for larger blades with higher deflections
due to its linearity and the limited number of DOFs.
•
• Multibody dynamic approach: Several (usually flexible) beam elements are
interconnected by kinematic joints or force elements. This method allows for
increased DOFs to be considered in the model.
•
• 1D FEM approach: The beam model is represented as a number of finite ele-
ments (FEs) connected by nodes. It requires higher computational effort but
provides a good approximation for the deformation of the structural component.
As the implementation of a dynamic structural model for aeroelastics can vary
significantly with respect to mathematical implementation as well as computational
complexity, only a brief overview will be presented in the following. Most beam
models rely on the assumption of small deflections, which may not be valid for large
and increasingly flexible rotor blades. Therefore, more complex theories, such as the
geometrically exact beam theory (GEBT), have been developed that account for the
geometric non-
linearities occurring at large deflections [53]. For further implemen-
tation guidance, the reader is encouraged to have a deeper look into some example
implementations of structural models for aeroelastic simulation [54–56].
2.3.4.2
Parametrisation of beam models
The parametrisation of a blade (or tower) beam model requires information on the
stiffness and mass distribution along spanwise stations of the beam, which are usu-
ally an input to the aeroelastic simulation model and can be computed with specific
preprocessors or blade/tower design tools from the respective geometry and design.
The mass distribution may also contain additional masses not originating from the
blade structure itself, such as blade ice accretion. The structural properties of a spe-
cific cross section of the rotor blade can be described by a
6 6mass matrix and
stiffness matrix, respectively. The stiffness matrix Kdescribes the relation between
the vector of forces and moments Fand the vector of elastic deflections and rota-
tions
uat the cross section:
F = Ku (2.50)
The stiffness matrix contains information on axial stiffness EA
, shear stiffness
kGA
,
bending stiffness EIand torsional stiffness
GJ
. Typically, some of the DOFs show a
structural coupling, so that the stiffness matrix can contain off-
diagonal terms, e.g.,
for structural bend–twist coupling.
The mass matrix Mdescribes the inertial behaviour of the cross section. It relates
the vector of forces and moments Fto the linear and angular accelerations R
u
, as given
in (2.51).
F = MR
u (2.51)](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-85-320.jpg)
![Models and simulation 57
Figure 2.7 shows a cross section of a blade and the most important geometrical
properties. The elastic centre is the point at which a normal force pointing in the
main beam direction does not lead to additional bending of the beam. Analogously,
the shear centre is the point at which a transversal force does not result in torsional
deflection but in pure bending. Pure bending about an applied bending moment will
occur if the bending moment is aligned with one of the principal axes.
2.3.4.3
Equation of motion
The generalised equation of motion of a discretised mechanical system can be writ-
ten as follows [54]:
MR
q + CP
q + Kq = Fg (2.52)
Here, M,
Cand Kare the mass matrix, damping matrix and stiffness matrix, respec-
tively, while
Fgis the generalised vector of external forces acting on the system. If
the generalised loads are known, e.g. from solving the aerodynamic problem, (2.52)
can be solved for the system accelerations R
q
. This set of generalised coordinates
q
corresponds to the DOFs of the system.
By representing the deflection shapes as linear combinations of the sys-
tem’s eigenmodes, the number of DOFs can be reduced to increase the com-
putational speed of the simulation. This is a common approach in aeroelastic
codes, where typically the first 6–12 eigenmodes of the blades are used and
4–10 eigenmodes of the tower. However, these numbers should only be seen
as a rule of thumb, since they depend strongly on the modelled system and the
simulation purpose. For such a modal approach, each generalised coordinate qi
corresponds to a (modal) deflection shape
ui
. The Craig–Bampton method [57]
is often used to identify the eigenmodes and corresponding mode shapes of
the system. The total deflection state can be described by superposition of the
selected number of considered mode shapes N
:
u(q) =q1u1 + q2u2 + : : : + qNuN (2.53)
Figure 2.7 Structural cross section of a blade ©Fraunhofer IWES](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-86-320.jpg)
![58 Wind turbine system design
Besides the described modal reduced approach, some codes use a full (1D) FEM
representation of the blade (and tower) beams. As explained above, this method
increases the accuracy of the model as well as computational costs and is becom-
ing more and more the state of the art for design load calculations of modern WTs
[51]. The derivation of the dynamic equations for such a model is summarised in
Reference 36. More details on the non-
linear equations of motion and the corre-
sponding implementation as a state-
space model can also be found in Reference 58.
An implementation of a multibody discretisation is shown in Reference 59.
2.3.5
Modelling of other components
Besides the already discussed component models, the most visible parts of a
WT, which have not been discussed yet, are the hub and the nacelle system. In
aeroelastic simulations used for design load calculation, these parts are usually
assumed to be rigid. Therefore, lumped masses are used to model the inertia of
the hub, including pitch system and blade bearings, as well as for the nacelle
system with the masses of main shaft supports, gearbox housing, generator sta-
tor and the electrical installations in the nacelle.
The drag of the rotor hub and the nacelle is typically accounted for by apply-
ing simple drag coefficients. The same holds for the tower drag, which includes the
BEM induction factors as described in section 2.3.2.2.
The electrical system of a WT is usually not considered in aeroelastic simula-
tions. The influence of grid faults, such as sudden power losses, however, can be a
relevant source for fatigue loads on the rotor-
nacelle assembly. To account for such
grid faults, a sudden drop of the generator torque can be applied in the aeroelastic
simulation model. For short-
time grid losses, so-
called fault ride through events,
the WT does not shut down during the generator torque loss, so that high loads can
occur. Depending on the type of the electrical configuration of the WT, this can rep-
resent the mechanical excitation of the system in an adequate way to reproduce the
structural loads acting on blades, hub, tower and foundation.
2.4
Detailed modelling of wind turbine drivetrains
The drivetrain of a WT is located inside the nacelle and consists of all components
from the main bearing to the electrical generator and power conversion system. It is
responsible for converting rotor kinetic energy into electrical power. Modern WTs
are designed to have a minimum operational life of 20 years with high reliability.
Unfortunately, this has not been the case, as studies have revealed high rates of
failure in WTs, with drivetrain-
related failures causing the longest downtimes [60].
Furthermore, due to the complexity of WTs disassembling procedures and, in many
cases, remote locations with difficult access, such frequent failures often result in
costly repairs. As a result, understanding the causes of such failures and reducing
their occurrence to improve system reliability have been the most important areas of
research in recent years.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-87-320.jpg)
![Models and simulation 59
NRELconducted extensive case studies in the Gearbox Reliability Collaborative
(GRC) project to determine the root causes of drivetrain-
related failures. One of the
major causes of premature drivetrain failures, according to these studies, is a lack
of a system-
level approach and incorrect estimation of drivetrain loads during the
design process [61, 62]. Limiting uncertainties in drivetrain component load estima-
tions during the design phase can aid in the development of reliable designs. Hence,
it is critical for design engineers to have simulation models that can provide realistic
load estimates. The following sections discuss the modelling methods for WT drive-
trains that are widely used in research and the wind industry.
2.4.1
General modelling approaches, methods and tools
Before going into details, it is worth mentioning the general goals or requirements
that determine the modelling approach or method. The selected modelling approach
might depend on the following criteria:
•
• Analysis scope
–
– Material-level analysis
–
– Component-level analysis
–
– System-level analysis
•
• Analysis types
–
– Static vs dynamic
–
– Linear vs non-
linear
–
– Explicit vs implicit
•
• Physical domain
–
– Mechanical
–
– Electrical
–
– Hydraulic
–
– Thermal
–
– Fluid
•
• Analysis goals
–
– Load response
–
– Durability study
–
– Noise, vibration and harshness (NVH)
–
– Control
–
– Condition monitoring
•
• Coupling type
–
– Fully coupled analysis
–
– Decoupled analysis
–
– Partially coupled analysis
Each of the aforementioned criteria requires specific modelling methods and has
various fields of application and research domains. For the purpose of WT drivetrain
design,a system level approach with sufficient modelling depthis necessary for complete](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-88-320.jpg)
![60 Wind turbine system design
depiction of the important drivetrain dynamics and load response. The available meth-
ods for modelling WT drivetrain can be classified into the following categories:
•
• Finite element method
•
• Multibody simulation
•
• Bond graph methods
•
• Block modelling
2.4.1.1
Finite element method
The FEM involves discretisation of the physical domain into a finite set of nodes
and elements [63]. FEM models can have DOFs varying between hundreds and
several millions depending on the model size and complexity. Some of the commer-
cially available general purpose FEM software packages include Abaqus Unified
FEA, Ansys®
, Altair HyperWorks®
and COMSOL Multiphysics®
. The FEM model-
ling method is suitable for detailed analysis of stresses, strains, contact and damage
for systems undergoing small deformations and motions. It is the primary method
for detailed component-
level analysis of the drivetrain subsystems. However, FEM
models require long computation times and are not suitable for performing system-
level dynamic simulations of the entire drivetrain system.
2.4.1.2 Multibody simulation
In MBS, the system is modelled as a combination of rigid bodies that interact with their
surrounding via constraints and force elements. Each rigid body is defined as a lumped
mass having a maximum of six DOFs. The flexibility of the individual components
is modelled as spring elements with estimated stiffness of the components. The final
system results in a set of ordinary differential equations. MBS is a widely practiced
method for dynamic simulation of mechanical drivetrains as it can accurately replicate
system dynamics and load effects while requiring little computational resources.
The standard rigid body approach in MBS can be further enhanced with flexible
bodies using modal reduction techniques. This optimally blends the advantages of the
FEM and MBS approaches. The flexible MBS is currently the most popular method
for modelling WT drivetrain systems. The commercially available MBS softwares
like Adams™, Simpack and RecurDyn include the flexible MBS approach.
2.4.1.3
Bond graph methods
Bond graphs are a graphical modelling method for dynamic physical systems. Bond
graph models consist of elements (representing certain physics) connected together
via bonds (representing power flow between the elements). It features a uniform
notation for all types of physical systems. A major advantage of bond graph methods
is that it is domain neutral, and system-
level modelling of multi-
physical domains
can be seamlessly performed. Softwares like 20-
sim, SYMBOLS 2000 and CAMPG
provide bond graph modelling features. Bond graph techniques have been used for
modelling WT drivetrains [64, 65], but their application in WT drivetrains is very
limited as compared to MBS method.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-89-320.jpg)
![Models and simulation 61
2.4.1.4 Block models
Another method for simulating and analysing multi-
domain dynamic systems is
block modelling. Each block represents a model component with mathematical
equations describing the system’s dynamic behaviour. Commercially available
softwares with block modelling features include Simulink®
, Dymola®
, and 20-
sim.
Simulink®
offers built-
in blocks for WT drivetrain and component models [66, 67].
Block modelling methods are useful for developing real-
time system models, con-
trollers and hardware-in-the-loop applications.
2.4.2
Different approaches of modelling a wind turbine drivetrain
WT drivetrains can be classified into two broad categories: geared drivetrains and
direct drives. The geared drivetrains have a gearbox that increases the rotational
speed of the rotor, allowing high-
speed generators to be used. In direct drives, the
generator is directly connected to the rotor shaft and runs at the same speed as the
rotor. As a consequence, direct drives typically require larger generators with more
pole pairs as compared to geared drivetrains. Although recent years are showing an
increasing number of direct-
drive WTs in offshore applications, geared drivetrains
still account for the majority of WTs deployed worldwide.
Based on the generator speed range, geared drivetrains are further divided into
two categories: medium-
speed geared and high-
speed geared. The operating speeds
of medium-
speed geared drivetrain generators are around 90–500 rpm, whereas
high-
speed geared drivetrain generators operate around 1500–1800 rpm [68]. The
high-
speed geared drivetrains allow for smaller size generators and are popular in
Generator
Brake disc
Gearbox
Main bearing
Main sha
Torque arms
Generator
sha
Generator
foongs
High speed
sha
Bedplate
High speed coupling
Figure 2.8
Main features of a typical three-
point suspension drivetrain
©Fraunhofer IWES](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-90-320.jpg)
![Models and simulation 63
where fnis the undamped eigenfrequency, JRotoris the inertia of the rotor, JGenis
the inertia of the generator and
nis the gear ratio.
Multi-
mass torsional models are an extension of the two-
mass model, as each
component has one DOF in these models, allowing for a more detailed analysis of
the drivetrain. Figure 2.10 represents a case with the drivetrain modelled as a multi-
mass spring system.
For the undamped free vibration case, the resulting equation of motion follows
the following form:
J R
+ K = 0 (2.56)
where
J
, K and represent the system inertia matrix, total stiffness matrix and the
vector of rotational displacements, respectively. The eigenfrequencies (!
) of the
system are obtained by solving the following eigenvalue problem:
!2 = eig(J1K) (2.57)
Torsional models are mostly used in aeroelastic codes during the initial development
phase of a new WT, when a global analysis of the WT is carried out for all relevant
load cases to estimate the drivetrain design loads. These estimated loads provide a
starting point for designing the internal components of the drivetrain, such as the
main shaft and gearbox [69].
Although torsional models are useful for estimating external loads on the drive-
train and avoiding natural frequency excitation of external WT components, they do
not capture the details of the complete drivetrain dynamics and internal excitation
sources. More details on the limitations of these models have been addressed in
References 70–72 with suggestions for using higher-
fidelity drivetrain models to
overcome these issues.
2.4.2.2
Rigid multibody model with six degrees of freedom
The next level of modelling fidelity involves rigid bodies (having lumped
masses and six DOFs) that are connected by joints and flexible elements (see
Figure 2.11).
The bearings are modelled as spring elements representing the estimated stiff-
ness of the roller elements. The stiffness values are estimated using an analytical
Figure 2.10
Drivetrain modelled as a multi-
mass spring system
©Fraunhofer IWES](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-92-320.jpg)
![Models and simulation 65
2.4.2.3
Mixed rigid flexible multibody model with six degrees of
freedom
A more realistic model of the drivetrain system takes component flexibility into
consideration. The components can be modelled as flexible bodies using FEM.
However, this results in models with very large number of DOFs that are not suit-
able for dynamic drivetrain simulation. This issue is overcome by implementing
modal reduction techniques via component mode synthesis (CMS). This method
involves modal reduction of FEM models into flexible bodies that have the same
dynamic behaviour as their parent FEM models but with significantly less DOFs.
The Craig–Bampton method [57] is the most widely used CMS method for gen-
erating flexible bodies in MBS. This method performs modal reduction based on
the fixed interface normal modes and the constraint modes. As a result, the overall
structure dynamics is a linear combination of these fixed interface normal modes
and the constraint modes.
To add modal reduced flexible bodies in MBS, the CMS procedure is first
applied to the meshed models in the FEM software. In this procedure, an appro-
priate set of component modes is selected, and interface nodes are defined at
locations where the flexible body can interact with the external environment.
These nodes are constrained to slave nodes on the surfaces of the body, where
the load is distributed (see Figure 2.12). After the modal reduction procedure,
the flexible body is imported in MBS, where it has connections with other bod-
ies, force and constraint elements. The imported flexible body allows for lin-
ear elastic deformation as well as large rigid body motion. The resulting MBS
model leads to a very realistic representation of the drivetrain system in terms
of deformations.
FE models
Reduced
flexible bodies
Interface definion
Modal reducon
Nodal coordinates Modal coordinates
Slave
node
Kinemac
constraint
Interface
node
Figure 2.12
Modal reduction procedure for flexible body generation
©Fraunhofer IWES](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-94-320.jpg)
![66 Wind turbine system design
2.4.2.4
Multi-physical high-fidelity model
Including high level of modelling details can lead to very complex drivetrain
models. Such high-
fidelity models can simulate detailed contacts and flexibili-
ties of machine elements. Multi-
physical models can be coupled with the MBS
model via co-
simulation. The resulting system model can be used, e.g. to simulate
electromagnetism of the generator, the control systems, the acoustic emissions,
lubrication analysis and thermal analysis. Figure 2.13 depicts the possibilities to
conduct multi-
physical simulations for a drivetrain.
This type of advanced modelling approach incorporates system complexity
as much as is technologically possible. Of course, such models require very long
computation times. These types of models usually find applications in research and
development. Figure 2.14 illustrates the different modelling approaches in order of
increasing modelling depth.
2.4.3
Modelling recommendations and best practices
In the early years of modelling WT drivetrains, a purely torsional system with
lumped masses and stiffness elements was a common practice in industry and
a basic requirement from design standards. However, such a model lacked the
ability to describe loads and dynamics at the component level [73]. With the
Figure 2.13
Multi-
physical high-
fidelity model of the drivetrain ©Fraunhofer
IWES](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-95-320.jpg)
![Models and simulation 67
increasing issues of WT system reliability, numerous research projects have been
conducted to evaluate the existing design practices and determine the optimum
method for dynamic modelling of drivetrains. According to the case studies con-
ducted by NREL in the GRC project, a purely torsional model is insufficient to
adequately describe gearbox loads. Based on their findings, new recommendations
for drivetrain modelling fidelity have been proposed [74]. These recommendations
suggest using MBS model for the mechanical drivetrain with multi-
DOFs and a
mix of flexible and rigid bodies to represent the internal drivetrain components.
Table 2.2 lists the recommended modelling fidelity based on the findings of the
GRC project. Following these recommendations has shown good correlation with
experimental results.
Computaon me
Model complexity
Very short Very long
Lowest Highest
Torsional model Rigid MBS model Mixed rigid-flexible MBS model Mul physical / high fidelity model
Figure 2.14
Comparison of different modelling methods for WT drivetrain
©Fraunhofer IWES
Table 2.2 Minimum model fidelity for drivetrain dynamics by NREL [74]
Component Recommended approach Required DOFs
Hub Rigid body, lumped mass N/A
Main shaft Flexible, FE beams From convergence study
Main bearing Stiffness matrices 5 (exclude rotation)
Gearbox housing Flexible, condensed FEs From convergence study
Planet carrier Flexible, condensed FEs From convergence study
Gearbox shafts Rigid shafts N/A
Gearbox support Stiffness matrices 6
Gears Rigid body and contact stiffness 6
Gearbox bearings Stiffness matrices 5 (exclude rotation)
Spline Stiffness matrices 2 (tilting)
Bedplate Rigid body or condensed FEs N/A
Generator coupling Stiffness matrices 5 (exclude rotation)](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-96-320.jpg)
![68 Wind turbine system design
2.5
Conclusion and summary
WT system design strongly depends on the capability of simulation models to accu-
rately predict the loads that a WT will experience during its lifetime. Within this
chapter, the most relevant modelling approaches for all the different aspects that are
required for WT design load estimation are introduced and summarised.
Suitable models for the environmental conditions are necessary for predicting
the loads and operating conditions the WT system will have to withstand. However,
the stochastic nature of environmental conditions makes it difficult to always cor-
rectly represent these in numerical approaches. Therefore, accurate and reliable
measurement data is highly supportive for realistic representation during simula-
tions. Various models for environmental conditions and corresponding impacts are
available with different levels of detail and fidelity, which always have to be selected
according to the point of interest of the corresponding analysis. Furthermore, the
modelling approaches should be tailored to the applicable specific conditions, such
as the determined installation site or the WT type.
To calculate the aerodynamics of a WT, the BEM theory can be combined with a
multibody-
based structural representation of the WT to perform design load calcula-
tions given the environmental conditions. Existing aeroelastic codes have generally
achieved very good levels of validation, but some challenges remain to be solved,
such as the accurate prediction of torsional deformations of rotor blades in certain
load simulations. Therefore, as computational power and capacity become afford-
able, traditional modelling approaches may be supplanted by more complex models
in the near future, even for design load simulations. These advanced models can
include non-
linear FEM beam models for the structural components as well as vor-
tex wake or actuator type models for the aerodynamics.
WT drivetrains can be modelled using a wide range of approaches with various
levels of detail. A balance between the required dynamic details, the correspond-
ing modelling effort and the resulting computation time are needed. Sufficient
accuracy in drivetrain dynamics can be achieved by following the NREL recom-
mendations, as this modelling approach has shown good validation with experi-
mental results. However, integration of high-
fidelity drivetrain models into load
calculation models will still remain computationally too expensive for the near
future, so this fully coupled approach will only be used for some specifically
selected load cases.
References
[1] International Electrotechnical Commission. ‘Wind energy generation
systems–Part1:Designrequirements’.4thed.No.IEC61400-
1inInternational
Standard; Geneva, Switzerland, International Electrotechnical Commission,
2019.
[2] International Electrotechnical Commission. ‘Wind energy generation
systems – Part 3-
1: Design requirements for fixed offshore wind turbines’.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-97-320.jpg)
![Models and simulation 69
1st ed. No. IEC 61400-
3-
1 in International Standard; Geneva, Switzerland,
International Electrotechnical Commission, 2019.
[3] International Electrotechnical Commission. ‘Wind turbines – Part 4:
Design requirements for wind turbine gearboxes’. 1st ed. No. IEC 61400-
4 in International Standard; International Electrotechnical Commission,
2012.
[4] DNV GL Loads and site conditions for wind turbines DNVGL-
ST-
0437. 2016.
Available from https://www.dnv.com/energy/standards-guidelines/dnv-st-0437-
loads-and-site-conditions-for-wind-turbines.html
[5] Poeppel T, Bockstedte A. ‘Certification requirements for the drive train dy-
namics of wind turbines’. Conference for Wind Power Drives (CWD); 2015.
pp. 85–97.
[6] GL Renewables Certification. GL Rules and Guidelines - IV Industrial
Services - Part 1 - Guideline for the Certification of Wind Turbines. Hamburg,
Germany: Germanischer Lloyd Industrial Services GmbH; 2010.
[7] GL Renewables Certification. GL Rules and Guidelines - IV Industrial
Services - Part 2 - Guideline for the Certification of Offshore Wind Turbines.
Hamburg, Germany: Germanischer Lloyd Industrial Services GmbH; 2012.
[8] American Gear ManufacturersAssociation AGMA 6006 - Standard for Design
and Specifications of Gearboxes for Wind Turbines. AGMA; 2003.
[9] Kaimal J.C., Wyngaard J.C., Izumi Y., Coté O.R. ‘Spectral Characteristics of
Surface-Layer Turbulence’. Quarterly Journal of the Royal Meteorological
Society. 1972, vol. 98(417), pp. 563–589. Available from http://doi.wiley.
com/10.1002/qj.v98:417
[10] Mann J. ‘Wind field simulation’. Probabilistic Engineering Mechanics. 1998,
vol. 13(4), pp. 269–282.
[11] Mann J. ‘The spatial structure of neutral atmospheric surface-
layer turbu-
lence’. Journal of Fluid Mechanics. 1994, vol. 273, pp. 141–168.
[12] Jonkman B.J. TurbSim user’s guide v2.00.00: DRAFT VERSION [online].
National Renewable Energy Laboratory; 2016. Available from https://www.
nrel.gov/wind/nwtc/assets/downloads/TurbSim/TurbSim_v2.00.pdf
[13] von Kármán T. ‘Progress in the Statistical Theory of Turbulence’. Proceedings
of the National Academy of Sciences of the United States of America. 1948,
vol. 34(11), pp. 530–539.
[14] Andersen O.J., Løvseth J. ‘The Frøya database and maritime boundary layer
wind description’. Marine Structures. 2006, vol. 19(2–3), pp. 173–192.
[15] American Bureau of Shipping. Design Standards for Offshore Wind Farms:
Final Report: Contract M10PC00105.ABS. 2011.Available from https://www.
bsee.
gov/sites/bsee.gov/files/tap-technical-assessment-program/670aa.pdf
[16] DNV GL ‘Environmental conditions and environmental loads’. August 2017
ed DNVGL-
RP-
C205. DNV GL AS; 2017.
[17] Jensen N.O. A note on wind generator interaction. Risø-M: Risø National
Laboratory; 1983.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-98-320.jpg)
![70 Wind turbine system design
[18] Frandsen S., Barthelmie R., Pryor S, et al. ‘Analytical modelling of wind
speed deficit in large offshore wind farms’. Wind Energy. 2006, vol. 9(1–2),
pp. 39–53.
[19] Rathmann O., Barthelmie R., Frandsen S. ‘Turbine wake model for wind
resource software’. EWEC 2006 Wind Energy Conference and Exhibition;
European Wind Energy Association (EWEA), 2006. pp. 1–12.
[20] Larsen G.C., Madsen H.A., Thomsen K., Larsen T.J. ‘Wake meandering: a
pragmatic approach’. Wind Energy. 2008, vol. 11(4), pp. 377–395. Available
from http://doi.wiley.com/10.1002/we.v11%3A4
[21] Madsen H.A., Larsen G.C., Larsen T.J., Troldborg N., Mikkelsen R.
‘Calibration and Validation of the Dynamic Wake Meandering Model for
implementation in anAeroelastic Code’. Journal of Solar Energy Engineering.
2010, vol. 132(4), 041014.
[22] Larsen T.J., Madsen H.A., Larsen G.C., Hansen K.S. ‘Validation of the dy-
namic wake meander model for loads and power production in the Egmond
aan Zee wind farm’. Wind Energy. 2013, vol. 16(4), pp. 605–624.
[23] Méhauté B. An Introduction to Hydrodynamics and Water Waves. Berlin
Heidelberg: Springer; 1976.
[24] Wheeler J.D. ‘Method for Calculating Forces Produced by Irregular Waves’.
J Pet Technol. 1970, vol. 22(3), pp. 359–367.
[25] Det Norske Veritas.Wind turbine design software - Bladed. 2022.Available from
https://www.dnv.com/services/wind-turbine-design-software-bladed-3775
[26] DTU wind energy. HAWC2. 2021. Available from https://www.hawc2.dk/
[27] Øye S. ‘FLEX4: Simulation of Wind Turbine Dynamics’. Proceedings of 28th
IEA Meeting of Experts Concerning State of the Art of Aeroelastic Codes for
Wind Turbine Calculations; Lyngby, Denmark: International Energy Agency,
1996. pp. 71–76.
[28] National Renewable Energy Laboratory OpenFAST Documentation. 2022.
Available from https://openfast.readthedocs.io/en/main/
[29] Betz A. ‘Das Maximum der theoretisch möglichen Ausnützung des Windes
durch Windmotoren’. Zeitschrift für das gesamte Turbinenwesen. 1920, vol.
26, pp. 307–309.
[30] Glauert H. ‘Airplane propellers’ in Durand W.F. (ed.). Aerodynamic Theory.
Berlin, Heidelberg: Springer; 1935. pp. 169–360.
[31] Ning S.A. ‘A simple solution method for the blade element momentum equa-
tions with guaranteed convergence’. Wind Energy. 2013, pp. 1327–1345.
[32] Hansen M.O.L. Aerodynamics of wind turbines. 2nd ed. London, UK:
Earthscan; 2008.
[33] Snel H., Schepers J.G. ‘Joint investigation of dynamic inflow effects and
implementation of an engineering method’. [ECN-
C–94-
107. ECN] 1995.
Available from https://publications.ecn.nl/E/1995/ECN-C--94-107
[34] Øye S. ‘Dynamic stall simulated as time lag of separation’. Proceedings of the
4th IEA Symposium on the Aerodynamics of Wind Turbines; Rome, Italy, 1991.
[35] Hansen M.H., Gaunaa M., Madsen H.A. ‘A Beddoes-
Leishman type dy-
namic stall model in state-
space and indicial formulations’ [online]. No.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-99-320.jpg)
![Models and simulation 71
Risø-R-
1354(EN) in Risø R, Report, Risø National Laboratory. 2004.
Available from http://iis-03.risoe.dk/netahtml/risoe/publ_uk.htm
[36] Hansen M.O.L, Sørensen J.N, Voutsinas S, Sørensen N, Madsen H.A. ‘State
of the art in wind turbine aerodynamics and aeroelasticity’. Progress in
Aerospace Sciences. 2006, vol. 42(4), pp. 285–330.
[37] Leishman J.G, Beddoes T.S. ‘A Semi-Empirical Model for Dynamic Stall’.
Journal of the American Helicopter Society. 1989, vol. 34(3), pp. 3–17.
[38] Leishman J.G. ‘Challenges in modelling the unsteady aerodynamics of wind
turbines’. [online] Wind Energy. 2002, vol. 5(2–3), pp. 85–132. Available
from http://doi.wiley.com/10.1002/we.v5:2/3
[39] Schepers J.G., Snel H. ‘Dynamic inflow. Yawed conditions and partial span
pitch control’. ECN-C-95-056. Netherlands Energy Research Foundation.
1995.
[40] Lanzafame R., Messina M. ‘BEM theory: How to take into account the radial
flow inside of a 1-
D numerical code’. Renewable Energy. 2012, vol. 39(1),
pp. 440–446.
[41] Peters D.A, He C.J. ‘Correlation of Measured Induced Velocities with a
Finite–State Wake Model’. Journal of the American Helicopter Society. 1991,
vol. 36(3), pp. 59–70.
[42] Sørensen J.N, Mikkelsen R.F, Henningson D.S, Ivanell S, Sarmast S,
Andersen S.J. ‘Simulation of wind turbine wakes using the actuator line tech-
nique’. Philosophical Transactions. Series A, Mathematical, Physical, and
Engineering Sciences. 2015, vol. 373(2035), 20140071.
[43] Boorsma K., Wenz F., Lindenburg K., Aman M., Kloosterman M. ‘Validation
and accommodation of vortex wake codes for wind turbine design load calcu-
lations’. Wind Energy Science. 2020, vol. 5(2), pp. 699–719.
[44] Morison J.R., Johnson J.W., Schaaf S.A. ‘The Force Exerted by Surface
Waves on Piles’. J Pet Technol. 1950, vol. 2(5), pp. 149–154.
[45] Sumer B.M., Fredsøe J. ‘Hydrodynamics Around Cylindrical Strucures’ in
Advanced series on ocean engineering. Revised Edition vol. 26. Singapore:
World Scientific Publishing Co Pte Ltd; 2006 Sep.
[46] Clauss G., Lehmann E., Östergaard C. Offshore Structures. London:
Springer London; 1992. Available from http://link.springer.com/10.1007/
978-1-4471-3193-9
[47] MacCamy R.C., Fuchs R.A. ‘Wave Forces on Piles: A Diffraction Theory’.
Contribution from the Scripps Institution of Oceanography. U.S. Beach
Erosion Board. 1954. Available from https://books.google.de/books?id=
M0K5GQAACAAJ
[48] Benaroya H., Nagurka M.L., Han S.M. Mechanical vibration: Analysis, un-
certainties, and control. 4th edn. Boca Raton and London and New York:
CRC Press Taylor Francis Group; 2018. Available from http://www.crc-
press.com/p/book/9781498752947
[49] Bauchau O.A., Craig J.I. Structural analysis: With applications to aerospace
structures. Vol. 163. of Solid mechanics and its applications. Dordrecht:
Springer; 2009.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-100-320.jpg)
![72 Wind turbine system design
[50] Timoshenko S.P. ‘LXVI. On the correction for shear of the differential equa-
tion for transverse vibrations of prismatic bars’. The London, Edinburgh, and
Dublin Philosophical Magazine and Journal of Science. 1921, vol. 41(245),
pp. 744–746.
[51] Lamei A, Hayatdavoodi M. ‘On motion analysis and elastic response of
floating offshore wind turbines’. Journal of Ocean Engineering and Marine
Energy. 2020, vol. 6(1), pp. 71–90.
[52] Wang L., Liu X., Kolios A. ‘State of the art in the aeroelasticity of wind
turbine blades: Aeroelastic modelling’. Renewable and Sustainable Energy
Reviews. 2016, vol. 64, pp. 195–210.
[53] Hodges D.H. ‘Geometrically Exact, Intrinsic Theory for Dynamics of Curved
andTwistedAnisotropicBeams’.AIAAJournal.2003,vol.41(6),pp.1131–1137.
[54] Shabana A.A. Dynamics of Multibody Systems. Cambridge University Press;
2009.
[55] Kim T., Hansen A.M., Branner K. ‘Development of an anisotropic beam finite
element for composite wind turbine blades in multibody system’. Renewable
Energy. 2013, vol. 59, pp. 172–183.
[56] Wang Q., Sprague M.A., Jonkman J., Johnson N., Jonkman B. ‘BeamDyn:
a high‐fidelity wind turbine blade solver in the FAST modular framework’.
Wind Energy. 2017, vol. 20(8), pp. 1439–1462.
[57] Mapa L. de P.P., das Neves F. de A., Guimarães G.P. ‘Dynamic Substructuring
by the Craig–Bampton Method Applied to Frames’. Journal of Vibration
Engineering Technologies. 2021, vol. 9(2), pp. 257–266.
[58] Veers P. Wind energy modeling and simulation. vol. 125 of IET Energy
engineering series. London: The Institution of Engineering and Technology;
2019. Available from https://digital-library.theiet.org/content/books/po/
pbpo125f
[59] Mo W., Li D., Wang X., Zhong C. ‘Aeroelastic coupling analysis of the flex-
ible blade of a wind turbine’. Energy. 2015, vol. 89, pp. 1001–1009.
[60] Faulstich S., Hahn B., Tavner P.J. ‘Wind turbine downtime and its impor-
tance for offshore deployment’. Wind Energy [Online]. 2011, vol. 14(3), pp.
327–337. Available from http://doi.wiley.com/10.1002/we.v14.3
[61] Keller J., Wallen R. Gearbox Reliability Collaborative Phase 3 Gearbox 3 Test
Report. No. NREL/TP-
5000-
67612 Technical Report. National Renewable
Energy Laboratory; 2017.
[62] Musial W., Butterfield S., McNiff B. ‘Improving Wind Turbine Gearbox
Reliability NREL/CP-
500-
41548’. National Renewable Energy Laboratory,
2007.
[63] Szabó B., Babuška I. Introduction to Finite Element Analysis: Formulation,
Verification and Validation. John Wiley Sons; 2011.
[64] Sanchez R., Medina A. ‘Wind turbine model simulation: a bond graph ap-
proach’. Simulation Modelling Practice and Theory. 2014, vol. 41, pp. 28–45.
[65] Torsvik J. Dynamic Analysis in Design and Operation of Large Floating
Offshore Wind Turbine Drivetrains [Doctoral thesis]. NTNU; 2020.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-101-320.jpg)
![Models and simulation 73
[66] Apostoaia C., Scutaru G. ‘A Dynamic Model of a Wind Turbine System’. In:
The 10th International Conference OPTIM’06; Transilvania University Press,
2006. pp. 261–266.
[67] Benbouzid M.E.H., Diallo D. ‘Development of a Matlab/Simulink®-based
wind turbine prototyping software through undergraduate student projects’.
ComputerApplications in Engineering Education. 2012, vol. 20(1), pp. 78–87.
Available from http://doi.wiley.com/10.1002/cae.v20.1
[68] Manwell J.F., McGowan J.G., Rogers A.L. Wind energy explained: Theory,
design and application. 2nd ed. Chichester: John Wiley; 2009 Mar. Available
from https://onlinelibrary.wiley.com/doi/book/10.1002/
9781119994367
[69] Vanhollebeke F. Dynamic Analysis of a Wind Turbine Gearbox Towards
Prediction of Mechanical Tonalities [Ph.D. Thesis]. Belgium: Katholieke
Universiteit Leuven; 2015.
[70] Peeters J., Vandepitte D., Sas P. ‘Flexible multibody model of a three-
stage
planetary gearbox in a wind turbine’. Proceedings of ISMA2004; Katholieke
Universiteit Leuven, 2004. pp. 3923–3942.
[71] Peeters J. Simulation of dynamic drive train loads in a wind turbine [Ph.D.
Thesis]. Belgium: Katholieke Universiteit Leuven; 2006.
[72] Peeters J., Goris S., Vanhollebeke F, et al. ‘A need for advanced and validated
multibody models as a basis for more accurate dynamic load prediction in
multi-
megawatt wind turbine gearboxes’. International Conference on Noise
and Vibration Engineering; 2008. pp. 2097–2112.
[73] Helsen J., Vanhollebeke F., Marrant B., Vandepitte D., Desmet W. ‘Multibody
modelling of varying complexity for modal behaviour analysis of wind tur-
bine gearboxes’. Renewable Energy. 2011, vol. 36(11), pp. 3098–3113.
[74] Guo Y., Keller J., La Cava W, et al. ‘Recommendations on Model Fidelity
for Wind Turbine Gearbox Simulations NREL/CP-
5000-
63444’. National
Renewable Energy Laboratory, 2015.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-102-320.jpg)
![1
Fraunhofer Institute for Wind Energy Systems, Bremerhaven, Germany
2
Nidec SSB Wind Systems GmbH, Salzbergen, Germany
Chapter 3
Pitch system concepts and design
Karsten Behnke1
, Arne Bartschat1
, Eike Blechschmidt1
,
Matthis Graßmann1
, Florian Schleich1
, Oliver Menck1
, and
Heiko Jungermann2
The pitch system allows for turning the rotor blades of the wind turbine about their
longitudinal axes. This turning movement is commonly called pitching and con-
trols the aerodynamic loads on the blades and thus the power output, the rotational
speed and structural loads of the turbine. Feathering of the blades means that the
blades are turned to reduce the lift force and hence the torque. It leads to a stop of
the turbine. Hence, the pitch system is one of the safety systems of a wind turbine
and thus always designed in a redundant manner. Design guidelines like the one
from Det Norske Veritas (DNV) make a redundant pitch system a mandatory pre-
requisite for a turbine certification [1]. Redundancy means that at least two of the
blades have independently operating pitch systems. Common wind turbines have
three blades. This leaves the designer the choice to have two pitch systems, one
for two and another one for one blade, or three pitch system, each for one blade.
In terms of design effort, procurement and operational costs, three identical pitch
systems are better, which makes them the common choice in today’s commercial
wind turbines.
The pitch system of one blade consists of the pitch bearing (Section 3.1) and
the pitch actuator (Section 3.1 and Chapter 8). In addition, a lubrication system
(Section 7.3), sensors for position and more, a power supply, backup power, connec-
tion to the main turbine controller, and a local backup controller complete the pitch
system. Interfaces of the blade bearing towards the blade and hub are usually bolted
connections.
The design of a pitch system is a complex task since many aspects have an
influence on it. Furthermore, the pitch system design affects the turbine design.
The blade and hub are directly connected to the blade bearing. Their design](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-104-320.jpg)
![Pitch system concepts and design 77
must therefore be considered for the pitch system, and vice versa, the pitch
system design must be considered in the blade and hub design. Hence, a pitch
system design cannot happen individually. Technical feasibility, experiences in
the company and costs are main drivers for any decision. However, to find a
start, Table 3.1 gives an overview of possible steps for such a project. The table
starts with the blade bearing design. It connects blade and hub and defines con-
sequently their dimensions. It must cope with the acting loads and ensure the
ability to pitch. In addition, it determines possible other choices for the remain-
ing components of the pitch system, like the actuator and auxiliary systems. The
following sections give detailed information for each step of the pitch system
design. These steps should not be understood as a gradual process. Calculations
or other decisions could lead to changes in a previous design stage. It is possible
that many iterations are mandatory. The individual steps will include examples
of possible design choices for the IWT7.5-
164, a 7.5 MW reference wind turbine
designed by Fraunhofer IWES [2]. The turbine is introduced in Chapter 1.
Step Inputs Objectives
3.1.6 Calculation and
dimensioning
• FE model from the
previous sections
• Load time series
containing: pitch angle,
moments and forces at the
blade root, and extreme
loads
• Perform static evaluation:
use the maximum load
situations for overload,
truncation and core
crushing check
• Perform fatigue
evaluation: calculate
rolling contact, structural
and bolt fatigue lifetime
3.1.7 Lubrication • Bearing dimensions and
type
• Operating conditions
• Choose grease type
• Design grease in- and
outlets and sealing
3.1.8 Coating • Specific site conditions
• Turbine location
• Define the type of
coating and layer
thickness to protect
against corrosion
3.2.1 Electrical actuator • Bearing interfaces and
position of the blade
(inner or outer ring)
• Acting wind loads,
friction torque and inertia
of the blade and other
rotating parts
• Calculate the torque,
which is needed to pitch
the blade
• Choose drive size
• Choose kind of motor
and back-
up power
• Determine back-up
capacity for emergency
drive
Table 3.1 Continued](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-106-320.jpg)
![78 Wind turbine system design
3.1 Blade bearing
Blade bearings connect to the rotor blade and the rotor hub of a wind turbine by
means of bolted connections. Their rings tend to be made of steel, with 42CrMo4-
type steels being the most common choice of material. Rolling elements be likely
made from 100Cr6. The rotor blade is mostly made of glass-
fibre-
reinforced plastic
(GFRP). Its weight and the acting wind loads cause the loads on the blade bear-
ing. The blade works as a huge lever, hence the dominant load on a blade bearing
is a bending moment. The bearing also sees loads in every other degree of free-
dom, however, their influence pales in comparison to that of the resulting bending
moment caused by the tilting of the blade.
Thus, the moments and forces at the blade root are input values for the bearing
design. All of them vary according to the wind conditions. Values for said loads
typically stem from the aero-
elastic simulations (cf. Chapter 1). These can have the
form of time series (see, e.g., Reference [3]) or postprocessed results like extreme
loads as defined in References [4] and [5], bin counts, damage equivalent loads, rain
flow counts for the last, see Reference [6]. A further example is a load revolution
distribution (LRD) of a rolling contact fatigue (RCF).
An aero-
elastic simulation time series of IWT7.5-
164 reference turbine is avail-
able online and can be downloaded under [3]. It can be used to understand the loads
that are acting on the blade bearing and the movements the bearing does.
As mentioned before, the blade bearing design process is iterative as it influ-
ences and is influenced by the other systems such as controller and blade design.
A heavy use of individual pitch control, e.g., reduces turbine loads significantly
and therefore lowers costs in structural components. This on the other hand low-
ers the lifetime of the blade bearing with respect to RCF, which can lead to
changes in rolling body diameter or even bearing size. Especially the latter is
commonly not wanted as it would influence the design of the blade root. A lower
use of individual pitch control leads to higher loads on the structural components
especially on the blade that would need more material to withstand. The higher
masses increase, in turn, the bending moment acting on the blade bearings and
could lead to higher RCF again. The aim during the design phase is to optimise
for overall system costs.
Figure 3.1 shows the iterative bearing design process. In the first iteration,
it will be necessary to make some assumptions until the bearing completely
fulfils the requirements. The design starts with the bearing geometry, which
is described in Sections 3.1.1–3.1.3. The geometrical properties are necessary
to create a FE bearing model to calculate the internal loads of the bearing (cf.
Section 3.1.4). The rotor star simulates the load distribution of the bearing for
realistic surrounding structures and turbine loads (cf. Section 3.1.5). The cal-
culated bearing loads are used for the static evaluation and RCF calculation in
Section 3.1.6. The calculation results may require adaptions in the dimensions of
the bearing. When the geometry of the bearing changes, a new FE model needs
to be created to run new simulations and recalculate the static evaluation and](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-107-320.jpg)
![Pitch system concepts and design 85
stated as in the following steps. Later it can be necessary to change these variables
depending on the results in Section 3.1.6.
The dominant load of a pitch bearing is the resulting moment (Mres
). As 3.1
shows, it is the combination of the edge- and flapwise bending moment.
Mres =
q
Mx
2
+ My
2
(3.1)
Radial (Fr
) and axial forces (Fax
) will influence the rolling body loads as well, but
play a minor role. Figure 3.6 shows a typical load distribution. Due to the bending
moment, there are two opposite regions where the loads are transferred. One side is
called traction side, the other, compression side. Between the peaks, the individual
rolling element loads are lower.
Under the assumption of stiff rings and interfaces, the bearing rows (m) share
the load equally and the highest contact force (Qmax
) can be calculated as 3.2
shows.
Qmax =
2 Fr
m z cos(˛)
+
Fax
m z sin(˛)
+
2 Mres
m r sin(˛) z
(3.2)
Here, z is the number of rolling elements per row, ˛is the nominal contact angle of
a ball bearing (0° for pure radial and 90° for pure axial contact) and r is the pitch
radius. This equation is based on DG03 [7], but the number of rows was added.
It assumes of rolling elements with linear stiffnesses and stiff surrounding struc-
tures; however, balls and rollers are non-
linear in their deformation behaviour and
the surrounding structures behave elastically. A real load distribution results into
Figure 3.6
Characteristic load distribution of a double-
row four-point contact
ball bearing © Fraunhofer IWES](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-114-320.jpg)
![86 Wind turbine system design
higher maximum forces. They are typically 10−25% above those from 3.2 and can
be even larger, depending on the specific deformation behaviour of the load case.
For first approximations, a corrective factor of 1.1−1.25 applied to Qmax
may thus
be useful.
For 3.2, m defines the number of rows that would carry loads under a pure axial
force. In double-
direction two-
row roller bearings, only one row carries an axial
load in a given direction, and therefore m should be chosen to be 1 in this case. For a
double-
row four-
point bearing, on the other hand, two rows carry an axial load, and
consequently, m should be 2.
The rolling element loads are then used in Hertzian calculations for contact
pressure of a single contact. The Hertzian calculations cannot be solved analytically.
Hence, equations for an approximation were well established. For example, Houpert
published a method. The maximum pressure (Pmax
) follows according to 3.3, see
Reference [8]. The Hertzian contact ellipse radii (a and b) are required inputs for
the equation. Here, another iterative process comes into play. The contact pressure
decreases with larger rolling elements since the contact area is larger. But if one
uses larger rolling elements, less of them fit into the available space on one raceway,
and hence each ball or roller has to carry a higher load. Hence, an optimum will be
a trade-
off.
Pmax =
1.5 Qmax
a b
(3.3)
According to Reference [9], maximum permissible contact stresses are 4.2 GPa for
ball bearings and 4.0 GPa for roller bearings. However, since these pressures are
related to extreme wind conditions, typical contact pressures during normal opera-
tion of the turbine are lower. Instead of 4.2 GPa for a ball contact or 4.0 GPa for a
line contact, it is recommended to aim at 2.5 GPa for a PC and 2.0 GPa for a line
contact. The experience shows that the results from fatigue or core crushing calcula-
tion then finally determine which loads are permissible during normal operation. In
addition, the initial contact angle of a four-
point bearing increases with higher loads.
A higher load could lead to contact ellipse truncation at the edge of the raceway
before the maximum permissible contact stress is reached. Truncation should be
avoided as well.
The IWT7.5-
164 blade bearing design is a double-
row four-point contact ball
bearing. The pitch diameter and the rolling element diameter determine the possible
number of rolling elements. It is necessary to have about 1–2 cm space between each
rolling element for a cage or spacer. A possible rolling element diameter for this
turbine class is 80 mm. In the following steps, it will be necessary to evaluate this
first approach. The bearing has 147 balls per row with a diameter of 80 mm each.
The number of rolling elements can be derived with the pitch diameter and the roll-
ing element size.
The IEC 61400 defines different design load cases (DLCs) [4]. For the following
calculations, the maximum load (Mres_max
= 30.3 MNm) in DLC 2.2 is used. DLC 2
are power production load cases with an occurrence of a fault. For this DLC, a partial](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-115-320.jpg)
![Pitch system concepts and design 87
safety factor for loads (γf
= 1.1) must be considered [4]. According to 3.2, it leads to a
contact force of 138 kN.
Qmax =
2 7158 kN
2 147 cos
45
+
1245 kN
2 147 sin
45
+
2 30.3 MNm
2 147 2.345 m sin
45
!
1.1
= 138kN
(3.4)
This value again is the input to calculate the Hertzian parameters. Due to the
simplifications of Equation 3.2, a safety factor of 1.25 is added, which leads to
Qmax
= 172 kN. With Equation 3.3, the bearing has a maximum contact pressure
of about 3.0 GPa. Table 3.3 gives the missing values to calculate the contact
pressure. It also introduces the ball grove conformity. It describes the ratio of
the radii from the rolling element and the raceway. The last one is slightly larger;
typically, values for blade bearings are 0.52–0.53. The conformity influences
the contact size and hence the contact pressure. With a higher conformity, the
contact pressure decreases, but on the other hand, a higher frictional work and a
higher friction are consequences. Other values from Table 3.3 stem from litera-
ture values for steel.
The calculated contact pressure is lower than the permissible. Typical loads
during normal operation behaviour are even lower and lead to lower contact
pressures. The bending moment during power production (DLC1.1) reaches val-
ues from about 22 MNm, which again lead to contact pressures in the range of
2.5 GPa. To sum up, a rolling element diameter of 80 mm has satisfactory results
in terms of ultimate loads and seems to fulfil the recommendations, explained
before.
The choice between spacers or a cage is the next topic. Both separate the rolling
elements and keep them equidistant. Table 3.4 lists advantages and disadvantages
for both types.
The IWT7.5-
164 has a steel cage. It is made of S235 steel. Figure 3.7 shows an
exemplary cage.
The raceways of the bearing are inductively hardened to increase the resilience
of the raceway. The hardening process induces currents that heat the raceways
locally and thus harden the bearing steel. Figure 3.8 shows two case-
hardened race-
ways of a four-
point bearing. As the magnetic field is not homogeneous, it is almost
impossible to harden one whole raceway without superposing the starting point.
Table 3.3 Bearing properties for Hertzian calculation
Value Unit
Conformity inner ring 0.53
Conformity outer ring 0.53
Hertzian ellipse radius a 2.05 mm
Hertzian ellipse radius b 13.00 mm
Young’s modulus E 210000.00 N/mm²
Poisson’s ratio ν 0.30](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-116-320.jpg)
![Pitch system concepts and design 89
3.1.3
Preliminary design of the bolted connections
This section will give a rough description for a preliminary and first design of the
bolted connections between blade, bearing and hub.Adetailed calculation of the bolted
connection has to be performed according to applicable standards like the DIN VDI
2230 in any case [10]. Usually, the blade bearing is connected to the blade root and to
the hub flange by using bolted connections. There are several requirements that must
be fulfilled with these connections. The most important requirement is to ensure a
permanent and safe connection, which can withstand the dynamic and ultimate loads
of the complete service life of the turbine. Besides this, the bolted connection allows to
pre-
load the rolling elements for specific bearing designs. These kinds of bearings have
separated rings such as three-
row roller blade bearings.
In general, there are two flanges with bolted connections, which differ slightly in
terms of demands on the design: the connection of blade bearing ring and blade root
and the connection between blade bearing and hub flange. Typically, the connection
of blade root and blade bearing can be characterised as follows:
•
• through holes on the bearing ring
•
• connection of different materials such as steel (bearing or stiffener plate) and
fibre-
reinforced plastics (blade root) with different stiffness properties
•
• possibility of having multiple joints within the clamp length due to additional
parts such as stiffener plates, ring extenders or others
•
• long clamp length due to the bearing geometry and special requirements of the
blade root especially for T-
Bolt connections due to bearing stresses in the radial
bore holes
•
• specific friction coefficients in the joint due to different materials
•
• additional sealing material in the contacts to prevent water from entering espe-
cially in harsh environments
Figure 3.8 Case-
hardened raceway © Fraunhofer IWES](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-118-320.jpg)
![90 Wind turbine system design
The connection of blade bearing and hub flange can be considered to have the
following characteristics:
•
• through-
holes or blind-
holes on the bearing ring
•
• shorter clamp length
•
• similar or comparable material properties of both connecting parts
•
• possibility of having multiple joints within the clamp length due to additional
parts such pitch system carriers, ring extenders or others
As the design of the blade root usually includes a detailed consideration of the
bolted connection due to the specific requirements of the lightweight construction,
the bearing design could be adapted to the needs of the blade root in terms of number
and size of bolts. The following preliminary design for the outer ring is an iterative
approach, which has to balance the requirements. If no information is available, the
first step towards the definition of a preliminary design of the bolted connection should
be the number of bolts. This number must be defined by the available space and the
expected ultimate loads which have to be carried by the bearing flange. In case of
the example in this section, the bolt circle diameter on the outer ring is 4880 mm (cf.
Section 3.1.1). To define the number of bolts, the designer also must consider the
tools needed to tighten the bolts. Usually, the bolted connections on a blade bearing
are tightened with hydraulic torque wrenches or bolt tensioners, which need enough
space to be applied to the bolts. In this example, 120 bolts (cf. Section 3.1.1) lead to
a distance between the bolts of roughly 128 mm on the outer ring, which seems rea-
sonable at first glance. The ultimate loads acting on the flange have to be transferred
into an axial force carried by the bolts. Therefore, 3.2 is adapted slightly. The loads
used for this calculation are based on DLC 2.2, according to the ultimate load analysis
(cf. Section 3.1.2). According to IEC-
61400-
1 [4], different partial safety shall be con-
sidered for the DLCs. In case of DLC 2.2, it is a safety factor of
f = 1.1
.
FB =
Fax
nB
+
2 Mres
r nB
f =
0
B
@
1245 kN
120
+
2
q
29.265 MNm
2
+
7.827 MNm
2
2.44 m 120
1
C
A1.1 = 239 kN
(3.5)
The clamping force needed to provide enough resistance against the opening of the
contact can be estimated as follows.
FK = s FB = 1.5 239 kN = 358.5 kN (3.6)
The factor s is set to 1.5, which is typical for dynamic load situations. In general,
the clamping force FK
should be higher as the working load FB
to ensure a positive
residual clamping load. Both loads are used to estimate the needed stress cross sec-
tion for the bolted connection.
AS
FB + FK
Rp,0,2
kkA
ˇ E fz
lk
(3.7)
This estimation is based on Reference [11] and needs some more values. kA
deter-
mines a factor for the tightening method,
kis a reduction factor based on the type of](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-119-320.jpg)
![Pitch system concepts and design 91
bolt,
ˇa factor for the elastic resilience of the used bolt, Ethe Young’s modulus for
steel, fza factor for the amount of embedding,
lkthe clamping length and Rp,0,2the
proof stress of the used bolt. For the preliminary design in this section, the following
values are chosen according to Reference [11]:
•
• Rp,0,2 = 900 MPaaccording to bolts of type 10.9
•
•
kA = 1.2for hydraulic tensioning
•
•
k = 1.15for setscrews
•
• fz
= 0.011 mm
•
•
ˇ = 1.1for set screws
•
•
lk = 450 mmpreliminary assumption according to the height of the bearing ring
of 284 mm (see Section 3.1.1), pitch carrier plates and the hub properties
This calculation is based on some estimations. Although these estimations only
aim towards a rough preliminary calculation of the bolted connection between the
hub and the outer ring of the blade, the designer already needs at least some basic
information regarding the major geometry. Based on the values estimated for this
preliminary design, the needed stress cross section according to (3.8) is
AS,min
239 kN + 358.5 kN
900 MPa
1.151.2
1.1 210 GPa 0.011 mm
450 mm
= 924.3 mm2
(3.8)
The stress cross section of the used bolt shall be equal or larger than the one esti-
mated with (3.8). In this case, the next ISO metric screw size that has a slightly larger
stress cross section is M42. The wrench size for bolt size M42 is 65 mm. Therefore,
the mean distance between the bolts of 128 mm should be sufficient. However, a
detailed analysis of possible collisions of the assembly must be performed according
to the dimensions of the tools used for tightening. Therefore, the preliminary design
of the bolted connection between outer ring of the bearing and the hub shall consist
of 120 bolts of the size M42. For an estimation of the pretension, the designer needs
to know the resilience of the bolt and the connected parts of the assembly. The val-
ues used in this example lead to the following resilience of the M42 bolts.
ıS =
lk
Es AS
=
450 mm
210 GPa 1121 mm2
1.91e6 mm
N
(3.9)
The resilience of the bearing outer ring, pitch carrier and the hub is estimated as follows.
First the designer needs the substitutional area of the connected parts calculated with
Asub =
4
d2
w d2
h
+
8
dw
DA dw
3
qlKdw
D2
A
+ 1
2
1
(3.10)
Using dw = 65 mm for bolt size M42, dh = 45 mm as a usual bore hole diameter for
M42, the distance between the bolts Da = 128 mm and lk = 450 mm, the substitu-
tional area is](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-120-320.jpg)
![Pitch system concepts and design 93
rather complex, a detailed analysis of the bolted connection should be performed
towards the ultimate strength of the connection and the contacts and interfaces of the
parts. Gaps and an opening of the contacts between the flanges should be avoided
and shall be verified within the FE analysis. In addition, the fatigue strength of the
connection has to be verified by more detailed calculations according to the stan-
dards like DIN VDI 2230.
3.1.4
FE blade bearing model
For the calculations in Section 3.1.6, the accurate contact force or pressure of the
rolling elements in the bearing is essential. Because analytical calculation methods
cannot consider the tilting of the bearing rings and complex load situations including
the surrounding structures, finite element analysis (FEA) is the only means to deter-
mine the load distribution realistically. FE models that contain the full bearing and
potential surrounding structures like blade and hub are called global models in the
following. Contact forces and angles as well as the deformation of the bearing rings
are possible results of global bearing models. The contact pressure can be calculated
with the contact force and the associated, analytically calculated, contact area.
In this section, the preliminary bearing design from the previous sections is
used to create the FE bearing model. After a plausibility check, the bearing is imple-
mented in a full or one-
third rotor star model in Section 3.1.5. There are many dif-
ferent approaches to create a FE model of a blade bearing. One way is to create a
detailed 3D model including all required geometrical properties like osculation and
bearing preload. An FE software then meshes the model automatically. That would
provide a bearing model with three-
dimensional solid rolling bodies. Characteristic
of any rolling bearing is that the rolling bodies roll in the raceways of the bearing
rings. The FEM considers that with the definition of frictional contacts between
the touching components. The blade bearing of the IWT7.5-
164 wind turbine is a
double-
row four-
point contact ball bearing. That means, in initial state, every ball
has four touching spots with the raceways. With 147 balls per row that sum up in
a total number of 1,176 contact definitions. Furthermore, the calculation of cor-
rect stresses in the material requires a sufficient fine mesh, especially in the contact
areas. All in all, this approach leads to an FE bearing model with a high level of
detail, many contact definitions and a very large number of nodes and elements.
That results in a significant high computational effort and requires a lot of simula-
tion time.
In the design process, many simulations are necessary to investigate differ-
ent load cases and related load distributions in the bearing. Long simulation run-
times due to high number of nodes and elements are obstructive. Manufactures
and researchers have developed and published different approaches to reduce the
complexity of the bearing model by decreasing the number of defined contacts
and elements. They have in common that they use non-
linear spring elements
to model the ball-
raceway contacts instead of modelling three-
dimensional solid
balls. A force-
deformation curve, calculated according to Reference [8], controls
the behaviour of the spring elements. For a ball bearing, the following equation](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-122-320.jpg)
![94 Wind turbine system design
calculates the deformation ıfor a given ball force
Q
, ball diameter DW and oscu-
lation s[8, 12].
ı = 8.97 104
1 s
0.1946 Q2/3
D1/3
W
(3.15)
The resulting force deformation behaviour of the blade bearing of the IWT7.5-
164
reference wind turbine is shown in Figure 3.10.
Gao [13], Smolnicki [14] and Daidié [12] published different approaches to
replace the solid ball with non-
linear spring elements to represent the ball-
raceway
contact behaviour of four-
point contact ball bearings. Each spring element repre-
sents one force transmitting diagonal of the ball. Therefore, two springs model one
ball with its four contact points with the raceways. The direction of the undeformed
spring elements represents the initial contact angle. The initial contact angle is the
angle between ball and raceway under unloaded conditions. For the blade bearing
of the IWT7.5-
164 wind turbine, the initial contact angle is 45°. In [13], the springs
are directly connected to the raceways. In [14], rigid beam elements connect the
springs to the raceways. Starting and ending point of a spring are the centres of two
opposite raceways. In [12], rigid shell elements, which roughly represent the size
of the contact ellipse for a reference contact pressure, are located on the surface
of the raceways. Rigid beam elements then connect these shell elements with the
spring elements. Here, the springs are also placed between the centres of two oppo-
site raceways. Figure 3.11 shows the arrangement of the spring (blue) and beam
(orange) elements for a modelled ball in a four-
point contact ball FE bearing. With
this approach, a compression force on the ball leads to a tensile force in the spring.
Figure 3.10
Force-
deformation curve of the blade bearing of the IWT7.5-
164
reference wind turbine © Fraunhofer IWES](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-123-320.jpg)
![Pitch system concepts and design 95
A ball-
raceway contact cannot transmit any tensile forces. According to that, the
springs must not carry any compression forces.
For the sake of completeness, Stammler [16] and Wang [17] published approaches
to model roller bearings with non-
linear spring elements. However, the focus in this sec-
tion is a double-
row four-
point contact ball bearing that is used as blade bearing for the
IWT7.5-
164 wind turbine and FE models of roller bearings are not considered further.
The preliminary design is explained in Sections 3.1.1 and 3.1.2. Figure 3.12 visualises the
FE bearing model that is used for the upcoming investigations in a cross-
sectional view.
A large rolling bearing has more important functional parts than the rolling ele-
ments. For example, bolt holes, sealing surfaces, bore holes to allow lubrication,
and for ball bearings fill plugs. However, not all parts are important to be consid-
ered in a global FE bearing model to analyse the load distribution on the raceways.
Bolt holes in the inner and outer ring allow to mount the bearing to its surrounding
structures. Tightened bolts introduce additional stresses to the bearing rings and
with that also influence the properties of the contacts between ball and raceway
(e.g., contact pressure and ball force). To obtain realistic load distributions when the
bearing is simulated with its surrounding structure, bolt holes and modelled bolts
need to be considered. The other functional components do not greatly influence the
load distribution and contact angle evolution of the bearing and can be left out for
the FE bearing model. That decreases the complexity of the model and reduces the
computational effort.
Figure 3.11
Modelled ball with non-
linear spring elements in four-
point contact
ball bearing [15]](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-124-320.jpg)
![98 Wind turbine system design
are compressed should transfer any load. Next, a bending moment is applied to the
bearing. Unlike for the comparison with the Hertzian theory, the tiling of the bearing
rings is not prohibited for these simulations. When a complex bearing deformation
is permitted, the boundary conditions must not be directly applied to the bearing
flanges. The clamping of the outer bearing flange and loading the inner flange leads
to a very uneven load distribution of the raceways. Generic surrounding structures
or simple steel rings that are mounted to the bearing flanges ensure a more realistic
load distribution of the bearing model. Characteristic of a four-
point contact ball
bearing is a change of the loaded diagonals between traction and compression side
(cf. Figure 3.6). The simulation of bending moments can verify the plausibility of
the qualitative load distribution. At the traction and compression side are always
two diagonals (one of each ball) loaded. Only in regions where the load changes
the raceway, all four diagonals carry load. Typically, the maximum ball force at the
compression side is higher than at the traction side. The qualitative progression of
the contact angle should be similar to the load distribution.
3.1.5
FE simulation of internal blade bearing loads
Once the bearing FE model is generated and the resulting contact forces and contact
angles have been checked for plausibility, it must be mounted into a wind turbine
rotor star model. A rotor star model containing the bearing’s surrounding structures
enables the evaluation of realistic internal loads. When blade bearings become
larger, the pitch diameter becomes disproportionately larger compared to the cross-
sectional area of the bearing rings. Because of that, large bearings are structurally
softer than small bearings and more sensitive towards the stiffness of their surround-
ing structures. In turn, the resulting internal loads of a large blade bearing are sig-
nificantly influenced by the stiffness of the rotor blade and the rotor hub [18]. As the
blade flange and the rotor hub have inhomogeneous stiffnesses along their circum-
ference, large blade bearings are exposed to complex load distributions. In addition,
the overall stiffness is relatively low. Depending on the outer load, the load of every
rolling element is different, and the specific load distribution needs to be considered
in the bearing’s design process.
It is best practice to model wind turbine rotor blades with shell elements to reduce
computational time [19]. In order to connect such a model to the rest of the rotor star
properly and to implement the opportunity to consider the bolted connection, a solid
root has to be added to the blade model. The blade root, modelled with solid elements,
should contain structural details like the metal inlets or T-
Bolts. This allows the defini-
tion of frictional contacts between the flange surfaces and the implementation of the
bolted connection. Even if no sliding or separation of the flange surfaces occur, consid-
ering the bolted connection can slightly change the bearings internal load distribution
compared to a model with bonded contacts based on multipoint constraints. It is a suffi-
cient way to model the bolts with beam elements that are connected to the surrounding
structure with node couplings.
The one-
third IWT7.5-
164 rotor star model consists of the main components, mean-
ing a one-
third of the rotor hub, the blade bearing, the stiffener plate and the rotor blade.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-127-320.jpg)
![Pitch system concepts and design 99
A part of it is exemplary shown in Figure 3.14. The outer ring of the bearing connects
directly to the flange of the hub. The bearing’s IR connects to the blade with the stiff-
ener plate in between. Using stiffener plates is a common way to reduce the ovalisation
of the bearing, which is caused by the loads and blade design. Further details of the
IWT7.5-
164 rotor star model can be taken from Reference [20]. In case it is foreseen to
strengthen the connection of rotor blade and rotor hub with additional components like,
e.g., ring extenders, those components need to be considered as well to end up with a
realistic assembly situation for the blade bearing.
Using only a one-
third rotor star model greatly reduces the computational time as
only one bearing model is implemented, which typically is the main driver for the com-
putational effort. A cyclic constraint is applied to the cutting planes of the one-
third
model of the rotor hub. Doing so makes the model behave symmetrically, meaning that
it is not possible to consider, that the three blades are loaded differently. The influence of
this simplification on the bearing’s internal load distribution is depended on the asym-
metry of the load acting on the entire rotor star. For load cases in which all blades are
loaded in a similar way the effect of this simplification is negligible. However, for simu-
lating the load distribution and contact angle variation in the blade bearing for the totality
of all possible load combinations it is recommended to build up a full rotor star model
containing the same components on all three flanges. A full rotor star model enables a
correct consideration of the hub’s deformation behaviour for asymmetric load condi-
tions that can affect the complex load distribution in the blade bearing at the flange of
interest. In both cases, the downwind flange of the hub model is completely fixed and the
rotation about the pitch axis is prevented by a very stiff torsional spring that is connected
between bearing’s inner and outer ring. Fixing the rotation about the pitch axis is crucial
as a rotation of inner and outer ring against each other would lead to incorrect results
regarding postprocessed contact forces when using a simplified FE bearing model that
bases on fixed spring elements, e.g., the Daidié approach shown in Section 3.1.4.
Figure 3.14
Finite element model of 1/3 IWT7.5-
164 rotor star
© Fraunhofer IWES](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-128-320.jpg)
![100 Wind turbine system design
To simulate the internal blade bearing loads, the loads acting on the rotor blade
are applied to the blade model in the blade coordinate system [1]. For this purpose,
two master nodes are created on the pitch axis of the rotor blade at two different
positions as shown in Figure 3.15. Each master node connects with force-
distributed
constraints (FDC) to slave nodes that are located at the spar caps. No bending
moments, only axial and radial forces are applied to the master nodes and the FDCs
transmit the loads to the structure of the blade via the slave nodes. The position of
the load application nodes should not be too close to the blade root flange as this
can affect the structural deformation behaviour in that region, which could lead to
unrealistic blade bearing loads. It is a good approach to use at least the first quarter
of the rotor blade length in order to consider the characteristic rotor blade behaviour
properly. For the exemplary one-
third IWT7.5-
164 rotor star model, the z-
positions
for the load master nodes are L1 = 15 m and L2 = 20 m.
By using two load application points, no fixed relation between bending
moment and radial forces is maintained and, in turn, this configuration allows
the generation of any desired load situation for the blade bearing even with fixed
z-
positions of the load master nodes. Based on the desired load situation at the
blade root, the radial loads that need to be applied at the master nodes are calcu-
lated as follows:
Fx2 =
My,root Fx,root L1
L2 L1
(3.16)
Fx1 = Fx,root Fx2 (3.17)
Fy2 =
Mx,root + Fy, root L1
L1 L2
(3.18)
Fy1 = Fy,root Fy2 (3.19)
Figure 3.15
Load application at the blade of the 1/3 IWT7.5-
164 rotor star FE
model © Fraunhofer IWES](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-129-320.jpg)
![Pitch system concepts and design 101
A first plausibility check of the load application can be done by using internal func-
tions to sum the nodal force and moment contributions of the elements at the bearing
flange surface. The summation point must be in the same position of the load refer-
ence point. Further plausibility checks for the total deformation behaviour of the
generated full rotor star model can be done by some synthetic load cases. Applying
the same loads to three blades should result in the same deformation at all flanges.
High loads (e.g., Mres,max
) only at one blade, while the others are not loaded, lead to
an overall tilting of the entire rotor star model.
After the plausibility check is done and it is ensured the rotor star model
behaves realistically, further reduction techniques can be applied in order to obtain a
more computationally efficient model. Superelement techniques based on condens-
ing a selection of finite elements into one unique element, named superelement.
Implementing the blade bearing between a blade-
sided superelement and a hub-
sided superelement allows a significant reduction in degrees of freedom and, in turn,
computational time with only a very little loss of accuracy [21].
3.1.6
Calculation and dimensioning
At this step of the design process, it must be verified whether the bearing withstands
the loads that it will see during the entirety of its lifespan. This consists, most impor-
tantly, of a statical and a dynamical verification. The statical calculation aims to
ensure that the bearing can survive a certain maximum load at least once, whereas
the dynamical verification aims to ensure that the entirety of loads during its lifetime
will not lead to fatigue failure of the raceways, rings or bolts.
Bearing verifications of this kind are typically performed using ISO 76 [9] or
ISO 281 [22]. These are intended for bearings placed in stiff surrounding condi-
tions. Blade bearings are, however, typically surrounded by rather flexible struc-
tures, which necessitates the FE models described in the previous section. The use of
these FE models is highly recommended for the following calculations as well. Any
formulae not including FE calculations risk significantly underestimating the actual
loads that occur, both statically and dynamically.
Static calculations
The statical calculations aim to ensure that the blade bearing can withstand the maxi-
mum load situations occurring for all operating conditions of the turbine. According
to Reference [1], several DLCs shall be considered to verify the structural integrity
of this component. The guideline provides a table with design situations and DLCs
and specifies safety factors for the loads.
The calculation starts with the analysis of the load time series for the DLCs in
order to identify the maximum resulting bending moment, according to 3.1. For the
IWT7.5-164, Mres,max
occurs for DLC 2.2 (cf. Section 3.1.2). All load components of
this load case are multiplied with the safety factor of 1.1 and afterwards simulated
with the developed FE rotor star model (see Section 3.1.5). The highest contact
force Qmax
obtained with the global bearing FE model is used to calculate the highest
resulting Hertzian pressure Pmax
. It is important to consider that the highest resulting](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-130-320.jpg)
![102 Wind turbine system design
bending moment does not necessarily lead to the highest contact pressure between
rolling element and raceway. As the structural behaviour of the rotor blade affects
the bearings internal load distribution, a different load angle or a specific combina-
tion of pitch angle and load angle can lead to higher rolling element forces and, in
turn, slightly higher contact pressure. For that reason, the occurring load angle for
the resulting bending moment needs to be calculated as well. In case the analysed
load time series contain resulting bending moments slightly lower than Mres,max
for
significantly differing load angles, these load cases should also be simulated with
the FE model.
As stated in Section 3.1.1, according to Reference [9], the maximum permis-
sible contact stresses shall be less than 4.2 GPa for ball bearings and 4.0 GPa for
roller bearings. This check can be done with the rolling element forces, obtained
from the global bearing model, which are used to calculate the maximum con-
tact pressure and size of the contact area. For double-
row four-
point contact ball
bearings the postprocessing of the resulting contact angles is required in order to
check for possible contact ellipse truncation. With the maximum contact angle,
it can be evaluated if contact ellipse truncation at the edge of the raceway will
be a severe problem for the current blade bearing design. It is crucial to prevent
the occurrence of contact ellipse truncation, as it goes along with an intensive
increase of the contact pressure. However, not only high contact pressures but also
high contact angle variations favour truncation. For that reason, it is not sufficient
only to take a look at the highest contact forces but also to check for truncation
in the region where the highest contact angle variations occur. In a double-
row
four-
point contact ball bearing that is loaded with a bending moment, the contact
angles on the traction side are usually higher than on the compression side of the
bearing. In case the contact ellipse comes very close to the raceway edge, a further
stress analysis with a detailed submodel (like exemplarily shown in Figure 3.13)
should be carried out.
Next step of the static calculations is the core crushing check according to
Reference [7]. As large blade bearings are only case-
hardened (see Figure 3.8),
there is a rapid decrease of the hardness at the transition zone between case (race-
way hardness) and core (ring material hardness). It is assumed that the core hard-
ness starts at 110% of the case depth. For that reason, it has to be ensured that the
resulting subsurface shear stresses do not reach down too deep into the material and
exceed the yield stress in shear or the limit shear stress in fatigue of the core mate-
rial. Therefore, the following equation has to be fulfilled:
shear,actual
shear,allowable
1
(3.20)
The allowable shear stress can be calculated either by means of the material’s ulti-
mate tensile strength and the given factor of 0.425
shear,allowable = 0.425 UTS (3.21)
or with an available value for the core hardness and the listed correlation to the yield
strength in shear (cf. Reference [6]). For the common material 42CrMo4, a typical](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-131-320.jpg)
![Pitch system concepts and design 103
value for the core hardness is around 25 HRC. The actual occurring subsurface shear
stress at the transition zone of case hardness and core hardness can be calculated by
the following equation:
shear, actual =
b
P
1.8754 105
(3.22)
This equation implies the semi-
minor axis of the contact ellipse b, the cumu-
lative measure of curvature
† and an interpolated subsurface shear stress
parameter
, which depends, among others, from the hardening case depth z.
For the IWT7.5-
164 blade bearing, the hardening case depth is z = 8 mm. The
maximum contact force Qmax
is postprocessed from the rotor star FE model
(cf. Section 3.1.5) loaded with the Mres,max
load case. On the inner ring raceway,
the resulting contact pressure is slightly higher than on the outer ring raceway,
which leads to two different values for the semi-
minor axis of the contact ellipse
b. Even though
shear, actualis slightly higher on the inner ring, it can be more criti-
cal if the calculated shear stress on the outer ring comes close to
shear,allowable
as the outer ring is more exposed to tensile stresses. Failure to satisfy this core
crushing check requires a reduction in the maximum load or a proper increase
in hardened case depth.
Fatigue
Any fatigue life calculation starts with the lifetime loads acting on the component
in question. According to Reference [1], fatigue calculations of any turbine com-
ponent include DLC 1.2, 1.7, 2.4, 3.1, 4.1, 6.4 and 8.3, weighed based on their
share of turbine operational time. DLC 1.2 represents the normal operation of the
turbine throughout its lifetime and is therefore undoubtedly the most important of
these.
Various safety factors can be applied to the calculation. IEC 61400 combines all
uncertainties into one material factor γm
and one load factor γf
, (γf
= 1 for fatigue states)
and DNV GL [4] additionally uses a factor related to the consequences of failure γn
,
which are to be applied to the cyclic stress or strain in each fatigue cycle. DNV GL
recommends safety factor γm
as shown in Table 3.5. The choice of parameters therefore
depends on the specifics of the turbine design, but in many cases, rolling contact fatigue
will use a safety factor of 1.0, if the turbine can safely endure the rotational failure
Table 3.5 Safety factors γm
according to DNV GL [5]
Component failure results in
the destruction of the wind
turbine or endangers people
Component failure results
in wind turbine failure or
consequential damage
Component failure
results in interruption of
operation
1.25 1.15 1.0](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-132-320.jpg)
![104 Wind turbine system design
of one bearing. Ring fatigue can cause severe subsequent damage and will therefore
mostly require factor 1.25. Bolted connections will likely use γm
= 1.15 because they
can be inspected and detected before a complete failure of the connection.
Rolling contact fatigue
For the blade bearing calculation, DLCs without any bearing movement may be
skipped as their effect on the lifetime calculation is assumed to be non-
existent with
all common calculation approaches.
As written above, fatigue lifetime (also referred to as rating life) of bearings is
typically performed according to ISO 281 [22]. For blade (and yaw) bearings, DNV
GL [5] specifically recommends ‘the ISO 281:2007 rating life calculation shall be
modified according to NREL Wind Turbine Design Guideline DG03, Section 4’.
This is due to a number of details that arise with blade bearings that ISO 281 does
not sufficiently factor in: in particular, blade bearings are primarily loaded with a
bending moment, which ISO 281 does not account for, and they move in small oscil-
lations rather than by rotating as assumed in ISO 281. The process illustrated in the
following is inspired by, but not identical to, NREL DG03 [7] but leans more closely
towards that described in Reference [20] which aims to improve some aspects of the
former.
There is not much available literature on the accuracy of fatigue lifetime cal-
culations for large slewing bearings, but some published calculations [20, 23] sug-
gest that the lifetime calculated will be lower than the actual usable lifetime of a
blade bearing. At the time of this writing, fatigue lifetime calculation is an ongoing
research topic. If readers have access to non-
published lifetime data, it is recom-
mended to compare this to the calculations performed in the following to be able to
correct any inaccuracies contained therein. Particular attention should be paid to the
fact that rating life is supposed to describe the point at which the first spall appears
on the raceway; for large slewing bearings, operation may be possible far longer
than this point.
In this case, we will focus on one of the bearings only, as fatigue loads will not
differ significantly between the three blade bearings typically used in modern wind
turbines. Required values from the DLCs are Mx
, My
, Fx
, Fy
and Fz
at the blade root
and the respective multipliers of each DLC.
Time series must be turned into a table of oscillations. Reference [6] describes
the process of a range-
pair count on a data set; for the following RCF calculations,
a rain flow count may be used as well and is recommended by NREL DG03. Both
approaches will result in similar outcomes. Figure 3.16 shows examples for a rain
flow count and a range-
pair count. Following the cycle count, a bin count akin to
those shown in References [6, 20] is carried out.
The IEC-
61400-
4 [24] recommends the usage of LRDs for gearbox bearings.
The following approach is very similar, but not identical to that of using LRDs, and
LRDs may be used instead of the following bin counting, too.
A generalised mean of each load component M (i.e., both forces and moments)
within each bin should be performed over all time steps t = 1 … T of the bin accord-
ing to](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-133-320.jpg)
![Pitch system concepts and design 105
Mweighted =
0
B
B
@
T
P
t=1
Mp
t
T
1
C
C
A
1/p
,
(3.23)
with exponent p = 3 for ball bearings and p = 10/3 for roller bearing as also used
below for the lifetime calculation. Moreover, a mean oscillation frequency Nb
, the
mean oscillation angle θb
and the time tb
spent in each bin b will be required. These
can be used to determine the accumulated movement lb
= 4·Nb
·θb
·tb
within the bin,
measured in degrees.
The above equation holds true for the oscillation angle θ defined as per
Figure 3.17. The oscillation angle will later also be required to correct for the oscil-
latory behaviour of the bearing.
Within each bin, it is then possible to calculate the lifetime L10
, which gives the
statistical point at which 10% of bearings are expected to have a first spall on the
raceway,
L10 = acorr
Ca
Pa
p
(3.24)
where p = 3 and p = 10/3 for ball and roller bearings, respectively. The factor acorr
can represent different corrective factors for oscillation, lubrication, etc., all of which
are multiplied with each other. Ca
refers to the dynamic axial load rating, which is
defined according to ISO 281 [22]. It is a function of geometrical parameters of the
bearing and can be determined by the manufacturer, irrespective of operating condi-
tions, and it is hence constant over all bins. For ball bearings with ball diameters of
above 25.4 mm and contact angles lower than 90°, it is defined as
Ca = 3.647bm fc (cos ˛)0.7
tan ˛ Z2/3
D1.4
W (3.25)
for other bearings refer to References [22] and [7]. The variable Z refers to the num-
ber of rolling elements per row, and DW
refers to the diameter of the rolling elements.
The contact angle is α and equals 90 for pure axial loads, in which case a different
equation for Ca
should be used, (cf. Reference [22]). The factor bm
= 1.3, and the
number of rows is given by i. The value fc
is either calculated according to Reference
Figure 3.16 Rainflow count (left) vs. range-
pair count (right) [15]](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-134-320.jpg)
![106 Wind turbine system design
[25] or interpolated. NREL DG03 [7] includes values for an interpolation of fcm
= fc
bm
using an osculation of 0.53.
The variable Pa
in the equation above is the equivalent load. It is ideally cal-
culated via ISO/TS 16281 [26] through the process described in Reference [20],
Section 2.6.3. In that case, an equivalent load for each race, denoted Qei
for races
on the inner and Qee
for ones on the outer ring, is, respectively, calculated using the
contact forces Qj
taken from finite element simulations. This is the most accurate
procedure to obtain the lifetime L10
.
Reference [7] proposes two simplified formulae that are analysed in Reference
[20]. One of them, denoted NREL 2, is a simplified variation of ISO/TS 16281 and
uses the force Qj
on two contact Points A and B on a race.
Pa =
1
ZNREL
PZNREL
j=1
QjA + QjB
3
1/3
ZNREL sin ˛
(3.26)
where ZNREL
= Z i. Alternatively, without using finite element simulations and con-
sequently the most inaccurate, one may use NREL 1 for an estimate, which goes
Pa = 0.75Fr + Fa + 2.5M
dm (3.27)
using an adjustment proposed by Reference [20]. Ideally, the factor 2.5 should be
validated with FE simulations by comparing to ISO/TS 16281 or NREL 2.
Figure 3.17 Angle definition used for θ © Fraunhofer IWES](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-135-320.jpg)
![Pitch system concepts and design 107
Pa
must be determined individually for each bin. Ideally, this means run-
ning finite element simulations for the weighted mean values in each bin.
Practically, this may not be possible, depending on the number of bins used.
In this case, one may simulate only a specified grid of points. Menck et al. [20]
propose an approach for a regression analysis of contact forces; an interpola-
tion between the simulated data points is possible, too. For simple calculations,
usage of NREL 1 can remove the need for finite element simulations entirely or
reduce its number significantly if some simulations are performed to verify the
formula.
The above calculations result in a lifetime L10
that describes a lifetime in full
revolutions. In order to account for the oscillatory behaviour of the blade bearing,
an oscillation factor aosc
is needed. Reference [27] proposes the curve-
fitted (cf)
factor
aosc_cf = aosc_IR_cf
1 +
aosc_IR_cf
aosc_OR
e
Bcf
1/e
1 + Bcf
1/e
(3.28)
using for the (curve-
fitted) IR oscillation factor aosc_IR_cf
for point contact
aosc_IR_PC_cf = f_crit_i 90
deg
1 0.09381 0.30679
,
(3.29)
where θdeg
is the oscillation angle of the respective bin in degrees, and with
aosc_OR = f_crit_o 90
deg
,
(3.30)
where fθ_crit_i,o
= 1 is recommended for typical blade bearings. Moreover, the load zone
parameter ε for each combination of inner and outer raceways can be estimated as
loaded
1 cos
NLB
Z
2
(3.31)
provided that NLB
Z, where NLB
is the number of loaded balls per row. The param-
eter Bcf
for the PC case is determined via
BPC_cf =
1
1 +
5.5958
fo
2fi 1
fi
2fo 1
!1.5153
1 + 0.11491 0.36257
(3.32)
The factor aosc
turns the lifetime L10
, measured in revolutions, into a lifetime L10osc
,
measured in oscillations. Since the following formulae will use revolutions rather
than oscillations, we thus introduce a corrective factor
acorr =
aosc
aHarris
=
aosc
90
deg.
(3.33)
The above equation corrects the oscillation factor aosc
by the factor
aHarris = 90/deg,
which is the oscillation factor that is implicitly assumed by using LRDs or ‘summing
up oscillations’ and dividing them over by 360° to obtain an equivalent number of rota-
tions. acorr
should reduce the lifetime for rotor blade bearings by about 10% if applied
correctly.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-136-320.jpg)
![108 Wind turbine system design
The load zone parameter ε can only be determined for one row (or combination
of inner and outer raceways, in a four-
point bearing). This means that the factor aosc
can only be determined for one row (or inner–outer raceway pair). If ISO/TS 16281
is used for the calculation of Pa
, each raceway pair can indeed use a separate aosc
,
but if NREL1 or NREL2 are used, one may use the lowest value of aosc
out of all the
raceways as a conservative estimate.
At this point, a lifetime L10b
can be calculated for each of the bins b. These are
then combined into a lifetime of the bearing over all bins L10
. This is achieved using
the Palmgren–Miner rule,
L10 =
1
lcoll
X
b
lb
L10,b
!1
(3.34)
where lb
is the movement in each bin b summed up (in degrees) over the entire
operating time of the turbine, and lcoll
is the sum of all movement in all bins, i.e.
lcoll
= Σb
lb
.
This finally results in a lifetime L10
, measured in revolutions, i.e., 360° move-
ments, of the bearing. Now an average rotational speed navg
(in rev/h) over the
entirety of operational time must be calculated by using
navg =
P
lb
P
tb
(3.35)
The final bearing lifetime in hours L10m,h
is then finally calculated as
L10m,h =
L10 106
navg
(3.36)
The lifetimes L10b
above can be modified to account for the influence of lubrica-
tion and contamination by determining the modified lifetime according to ISO 281.
NREL DG03 gives a summary of this procedure with a focus on blade bearings.
Using the procedure as given in NREL DG03, however, results in extremely low
values of aISO
, typically giving the lowest possible absolute value of aISO
= 0.1 [7,
20, 23]. It should thus be noted that ISO 281 specifically allows for a correction
of the viscosity ratio κ for lubricants that contain effective EP-
additives in applica-
tions in which poor performance would otherwise be expected. This is possible if
the additives have been tested under realistic lubrication conditions, e.g., in a real
application or by performing an FE8-
Test according to DIN 51819-
1. If this condi-
tion has been fulfilled, κ can even be set to 1, which results in a significant increase
of aISO
. This fact highlights the importance of an appropriate grease with effective
EP-
additives for the best performance of a blade bearing not just with respect to
wear but also raceway fatigue.
If the raceways are not hardened to 58 HRC, a correction of the lifetime, e.g.,
as given by NREL DG03, is necessary. Likewise, the case depth of the raceways
must be chosen appropriately such that at the boundary between core and hardened
raceway, the allowable fatigue shear stress of the core (i.e., the fatigue strength of](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-137-320.jpg)
![Pitch system concepts and design 109
the core with respect to shear stress) is not exceeded, otherwise, an adjustment as
shown in NREL DG03 must be undertaken.
Ring stresses
DNV GL requires the tangential hoop stresses of the bearing ring to be ‘analysed and
documented’. Compressive hoop stresses can increase the raceway life of a bearing
[28], but tensile hoop stresses can have the opposite effect.
If the ring stresses during fatigue loading are significantly above the fatigue
limit of the ring material, a fatigue lifetime calculation of the rings may be necessary
to prevent ring failure. Specifically, around the bolt holes and filler plugs if present,
high stresses are expected. DNV GL 2016 [5] requires a fatigue calculation of the
bolt holes.
Bearing rings can be considered as a structural component since they are part of
the wind turbine structure that transfers significant loads. For structural components,
Reference [4] refers to ISO 2394 [29]. ISO 2394 highlights the importance of accu-
rate models that particularly take into account those properties of the component that
have a high effect on its behaviour. For blade bearings, this particularly means that
FE models with an accurate representation of surrounding structures are very use-
ful; otherwise, high safety factors must be used due to high model uncertainty. Time
series from aeroelastic simulations must be turned into time varying sequences of
ring stresses, which, in turn, have to be transferred to bins of loads cycles using, e.g.,
the rain flow count approach. With these bins, S–N curves are commonly analysed
using the Palmgren–Miner rule, as is also done in the following sections for bolts.
S–N curves for unhardened 42CrMo4 steel taken directly from a blade bearing can
be found in the literature, e.g., in Reference [30], or experimentally determined.
According to ISO 2394, the calculation has to take into account various safety fac-
tors for possible uncertainties in the load level and model, material model and the
damage accumulation rule itself. These can be used as defined above by DNV GL
or IEC.
Bolt fatigue
Due to the mostly axial, dynamic loading at the bearing flanges, bolts are at risk of
fatigue failure. A bolt fatigue calculation is hence required by DNV GL. S–N curves
of the bolts depending on the manufacturing process of the bolts and the number
of load cycles are given in Reference [4] for a survival probability of over 97.7%.
The S-
N curves are given in Table 3.6 in terms of detail categories according to
Reference [31] and negative inverse slope.
The detail category of a bolt gives the reference fatigue strength in MPa, which
corresponds to the point of the S-
N curve that results in N = 2 million cycles. For
bolts with a nominal diameter d 30 mm, the S-
N curve is reduced by the factor
ks =
30
d
0.25
(3.37)](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-138-320.jpg)
![110 Wind turbine system design
According to References [4] and [31]. Both sources also highlight the importance of
considering bending stresses resulting from prying effects and other sources. This
may be omitted if detail category 36 is used instead.
Stress histories can be evaluated using a rain flow count or the reservoir method,
(cf. Reference [31]). Like with other fatigue calculations, different load amplitudes
can be combined using the Palmgren–Miner rule
D =
P
b
nb
Nb
(3.38)
where nb
is the number of load cycles that occurred in bin b, and Nb
is the number
that can occur according to the S–N curve given above at the specified stress ampli-
tude. After including all bins b, the result D ≤ 1 fulfils the design requirements. The
number of bins b should be high enough for sufficient accuracy, [32] recommends
at least 20.
Bolt loading along the circumference of the bearing flange will vary signifi-
cantly, and the highest bolt loads are expected in the centre of the tension side of
the bearing. Here, a sufficiently large section of bolts should be validated against
fatigue failure.
3.1.7
Lubrication system
The lubrication mainly ensures the function of the pitch system. In modern wind tur-
bines, a system supplies the lubricant automatically and in predefined intervals. The
main principals for pitch and yaw systems are similar. To avoid doubling, Chapter 7.3
gives further information regarding lubrications systems. This section explains charac-
teristics for pitch systems. The gearbox itself or bearings inside the motor and gearbox
must be lubricated as well. The gearbox is filled with oil, which circulates with the
Table 3.6 S-
N curves for different bolt types according to DNV GL [1]
Detail category Negative inverse slope Bolt type
For N 5·106
cycles
For N ≥ 5·106
cycles
71
2
FS max
F0,2 min
85
FS max= max. bolt
force
F0,2 min= bolt force
at 0.2 % elastic
strain limit
6 11 Rolled after heat treatment, no
thermal coating
71 3 5 Rolled before heat treatment,
no thermal coating
50 3 5 Rolled threads, thermal
coating](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-139-320.jpg)
![116 Wind turbine system design
Gearbox
A gearbox increases the torque and reduces the speed from the motor. Typically,
gearboxes have multi-
stage planetary gears to realise a high transmission within
limited space. It is filled with an oil to lubricate the gears and bearings. The output
pinion drives the blade bearing rotating ring and thus the blade.
Performance level
For the performance level (PL), there are two main functions that need to be fulfilled
to ensure a safe stop of the turbine during an emergency operation:
•
• safe movement to feather position
•
• safe stop in feather position
The function ‘safe movement to feather position’ is triggered by the safety
chain, the turbine controller or by the pitch system itself. The safety chain contains
two channels via the slip ring and is connected to a safety relay inside the pitch sys-
tem. This safety relay initiates the emergency function of each perfect pitch axis by
disconnecting 24V DC of the emergency feather command (EFC) input. The emer-
gency operation will be stopped when the first limit switch is reached by the func-
tion ‘safe stop in feather position’. According to the recommendation of DNV-
GL
Guideline [4], the PL should be ‘d’ for the function ‘Protection against excessive
rotor speed’. As mentioned, this is a recommendation. The PL is categorised from
‘a’ to ‘e’ and is defined in the DIN EN ISO13849-
1 [33], where ‘a’ is showing the
lowest probability of a dangerous failure per hour and ‘e’ is the highest. The real
required safety level needs to be evaluated during the risk assessment of the turbine.
Safety-
related parts of control systems (SRP/CS) are divided into subsystems
(input, logic, output) according to DIN EN ISO 13849 [34]. In this content, the SRP/
CS provides the safety function including a PL, which effectuates the necessary risk
minimisation. By providing the safety function, the layout of the SRP/CS is part of
the strategy of the risk minimisation.
Table 3.7 Pros and cons for AC and DC motors
Pros Cons
AC • Low costs for motor and
converter.
• Motors cannot be directly
powered by a battery in case
of converter failure.
DC • Motor can be directly
powered by batteries in case
of a converter failure.
• High torque at low speeds.
• Brushes on motor must be
frequently replaced.
• DC motors are costly.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-145-320.jpg)
![118 Wind turbine system design
3.2.2
Operating conditions
The pitch actuator rotates with the hub. Hence, a centrifugal force is acting at the
actuator. Shocks by gusts effect the pitch actuator as well. Interfaces like bolted con-
nections and electrical connectors and the components themselves must withstand
this load. Furthermore, it influences liquids, like the gearbox oil. Lubrication, which
includes the sealings, must fit to the rotating operation. The cut-
out wind speed of
the turbine and the position of the actuator in the hub allow an estimation of the
centrifugal force. Under normal operation, the cut-
out wind speed is the highest
rotational speed. The distance between the rotational axis and the actuator is the
lever arm. The IWT7.5-
164 has a cut-
out wind speed of 10 rev/min. The distance
from the rotational axis towards the motor and gearbox is about 2 m.
Environmental conditions effect the actuator as well. It is necessary to distin-
guish between the position of the blade if it is mounted to the inner or outer ring. In
the second case, the actuator must be mounted to the outside of the hub and protec-
tion against water or saltwater become more important. For both types, a protection
against dust and salt in the air is necessary. Temperature conditions can derivate
from the turbine temperature conditions, e.g., according to IEC 61400-
1 [4].
Due to active loads inside of the pitch cabinet, the experience shows a 10–15
K higher inside temperature compared to the ambient temperature. This delta can
be reduced to 5–10 K if those cabinets will be equipped with a fan. Measures to
keep the need protection grade of IP54 need to be done in this case. Table 3.8
lists the temperature conditions. Generally, three different temperature versions
are defined:
•
• normal climate version
•
• cold climate version
•
• hot climate version
3.2.3
Calculation and dimensioning
As mentioned in Section 3.1, loads for the dimensioning typically stem from the
aero-
elastic simulations (cf. Chapter 1). An aero-
elastic simulation time series of
IWT7.5-
164 reference turbine is available online and can be downloaded under [3].
The most important load component is the moment (Mz
), which would rotate
the blade along its longitudinal axis. The pitch actuator and the brake must be able
Table 3.8 Temperature conditions for the pitch actuator
Normal conditions Extreme conditions
T min −10°C −40°C
T max +40°C +50°C](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-147-320.jpg)
![Pitch system concepts and design 119
to counteract this moment. The other loads at the blade root influence the load in
the bearing and thus the friction torque. Besides the loads, the pitch speed
P
'
is a
mandatory input for the dimensioning of the pitch actuator, as the speed and torque
define the power.
Power calculation
The first step is the calculation of the required power of the pitch actuator. The torque
(M), which is necessary to turn the blade, consists of several parts. The actuator must
counteract the friction torque
Mfric
, the acting wind loads
Mz
and the inertia (J
), as
3.39 shows. M is the torque at the blade bearing ring.
M = Mz + Mfric + J '
(3.39)
To calculate the motor torque (MM
) the transmission from the bearing ring towards
the gearbox pinion (ibearing
) and the gearbox transmission (igearbox
) needs to be con-
sidered, as 3.40 shows. Information about the gearbox transmission is not available
yet. The choice of a gearbox is probably an iterative process and must balance with
the motor.
Furthermore, losses in the toothing, the gearbox, motor and converter increase
the needed power as well. They can be considered with efficiency factors.
MM = M
igearbox ibearing
(3.40)
The blade weight and its design have the highest influence on the inertia since it is
the heaviest part. Nevertheless, the inertia of the rotating ring of the blade bearing
and further parts like stiffener plates add to the total weight. If the design is fixed, it
is possible to determine the inertia of all parts. The wind loads are dependent on the
current pitch angle and the wind conditions. They can be simulated and hence esti-
mated in an aeroelastic simulation. Dependent on the direction of the pitch move-
ment Mz can be supportive or not. However, for an extreme load calculation, it is
considered as an additional resistance. The friction in the blade bearing comes with
the highest uncertainty. It consists of several parameters. The friction in the roll-
ing contact is determined by the type and number of rolling elements and it is load
dependant. The sealings, the cage and grease displacement have an influence as well.
With ongoing degradation of the bearing the friction torque will increase. Even if the
bearing is damaged, often it still can turn, just with a higher torque. A comprehen-
sive overview over several test results is presented in Reference [35]. It is possible
to use bearing manufacturers equations to estimate the friction torque, but often, the
bearings do not exactly reflect the behaviour predicted by the models. Furthermore,
the friction torque models in the literature differ as shown in Reference [36]. In
addition, it is questionable which one fits best to the application. Hence it may be
reasonable to add safety factors.
The required torque and the required pitch speed then can be used to choose a
motor type, its power-
class and the gearbox. Equation 3.41 can be used to calculate
the mechanical power.
P = MM P
' (3.41)](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-148-320.jpg)
![120 Wind turbine system design
Besides the motor curve, a decision needs to be made for the type of motor, fitting best
to the turbine needs (DC, AC asynchronous, AC synchronous). Mechanically, the motor
flange and shaft need to be aligned with the gearbox flange and shaft. As pitch system
suppliers not always also deliver the gearbox, this must be considered on both sides.
Emergency shut down
An emergency shut down of the turbine requires turning the blades to feather posi-
tion. According to the load simulation, the best emergency shut down speed will be
defined and is implemented into the pitch drive. For a detailed layout of the back-
up
system, the load requirements and the speed requirements during the emergency
operation need to be known and considered. Typically, the required EFC speed is
defined by the back-
up voltage and the needed torque will be considered in the
installed capacity. Especially for UC back-
up systems this is more relevant, as the
design considers, with a small safety margin, the exact needed energy. Figure 3.21
shows an example of the decreasing voltage level, during an EFC operation. Here it
must be made sure that the needed voltage level for the required EFC speed is kept.
Temperature
The IEC 60034 [37] gives guidelines for rotating electrical machinery, like the pitch
drive. Rising temperatures, because of high currents inside the drive and the motor,
will be monitored by the pitch system itself. The drive temperature will be measured
Figure 3.21
Decreasing voltage level of a capacitor during an EFC operation
© Nidec SSB Wind Systems GmbH](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-149-320.jpg)
![Pitch system concepts and design 121
by an internal temperature sensor as well as a Ixt-
function to reach the full dynamic
of the drive. The Ixt-
function displays the thermal load of the electronic components
inside the drive, which allows to use the full dynamic without risking the drive to
fail on thermal reasons.
For the motor, there are typically a temperature sensor showing the real tem-
perature in the winding and additionally a positive temperature coefficient close to
the winding temperature limit.
Brake
The brake prevents unintended changes of the pitch angle. Typically, the brake is
mounted at the motor and holds the motor shaft, hence acting loads must be trans-
mitted to the motor. The highest Mz
from the simulation data is 729 kNm. It occurs
in DLC 6.2, where the turbine is in parked position and the wind speed is extreme.
A safety factor of 1.1 applies for this load case.
If the brake is active, speed-
depended terms from 3.39 can be omitted since the
blade does not pitch. Furthermore, and in contrast to an intended direction change of
the pitch movement, here the pitch bearings friction helps to keep the blade in position.
Gearbox
The gearbox design must fit to the motor. The gearbox is probably an outsourced
item. Nevertheless, a static and a fatigue strength verification according to DNV
GL [5] are necessary. For the calculations of the components the individual norm
applies. For example, for the bearings ISO 76 [9] and ISO 281 [22] or for the gears
ISO 6336 [38]. The usage as a part of a wind turbines pitch drive comes with one
additional challenge. The pitch angle adaptions are often just a few degrees. Besides
a frequent direction change, it leads to an unsymmetrical load of the teeth. A small
adaption of the pitch angle means that just a few teeth transmit the load. This affects
especially for the stages with a low rotational speed, the pinion and the blade bearing
itself. These circumstances could be considered with a correction factor.
References
[1] DNV GL ‘Loads and site conditions for wind turbines’. 2016.
[2] Popko W. and Thomas ‘IWES Wind Turbine IWT7.5-
164. Rev 4’. 2018.
[3] Popko W. ‘Aero-
Elastic Simulation Time Series of IWT7.5 Reference
Turbine’. 2019.
[4] IEC 61400-
1 ‘C wind turbines - part 1: design requirements’. 2019.
[5] DNV GL ‘Machinery for wind turbines’. 2016.
[6] Stammler M., Reuter A., Poll G. ‘Cycle counting of roller bearing oscillations
– case study of wind turbine individual pitching system’. Renewable Energy
Focus. 2018, vol. 25, pp. 40–47.
[7] Harris T., Rumbarger J.H., Butterfield C.P. ‘Wind Turbine Design Guideline
DG03: Yaw and Pitch Rolling Bearing Life’. 2009.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-150-320.jpg)
![122 Wind turbine system design
[8] Houpert L. ‘An Engineering Approach to Hertzian Contact Elasticity—Part I’.
Journal of Tribology. 2001, vol. 123(3), pp. 582–588.
[9] DIN ISO 76:2009-
01 ‘Rolling bearings - static load ratings (ISO 76:2006)’.
2006.
[10] VDI Verein Deutscher Ingenieure ‘Systematische Berechnung Hoch-
beanspruchter Schraubenverbindungen Zylindrische Einschraubenverbind-
ungen’. VDI. 2015.
[11] Wittel H., Jannasch D., Voßiek J., Spura C. ‘Roloff/MatekMaschinenelemente:
Normung, Berechnung, Gestaltung’. in Springer Vieweg; 2019.
[12] Daidié A., Chaib Z., Ghosn A. ‘3D simplified finite elements analysis of load
and contact angle in a slewing ball bearing’. Journal of Mechanical Design.
2008, vol. 130(8).
[13] Gao X.H., Huang X.D., Wang H., Chen J. ‘Modelling of ball-
raceway contacts
in a slewing bearing with non-
linear springs’. Proceedings of the Institution of
Mechanical Engineers, Part C. 2011, vol. 225(4), pp. 827–831.
[14] Smolnicki T., Rusiński E. ‘Superelement-
based modeling of load distribu-
tion in large-
size slewing bearings’. Journal of Mechanical Design. 2007,
vol. 129(4), pp. 459–463.
[15] Stammler M. ‘Endurance test strategies for pitch bearings of wind turbines’
Fakultuät für Bauingenieurwesen und Geodäsie, Gottfried Wilhelm Leibniz
Universität Hannover; 2020.
[16] Stammler M., Baust S., Reuter A., Poll G. ‘Load distribution in a roller-
type
rotor blade bearing’. Journal of Physics: Conference Series. 2018, vol. 1037.
[17] Wang H., He P., Pang B., Gao X. ‘A new computational model of large three-
row roller slewing bearings using nonlinear springs’. Journal Mechanical
Engineering Science. 2017, vol. 231(20), pp. 3831–3839.
[18] Chen G., Wen J. ‘Load performance of large-
scale rolling bearings with sup-
porting structure in wind turbines’. Journal of Tribology. 2012, vol. 134(4).
[19] Laird D., Montoya F., Malcolm D. ‘Finite element modeling of wind turbine
blades’. 43rd AIAA Aerospace Sciences Meeting and Exhibit [online]; Reno,
NV, Reston, VA, 2018. Available from https://arc.aiaa.org/doi/book/10.2514/
MASM05
[20] Menck O., Stammler M., Schleich F. ‘Fatigue lifetime calculation of wind
turbine blade bearings considering blade-
dependent load distribution’. Wind
Energy Science. 2020, vol. 5(4), pp. 1743–1754.
[21] Plaza J., Abasolo M., Coria I., Aguirrebeitia J., de Bustos I.F. ‘A new finite ele-
ment approach for the analysis of slewing bearings in wind turbine generators
using superelement techniques’. Meccanica. 2015, vol. 50(6), pp. 1623–1633.
[22] DIN ISO 281:2010-
10 ‘Rolling bearings – Dynamic load ratings and rating
life (ISO 281:2007)’. 2007.
[23] Schwack F., Stammler M., Poll G., Reuter A. ‘Comparison of Life calcula-
tions for Oscillating Bearings Considering Individual Pitch Control in Wind
Turbines’. J. Phys.: Conf. Ser.. 2016, vol. 753, p. 112013.
[24] International Electrotechnical Commission ‘Wind turbines: part 4: Design
requirements for wind turbine gearboxes, IEC 61400-
4’. 2012.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-151-320.jpg)
![Pitch system concepts and design 123
[25] ‘DIN, DIN SPEC 1281-
1:2010-
05: Wälzlager - Erläuternde Anmerkungen
zur ISO 281 - teil 1: Dynamische Tragzahlen und nominelle Lebensdauer
(ISO/TR 1281-
1:2008 + cor. 1:2009)’. 2009.
[26] ‘DIN, DIN 26281:2010-
11: Rolling bearings – Methods for calculating
the modified reference rating life for universally loaded bearings (ISO/TS
16281:2008 + cor. 1:2009)’. 2009.
[27] Houpert L., Menck O. ‘Bearing life calculations in rotating and oscillating
applications’. Journal of Tribology. 2022, vol. 144(7), pp. 1–31.
[28] Oswald F.B., Zaretsky E.V., Poplawski J.V. ‘Relation between residual and
hoop stresses and rolling bearing fatigue life’. Tribology Transactions. 2014,
vol. 57(4), pp. 749–765.
[29] ‘ISO, ISO 2394:2015-
03: general principles on reliability for structures’.
2015.
[30] Friederici V., Schumacher J., Clausen B. ‘Crack propagation modelling
for service life prediction of large slewing bearings’. Procedia Structural
Integrity. 2014, vol. 35, pp. 106–114.
[31] ‘European committee for standardisation, EN 1993-
1-
9: eurocode 3: design
of steel structures-
part 1-
9: fatigue’.
[32] Det norske veritas ‘DNV-
RP-
C203: Fatigue Design of Offshore Steel
Structures’. 2021.
[33] ‘DIN EN ISO 13849-
1:2016-
06: Sicherheit von Maschinen-
sicherheitsbezogene
teile von Steuerungen-
teil 1: Allgemeine Gestaltungsleitsätze, DIN ISO’. 2016.
[34] DIN ISO ‘ISO 13849-
1:2015: Safety of machinery — Safety-
related parts of
control systems — part 1: General principles for design’. 2015.
[35] Menck O., Behnke K., Stammler M., Bartschat A., Schleich F., Graßmann M.
‘Measurements and modeling of friction torque of wind turbine blade bear-
ings’. Journal of Physics. 2022, vol. 2265(2), p. 022087.
[36] Stammler M., Schwack F., Bader N., Reuter A., Poll G. ‘Friction torque of
wind-
turbine pitch bearings – comparison of experimental results with avail-
able models’. Wind Energy Science. 2018, vol. 3(1), pp. 97–105.
[37] IEC ‘Rotating electrical machines: part 1: Rating and performance’. [IEC
60034-
1] 2022.
[38] DIN ISO ‘ISO-
1:2006-
09: Tragfähigkeitsberechnung von Gerad- und
Schrägver zahnten Stirnrädern - teil 1: Grundnorm, einführung und allge-
meine Einflussfaktoren’. 2006.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-152-320.jpg)
![Yaw system concepts and designs 127
addition, it provides friction torque during yawing to reduce the alternating
loads on the yaw drivetrain. The yaw brake system usually consists of a set of
yaw brakes mounted on the nacelle and acting on a yaw brake disc mounted
between the yaw bearing and the tower. However, there are also yaw system
concepts in which the nacelle is held in position solely by the yaw motor brakes
or by the yaw motors themselves (see section 4.3).
•
• A yaw control system: It processes the signals from the wind direction sensors
and the yaw system sensors such as the nacelle position. The control system
manages, commands, directs or regulates the behaviour of the yaw system.
Downwind turbines have the theoretical advantage that they may be built with
passive yaw systems since the rotor itself is able to yaw the nacelle into the wind.
However, if a wind turbine yaws passively in the same direction for a long period
of time, the cables between the nacelle and the tower are twisted impermissibly. For
that reason, at least a few yaw drives are required for untwisting the cables.
In addition, passive yaw systems for multimegawatt wind turbines have to be
designed in a way that the nacelle does not follow every change in wind direction and the
yaw speed does not become too high to avoid high gyroscopic loads. As a result, passive
yaw systems are basically similar to active yaw systems as described above.
When the rotor is aligned with the wind direction or when the nacelle is main-
tained, the yaw system has to ensure that the nacelle holds its current position with
the holding torques of the yaw system components. These torques can be:
•
• friction torque of the yaw bearing
•
• friction torque of the yaw brake system
•
• friction torque of the yaw motor brakes
•
• driving torque of the yaw motors
If the external load exceeds the holding torque, the yaw system slips. Such yaw
slippage events have to be carefully assessed because they can lead to critical situa-
tions for the wind turbine and the yaw system components (see section 4.2.6).
For maintenance and repair activities, a yaw locking device is required in case the
holding torque is not sufficient, with which the yaw system is locked in place to exclude
the risk of personal injury. According to the DNV standard [1], a yaw locking device is
not needed if there are two independent brake systems, e.g., the yaw brake system and
the yaw motor brakes, and each brake system is able to hold the nacelle in position.
When the control system detects an average difference between wind direction
and nacelle position that exceeds a certain value in a certain period, it initiates a
yawing process. There are typically different criteria for adapting the frequency of
the yawing process to the respective wind turbine operating status, such as:
•
• idling of the wind turbine
•
• power production at low wind speeds (partial load range)
•
• power production at high wind speeds (full load range)](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-156-320.jpg)
![130 Wind turbine system design
To sum up, wind direction distribution and wind direction changes vary greatly
between locations. As a result, the number of operating hours and the number of
yawing processes also vary greatly.
The difference between nacelle direction and wind direction is called yaw misalign-
ment. During power production, the yaw misalignment corresponds approximately to a
normal distribution. The standard deviation depends on the yaw control system and its
criteria for initiating a yawing process. It is usually less than 10° (Figure 4.3).
Yaw misalignment results in a lower efficiency of the rotor. Figure 4.4 shows the
decrease in the power coefficient CP
with increasing yaw misalignment. According
to the momentum theory for a wind turbine rotor in steady yaw, the power coeffi-
cient CP
is given by the following equations, in which γ represents the yaw misalign-
ment [2]:
CP = 4a(cos a)2
(4.1)
CPmax =
16
27
cos3
at a =
cos
3
(4.2)
• power coefficient [–]
• max. power coefficient [–]
• yaw misalignment [°]
CP
CPmax
This cos3
γ rule is commonly used for evaluating the influence of yaw misalign-
ment on the energy yield.
Figure 4.3 Exemplary yaw misalignment distributions](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-159-320.jpg)
![Yaw system concepts and designs 131
If the rotor is imperfectly aligned with the wind direction, even in a steady wind,
the angle of attack on each blade is continuously changing as it rotates. As a result, the
loads on the rotor blades are fluctuating, causing higher fatigue loads for the blades and
thus for the wind turbine. For that reason, yaw misalignment needs to be considered in
the load simulation.
In the partial load range of the power curve, yaw misalignment leads to yield
losses and higher fatigue loads. In the full load range, however, the electrical power
of the wind turbine is regulated by adjusting the pitch angles of the rotor blades. The
pitching of the rotor blades only begins after the nominal power has been reached.
As a result, the poorer efficiency of the rotor has no influence on the energy yield.
However, the wind turbine is still subject to higher fatigue loads.
4.1.3
Typical key data
In the following, some typical key data of active and friction-
damped yaw systems are
discussed. Since these depend heavily on the location, the yaw system concept and the
yaw control strategy of the respective wind turbine, only a general overview can be
given.
The number of operating hours depends heavily on the wind direction distribu-
tion, the wind direction changes (see section 4.1.2), the yaw speed and the control
system. In its standard [1], the DNV requires at least 10% of the turbine’s service
life for the number of operating hours, which corresponds to an average of 876
hours per year. This seems to be a conservative approach, because at many sites the
operating hours are well below 5% of the turbine life. For the design of yaw systems,
Figure 4.4 Power coefficient variation with yaw misalignment](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-160-320.jpg)
![Yaw system concepts and designs 133
teeth is added to this. The minimum normal backlash is usually 0.03–0.04 times the
gear module and needs to be converted to the circumferential backlash.
The efficiency of the yaw gearboxes, which usually consists of four to five
planetary stages, can be assumed to be 97.0–97.5% per gear stage. The efficiency
between output pinion and yaw bearing teeth lies in a similar range. The overall
efficiency of the yaw system is thus roughly between 83% and 88%.
4.2 Design loads
Basic knowledge of the loads acting on the yaw system is important for understand-
ing the different yaw system concepts, the component designs and the dimensioning
of yaw systems. For this reason, this section deals with the design loads.
After a short introduction to the consideration of yawing processes and yaw
misalignments in the load simulation, the loads acting on the yaw bearing and the
yaw drivetrain are described. Then the focus is on the loads acting on the yaw drive-
train and how the aerodynamic loads are modified to consider all loads in the design
of the yaw drivetrain. Finally, some special load situations that can occur during
yawing processes are discussed.
The Fraunhofer IWES wind turbine IWT-
7.5-
164 serves as common thread for this
book. The loads for this turbine were simulated between 2013 and 2017. The focus was
not on the yaw system. As a result, the loads on the yaw system are quite high compared
to today.
Considering today’s possibilities for turbine load reduction and optimising the
design-
driving load cases, the yaw system loads would be significantly lower. For
this reason, the authors of this chapter have decided to use the yaw system loads of a
comparable 7 MW wind turbine with a rotor diameter of about 170 m.
4.2.1
Introduction
For wind turbine certification, system and component loads are to be simulated accord-
ing to IEC 61400 [3] or DNV standard [4]. The standards specify the minimum require-
ments for the design load cases (DLCs) for the assessment of ultimate and fatigue
strength of the wind turbine. Further information can be found in Chapters 1 and 2.
Yawing processes are usually not part of the DLCs. Only a few DLCs may
require yawing, depending on the definition of the DLCs by the wind turbine original
equipment manufacturers (OEMs). Yawing processes are therefore often modelled
in a simplified manner to simply meet the certification requirements and to reduce
the turbine loads in the simulation, e.g., by yawing with constant speed. In addition,
the yaw system is often modelled rudimentary in current simulation software.
As a result, the yaw system loads are often given for the non-
yawing wind tur-
bine. For the dimensioning of the yaw system and its components, the loads must
therefore be modified to consider yawing processes (see section 4.2.5).
The DLCs are simulated for different mean yaw misalignments, depending on
the yaw control system and possible failure load cases. In the fatigue load post-
processing, the time series are weighted according to their frequencies.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-162-320.jpg)
![134 Wind turbine system design
Appropriate assumptions are to be made for yaw misalignment. If values for
the wind turbine cannot be specified, yaw misalignment of −8°, 0° and +8° evenly
distributed in ±8° shall be applied in DLC 1.2 according to the DNV standard [4]
(Table 4.1). Other yaw misalignments, e.g., −5°, 0° and +5°, and other percentages
of time, e.g., 25%, 50% and 25%, can be found for DLC 1.2.
A higher number of values for the yaw misalignment and the percentage
of time would represent the yaw misalignment distribution (Figure 4.3) better.
However, the number of values is generally kept small to keep the number of
required load simulations and thus the simulation time low. Nevertheless, an
attempt should be made to map the behaviour of the yaw control system with its
different criteria for adapting the frequency of the yawing process to the wind
turbine operating status.
The yaw system loads are usually given for the yaw bearing coordinate system
(Figure 4.5). These loads include:
•
• aerodynamic loads
•
• mass and inertia loads
•
• gyroscopic loads
•
• reaction forces, e.g., of the powertrain
In Figure 4.5, the yaw bearing loads are drawn in directions according to a con-
ventional coordinate system and not in the direction they usually have:
XK
• horizontal in direction of the rotor axis
• ZK vertically upwards
• YK horizontally sideways, so that XK, YK, ZK rotate clockwise
• FXK radial force Fx
• FYK radial force Fy
• FZK axial force Fz
• MXK roll moment Mx
• MYK tilt moment My
• MZK yaw moment Mz
In the following subsections, the yaw system loads for the non-
yawing wind
turbine are discussed.
Table 4.1 Yaw misalignment in DLC 1.2 [4]
Yaw misalignment Percentage of time
−8° 33.3%
0° 33.3%
+8° 33.3%](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-163-320.jpg)
![Yaw system concepts and designs 135
4.2.2
Yaw bearing loads
Yaw bearings are usually dimensioned by ultimate loads. Above all, failures in the
pitch system that cause high aerodynamic imbalances, e.g., if a rotor blade is stuck,
or large yaw misalignments at high wind speeds, can lead to very high loads on the
yaw bearing. Which DLCs ultimately lead to the highest loads depends on several
factors, such as the wind turbine design and the control and safety architecture.
Table 4.2 shows an example of the yaw bearing ultimate loads of a wind turbine
of the 7 MW class based on a statistical analysis of all DLCs for ultimate assess-
ment. For each load component, the loads are listed at the point of time at which the
respective load component has its maximum or minimum value. It should be noted
that the values in the extreme load table are mean values of the maximum or mini-
mum values from a set of time series of a certain DLC.
The magnitude of the forces is up to about 5,000 kN. The moments reach values
up to about 24,000 kNm. The bending moments Mx
and My
are particularly relevant
for the design of the yaw bearing as they lead to high contact stresses in the yaw
bearing. As the entire nacelle rests on the yaw bearing, it is also subject to high axial
loads. The radial forces play a rather subordinate role.
Figure 4.6 exemplifies the course of the resulting bending moment Mxy
in DLC
1.3 (power production with extreme turbulence model) at a wind speed of 19 m/s.
Figure 4.5 Yaw bearing coordinate system [4]](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-164-320.jpg)
![Table 4.2 Yaw bearing ultimate loads
Mx
My
Mxy
Mz
Fx
Fy
Fxy
Fz
γF
Load case
[kNm] [kNm] [kNm] [kNm] [kN] [kN] [kN] [kN] – –
Mx Max 16,100 10,030 19,580 9,650 890 –230 920 –4,730 1.35 dlc1p3
Min –6,890 –2,250 7,340 –3,720 580 150 600 –3,770 1.10 dlc2p2g
My
Max 10,730 24,110 26,510 –7,350 410 –240 490 –4,490 1.35 dlc2p1a
Min 8,180 –20,440 22,030 4,070 260 100 290 –3,830 1.10 dlc2p2b
Mxy
Max 12,260 23,940 26,940 –3,920 730 –170 760 –4,550 1.35 dlc1p3
Min 10 –10 20 –2,280 60 –60 90 –3,060 1.10 dlc7p1b
Mz Max 10,880 4,040 11,920 19,190 350 –190 400 –3,800 1.10 dlc2p2b
Min 7,380 –7,200 10,450 –20,120 230 190 320 –4,360 1.35 dlc1p4
Fx Max 12,520 5,380 13,650 –990 2,050 –20 2,050 –4,850 1.35 dlc1p4
Min –1,390 –4,430 4,640 –2,650 –1,380 20 1,380 –3,640 1.10 dlc2p3
Fy
Max –2,530 1,140 2,900 –6,350 –150 1,080 1,090 –3,590 1.10 dlc6p2
Min 3,390 –3,580 5,180 7,410 –160 –1,210 1,220 –3,440 1.10 dlc6p2
Fxy
Max 12,520 5,380 13,650 –990 2,050 –20 2,050 –4,850 1.35 dlc1p4
Min –60 –3,000 3,000 140 10 –10 10 –3,430 1.10 dlc2p2b
Fz Max 1,260 5,010 5,360 2,690 110 –410 420 –2,860 1.10 dlc6p2
Min 12,810 5,370 14,230 7,060 800 –140 810 –5,050 1.35 dlc1p3](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-165-320.jpg)
![154 Wind turbine system design
downwind position, the nacelle must also be locked so that it maintains its current
position (see section 4.2.3).
The active yaw system is equipped with torque producing components able to
rotate the nacelle against the stationary tower. As described in section 4.1.1, it con-
sists of:
•
• A yaw bearing that rotatably connects the nacelle with the stationary tower and
transfers the loads from the nacelle into the tower.
•
• A set of yaw drives that provide the driving torque to turn the nacelle and the
holding torque to hold the nacelle in position.
•
• A yaw brake system that provides holding torque to lock the nacelle and friction
torque to dampen the aerodynamic loads.
•
• A yaw control system that manages, commands, directs or regulates the behav-
iour of the yaw system.
There are a few downwind horizontal-
axis wind turbines in the multimegawatt
range on the market. This wind turbine configuration has the theoretical advantage
that it may be built with passive yaw systems since the rotor itself is able to yaw the
nacelle into the wind. However, a few active yaw drives are needed to untwist the
cables and to turn the nacelle when the rotor is not turning (see section 4.1.1).
The passive yaw system utilises the wind force to adjust the wind turbine rotor
into the wind. Small upwind turbines up to 100 kW can also be equipped with pas-
sive yaw systems consisting of a yaw bearing and a tail fin (or wind vane). The
tail fin is designed in such a way that it turns the rotor into the wind by applying a
‘corrective’ torque to the nacelle. However, this low-
cost and reliable solution is
not able to cope with the high aerodynamic loads in large wind turbines. For this
reason, passive yaw systems in multimegawatt wind turbines can only be found in
downwind turbines.
In large wind turbines, passive yaw systems are basically similar to active yaw
systems, as they have to be designed in that way that the nacelle does not follow
every change in wind direction and the yaw speed does not become too high to avoid
high gyroscopic loads. This is achieved by a high yaw friction torque. In addition to
the friction torque of the yaw bearing and the yaw brake system, the yaw drives can
provide additional damping torque to control the yaw movement.
Downwind offshore turbines mounted on a floating structure may allow to omit
the yaw system, as the whole turbine including tower and floating structure are
aligned to the wind direction. A rotatable connection between nacelle and tower
is then not required. One example is the Nezzy² floating wind turbine, developed
by Aerodyn Engineering GmbH and tested by EnBW Energie Baden-
Württemberg
AG [5].
In the event of a grid loss, downwind turbines with a passive yaw system have
the advantage that the rotor aligns itself with the wind direction, whereas turbines
with an active yaw system can no longer be yawed. This can lead to very large yaw
misalignments and thus high turbine loads that can be dimensioning for certain wind
turbine components.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-183-320.jpg)
![Yaw system concepts and designs 155
One way of avoiding these high loads is to install an energy storage device that
enables the active yaw system to continue to operate for a certain period. Another
option would be to bring the rotor into the downwind position and to switch from
active to passive yawing.
As shown in section 4.2.3, a yaw friction torque is needed to dampen the aero-
dynamic loads. It can be provided in the following ways:
•
• Yaw brake system: In the case of an active yaw brake system, the pressure on
the brake pads is typically reduced so that a brake torque is also generated dur-
ing yawing. The pressure is increased again after the yawing process is finished.
It would also be possible to vary the pressure depending on the load during
yawing. In the case of a passive yaw brake system, the brake torque cannot be
changed. If the passive yaw brake system is designed to hold the nacelle in posi-
tion, the yaw drives have to work against a high friction torque.
•
• Yaw bearing: A pretensioned sliding bearing has a much higher friction torque
than a roller bearing. It normally provides so much friction torque that there is
no need for an additional yaw brake system. The sliding bearing can be active
or passive. The active one allows to reduce the friction torque during yawing.
With the passive sliding bearing, the yaw drives have to work against a high
friction torque.
There are yaw systems with electro-
mechanical yaw drives and a low yaw fric-
tion torque (section 4.3.8). The yaw motion is then damped a little by the yaw drives.
To eliminate the torsional backlash, the yaw drives are then usually pretensioned.
The backlash is still present but is no longer passed through at the same time.
During non-
yawing operation, the yaw system can be either stiff or soft. While
the yaw angle remains constant in stiff yaw systems, limited damped yaw motion is
possible in soft yaw systems. Studies have shown that soft yaw concepts can lead
to considerable reduction in fatigue and ultimate load of the yaw moment Mz
[6].
Other components, such as rotor blade and tower base, can also benefit a little from
soft yaw systems.
Yaw systems with a yaw brake system are usually stiff during non-
yawing oper-
ation since the brake system fixes the yaw angle. The yaw system may only slip in
extreme load events (see section 4.2.6).
In upwind turbines, yaw systems without a yaw brake system, in which the
nacelle is held in position by the yaw motor brakes of the pretensioned electro-
mechanical yaw drives, can also be considered to be stiff. If the yaw friction torque
is exceeded, only very small yaw movements occur along the torsional stiffness
curve of the yaw system.
Soft yaw systems dampen wind gusts by yielding to wind loads. Depending on
the external loads and the soft yaw system design, the yaw angle can change up to
five degrees. Passive yaw systems in downwind turbines often include the soft yaw
function. In upwind turbines, soft yaw can be achieved by hydraulic yaw systems as
presented in Stubkier [6] or by electric yaw systems, in which the nacelle is held in
position by the energised yaw motors.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-184-320.jpg)
![Yaw system concepts and designs 183
addition, several variants are examined to come to the best solution. The system and
component designs are further detailed in several loops. The number of variants is
reduced in the process.
The auxiliary systems are dimensioned at a later stage. Detailed information
about the other components is often required here, which are not available at the
beginning of the development process. Of course, the auxiliary systems must be
considered when deciding on a variant.
Changes in component designs can affect other yaw system components and
the companion structure. Changes from the outside, such as new loads, changed
installation space or new requirements, can lead to major changes. For these reasons,
there is often an iterative jump back and forth to one or more work steps.
Table 4.11 summarises the key data of the Fraunhofer IWES wind turbine IWT-
7.5-164, Table 4.12 the requirements for the yaw system. These are explained in
more detail below.
Table 4.11 Wind turbine data IWT-
7.5-
164 [7]
Parameter Value
Rated power 7.5 MW
Rotor diameter 164 m
Hub height 120 m
Rated rotor speed 10 rpm
Drivetrain concept Direct drive
Design life 20 years
Operating temperature −20 to +40°C
Survival temperature −30 to +50°C
Yaw moment of inertia 5.531E+07 kgm²
Table 4.12 Boundary conditions for the yaw system
Parameter Value
Active/passive Active
Soft/stiff Stiff
Damping Friction-damped
Type of yaw drives Electro-
mechanical drives with asynchronous motors
Yaw system loads Section 4.2, maintenance loads see section 4.4.2
Operating hours 10% of turbine design life
Yaw speed 0.25–0.50°/s
Yaw slippage Allowed to a limited extent in case of extreme load cases
Power supply 400 V, 50 Hz
Max. machine frame width 4.2 m
Available installation space for
yaw drives
225° of 360°
(left, right and rear)](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-212-320.jpg)
![184 Wind turbine system design
A turbine design life of 20 years is assumed based on the given loads (see section 4.2).
The temperature range is not specified for the IWT-
7.5-
164 [7]. Since the temperature
range has only a minor influence on the yaw system design and the loads were simulated
for normal temperatures, a normal temperature application is assumed.
Cold climate or hot climate applications must be considered in the component
designs accordingly. For example, materials, seals and lubricants must be suitable
for the extreme temperatures.
The yaw system for the IWT-
7.5-
164 has not yet been detailed. Since it shall be based
on common yaw system concepts (see section 4.3.8), the yaw system shall be an active,
stiff and friction-
damped one using electro-
mechanical drives with asynchronous motors
(Table 4.12).
As explained and presented in section 4.2, loads of a comparable 7 MW wind
turbine with a rotor diameter of about 170 m are used instead of the IWT-
7.5-
164
loads. These loads correspond better to the current state of the art.
The operating hours of the yaw system are assumed to be 10% of the turbine design
life. This corresponds to the conservative specification from the DNV standard [1].
The yaw speed of the IWT-
7.5-
164 is given as 1°/s [7]. This is comparatively
high for such a large wind turbine (see section 4.1.3). For this reason, the target cor-
ridor is set to 0.25–0.50°/s.
Yaw slippage (see section 4.2.6) is only allowed to a limited extent in extreme
load cases and should be avoided if possible. A simulation of yaw slippage events is
not carried out as part of the example dimensioning.
The outer diameter of the tower top flange is specified as 3 m [7], which is com-
paratively small for a 7 MW wind turbine with the given loads. With a larger diam-
eter, more yaw drives and yaw brakes would fit in the wind turbine. In addition, one
yaw drive and one yaw brake generate a higher driving and holding torque due to the
larger diameter, which reduces the number of drives and brakes required.
A larger diameter is also beneficial for the yaw bearing. Smaller balls can be
arranged on a larger raceway diameter. This reduces the bearing height and improves
the lubrication conditions in the bearing. For the same adjustment angle, the balls cover
a greater distance in relation to their diameter compared to a smaller bearing with larger
balls. As a result, the grease is better distributed in the yaw bearing and is thus better
mixed. The lubricating film is also renewed more quickly, thereby reducing wear.
For these reasons, the specified diameter of the tower top flange is not used here.
It should rather result from the yaw system design.
The IWT-
7.5-
164 is an onshore wind turbine. Since this is a reference wind
turbine, no target markets are defined from which requirements for the permissible
nacelle dimensions could be derived. However, to be able to transport the nacelle
as cost-
effectively and flexibly as possible, a certain width should not be exceeded.
Larger widths are possible, but limit means and routes of transportation. This makes
transportation more expensive and reduces the number of possible wind turbine sites.
For the example dimensioning, the maximum permissible width of the machine
frame is set at 4.2 m. This allows road transport in many markets as well as rail
transport in the United States of America. It is assumed that the nacelle cover and
secondary structure can be transported separately if necessary.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-213-320.jpg)
![Yaw system concepts and designs 185
Machine frame design and drivetrain concept have a major impact on the avail-
able installation space for the yaw drives. Wind turbines with a geared drivetrain
usually only allow yaw drives to be arranged to the left and the right of the pow-
ertrain since rotor shaft and bearing assembly, main gearbox and generator require
a lot of space in the nacelle above the yaw system. In comparison, direct drive wind
turbines, where the generator hangs in front of the tower, often offer more space for
yaw drives.
For the direct driven IWT-
7.5-
164, an available installation space of 225° is
assumed for simplification. The yaw drives are located on the left, right and rear. This
would be specified in more detail as part of an overall wind turbine development.
The standards IEC 61400-
1 [3] and DNVGL-
ST-
0361 [1] specify the minimum
requirements for the design of wind turbines and yaw systems intended to ensure an
acceptable safety level. Any of the requirements of these standards may be altered if
it can be suitably demonstrated that the safety of the system is not compromised. The
example sizing of the yaw system, however, follows the minimum requirements of
IEC 61400-1.
4.4.2
Step 1: yaw system, holding torque and driving torque
As described in section 4.3.7 and shown in section 4.3.8, wind turbine manufac-
turers often specialise in one yaw system concept. A detailed selection of the yaw
system concept in this artificial example dimensioning process makes no sense and
is beyond the scope of this chapter. Therefore, it is determined that the yaw system
shall correspond to Concept 3 consisting of a double-
row four-
point contact ball
bearing, a hydraulic yaw brake system and electro-
mechanical yaw drives operated
with frequency converters.
In the following, the required minimum driving and holding torques are deter-
mined based on the given loads, which form the basis for the dimensioning of the
yaw brake system and the yaw drive system. According to the DNV standard [1],
the two brake systems (hydraulic yaw brakes and yaw drives) shall be designed in
such a way that each brake system is able to hold the nacelle in position during wind
turbine installation, service and maintenance. Therefore, a yaw locking device is not
required.
The relevant yaw moment Mz
values are summarised below, with
the first value given in Table 4.2:
• Max. yaw moment Mz with partial load factor: 20,120 kNm
• Max. yaw moment Mz without partial load factor: 17,440 kNm
• Max. yaw moment Mz in individual time series: 19,220 kNm
• Max. yaw moment Mz during installation/maintenance: 8,720 kNm
• Max. yaw moment Mz in LDD: 15,600 kNm
Since yaw slippage should be avoided in this example, the total holding torque
of the yaw system should be greater than or equal to 20,120 kNm. If yaw slippage is](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-214-320.jpg)
![186 Wind turbine system design
allowed, the minimum total holding torque should be carefully determined, and the
yaw slippage events should be simulated and analysed in detail (see section 4.2.6).
It should be noted that the values in the extreme load tables are mean values of
the maximum values from a set of time series for a specific DLC. This means that
in the individual time series values can occur that are higher than the maximum yaw
moment Mz
without partial load factor (see above).
For maintenance purposes, the yaw brake system and the yaw drives must gener-
ate a minimum holding torque of 8,720 kNm. If the hydraulic brake system has leak-
age or is maintained and thus pressure-
less, the yaw drives hold the nacelle safely in
position. It is very unlikely that all yaw drives fail at the same time. However, should
this be the case, the hydraulic yaw brake system keeps the nacelle in its position.
During turbine installation, it needs to be checked whether and when the hydrau-
lic yaw brake system can provide its holding torque. In some cases, it is possible to
generate the required hydraulic pressure with a manual pump.
Permissible wind speeds for wind turbine installation as well as service and
maintenance activities are usually specified considering the requirements from IEC
[3] and DNV [1]. In principle there are two possibilities. Either the allowable wind
speeds are specified, and the maintenance loads are simulated for these wind speeds,
or the permissible wind speeds are determined on the basis of the available holding
torques of the yaw system.
Another criterion could be how often the holding torque of the hydraulic yaw
brake system is exceeded in the fatigue DLCs. In this example, this is less than 20
hours in 20 years (Figures 4.14 and 4.15), which is considered acceptable.
Therequiredminimumholdingtorquesaresummarisedbelow.Itisobviousthatthetwo
holdingsystemsneedtogeneratehigherholdingtorquestoachievethetotalholdingtorque.
• Total holding torque: ≥ 20,120 kNm
• Holding torque of yaw brake system: ≥ 8,720
• Holding torque of yaw drives: ≥ 8,720
The determination of the partial brake pressure during yawing is always a com-
promise. The higher the partial brake pressure, the higher the damping of the exter-
nal yaw moment Mz
, but also the higher the load on yaw drive and yaw bearing teeth.
In this example, a constant partial brake pressure is assumed. It is often in the
range between 10% and 25% of the full brake pressure during non-yawing operation.
This means that in this example the yaw brake friction torque would be
at least 872–2,180 kNm. Figure 4.15 shows that the yaw brake friction torque
would then be higher than the external yaw moment Mz
between about 50% and
85% of the time. In this example, 10% is set as the starting point. The actual yaw
brake friction depends on the yaw brake used and is determined in section 4.4.2.
• Partial yaw brake friction: ≥ 872 kNm](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-215-320.jpg)
![188 Wind turbine system design
With the specification of the holding and driving torques to be achieved, the
basic requirements for the yaw system are defined. The subsystems and components
can now be dimensioned (see following subsections). In section 4.4.8, it is finally
checked whether the yaw system meets the requirements (Table 4.13).
4.4.3
Step 2a: yaw bearing, yaw brake and yaw drive
Due to the interdependencies, yaw bearing, yaw brake system and yaw drive system
are usually dimensioned in parallel. Depending on the boundary conditions, a differ-
ent starting point can be useful. In this dimensioning example, the yaw brake system
seems to dimension the size of the yaw system. Hence, the design is started with this
system. Before that, however, the components used are presented.
As described in section 4.3.3, yaw brakes from different suppliers have stan-
dardised connecting dimensions. In multimegawatt wind turbines, two different
brake sizes are common:
•
• a small brake with two pistons with a diameter of 90 mm
•
• a large yaw brake with three pistons with a diameter of 120 mm
The larger one shall be used for the IWT-
7.5-
164. With the large yaw brake, a
higher brake torque can be generated compared to the small one on the same diam-
eter. This allows to keep the yaw system diameter small. The main data of the yaw
brake used in this dimensioning example are summarised in Table 4.14.
There are no standardised yaw gearboxes. However, certain gearboxes are used
more often than others by several wind turbine OEMs. To benefit from the cost
advantages, it is worth checking whether existing gearboxes that are produced in
large quantities can be used. In this dimensioning example, two different yaw gear-
box sizes are considered to show the influence on the yaw system design. The main
data are shown in Table 4.15.
The asynchronous motors are standardised according to the IEC 60034 series.
Suitable yaw motors are selected so that the static load capacity of the yaw gearbox
and yaw bearing teeth are not exceeded. The maximum motor torque is limited to
a maximum of 1.8 times the nominal torque by a frequency converter. The yaw
bearing teeth usually limit the maximum allowable motor torque, which is already
considered here. The motor data are summarised in Table 4.16.
Table 4.13 Yaw system – fatigue and static strength analysis
(Sub-)component Standards, guidelines, methods and criteria
General IEC 61400-
1 [3], DNVGL-
ST-
0361 [1]
Yaw system Holding torques (yaw slippage events), driving torque (overload
events, availability of yawing), yaw start and stop events
Components See sections 4.4.4–4.4.6](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-217-320.jpg)
![Yaw system concepts and designs 191
FB = FC 2 static or dynamic (4.4)
MB = a FB
Deff
2 (4.5)
• number of yaw brakes [–]
• number of pistons per calliper half [–]
[–]
• effective braking diameter [m]
• piston diameter [mm]
• braking force [kN]
• clamping force [kN]
• braking torque [kNm]
• hydraulic pressure [bar]
• coefficient of friction
DPiston
Deff
aPiston
FB
FC
a
MB
pholding or yawing
static or dynamic
First, the clamping force FC
is calculated from the total piston area of a calliper
half and the hydraulic pressure. Second, the braking force FB
results from the clamp-
ing force, the number of friction pairs and the coefficient of friction. Multiplying
the braking force by the number of yaw brakes and the effective braking diameter
finally gives the braking torque MB
. The manufacturer’s documents usually contain
a specification for determining the braking diameter Deff
depending on the brake cal-
liper design and inner diameter of the brake disc.
When arranging the yaw brakes care must be taken to ensure that there is
sufficient space between the yaw brakes. The space is needed for tools and equip-
ment during assembly and disassembly of the yaw brakes. Ideally, the brake cal-
liper halves can be pivoted via one of the outer fastening bolts which remains in
place as pivot. This enables an easy brake pad replacement. In addition, sufficient
space must be provided between the yaw brake and the yaw bearing, e.g., for
lubrication lines.
It often makes sense that both brake systems each provide about 50% of
the holding torque. For this example, this means that the yaw brakes must
generate around 10,000 kNm of holding torque. This requires 14 yaw brakes
and a brake disc inner diameter of 2,625 mm. The results are summarised in
Table 4.18.
The hydraulic pressure is subject to tolerances. This is considered in Tables 4.14
and 4.18. For yaw slippage events, the minimum braking torque during non-
yawing
operation is of interest. For yawing, the maximum braking torque defines the load
on yaw gearbox and yaw bearing teeth.
The yaw brakes are standardised components that have to fulfil the requirements
of the IEC or DNV standard (Table 4.19). The strength of the brake callipers is
verified by finite element method (FEM) calculations and bench tests. The strength
verification of the bolted connections is also done using FEM.
The friction coefficients as a function of pressure, temperature and sliding
speed as well as the wear and yaw squeaking behaviour are examined in bench
tests. The tightness of the seals is also validated in bench tests. To prove the
suitability for cold climate applications, the tests are also carried out in climate
chambers.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-220-320.jpg)
![192 Wind turbine system design
4.4.5
Step 2c: dimensioning of the yaw bearing
When designing a yaw bearing, experience from other wind turbines can be used.
Based on the loads and the available installation space, the yaw bearing can be
pre-
dimensioned with little effort. In the further course, the yaw bearing design is
detailed, and extensive calculations are carried out.
In the previous subsection, the yaw brake system was dimensioned. The size
of the yaw brake system defines the inner diameter of the yaw bearing in this
Table 4.18 Yaw brake system
Parameter Value
General
Number of yaw brakes 14
Brake disc inner diameter 2,625 mm
Effective braking diameter 2,727.5 mm
Clamping force FC
Nominal clamping force – holding 610.73 kN
Minimum clamping force – holding 603.94 kN
Nominal clamping force – yawing 61.07 kN
Maximum clamping force – yawing 78.04 kN
Clamping force – untwisting 0 kN
Braking force FB
Nominal braking force – holding 549.65 kN
Minimum braking force – holding 543.55 kN
Nominal braking force – yawing 61.07 kN
Maximum braking force – yawing 78.04 kN
Braking force – untwisting 0 kN
Braking torque MB
Nominal braking torque – holding 10,494.25 kNm
Minimum braking torque – holding 10,377.65 kNm
Nominal braking torque – yawing 1,166.03 kNm
Maximum braking torque – yawing 1,489.92 kNm
Braking torque – untwisting 0 kNm
Table 4.19 Yaw brake – fatigue and static strength analysis
(Sub-)component Standards, guidelines, methods and criteria
General IEC 61400-
1 [3], DNVGL-
ST-
0361 [1]
Brake callipers Finite element method, bench tests (extreme load
test, fatigue load test)
Brake pads Bench tests (coefficient of friction, wear behaviour
and squeaking behaviour), heating calculation
Seals Bench tests (tightness)
Bolted connections Finite element method, VDI 2230](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-221-320.jpg)
![194 Wind turbine system design
Yaw bearing weight and amount of grease are usually determined by the bearing
manufacturer. Here, an estimate is made based on the CAD model and experience.
In the past, yaw bearings were designed purely analytically for extreme loads. This
approach is no longer up to date. The deformation of the companion structure and in
particular changes in stiffness have a major influence on the rolling element forces and
contact angles as well as on bearing ring and bolt stresses (see below). An analytical
calculation should only be used as part of a pre-
dimensioning of the yaw bearing.
The following shows how the yaw bearing raceway can be verified analytically.
The extreme load case with the highest yaw bearing Mxy
is used (Table 4.2), as this
leads to the highest rolling element force.
First, the maximum ball force is calculated using (equation 4.6). The maximum
contact angle is usually assumed to be 60°. The influence of the companion structure
is considered by the excess load factor by which the maximum ball force is conser-
vatively increased. The results are shown in Table 4.21.
Q =
4 Mxy
Dpw 2 Z
+
ˇ
ˇFZ
ˇ
ˇ
2 Z sin ˛
+
4
ˇ
ˇFxy
ˇ
ˇ
2 Z cos ˛
!
K
(4.6)
• Dpw raceway diameter [m]
• Fxy yaw bearing Fxy [kN]
• Fz yaw bearing Fz [kN]
• K excess load factor
• Mxy yaw bearing Mxy [kNm]
• Q maximum ball force [kN]
• Z number of balls per raceway
• maximum contact angle [°]
[–]
[–]
Table 4.21 Yaw bearing raceway verification
Parameter Value
Loads
Yaw bearing Mxy
26,950 kNm
Yaw bearing Fxy
760 kNm
Yaw bearing Fz
4,560 kNm
Bearing and calculation data
Raceway diameter Dpw
3,212.5 mm
Ball diameter Dw
60 mm
Number of balls per row Z 152
Max. contact angle α 60°
Constant cp
934.45 (N/mm²)2/3
Excess load factor K 1.30
Calculation results
Maximum ball force Q 214.10 kN
Permissible surface pressure pperm
4,200 N/mm²
Maximum surface pressure p 3,648.14 N/mm²
Safety factor 1.15](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-223-320.jpg)
![Yaw system concepts and designs 195
Then, the contact stress according to Hertz can be calculated using (equation 4.7).
The required constant cp
results from the radii and the material properties of the roll-
ing element and raceway. A calculation is not discussed in detail here.
p = cp 3
s
Q
D2
w
(4.7)
• cp constant [(N/mm²)2/3
]
• Dw ball diameter [mm]
• p contact stress [N/mm²]
• Q maximum ball force [N]
The maximum permissible contact stress is usually defined by the bearing
manufacturer considering material, surface hardness and hardness depth. Often, the
permissible surface pressure of 4,200 N/mm² from ISO 76 is used. The static safety
factor is the ratio between the permissible and actual contact stress. According to
the DNV standard [1], it shall be at least 1.1. The IEC standard [3] allows a safety
factor of 1.0.
The results are summarised in Table 4.21. The static safety factor is 1.15, which
should be sufficient. However, the finite element calculations considering the com-
panion structure need to confirm this.
The fatigue and static strength analysis according to the IEC or DNV stan-
dard is carried out using finite element calculations, considering the influence of
the companion structure (Table 4.22). The focus is on the raceways and the bolted
connections.
The structural integrity of the bearing rings should be verified as well. For a
sufficiently high number of unit load cases, the ring stresses are determined using
Table 4.22 Yaw bearing – fatigue and static strength analysis
(Sub-)component Standards, guidelines, methods and criteria
General IEC 61400-
1 [3], DNVGL-
ST-
0361 [1]
Bearing rings Finite element method (ring stresses and seal gap
widening)
Raceways Finite element method (rolling element forces,
contact angles, Hertzian pressure, pressure ellipse
truncation, plastic deformation, stresses in raceway
edge, core crushing and lifetime)
Gears ISO 6336 series (surface durability and tooth bending
strength)
Bolted connections Finite element method, VDI 2230, IEC 61400-
1 [3],
DNVGL-ST-0361 [1]](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-224-320.jpg)
![Yaw system concepts and designs 197
A truncation of the pressure ellipses at the raceway edge can also occur. This
means that the pressure ellipse cannot form completely (Figure 4.36). As a result,
there can be higher surface pressures and high stresses in the edge of the raceway.
This can result in plastic deformation and chipping of the raceway edge.
4.4.6
Step 2d: dimensioning of the yaw drive system
After the yaw brake system and yaw bearing have been designed, the required num-
ber of yaw drives can be determined. In principle, there are three criteria that must
be met:
•
• Holding torque: The yaw drives must provide sufficient holding torque so that
the requirements from section 4.4.2 are met.
•
• Driving torque: The yaw drives must provide sufficient driving torque so that
the requirement from section 4.4.2 is met. The required driving torque should
be redefined based on the actual yaw brake friction during yawing.
•
• Yaw bearing gear strength verification: The yaw bearing is usually the weakest
part. The minimum safety factors according to the IEC standard [3] must be
fulfilled.
The required yaw system parameters are summarised in Table 4.23. It also
shows that the nominal yaw speed is in the desired range stated in Table 4.12.
The difference between the required total holding torque and the holding
torque of the yaw brakes results in the required holding torque of the yaw drives
(Table 4.24). The holding torque per yaw drive is given by the multiplication of
motor brake torque, total ratio and static efficiency. With these two values, the
required number of drives can be easily determined.
The yaw brake friction torque during yawing is given in Table 4.18. The yaw
bearing friction depends on the loads acting on the yaw bearing. It can be estimated
using the equations from the bearing manufacturers. For reasons of simplification, a
Figure 4.36 Pressure ellipse truncation](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-226-320.jpg)
![Yaw system concepts and designs 201
If a load level of the load spectrum were to exceed the truncation torque, the
minimum safety factor for the fatigue strength analysis would be below 1.10. In
20 years, the truncation torque is only exceeded for 8.52 hours, which is 0.05%
of the time.
It must be ensured that the maximum driving torque per yaw drive (Table 4.26) is
smaller than the truncation torque (Table 4.27). A sufficient safety distance between
the torques depending on the yaw system concept and the yaw control system should
be considered. Overload events need to be detected and overloads in the yaw drive-
train need to be avoided. In this example, this is ensured via the frequency converters
and other sensors that measure the motor speeds, for example.
The arrangement of the yaw drives has an influence on the number of load cycles
of the most heavily stressed yaw bearing tooth. In the first calculation, all drives are
arranged close together. This is the most conservative variant. In the further course,
the arrangement is detailed considering the available installation space. The individ-
ual optimisation steps are not shown here. Only the results of the final arrangement
are shown in Table 4.27.
The fatigue strength analysis confirms the number of yaw drives. Fourteen small
or eight large yaw drives are needed to fulfil the requirements. There are higher safety
factors for the smaller drives.
The yaw system configurations seem to be well balanced. All criteria lead to
almost the same number of yaw drives. A higher partial brake pressure would lead
to higher loads and thus to a higher number of yaw drives in this criterion and in the
driving torque criterion.
The static strength analysis is done with the maximum holding torque per yaw
drive. However, the upper tolerance of the motor brake torque of 20% is consid-
ered. Since the motor brake was chosen to match the yaw gearbox and its pinion, the
required safety factors are achieved.
The fatigue and static strength analysis of the yaw gearbox is done according
to the IEC or DNV standard considering the calculation standards for the different
machine elements (Table 4.28). Structural components such as housing and planet
carrier are verified via finite element calculations.
Table 4.28 Yaw gearbox – fatigue and static strength analysis
(Sub-)Component Standards, guidelines, methods and criteria
General IEC 61400-
1 [3], DNVGL-
ST-
0361 [1], bench tests (extreme
load test, fatigue load test, measurement of backlash and
torsional stiffness)
Gears ISO 6336 series (surface durability and tooth bending strength)
Bearings ISO 76, ISO 281
Shafts DIN 743
Shaft-hub connections DIN 5466, DIN 6892, DIN 7190
Bolted connections VDI 2230, finite element method
Planet carriers, housing Finite element method, IEC 61400-
1 [3], DNVGL-
ST-
0361 [1]](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-230-320.jpg)
![202 Wind turbine system design
In addition, bench tests are often carried out to validate the fatigue and static load
capacity of the gearbox. Backlash and torsional stiffness are also measured.
Asynchronous motors and their motor shafts are standardised. So, there is no
need for a separate fatigue and static strength analysis (Table 4.29). Depending on
the load characteristic and the required operating hours, however, it can make sense
to have closer look at the bearing and the shaft-
hub connection. Torque-
speed curve,
current-
speed curve and the other motor data according to the IEC 60034 series are
determined in bench tests.
4.4.7
Step 3: auxiliary systems
The design and dimensioning of the auxiliary systems are not discussed in detail
here. The power supply is assumed to be given. The hydraulic system and the lubri-
cation system for the example yaw system are described in Chapter 7 of this book.
In the following, the sensors are briefly discussed in general.
In section 4.3.6, possible sensors are presented. The following signals shall be
measured in the example yaw system:
•
• yaw position and yaw speed
•
• zero-
degree position (north position) and end position of the cable loop
•
• current and voltage via the frequency converter
•
• motor winding temperature
•
• hydraulic pressure at the yaw brakes
•
• status information of the lubrication and hydraulic system
These signals ensure the proper function of the yaw system. In addition, over-
load events and failures can be detected.
4.4.8
Summary
The yaw system of the IWT-
7.5-
164 wind turbine is dimensioned in the previous
subsections. Some design and calculation aspects are discussed in more detail.
However, a complete and detailed design is not possible within the scope of this
book. The example dimensioning is therefore to be understood as a first insight
Table 4.29 Yaw motor – fatigue and static strength analysis
(Sub-)component Standards, guidelines, methods and criteria
General IEC 61400-
1 [3], DNVGL-
ST-
0361 [1]
Motor IEC 60034 series, bench tests
Bearings ISO 281, ISO 76
Bolted connections VDI 2230
Shaft-hub connections DIN 6892](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-231-320.jpg)
![Yaw system concepts and designs 205
References
[1] Machinery for wind turbines. DNV GL standard DNVGL-
ST-
0361, DNV GL
AS; 2016 Sep.
[2] Jenkins N., Burton T., Bossanyi E., Sharpe D., Graham M. Wind Energy
Handbook. Third edition. Hoboken, NJ: Wiley; 2021. Available from https://
onlinelibrary.wiley.com/doi/book/10.1002/9781119451143
[3] Wind energy generation systems – part 1: design requirements. Geneva,
Switzerland: IEC standard 61400-
1:2019-
02, International Electrotechnical
Commission; 2018.
[4] Load and site conditions for wind turbines. DNV GL standard DNVGL-
ST-
0437, DNV GL AS; 2016 Nov.
Table 4.30 Yaw system – summary
Parameter Value small drive Value large drive
General
Number of yaw drives 14 8
Number of yaw brakes 14 14
Total ratio 18,556.36 15,495.86
Required yaw speed 0.25–0.50°/s 0.25–0.50°/s
Nominal yaw speed 0.050 rpm
0.301°/s
0.061 rpm
0.368°/s
Dynamic efficiency 0.859 0.859
Static efficiency 1.000 1.000
Centre distance 1,831.5143 mm 1,889.2759 mm
Max. machine frame width 4.2 m 4.2 m
Actual machine frame width Approx. 4.15 m Approx. 4.15 m
Holding torque
Required total holding torque ≥20,120 kNm ≥20,120 kNm
Total holding torque 20,885.81 kNm 20,411.60 kNm
Required partial holding torque ≥8,720 kNm ≥8,720 kNm
Holding torque of yaw brakes 10,494.25 kNm 10,494.25 kNm
Holding torque of yaw drives 10,391.56 kNm 9,917.35 kNm
Driving torque
Required max. driving torque ≥7,560 kNm ≥7,560 kNm
Maximum driving torque 7,585.05 kNm 7,664.73 kNm
Exceedance of maximum torque 0.17% 0.17%
Rated driving torque 4,237.79 kNm 4,280.28 kNm
Exceedance of rated torque 9.97% 9.97%
Inertia
Inertia of tower head (rotor + nacelle) 5.531E+07 kgm² 5.531E+07 kgm²
Inertia of yaw motors 4.773E+07 kgm² 5.859E+07 kgm²
Ratio yaw motors/tower head 0.86 1.06](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-234-320.jpg)
![206 Wind turbine system design
[5] EnBW Energie Baden-
Württemberg A.G. lernt schwimmen in der ostsee [online].
2020. Available from https://www.enbw.com/unternehmen/presse/ windkraftan-
lage-nezzy-lernt-schwimmen-in-der-ostsee.html [Accessed 2 Apr 2022].
[6] Stubkier S. ‘Hydraulic Soft Yaw System for Multi MW Wind Turbines’. [Ph.D.
dissertation]. Aalborg University, Denmark, Institute of Energy Technology.
[7] Fraunhofer Institute for Wind Energy Systems. IWES wind turbine IWT-
7.5-
164
rev. 2.5. Bremerhaven: Fraunhofer Institute for Wind Energy Systems; 2017.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-235-320.jpg)
![210 Wind turbine system design
g. Depending on the concept, the electrically or hydraulically actuated rotor brake
can be installed on the low-
speed or high-
speed side of the DT. It is used to
brake the rotor from low speed or idle operation, e.g., in the event of a grid fault,
until standstill. Once stationary, it also acts as a parking brake—keep atten-
tion—this is not equivalent to a rotor-
lock system described above.
h. The gearbox normally steps up the rotational speed from the wind rotor and
transmits torque to one or splits between more generators (refer to Chapter 6),
and therefore is usually applied one to four gear stages with a reaction force
support at the housing (Figure 5.1). Depending on the concept, the gearbox has
a flange connection or torque arms to handle reaction forces. The gearbox in a
DT is optional (Figure 5.2).
Usually, the whole DT of WT is tilted upward in the rotor direction under an
angle of 4–6°, to serve enough space between blades and tower under operation. All
DT schematics shown in section 5.2 have a DT tilt of 5°.
5.2 Drivetrain concepts
5.2.1
Drivetrain diversification and classification
The simplest form of characterization or differentiation of WT DTs is that in gearless
and geared DT concepts, regardless of how many gear stages are used. Concepts with
continuously variable-
ratio transmission were developed and built as prototypes or in
smaller series but have not been established on the market until now. These concepts
combine a mechanical gear transmission solution with a hydraulic (the DeWind D8.2
WT from 2006 was equipped with such a hydrodynamic WinDriveTM
[1] transmis-
sion) or electrical actuator for setting the continuously variable ratio (Co. SET, refer
to Chapter 6). Since such solutions are no longer offered on the turbine market today,
Figure 5.2
Schematic of a WT nacelle with gearless, direct drive from the hub
with pitch system to generator, © Fraunhofer-
IWES](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-239-320.jpg)
![Drivetrain concepts and developments 211
detailed explanations are not given here. The interested reader is referred to specific
literature and company publications. Another special form of variable transmission by
means of the purely hydraulic solution using a closed-
circuit hydraulic system with
a variable displacement pump/motor combination is also not further considered, too.
However, these designs did not get beyond a demonstrator status.
Early WTs had rather “classic” DT concepts with three-
stage gearboxes to step
up the slow rotational speeds of the wind rotor to rated output speeds in the range of
1000–1800 rpm, corresponding to the main grid frequency of 50 or 60 Hz, for direct
grid coupled standard generators, either induction generators (IGs) or synchronous
generators (SGs) with fixed four, six, or eight poles or pole-
changing designs. Those
turbines were very limited regarding rotor speed variations and belong to the class of
stall-
controlled, so-
called fixed-
speed WTs. In these early years of commercial wind
energy utilization (1980s) with less turbine capacity, this type of DT had advantages;
the main reason for combining multi-
stage gearboxes and so-
called high-
speed gen-
erators was the availability of off-
the-
shelf industrial products for such components.
Since the 1990s, more and more variable-
speed WTs using full or partial
power converters for grid coupling and active pitch control became a standard.
With the followed, fast-
growing size (rotor diameter and rated power) of turbines
more and more frequently failures especially within suspension system and the
gearboxes cropped up and revealed the necessity of more detailed DT analysis
due to the complexity of site conditions, loads, and entire system dynamics [2]. In
contrast to the beginnings of the DT designs in the 1980s, for over 30 years, more
and more high-
fidelity models (FE Finite Element, Multi Body Simulation incl.
flex-
body parts, and multi-
domain/physics models) have been used, which now
include nearly the entire mechanical DT properties down to the rolling contact of
the bearings, tooth contact of the main gears, and non-
linear, distributed stiffness
and damping details of individual modules. However, this does not mean that
the usage of such tools automatically leads to reliable and more efficient designs.
Broad experience and comprehensive technical understanding remain indispens-
able and key elements for this. But these sophisticated models are nowadays
standardly applied in the design process for WT DTs, for load assessment and
strength verification under stationary, transient, and long-
term dynamic operating
states [so-
called DLC, design load cases, refer to Chapters 1 and 2, (IEC 61400-
1,
GL Rules and Guidelines Part 1 Wind turbines; 2010 and Part 2 Offshore Wind
Turbines; 2012, DNV GL loads and site conditions 2016 technical guidelines)], as
well as for controller synthesis, structural-
borne noise assessment, and optimiza-
tion in tailored variants.
In the first time of extensive multi-
megawatt turbine development, since the
late 1990s, the Original Equipment Manufacturers (OEMs) as well as the suppliers
react with specialized component designs for single types of turbines. So, in general,
the failure rates of the mechanical DT system decreased continuously with outliers
upward due to the extensive growth of the new WT sizes, introduction of new DT
concepts, and still lacking knowledge or underestimated parasitic effects. Increasing
rotor diameter causes inevitably exponentially growth of mechanical loading by
forces, bending moments, and torques applied on the DT in general, as well as the](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-240-320.jpg)
![212 Wind turbine system design
resulting deflections and displacements. The even higher loading on the mechanical
DT components required even more innovative and reliable concepts and compo-
nents within even shorter design periods.
The development of gearless DTs for WTs started in the early 1990s. The German
manufacturer ENERCON designs Direct-
Drive (DD) WTs, using multipole electrical
excited synchronous generators. So, the first E30/E40 turbines marked a historical point
from that point of view. Although these early gearless DTs were typically heavier than
those with a classical geared DT, they seemed to have superior reliability at that time.
The last stage of DD technology nowadays uses permanent magnet excited generators
[e.g., OEMs Siemens Gamesa Renewable Energy (SGRE), Goldwind, Lagerwey, and
Vensys], originally introduced by the inventor and wind pioneer Prof. Klinger in early
1997 with the Genesys 600-
kW turbine and developed further on to a first-
megawatt
series turbine by the company Vensys Energy. Later on, this IP serves the Chinese
OEM Goldwind as a technical basis.
This general differentiation leads to the simple DT classification mentioned
above, which, however, is less suitable for directly comparing the various designs.
Here a smarter differentiation seems helpful in order to correctly compare the applied
DT concepts of the OEMs and the technical details. Of course, finally, regardless of
the respective DT design, it is the entire turbine design, which counts as a whole,
e.g., in terms of tower head mass, serviceability, or technical reliability. In order to
describe different types of DTs and to compare their properties, the following clas-
sification, according to four main characteristics, can be useful.
1. gear transmission (corresponds to gear stages/ratio in general)
• No gear stage (i=1:1) → gearless DT/referred to as DD
•
one to two gear stages
(i=1:9–50)
→ medium-
speed DT/referred to as Hybrid-
Drive
Remark:
Latest OWT DT designs with three gear stages (all planetary, i=50–100) sometime
still referred to as Hybrid-
Drive
•
three to four gear stages
(i=1:60–150)
→ high-
speed DT / “classic” Geared-
Drive (GD)
• Variable-
ratio gear stage → variable-
ratio DT (eg, “WinDriveTM
” from
Voith)
2. Generator type (corresponds to type of electrical DT):
• IG + full rated converter
• doubly-
fed IG (DFIG)+ partially rated converter
•
synchronous generator build as permanent magnet excited (PMSG) or as electrical excited
(EESG) type + full rated converter
3. Rotor/-Main shaft suspension
• three-
point suspension → one main bearing at the front (upwind side),
second support bearing in the gearbox
• four-
point/dual suspension → classic dual bearing support, two
separated bearings
• (dedicated) bearing unit → dual bearing support, two bearings with
one shell
• single bearing → only one bearing support, sometimes
referred to as “Moment bearing”](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-241-320.jpg)
![Drivetrain concepts and developments 213
4. Level of integration:
• non-
integrated → all components separated, no functional
dual use of parts
• low-
integrated → at least one functional dual use of DT
parts
• semi-
integrated → multiple functional use of DT parts →
high compactness of the DT
• fully integrated → highest integration, hardly no single
functional use of DT parts
As already mentioned, roughly since the early 2000s, theWTmanufacturers have
entered the multi-
megawatt rated power class, but that does not mean that a favored
DT concept has emerged so far. But on the contrary, until mid of the 2010s, the num-
ber of DT concepts and variants further increased dramatically, which resulted in a
high diversification (refer to Figures 5.3–5.5). During that time, manufacturers have
mainly defined themselves and their innovative capacity by very special concepts.
The lot sizes of non-
variable parts were correspondingly low, sometimes with a
negative impact on the turbine costs and reliability. Not much later, since the mid of
the 2010s, there was a clear trend appearing, especially on the onshore market that
the WT became more of a commodity product with a rather low entry level, in terms
of risk, price, and technology. In consequence, even more players were entering
the market (e.g., Chinese OEMs were catching up). Various, quite PR-
driven terms
(e.g., “HybridDriveTM
,” “Compact Drive,” “PureTorqueR
) made the rounds, and for
the last time, some OEMs wanted to set themselves apart from one another in this
phase with explicit unique designs. Nowadays, standardization, modularization, and
platform strategies are common wordings also in the wind industry, whereas these
were frowned at OEMs in the former times and were sometimes ridiculed, when it
was still being discussed in the early 2010s. Thus, a design classification based on at
least some design features promise to bring more order to this diversity and can help
to compare various DT designs in a better way.
Figure 5.3 Diversification of WT DT concepts [3]](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-242-320.jpg)
![214 Wind turbine system design
So, the main reason to shift from more technology to a consistently
economical-
optimized-
based view was simple costs and the competition for mar-
ket shares. Within a MAKE Consulting report of 2011 (Figure 5.3), they called
it, “The need to push wind power towards grid parity while coping with the size
and cost increases associated with larger turbines is the core impetus for the diver-
sity of drivetrain choices available today.” On the other side, a strong technologi-
cal driver was the upcoming offshore wind applications. As an example, some of
the OEMs and system suppliers sought to combine the predicted advantages of
DD (higher overall reliability and high DT compactness) and “classic” geared
high-
speed DTs (track record, availability and experience of existing transmission
suppliers, and lower component weight) by developing medium-
speed gearbox
DTs, respectively, sometimes referred to as Hybrid-
Drives. These Hybrid-
Drive
designs should end up to reliable, lightweight, and compact DTs without using the
third (high-
speed) gear stage, typically a spur gear stage, of classical WT gear-
boxes, which is presumed to as the most error-
prone gear stage. At the electrical
part of these DTs, mid-
size, medium-
speed, reliable synchronous generators are
applied. In case of utilizing PM technology, this means the same performance
Figure 5.4
Combinations of main DT classes and different electrical subsystems
[5]](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-243-320.jpg)
![Drivetrain concepts and developments 215
Figure 5.5 Technical combinations of different key elements for DT concepts [4]](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-244-320.jpg)
![Drivetrain concepts and developments 219
There are two more design characteristics of this DT class that remain to be
discussed, which are the main shaft, respectively, rotor suspension system, and the
level of integration. The non-
integrated GD DT concept utilizes a dual-
bearing sup-
port, sometimes referred to as a four-
point suspension system (Figure 5.7), dedi-
cated to the rotor main shaft support. Typically, the design philosophy is to realize
a common and robust suspension system with a high level of stiffness for all non-
torque loads within a statically determined support configuration. Mostly, the first
bearing (upwind bearing next to the rotor) is a fixed one and the rear one (down-
wind side) is a floating bearing, which allows axial clearance for thermal expansion
of components [6]. To handle the torque reaction moments, the gearbox is usually
equipped with two torque arms (only one torque arm can cause additional bending
under torque loading) and top mounts realized, solely to support itself with respect
to rotation about the longitudinal gearbox axis and not to carry the dead-
weight
forces. Additionally, the trunnion supports are designed slightly elastic (elastomer
supports or bushings, respectively) for vibration damping and to provide structural-
borne noise decoupling.
The lowest level (absence) of integration of this concept is quite obvious,
each component has a specific functionality, and dual functionalities are strictly
avoided. An advantage of this concept is the good serviceability due to the sepa-
rated structures (e.g., for a gearbox change). Disadvantages are the comparatively
large mechanically DT dimensions and thus passive and active weights due to large,
solidly built machine beds, main shaft, critical tolerance chains, etc. The generator
and the gearbox output shaft are connected by a special high-
speed coupling that,
seen from a pure mechanical aspect, shall compensate for a certain range of static
and dynamic misalignments without the formation of constraining forces. Since the
generator is mounted separately on flexible and vibration-
damping supports on a
machine support (can be designed in one or more parts), mechanical tolerances,
misalignments during DT assembly, and operating vibrations can be neglected to
a certain extent. Thus, constraining forces between components, due to misalign-
ments, should be mitigated.
In order to clarify the difference, a related “classic” DT concept is presented
next, which uses a different type of main shaft suspension, according to Figure 5.8.
In order to save dead-
weight and installation space, a separated second main shaft
bearing (on the downwind side) is not used here. The bearing of the first gear stage
(e.g., usually the bearing of the planet-
wheel carrier) is used for this instead. Thus, it
has a dual functionality in this configuration. The support of the first planetary stage
and the function of the second main shaft support point on the downwind side (refer
to four-
point suspension) lead to a characterization as a low-
integrated concept. The
main bearing on the upwind side is designed as a fixed bearing, and the gearbox
has two torque arms with trunnion bearings [6]. These perform a comparable func-
tion as with the four-
point suspension, besides carrying dead-
weight force here, too.
In consequence, this DT concept is more compact, which saves weight (partially
compensated by the necessarily reinforced gearbox housing and support) but has a
more complex dynamic, due to parasitic loads in the gearbox and possibly induced
deformations within the first gear stage. The type of generator connection toward](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-248-320.jpg)
![Drivetrain concepts and developments 223
misalignment-
compensating slow-
speed shaft coupling (pure torque transfer as
the main objective) requires a great effort due to design and manufacturing com-
plexity. Furthermore, additional axial space is needed for integration, according to
Figure 5.11, which means less compactness of the entire DT, but at least creates a
mechanical determinated system.
Some DTs for WTs still apply a very special form of double-
bearing suspen-
sion. A double-
bearing support on a so-
called “king-
pin” structure, which is hollow-
type like rotor main shafts. The function of the main shaft is partly taken over by
the hub structure and, as shown in Figure 5.12, by a slim, flexible main shaft, which
serves to transmit torque and, due to its flexible characteristics, to decouple parasitic
loads from the gearbox input side. The gearbox is attached separately to a support
structure within the nacelle. This special main shaft takes over the additional task
of a flexible coupling on the low-
speed shaft, due to the dual use of low-
integration
level results per definition [7, 8].
The next logical step toward higher integration is to replace the bearing unit or
the dual bearing king-
pin solution with a single-
bearing solution, often referred to as
moment bearing. The idea of a moment bearing was introduced by bearing suppli-
ers. With an additional larger diameter and reduced overall length, two tapered roller
bearings that are set against one another in an O-
configuration (also referred to as
back-
to-
back, B2B) are integrated into one bearing within the SKF solution, the so-
called “NautilusTM”
bearing. Other bearing manufacturers (e.g., Co. ThyssenKrupp
Rothe Erde) used an alternative design and also created a single-
bearing solution
Figure 5.12
Classic” geared, high-
speed, low-
integrated DT concept with
four-
point suspension realized by a dual bearing rotor hub as a
common shell on a hollow king-
pin structure and an inner, angular
flexible, “PureTorqueR”
shaft (GE Alstom-
Ecotècnia ECO 3.0 MW)](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-252-320.jpg)
![226 Wind turbine system design
These DT concepts, sometimes referred to as Hybrid-
Drives, were originally
introduced from gearbox system suppliers (Winergy HybridDriveTM
and Moventas
FusionDriveTM
) and OEMs to increase the level of integration (higher compactness)
of WT DTs, i.e., to decrease tower head mass, increase reliability, and thus lower
“Levelized Cost of Energy” (LCoE). Commonly gearbox and medium-
speed gen-
erators are rigidly linked at their housings by flange connection, so the gearbox
housing carrying the dead weight of the generator too (Figure 5.15). An addition-
ally higher integration level of gearbox and generator is rather rare but technically
possible e.g. by using the rigid bearing unit concept for the support of the planet
carrier of the first gear stage. Even a generator without own bearing support would
be technical possible (“fly wheel concept”), but mostly, we see still self-
supported
concepts for each component. In general, this higher the risk of thermal run away of
bearings and unwanted constraining forces due to critical tolerance with the double
fit of shaft and housing flange connections. These issues can be solved by the use
of flexible shaft coupling, flexible bushings for flange connection, or at least a more
flexible bearing support for the generator shaft, which allows some angular mis-
alignments (“Spherical Roller Bearings,” SRBs). Other solutions could be designed
with an overall lower bending stiffness of the shaft [e.g., face-
to-
face (F2F) arrange-
ment and single-
bearing solution] or with enough bearing clearance to ensure a
stable (from a thermal point of view), static determinated suspension in connection
with the gearbox. Almost the same issues occur at the linkage between the bearing
unit, respectively, single-
bearing support and the gearbox. So, in the past, OEMs
Figure 5.15
Low-
integrated, geared, medium-
speed concept with double-
bearing suspension (bearing unit), rigid slow-
speed shaft/gearbox
coupling, and flange-
connected gearbox and generator](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-255-320.jpg)
![Drivetrain concepts and developments 231
their mechanical construction, weight, and thus cost aspects. On the other hand, from
a mechanical point of view, for inner rotor concepts, there is no big difference between
these generator types. However, differences pop up in the technical details and con-
straints such as slip ring systems, restrictions of pole pitch, controllability, and power
density. Common arguments for the use of EESG are that it has one more degree of free-
dom for control, due to variable electrical excitation and lower material costs (no rare
earth material needed). PMSG utilize rather expensive rare earth material (price shock
for neodymium and dysprosium started 2011 [5]), as a rule of thumb 600–900 kg mag-
net material (NdFeB, with ~30–40 % share of rare earth materials) per installed MW
for typical DD with nominal rotational speeds from 7 to 11 rpm in the multi-
megawatt
power range. If we discuss the possible characteristic features of DTs utilizing DDs,
we are starting again with the classic four-
point suspension (double-
bearing) system
that can be applied here, too. A 4 MW OWT (Figure 5.20) from GE in 2011 designed
by ScanWind with a back-
pack fly-
wheel inner rotor PM generator concept uses this in
combination with a long main shaft, two separated main bearings, fixed on the up-
wind,
floating on the down-
wind side, torque arms, and trunnions supports at the generator
housing to handle torque reaction forces. Thus, the dead-
weight forces of the wind rotor
and the generator affect opposite sides of the shaft and machine bed. Though some
advantages, this concept with separated components could not survive on the market
and was also comparatively heavy with a top head mass of 280 tons (including genera-
tor own weight of 84 tons). Just to compare it, this weight is fairly equal to the two-
stage medium-
speed 10 MNm gearbox of the Adwen AD8-
180 OWT from 2015. The
medium-
speed (medium voltage) PM generator (8.6 MW@~350 rpm, rated) has a dead
weight of ~32 tons. Figure 5.20 (bottom) presents the equivalent DD drivetrain concept
to PureTorqueR
design for geared DT, with a slim and therefore flexible main shaft to
decouple rotor load-
induced deformations from the generator and thus its air gap. Also,
the corresponding geared DT variant, according to Figure 5.12, achieves that by means
of a non-
torque flexible main shaft and separated support structure of the gearbox.
When it comes to WT DDs, ENERCON was undoubtedly the pioneer in intro-
ducing this concept in the 1990s. These first DD concepts were designed with an
electrically excited synchronous generator. Sometimes, it was referred to as ring
generators because of its outward appearance, with an internal rotor salient pole
design, a common double-
bearing solution for the wind rotor hub on a “long” king-
pin structure, and the hub structure rigidly coupled to the rotor of the generator
(refer to Figure 5.22). The type of bearing support and the inner rotor concept make
it inherently difficult to access the rotor hub via the nacelle. Later concepts that
apply other main suspension concepts offer clear advantages in direct comparison
here. The design of this rotor suspension system can best be compared with a kind
of hub integrated bearing unit, already presented for the gearbox concepts, but less
compact, and the outer common bearing shell is part of the hub structure. With a
suitable choice of bearings, it offers high axial rigidity and low bearing clearance.
The weight forces from the wind rotor can be advantageously distributed and dis-
sipated over the multiple-
part king-
pin structure using a comparable small bear-
ing diameter. The Achilles heel and therefore disadvantage of many DD concepts
compared to geared transmission concepts is the large influence of loads and thus
structural deformations on the air-
gap dimensions and geometry of the generator](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-260-320.jpg)
![Drivetrain concepts and developments 235
generator bearing. With only a slightly larger overall length, this alternative concept
offers the advantage of a broader supplier base for more standardized large bearings.
In the past, there have certainly been supplied bottlenecks for highly specialized
moment bearings [5]. The new Enercon EP5 and the Lagerwey LP4 use that concept
with EESG and PMSG respectively, but both with inner rotor design.
To avoid load-
induced deformation of the generator air gap, OEM Alstom (now
GE) extended their PureTorqueR
philosophy from geared DTs (refer to Figure 5.12)
toward DDs within the “Haliade” OWT. This DT concept (refer to Figure 5.24)
combines the classic dual-
bearing rotor in-
hub suspension on a solid king-
pin struc-
ture with a specific flexible coupling adapter (angular flexible, torque rigid but with
damping elements) between the rotor hub and the generator rotor. The concept shall
decouple the non-
torque loads from the separately supported (with additional bear-
ings) PM generator. Thus, the air-
gap sensitivity issue was solved here in the classic
way (mechanical decoupling). But consequently, this leads to a higher number of
Figure 5.23
Fully integrated DD (outer PM rotor) with integrated single-
bearing suspension, king-
pin support structure (SG-
3.0–14.0 MW
DD, Co’s Goldwind, EWT, Lagerwey, Leitwind, XEMC/Darwind,
and STX) with an inherent very low wind rotor load induced air-
gap sensibility](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-264-320.jpg)
![Drivetrain concepts and developments 237
uses an outer rotor PM generator as the central component in the hub on a king-
pin
support structure (Figure 5.25). The common rotor/generator suspension is realized
by the classic form of a robust double-
bearing solution, which can be realized with
comparable small bearing sizes (diameter). Finally, the full integration level in this
case is achieved by the fact that the generator rotor and the rotor hub merge into one
functional unit. The disadvantage of the concept, which due to the principle also has
a very low air-
gap sensitivity, is the larger rotor hub diameter required to integrate
the pitch systems in the radial direction above to the generator and to enable service
activities at them. Despite some advantages, this concept has not yet been imple-
mented with one exception, IMPSA IWP 100.
5.3
General design rules and procedures
The design of modern WTs and thus the design of DTs are based on the fundamental
technical rules and standardized procedures. Since the 1990s at the latest,WTs and their
structural, mechanical, and electrical subsystems and components have been designed
in accordance with national and international standards (ISO, IEC, DIN, DKE, and
ANSI/AGMA) and technical guidelines issued by certification companies [e.g., DNV
(formerly DNV-
GL), TÜV, Bureau Veritas, DEWI-
OCC, Wind-
FGW]. The standardi-
zation by means of binding, harmonized, and generally accepted technical regulations
must be clearly distinguished from the process of certification (refer to Figure 5.26).
In principle, the idea of certification is an independent review by an accredited author-
ity with regard not only to compliance with the binding regulations but also to the
best state-
of-
the-
art experience (technical guidelines, e.g., DNV GL 2010, VDI, and
Wind-
FGW) in the design, development, production, and grid integration. It serves
to establish product and operational safety as well as an assessment of quality and
technical risks, to ensure transparency and trust among all stakeholders, besides other
Figure 5.26
Key elements of WT certification process (different tasks/
assessments)](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-266-320.jpg)
![238 Wind turbine system design
things by carrying out accredited procedures, e.g., for tests and measurements (refer
to Chapter 9).
Fundamental requirements for predefined service lifetime design of a WT/DT
are as follows:
•
• mitigate or avoid loads due to proper basic design, e.g., by lightweight construc-
tion or through compliance with general design rules,
•
• withstanding loads (external, internal, and environmental conditions), which
means sufficient reliability as well as robustness with a predefined level of safety,
•
• control or manage loads (mechanically, electrically, by controls) with passive
(e.g., twist-
bend coupling, flex-
pin gearbox, and torque limiters) or active (e.g.,
individual pitch control IPC, peak shaving, and active damping) approaches,
•
• mitigation of environmental impact (e.g., noise emissions, shadow effects, ice
drop, environmental pollution by leakage of operating fluids, and esthetics),
•
• acceptable serviceability (monitoring and diagnostics, maintenance, and repair)
due to the fact that technical systems need service and can fail in general, all in
respect of the boundary condition of general manufacturability and operation
at the lowest life cycle costs, grid compliance, and the highest possible, site-
specific energy yield.
5.3.1
Safety, protection, reliability and control
In discussions, the term safety sometimes remains unclear due to its ambiguity and
its use in the context of protection. The simple rule in this context is protection
serves the operational safety. Safety in terms of design characteristics (design safety
margins) serves the reliability.
In general, all required strength verifications within the design process need
extensive calculations, analyses, and simulations as well as some validations through
field measurements and tests (material, component, system, and field tests) (refer to
Figure 5.27, [9]). This is standard to ensure safety and reliability. For even higher
reliability, an overall product validation (refer to Chapter 9) with a test and valida-
tion concept that accompanies the development, already starting with the first devel-
opment phase, is strongly recommended. Within the last 10–15 years, a rethinking
and higher acceptance of development-
accompanying and -embedded test strate-
gies can be noticed in the wind industry in general. Perhaps, this was triggered by
the increasing number of technical problems in the intensive growing phase (e.g.,
damage of gearboxes, bearings, and generators) of modern multi-
megawatt WTs
from the late 1990s onward. The causes were diverse and often complex, but from a
retrospective, e.g., by proper environmental testing, test bench campaigns, and test-
based model validations; these problems most probably could have been avoided.
In general, experts from OEMs, suppliers, and science are still discussing whether
the series defects are mainly arisen by the reason of the overall system development
and integration, thus at the OEMs side or at the supplier’s side, due to quality issues
of the mostly very special and large components (large bearings, gears) that are pro-
duced in relatively small quantities.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-267-320.jpg)
![Drivetrain concepts and developments 239
From a more independent perspective, however, the reason can be partially
pinned down to the lack of technical exchange between the OEMS and suppli-
ers and sometimes poor technical component specifications and less experimental
validation, too. Another aspect that should not be underestimated is the still ongo-
ing market pressure regarding costs, which we already know from other industries
(“López-
Effekt” 1993) and that can cause massive product quality issues.
From an operational but also from a design (down to component level) related
point of view, the control and protection system of a WT is determinant for load
management (minimizing material stresses, [10]) on the one side and on the other
side crucial for the turbine protection against hazardous situations in case of system
failures or extreme external conditions. Here a clear distinction between both of
these has to be made. The control system shall operate the turbine in a controlled
manner within the permissible operating range, e.g., during all normal operating
maneuvers, during start-
up, ramp-
up, partial, full, and transient overload operation
as well as during controlled shutdown or power reduction mode. This includes clas-
sic continuous and discrete controls (closed loop control) and condition-
based oper-
ation mode transitions (state machine with state transitions); therefore, the turbine
Figure 5.27
General design process with strength verification and model
validation starting from entire turbine servo-
aero-
elastic concept
simulation for interface load assessment to component load
analysis and optimization](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-268-320.jpg)
![240 Wind turbine system design
control system has interfaces to the primary controls of internal component and
system controllers (pitch controller, main converter, pre-
heaters, etc.). Due to that,
the task of the classic turbine control system covers no dedicated protection and
therefore safety aspects; however, in terms of load reduction, it increases WT safety
in terms of reliability.
Furthermore, it is important to understand that the load analysis and also a vibra-
tion analysis [11] for the DT should only be carried out with the turbine controls
taken into consideration, because the influences on most of the loads are essential
(e.g., [10, 12–15]). The inherent vibration characteristics of single components are
important but should only be used for the preliminary design phase in that detached
way. Cross-
couplings between the mechanical DOF and force feedback effects of
connected components/interfaces should never be underestimated in WT design.
The control system of a turbine usually has feedback of the following internal sig-
nals, especially but not only from the DT:
•
• rotational rotor speed respectively generator speed
•
• wind speed and direction from the nacelle anemometer
•
• nacelle/tower vibration level
•
• temperatures (external, internal, oil, generator, converter, coolant, etc.)
•
• voltage, current, and frequency and connection status at PCC, output power
•
• cable twist status
•
• yaw position, respectively, yaw error
•
• further binary status information, e.g., about brake wear, from hydraulic auxil-
iaries, lubrication and cooling systems (refer to Chapters 7 and 8)
The control system uses the following main actuators to operate the turbine in
stationary and transient operation:
•
• yaw drive system (drives and hydraulic brakes, refer to Chapter 4)
•
• pitch system (single-
blade angle adjustment, refer to Chapter 3)
•
• activation of the mechanical brake system (on low- or high-
speed shaft)
•
• electrical grid connection (controlled on and off of the auxiliary and main
switches, refer to Vol. 2)
○
○ switched on—after synchronization
○
○ switched off—at loss of grid connection or permanent grid voltage or fre-
quency, outside the prescribed limits
•
• Active, reactive power and generator torque adjustment (refer to Vol. 2) by
means of the main converter (B2B configuration of the generator side and grid
side converter internally connected by a DC-
voltage link).
Typical time constants of the actuators or dynamic response times, respectively,
for the entire turbine and thus the DT are shown in Table 5.1. On the other hand, the
dedicated turbine protection systems take over, if the control system, critical sen-
sors, subsystems, or components fail or as a result of an event, so that the turbine and
thus the DT are no longer operating in a normal range. Once activated, the protection](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-269-320.jpg)
![Drivetrain concepts and developments 241
system can be based on aerodynamic, mechanical, or electrical principles to ter-
minate an abnormal operation or hazardous situation, transfer the turbine to a safe
condition, and maintain the system in this condition. Therefore, a protection system
consists of a detection, an activation, and an actuating unit to fulfill this task. A fur-
ther key requirement to the protection system is its fail-
safe characteristic (e.g., in
case of power supply/grid failures). Furthermore, according to the IEC61400 stan-
dard [16], the requirement is to bring the WT rotor to a full stop from any hazardous
operation. The control and the protection system shall be independent systems as far
as this is technically possible. The protection system shall run on a dedicated safety
PLC and use dedicated redundant sensors and/or signal lines; furthermore different
physical principles should be used to avoid systematic failures.
As a standard, a redundant braking system must be implemented (at least one
system on the low-
speed shaft); in modern multi-
megawatt WT, this is usually real-
ized (and generally accepted) by the three independent blade pitch actuator systems
with energy storage units. The aerodynamic break effect of at least one rotor blade
in complete feather position must be sufficient to bring a turbine in uncritical idle
speed. The optional requirement to bring the turbine and therefore the DT in full
stop condition can be managed by a mechanical service braking system on the main
or generator shaft (usually from max. 10–15% of rated speed). After full stop, the
mechanical service break continues to act as a parking brake. An absolute minimal
requirement is to bring the turbine into a safe and controlled idle mode in a worst-
case multi-
failure scenario. In general, the protection system must be designed to
override the control system in any case; its design shall according to the necessary
safety class (safety integrity level) result from a combined process of failure mode
assessment and failure criticality assessment.
The detected level for safety system activation has to be defined; that design
limits are not exceeded, which means such a situation has to be simulated and ana-
lyzed intensively during the design process, especially if they can cause ultimate
loads at the turbine and its DT components (e.g., torque reversal or overload within
gear stages during emergency stop or grid loss [15, 17–20]). Such scenarios will be
identified (if not already specified by standards, e.g., IEC) and checked within safety
analysis process [failure mode and effect analysis (FMEA), failure mode, effects,
and criticality analysis (FMECA), and fault tree analysis (FTA)] and optional within
turbine type certification again.
Table 5.1 Main actuator systems of a WT and the ranges of dynamic response
Actuator Response time/dynamic Remarks
Yaw drive max. 1°/s (dynamic) Discontinuous operation (dead band)
Pitch drive max. 8–10°/s (dynamic) Discontinuous, collective pitch control
Continuous individual pitch control
Generator
torque
10–50 ms (response) Air-
gap generator torque by generator side
main converter
Grid side
converter
2–20 ms (response) Active and reactive power by grid side main
converter](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-270-320.jpg)
![Drivetrain concepts and developments 243
FTA describes a probabilistic safety analysis, which is based on Boolean alge-
bra and is used to determine the probability of a plant or overall system failure. The
method is described by the International Electrotechnical Commission as the inter-
national standard IEC 61025 (EN 61025) under the term “fault state tree analyses.”
FMECA is a methodology, developed originally to change from an approach
of “find failure and fix it” (reactive approach) to “anticipate failure and prevent it”
(preventive approach). The methods developed focused on qualitative and quantita-
tive risk identification for preventing failures. Therefore, FMECA involves quantita-
tive failure analysis. Within FMECA, a series of linkages between potential failures
(failure modes), the impact on the operation (effects), and the causes of the failure
(causes and mechanisms) is generated. The methods and techniques associated with
the FMECA were published in a series of Military Standards. MIL-
STD-
1629A
is the most prominent of these standards and is still in use today. It is inductive or
data-
driven and linking elements of a failure chain as follows: “Effect of Failure—
Failure Mode—Causes/Mechanisms.” The FMECA shall be performed prior to any
failure actually occurring. It analyzes risks, which are measured by criticality (as a
combination of severity and probability), to take preventive action and thus provide
an opportunity to reduce the possibility of failure and its criticality.
FMEA and FMECA are closely related tools. Each tool resolves to identify
failure modes that may potentially cause product or process failure. The FMEA is
qualitative, just exploring “what-
if scenarios,” where FMECA includes a degree of
quantitative input taken from a source of known failure rates.
Reliability as a measure of failure probability can be explained as follows.
During the lifetime of a structure or mechanical component, these are subjected to
loads and environmental effects. In consequence, this leads to a change of health
state condition, which means lifetime consumption up to the point where deteriora-
tion, microstructural damage, corrosion, or wear out causes a failure. The reliability
of a part can be defined by the probability of the part reaching a limit state and thus,
real or per definition, entering a state of failure [21]. There are several different
types of limit states, but two types are very common, the ultimate limit state, which
directly corresponds to the limit of the load-
carrying capacity of the structure or the
part and is characterized by material failure and fracture or related failure modes
(e.g., extensive plastic yield, brittle fracture, instability, and buckling). On the other
hand, it is the fatigue limit state. Here the failure probability increases as a function
of time. A prediction of the failure probability as a function of time and method
of equivalent loads can be used to predict the time when the failure probability
will exceed (consumption of fatigue budget) a critical threshold (by definition, e.g.,
a maximum acceptable failure probability). Additional limit states can be defined,
e.g., the serviceability limit state, which can imply that there are deformations out
of tolerance without material failure (e.g., cracks, wear, corrosion, permanent defec-
tion, or vibration), so the further operation becomes unsafe. This corresponds to a
failure mode without critical consequences if detected, and the turbine will be pre-
ventively and permanently shut down. The structure, component, or system enter the
state “unreliable,” which may result in a defined action, e.g., shut down or inspection
with subsequently revaluation if applicable.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-272-320.jpg)
![244 Wind turbine system design
Typically, the design safety of mechanical components and thus DT components
will be determined as described above, based on the material properties. However,
as these components are parts of complex mechanical systems, there are additional
aspects of safety to be evaluated. Due to the complexity, a further range of possible
influencing factors for aging, wear, and component failure (temperature, humidity,
salt mist, sand, etc.) have to be taken into account. The resulting, more complex
(combined) failure modes and a nonlinear temporal progression of damage make
it even harder, sometimes not feasible, to find a probabilistic approach to assess
mechanical safety. Experience-
based, empirical methods or condition monitoring
based on measurements must be used here, in order to record the current status
and the progress of the damage to assess whether the defined limit states have been
reached.
5.3.2
Loads and load cases
As part of the design process for subsystems and components, load cases, which
the turbine and its DT are exposed during the anticipated service life and thus are
relevant for its strength and characteristically for the planned site class (e.g., IEC
wind class), have to be defined as well as their probable proportion of the designed
service life. As already explained, all components have to withstand these loads
with a sufficient safety margin, defined by partial safety factors. Design load cases,
as already explained in Chapter 1, are predefined design situations under various
expected operations, typical WT maneuvers, and external conditions, mainly wind
events. The IEC 61400-
1 standard [16] gives a guideline and lists the design relevant
load cases with a description of condition, turbine maneuver, and other parameters.
The previously explained FMECA is a useful tool to decide which load cases are
design relevant for the turbine, to determine sectional strains for individual systems
and components, e.g., main shaft hub flange, main shaft suspension, gearbox, gen-
erator, of the entire DT.
It should be pointed out here that this identification of sectional strains for the
definition of load input functions especially for the DT should only be a first step in
the design and analysis. In fact, the state-
of-
the-
art tools for load simulation and thus
load calculation for the entire WTs are already quite powerful. Whereby for WT
sub-
systems like the DT, the level of detail for the models is often limited (e.g., DT
simplified as a two-
mass oscillator). These limitations affect the number of mechani-
cal DOF as well as the modeling depth (rigid, flex-
body, friction, and damping) and
thus the consideration of linear and nonlinear deformation (nonlinear stiffness, back-
lash, bearing clearance, point-
line contacts, position-
dependent parameter changes,
cross-
couplings, and micro movements). Conducting of detailed sub-
system simu-
lations (e.g., MBS of the DT, refer to Chapter 2) just using pre-
calculated (by the
use of simplified sub-
system models within a WT simulation [22]) sectional forces
and moments at system interfaces, e.g., on the main shaft flange, lead to systematic
errors. But it is still quite common for the previously calculated section loads to
be used for the detailed design, for example, for highly detailed FE-
calculations of
components.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-273-320.jpg)
![246 Wind turbine system design
Wind conditions can be roughly divided into normal and extreme wind con-
ditions and thus inflow condition for the WT. The condition and the assumption
if it is a normal (mean frequently used) operation mode or a temporary (may be
very rare or unique) situation lead to a specific kind of analysis, either fatigue or
ultimate strength related (refer to Chapter 1). As a rule of thumb, if the specific
turbine (means hardware and software of the turbine) operation situation can be
either characterized by a fault, respectively, malfunction, or the external conditions
(mostly wind) are extreme, then the situation belongs to the class of ultimate load
cases; otherwise, it is a fatigue load case. In general, the resulting DT load are as
follows:
•
• aerodynamic-
induced input loads, means 6-
DOF rotor, respectively, rotor hub
flange loading (torque, thrust, radial forces, plane bending moment); for more
details, refer to Chapter 2
•
• grid fault-
induced input loads, due to grid events dynamically coupled to the
generator air-
gap torque (torque peaks, reversals, oscillations, [17, 18 and 15])
Loads, here referred to as aerodynamic induced loads, are caused by aerody-
namic forces from the dynamic 3D-
wind field, structural properties causing aero-
elastic interaction, as well as by the blade-
pitch control concept for power control,
load reduction, and balancing. The aerodynamic forces and aero-
elastic modeling
with a detailed explanation of these loads are not intended to be the subject of this
book, but for some more details, refer to Chapter 2 [23–27]. Load case simulation
with the entire turbine model using commercialist tools, e.g., Bladed, Flex5, FAST,
as already described, will provide coupled aerodynamic load input data for the rotor
flange as time series, distributions sometimes referred to as load level distribution,
and extreme values. Besides that, dead-
weight forces, icing, wave (site specific),
internal excitation, or reaction loads are always present, and DT components are
exposed to them indirectly or directly. Briefly other static and dynamic loads result
from:
•
• gravity loads (rotor weight, in general components dead weight)
•
• centrifugal forces and Coriolis forces, forces of inertia (due to rotation and
unbalance of mass)
•
• gyroscopic forces (due to yawing)
•
• external excitation reaction forces (structural deformation or oscillation of the
bedplate and support structure)
•
• internal excitation forces (can cause oscillations due to generator cogging, uni-
lateral magnetic pull forces due to eccentricity or axial not centered generator
rotor, gear meshing, misalignments, imbalances, cross couplings, e.g., due to
kinematics of flexible shaft couplings, gear reaction forces)
One example of additional dynamic loading is strongly increasing gyroscopic
force effects for flexible main shaft or rotor shaft supports. In particular, gyroscopic
forces induced from the rotor mostly occur during the nacelle yawing operation](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-275-320.jpg)
![Drivetrain concepts and developments 247
(refer to Chapter 4). This happens in any case regardless of the level of flexibility of
rotor/main shaft suspension and leads to an additional yawing moment (only non-
three bladed rotors) about the vertical axis of the rotor plane and to a tilting moment
about its horizontal axis. Besides the rotor speed and inertia, the angular yaw velocity
affects this proportion. Therefore, yaw speed is typically limited, and strong acceler-
ation and deceleration (e.g., due to yaw drive jerks and friction slip-
stick effects) are
avoided by yaw drive design and control for modern multi-
megawatt WTs. Adding
then flexibility to the main shaft support can cause significant forces and oscillations
[28] (whirl effect) with dynamic feedback and possibly self-
reinforced loading on
bearings, bedplates, the main shaft, yaw bearing, and dependent from concept at the
gearbox, generator, oil lubrication system, if strong tower oscillations are induced.
5.3.2.1
Drivetrain with three-point main shaft suspension
To calculate, usually design relevant, the bending moments along the main shaft
as well as the resulting loads at the main bearings and the gearbox trunnions (gear-
box supports in general), the equations for force and moment balance along the
axes have to be established. The number of equations and their complexity can be
reduced, due to some boundary conditions and pre-
assumptions, which are neces-
sary to come to an explicit result. For a three-
point main shaft suspension (refer to
Figure 5.29), according to Figure 5.8, the following assumptions are made.
The main bearing does not support any moments (also bearing friction is
neglected):
MBy = MBz = MBx = 0 (5.1)
The generator coupling (flex coupling) does not carry any loads (forces and
moments), besides torque:
MCz = MCz = 0 (5.2)
Figure 5.29
DT forces and moments for classic three-
point rotor shaft
suspension as a basis to calculate balance of forces and
moments, with FcogXz
part of dead-
weight forces along the z-
axis
FcogXz = WXcos
, whereas
WX the dead-
weight force of the
component X (refer to Index).](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-276-320.jpg)
![Drivetrain concepts and developments 249
a) Gearbox, shrink disc, brake disc dead-
weight carried by the trunnions:
FGz + WG cos
+ WSh cos
+ WBD cos
= 0 (5.5)
b) Trunnion force in the y-
direction is negligible compared to the bearing forces:
FGx + WG sin
= 0 (5.6)
FGy FBfy, FBry (5.7)
5.3.3
Loads analysis and strength verification
Usually, the rotating parts of a DT are more prone to reliability issues than structural
parts. For the design and strength verification of DT components, one must distin-
guish between different analysis methods in general. There is the fatigue analysis,
well known for structural parts, which is usually conducted with aggregated damage
equivalent loads. Similarly, in the process but with another focus, the analysis of
extremes is supplemented by a quasi-
static (“worst-
case”) analysis, which covers
unique or rare ultimate loading cases. And somehow special for specific mechani-
cal parts (gears, bearings, sealings, etc.), the analysis regarding temporal propor-
tions of load levels or operating states utilizes load duration distributions (LDD) for
the calculation of aggregated operational loading or specific conditions, which not
exclusively address fatigue issues [29].
Mechanical parts can fail due to many different failure modes resulting from
different physical effects, often somehow related to classic fatigue, but not only
[30]. Typical failure modes are, e.g., bearing fatigue, bearing axial cracking,
gear-
tooth cracking or fracture, fretting corrosion, micro/macro pitting, scuff-
ing, true or false brinelling, cage fracture, element or ring fracture, skidding,
sliding and electrical discharge, surface wear-
initiated fatigue, subsurface/
surface fatigue, and fatigue spalling from Hertzian stress (max. shear stress)
under the contact surface. To deal with that, specific influencing factors, partial
safety, and safety factors are used for the individual failure modes or loads,
respectively. However, the use of general partial safety factors often leads to
rather conservative designs, especially where several factors are multiplicatively
linked. The advantage of smart validation strategies (refer to Chapter 9) should
be pointed out here again. The current partial safety factors were mostly deter-
mined on scaled components or specific material samples for each type of load.
Superposition of loads or influencing factors is then mostly considered conser-
vatively. Tests on real or smart scaled components (refer to Chapter 9) can help
to adjust the safety factors for the respective application and parts without losing
reliability but saving costs.
The required partial safety factor is defined as the ratio between allowable stress
and permissible stress, whereby the calculated actual stress (including influencing
factors, e.g., stress concentration, scaling, surface treatment, etc.) based on the nom-
inal stress should never exceed the permissible stress.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-278-320.jpg)
![Drivetrain concepts and developments 251
Figure 5.31
From material probe testing (here without mean stress, R = −1
[31]) to the principle S/N-
curve shape for steal parts with main
characteristics](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-280-320.jpg)
![Drivetrain concepts and developments 253
mean-
stress-
free S/N curve in the damage calculation. Here, the mean-
stress afflicted
cycle amplitudes are converted to the respective weighed mean-
stress-
free ampli-
tudes either using a linear function (“Goodman line”) based on the tensile strength
or the yield point (“Soderberg line”) or using the “Gerber parabola,” depending on
the material characteristics. Furthermore, notches and surface roughness have an
essential influence on the fatigue characteristics as well as the material itself. So,
in consequence, we come from the basic (synthetic) material S/N curve to modified
S/N curves for components, which includes additional factors (e.g., surface rough-
ness, scaling factors, treatments, stress concentration, material support rate).
For a more realistic assessment of specific stress cycle ranges near and below
endurance strength, e.g., the FKM Guideline [19, 31, 32] suggests three variants of
the Palmgren–Miner formula. The “miner elementar” uses all amplitudes also those
below endurance strength of the component S/N curve, the “miner original” neglects
those amplitudes, and the “miner consequent,” applies a flatter slope in the endur-
ance strength area than in the fatigue strength area, due to the previous damage.
In recent decades, continuous progress has been made in the reliable assessment
of the damaging effects of dynamical loads in components (Figure 5.33). We see a
development from the “classic” strength verification over state-
of-
the-
art FE simula-
tion with fully elastic (linear, nonlinear) material properties to even more modern
methods like general fracture mechanics with the local and the probabilistic crack
propagation concept.
Fracture mechanics describes the process of crack growth, crack propagation,
crack arrest ability, and fracture in a component or material under operating condi-
tions (load cycles) [33, 34]. By estimating the lifetime or remaining useful life of
Figure 5.33
Comparison of classic and advanced methods and approaches
performing the strength analysis of components: (a) load profile
over time with constant load limits vs. (b) with variable load limits
(real load time history) (c) synchronous phasing of the load shares
vs. (d) non-
synchronous, arbitrary phasing (e) ideal linear-
elastic
behavior vs. (f) elastic-
plastic and nonlinear material behavior
(i.e., hysteresis) (g) material free of cracks vs. (h) consideration of
defects (i.e., cracks) → fracture mechanics methods: LEFM and
YFM, probabilistic approaches](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-282-320.jpg)
![Drivetrain concepts and developments 255
times for different DLCs. These 10-
min periods have their origin in meteorological
modeling and bin classification of wind measurements. For some of the mechanical
DT component such description and clustering does not fall into place, as mentioned
above, even when most of the loading results from wind, but nevertheless it is com-
mon in use. So as mentioned within a 10-
min period, stochastic wind characteristics
such as the mean wind speed and the turbulence intensity at hub height are assumed
to be constant. During such short periods of stationary conditions, the load response
processes that finally cause the stress cycle in the considered components then can
be taken as stationary stochastic processes, too. Therefore, the wind field-
induced
stress range distributions under stationary condition are often seemed to have similar
distribution characteristics that are close to a Weibull distribution [21, 31, 32]. Also,
the principle of S/N-
curve modeling is based on the assumption that most fatigue
failure characteristics of materials for specific load conditions (stress cycle ampli-
tude and mean value) can be described by Weibull distribution, due to its inherent
“memory” properties:
FS
s
= 1 e
s/SA
ˇ
(5.9)
For the representation of stress range distribution, it will often be sufficient to apply
specific Weibull distributions. The simplest distribution model among this family of
distorted distribution is the three-
parameter Weibull distribution [21, 31]:
FS
s
= 1 e
sa
/SA
ˇ
(5.10)
The coefficients a, β, and SA
are distribution parameters, and they can often be
expressed as a function of wind characteristics (average wind speed and turbulence
intensity) [21]. When the short-
term (means here periods of 10-
min) load cycle dis-
tribution, conditions by the assumed wind regime have been established and the
long-
term distributions of average wind speed and turbulence intensity are known
for the site, and furthermore, the anticipated service life of the turbine has been
defined, then the cumulated distribution of all load cycle ranges during the lifetime
can be established. As already mentioned, this cumulated distribution of different
production modes times of a turbine can itself often be represented by a Weibull
distribution, which forms a significant contribution to the loads that cause fatigue
damage.
To obtain the distribution of all relevant fatigue loads in the design life, the
cumulated distribution has to be supplemented by the load cycle distributions from
transient conditions such as start–stop, standstill, idling, and abnormal yaw mis-
alignment. For this purpose, information, e.g., about the duty cycle or grid events
for the WT is necessary. The duty cycle is a repetitive period of operation, which
is characterized by a typical sequence of different modes of different durations for
consecutive 10-
min periods. Information about the duty cycle should, as a mini-
mum, contains the number of starts and stops during a representative period of time,
together with the corresponding wind characteristics [21].
Starting and stopping the turbine are potentially critical with respect to fatigue.
One reason for this is that connection (after synchronization) of the electric WT](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-284-320.jpg)
![256 Wind turbine system design
system to the grid can cause high transient loads in the entire DT. Specific loads dur-
ing standstill and idling below cut-
in wind speed with respective low aerodynamic
loading (especially thrust) can be significant in comparison to nominal operating
condition, due to rotor gravity forces, low speed, poor lubrication, and harmful kine-
matic conditions for gear and bearings. Yaw misalignment as well as yaw movement
can be critical in any case with respect to DT loading and fatigue. It should be noted
that cumulated stresses at particular locations considered critical with respect to
fatigue or static strength may be due to a combination of single stresses arising for
different reasons or load processes (wind and yaw movement). For a detailed stress
assessment on a component level, one approach is to assume load peaks (torque,
axial forces, bending, etc.) of two load processes appear simultaneously. Thus, high-
amplitude cycles can be combined with other high-
amplitude cycles, even if they
have slightly different frequencies. The mean values of two superposing load pro-
cesses should be combined in a kind of “worst-
case” scenario to create the largest
possible mean stress value. Once the load spectrum has been established over the
design lifetime, it is common to define so-
called damage-
equivalent load ranges S0
to be used with an equivalent number of cycles neq
. A damage-
equivalent load range
is defined as that constant load range S0
, which in neq
cycles will lead to the same
accumulated damage as the distributed load spectrum that consists of many different
load ranges. When the equivalent number of cycles and the specific material fatigue
characteristic (S/N curve) are chosen or specified, the equivalent load range can be
found as [21, 31, 32] follows:
S0 =
0
B
@
P
i
niSk
i
neq
1
C
A
1/k
(5.11)
in which k denotes the S/N-
curve slope. This definition of a damage-
equivalent load
range is well known in fatigue analysis and is often used in WT strength verification.
It should be noted here that this method is only applicable for materials whose S/N
curves can be described by one slope k (e.g., not valid for composite materials). As
discussed, load spectra are usually not known with certainty but are somehow pre-
dicted, e.g., from a limited number of time series of load responses, obtained from
aero-
elastic and servo-
mechanic simulations or field measurements. The already
introduced time series of 10-
min duration are used for this purpose and one estimate
of the equivalent load range can be calculated from each simulated time series of a
DLC. Then, time series of load response are generally simulated for various wind
characteristics, operation modes, and duty cycles. The already introduced cumulated
load spectra are established on this basis by appropriate weighting according to the
expected long-
term distribution over the design lifetime.
5.3.3.2
Statistical and deterministic extremes (ultimate loads)
When extreme load responses are of interest, such as for design against failure in
ultimate loading regimes, extreme value distributions are needed. So, let us assume](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-285-320.jpg)
![Drivetrain concepts and developments 259
rotor, for the current and future generation of WTs. It should be noted at this point
that the general designation “platform” or “platform strategy” for WTs and their DTs
is not a universal sign of product quality. The manufacturers sometimes pursue differ-
ent goals with it, and the degree of supplier involvement varies greatly. The industry
sector with the highest level of experience and know-
how in this area is probably the
automotive industry (e.g., MQB, referred to as “modularer Querbaukasten,” MSL,
MSQ, and MEB from VW). Their product lines are initially based on consistent, deep
modularization for a technical platform that can be used as broad as possible (in terms
of product range), with standardized module interfaces if applicable, the highest pos-
sible proportion of identical parts, as well as with the highest possible flexibility for
small individually adaptations, future developments, and products. Of course, “a WT
is not an automobile” is the most common response to comparisons with the automo-
tive sector with its consumer product focus, much higher production numbers, etc.,
but anyway, some WT OEMs are still rather away from the fundamental idea of a real
platform concept, except the wording.
From a generic perspective, a platform strategy for WTs and their DTs should
take into account the following technical product requirements, at which this list
does not claim to be complete:
•
• different WT applications, onshore, near-
shore, and offshore sites, etc.
•
• different wind regimes for specific sites or markets (mean wind speed, turbu-
lence intensity, extreme condition, e.g., typhoon)
•
• additional project (site)-
specific requirements (complex terrain and high alti-
tude), cold and hot climate (dry and/or humid), other customizations
•
• specific energy market conditions or requirements
•
• grid requirements (e.g., grid codes, ancillary grid system services)
•
• specific requirements for noise emissions (limits and flexible regimes) and other
environmental impacts
•
• market limitations regarding transportation, construction, and service
•
• market and competition requirements for flexible power upgrades
•
• cost-
efficient and short-
period future developments and optimizations
So, let us have a look at the more experienced automotive sector again. Here,
the supplier industry reacts accordingly and offers suitable modules for the different
OEM platform approaches. Suitable modules in this context mean the OEM defines
the interfaces of the modules, “standardized” them (just for the single OEM or in
agreement with other OEMs) if applicable. Then the suppliers can bring all their
innovation potential and specific know-
how into module design, but modules and
thus suppliers remain exchangeable. For nearly 10 years, the wind industry sector
starting increasingly using their own ideas of platform strategies [35, 36] of course,
focused to reduce manufacturing costs, optimize production and construction pro-
cesses, faster upgrades to stay competitive, and to achieve fast development cycles
for the markets in general.
In this process, fundamentally different approaches are conceivable with regard to
the role of the supplier industry. Especially for DTs, currently, an increasing change](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-288-320.jpg)
![264 Wind turbine system design
aerodynamic blade profile coefficients cL
(lift coefficient) and cD
(drag coefficient).
In mechanics, dimensionless parameters such as damping coefficients, e.g., Lehr’s
damping factor, as well as the description of vibrations using amplitude and ampli-
fication factors, as well as normalized mode shapes, are known generally. Electrical
engineering also uses dimensionless parameters, e.g., in the form of stray numbers,
winding factors, pole overlaps, relative short-
circuit voltages.
With the help of the similitude theory, scaling rules, also known as modeling
rules, can now be derived for the WT as a whole but, of course, also for its subsys-
tems and components [37]. In concrete terms, this can mean developing a model
family with different sizes and corresponding performance characteristics from an
existing (complete) system by scaling it up. In the simple but not trivial case, e.g.,
the (conceptual) upscaling of an existing 1 MW WT has already been tested in the
field to a multi-
megawatt turbine. The scaling as a rule has to be done with a scal-
ing factor φ according to strict geometrical rules. All mechanical dimensions of the
original plant design are then scaled with the same scaling factor, i.e., increased or
decreased accordingly, and only then result in a strict geometric similarity.
For certain operating parameters (e.g., nominal speed) of the original system,
there are necessary changes for the scaled systems if certain technical boundary condi-
tions (e.g., an unchanged top speed number, nominal wind speed, and wind class) of
the system for the sake of compliance with design features (e.g., rotor coefficients)
are physically required, due to the similarity theory. This scaling, considering bound-
ary conditions (laws of similarity and environmental conditions), then leads to corre-
sponding model laws (laws of scaling). These apply then to the model family under the
specific boundary conditions and can be regarded them as basic model laws. Starting
from this consideration of the overall system, the derivations (specific model laws) for
the subsystems and components within this application result, i.e., specifically for this
WT under these conditions. This understanding is important because, for example, a
direct comparison of systems with different sized rotors in the current development
does not necessarily aim at a higher turbine capacity (power), according to the proce-
dure described above regarding the basic model laws.
Increasingly, turbines are being designed with a similar design, but larger rotors
for better energy yield, respectively, a better capacity factor, at low-
wind sites, i.e.,
in the similarity analysis and thus the model laws, there are deviations from the basic
rules, due to a changed nominal rotational rotor speed, wind speed, as well as rated
generator power. The same applies, as an example, when reducing the design blade
tip speed for the noise-
optimized system design. Of course, a distinction must be
made as to whether this is a special operating mode of the basic design or whether
the system design as a whole has been optimized for this requirement.
The theoretical scaling in conjunction with the similarity theory makes it possible
to compare development lines from OEMs and sometimes also between OEMs with
regard to performance indicators [specific weight-
to-
power ratio (tons/MW), specific
power rating (W/m2
), specific torque-
per-
weight ratio (Nm/kg), and weight-
per-
swept
area ratio (kg/m2
)] without knowing all the design details. Due to a large number of
additional influencing variables and uncertainties, these comparisons usually do not
provide exact values, but at least usable qualitative statements (Table 5.2).](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-293-320.jpg)
![266 Wind turbine system design
5.3.5.1
Wind turbine performance indicators
Despite the clearly described scaling laws and similarity rules, it is not trivial to
compare turbines of different sizes, stages of development, and DT topologies with
one another in terms of their performance and level of innovation. As already men-
tioned, the reasons for this often lie in the subjectively small but nevertheless rel-
evant deviations of the individual main design variables (e.g., nominal wind speed,
tip speed at design speed factor, wind classes) but also in the partly inconsistent,
publicly accessible technical data sheet information regarding, e.g., nacelle and
tower head masses, nominal speeds, power curves, from the turbine manufactur-
ers or database providers. In the meantime, especially for the latest generations of
turbines in the onshore and offshore sectors, there are hardly any clear technical,
publicly available details, apart from the rotor diameter and nominal power of the
turbines. The details of nacelle weights and tower head weights are for the most
part based on indirect information from third parties, e.g., crane manufacturers and
other sources that are difficult to verify and are therefore always subject to a certain
degree of uncertainty. Clear reasons for these information restrictions on the part of
the OEMs are not clear; one assumption is that transparent and detailed informa-
tion would allow conclusions to be drawn about a possible future capacity of the
platforms and a general comparability between different OEM products. Probably,
these are not desirable from an OEM competitive point of view. Occasionally, how-
ever, there is technical information directly from the manufacturers, which provides
exemplary insights into the applied technologies and current physical limitations.
The articles written by Eize de Vries in the Windpower Monthly magazine
should be emphasized here, which always shine with a high degree of technical
details and represent an important and trustworthy source of information about WTs
and DTs. As an example, a short excerpt from an interview at “WindEurope 2019”
in Bilbao about details on the then brand new V164/174 OWT from MHI Vestas
shall be presented here. MHI Vestas CTO senior product manager Anders Bach
Andersen talks to Eize de Vries about technical innovations: “Because 10.0 MNm
gearbox input torque is a constant both turbines operate with 90 m/s rated tip speed,
the resulting rotor speeds are 9.9 rotations per minute for the V174 and 10.5 rpm
for the V164. Since power is a function of torque and rotational speed, …,” “the
bigger V174 rotor turns slower, and the maximum power output must be reduced.
To manage the generator temperature even during sustained operation with many
full-
load hours and/or in high-
temperature regions, the V174 gearbox step-
up ratio
is increased from 1:38.1 to 1:40.8. This offers higher generator speed with reduced
internal heat production and thus extra thermal reserves due to lower generator cur-
rents for a given output” [38].
This example of slightly adjustable gear ratios shows quite impressively a major
advantage of the geared DT concept compared to DDs. Furthermore, this example
also illustrates the essential influence, means benefits, and disadvantages of the
potential to increase the rated rotor-
blade tip speed, especially for offshore applica-
tions. The debate about proper blade tip-
speed limits has a longer history, predicting
that offshore-
turbine tip speeds would increase within years from 80–90 m/s of the](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-295-320.jpg)
![268 Wind turbine system design
the current state of technical development of the WT, the characteristic values of
the specific rotor power are suitable for a characterization, if calculated once about
the WT nominal wind speeds and once again with regard to the mean wind speed of
the respective IEC wind class of the WTs. If the ratio to the theoretical maximum
value is calculated for both values according to (5.13), differences for a specific tur-
bine show the degree of the design as a real low WT and, in the comparison between
two turbines, the possible performance difference between those two WTs.
Remark: Be aware the nominal specific rotor power of WT today (datasheet
value) is just given as the ration of rated power and rotor swept area, independent
of the WT IEC wind class design and its nominal wind speed, where rated power is
reached [24]. So, this value gives only a general design, no WT performance indica-
tion. Values around 350 W/m2
for offshore WTs and well below, as well as values
around and below 250 W/m2
for onshore WTs clearly speak for a today “low-
wind
design,” although there are no official low WTs for offshore application by definition
(no specific IEC wind class):
=
PWT_th
AR
= 1
2
air total cpmax v3
N = 0, 265 v3
N [W/m3
]
pAth.
(5.14)
tAth.
min., max.
= pAth. 1
!min.:::.!max.
[Nm/m3
]
(5.15)
This means an assessment of realized turbines with their data sheet-
specific infor-
mation for the specific rotor area output should be at least conducted with regard to
the calculated theoretical value pAth.
(5.14) at the nominal wind speed of the turbine.
According to the model laws for WT scaling, however, the specific rotor power
density has no dependencies on scaling. Therefore, it cannot be used for a charac-
terization of innovations in turbine design. However, in literature or studies [35],
sometimes, the classification “best in class” is used for the lowest values according
to the IEC design wind class and the nominal power of the turbine, which is quite
irritating and not reasonable.
The specific rotor torque rating, which is unusual in the literature, seems more
suitable for this purpose as a technological assessment. Due to the speed depen-
dency at a constant design blade tip speed, a proportional increase is to be expected
here when scaling WTs applying the same concept (no innovations), causing the
special torque-
dependent component weights to increase, e.g., gearbox, generator
(especially DD), and also the passive structures have also to be reinforced [39].
These component dead weights (masses) applying a pure scaling of existing WT
technologies, as already known, show a cubic increase. However, due to the lack of
mass reference (and thus implicit costs), the specific rotor torque rating does not also
have any pronounced evaluation properties but rather serves to ensure the general
comparability of designs, e.g., if the design nominal speed is reduced for noise-
optimized operation with the same rotor size and nominal wind speed. Such turbine
designs would be recognizable not only by their specific rotor power but also by
their comparatively high specific rotor torque tAth.
(5.15). In simplified terms, it can
be stated that the specific rotor power and the specific rotor torque serve more as a
general characterization of the turbine design.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-297-320.jpg)
![Drivetrain concepts and developments 269
Mass-
related parameters, on the other hand, are better suited as performance
indicators. The reason for this is that we generally expect a cubic increase of
nacelle and tower head mass. This, in turn, leads (for pure scaling) to a lin-
ear increase of tower head mass (rotor+hub+nacelle) to rotor swept area ratio,
which, viewed on its own, refutes the economic argument of larger WTs solely
through scaling without innovations (refer to [40]). Previous investigations
(refer to [24, 27 and 41]) have shown that when looking at different generations
of turbines, this ratio moves within certain, relatively narrow limits, which can
only be achieved through the consistent introduction of innovations in always
bigger WTs. Typical values here are in the range slightly below 20 kg/m2
(typi-
cally 18 kg/m2
); previous experiences show that it has not yet been possible to
fall below 15 kg/m2
for multi-
megawatt WT, at least when using the current
technologies. For that reason and because no really disruptive developments
can be identified on the market in the short and medium term, this should cur-
rently mark our lower limit (best point). It is even more astonishing that turbines
with ratios of 22 kg/m2
and more (~26 kg/m2
) are currently able to assert them-
selves on the market. The very simplified cost estimate, which increases linearly
with the weight, does not seem to play such a decisive role here or is superim-
posed and compensated, respectively, by other effects at the OEMs (production,
assembly, supply chain, logistics, etc.). Nevertheless, the ratio provides a good
first estimate of the resource consumption (level of utilization) of the respective
turbine technology (especially in the rotor and nacelle area) with regard to the
required specific use of material resources.
Another interesting parameter from the author’s point of view is the specific
torque density (Nm/kg), which is also used to assess the force density and thus the
specific performance of gears (range of high-
end WT transmission at 200 Nm/kg
and slightly higher). Just to emphasize the technological improvements and inno-
vation, let us state here on a component level, 10 years ago, 100 Nm/kg marked
the best point for WT gearboxes. For the entire turbine, this proposed performance
parameter (Nm/kg) is related to the tower head mass or nacelle weight and there-
fore includes the aerodynamic and structural design of the rotor as well as the DT/
nacelle design. Therefore, it means this key parameter is a strongly aggregated vari-
able. Within modern WTs, it is in the range of 11–25 Nm/kg (rotor torque to tower
head mass ratio), whereby the current values for DD concepts are currently roughly
estimated 10–15 % higher than those for the best-
geared WT and thus currently
mark the best-
in-
class point, in terms of general (not evaluated) resource efficiency
of the material. According to the laws of pure scaling, this performance parameter
should be quite constant. If we compare older with newer realized designs, we see
a clear trend in implementing innovations, a higher utilization of material, based
on the same and the introduction of new materials with higher strength properties,
documented in rising values over time (5–10 Nm/kg in 1996, 8–15 Nm/kg in 1999,
refer to [24]).
For the specific weight-
per-
power ratios (tower head mass/rated power) due to
the currently very similar blade tip-
speed interpretations, we would expect a linear](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-298-320.jpg)
![Drivetrain concepts and developments 271
5.4.1
ENERCON
ENERCON started producing WTs in the mid-
1980s. The first turbines
E-
15/16/17 in the power range of x0kW and a little later the E-
32 with at least
300 kW already equipped with a synchronous generator and full converter,
until that time ENERCON turbines were still equipped with two-
stage gear-
boxes. With the first gearless series WT E-
40 (500 kW) from 1992, the wind
pioneer ENERCON (the founder Aloys Wobben, respectively) achieved the
technical and economically breakthrough already in 1993, again 2 years later,
ENERCON played with the E-
66 in what was the former WT top class of 1.5–2
MW (the number behind the Enercon “‘E-” always defines the rotor diameter
of the WT type, respectively). Later on, ENERCON introduced the designation
EPx as an additional type designation, which indicated the belonging to a spe-
cific xMW platform (with the corresponding rated power range) of the turbine.
The E-
66 thus formally belongs to the EP1 class, the successor E-
70 2004 with
2–2.3 MW already to the EP2 class. Until 2017, ENERCON always relied on
the tried-
and-
tested basic nacelle and DT concept of the E66 for all new WT
developments, with the characteristic egg-
shaped nacelle, first made of GRP
and later on made of aluminum, since the introduction of the E-
82 series. This
consistency is also a reason why the practical verification of scaling laws can
be shown and verified particularly well on this WT up to 2017. During that
time, there were no fundamental changes in concept, but evolutionary, partial
improvements only (e.g., cooling conversion to partial water cooling within the
E-
82, 2006-
2014, EP2 series); Figure 5.22 (bottom) shows the basic representa-
tion of the long-
established ENECON “old” DT concept. This concept applied
a combined bearing of the rotor hub and the generator rotor on a “long” king-
pin structure with a two-
bearing bearing solution [small bearings in diameter
with large bearing spacing, see FEM Figure 5.22 (top)] in the hub. The laws of
scaling and the adherence to the generator technology as an electrically excited
internal rotor synchronous machine and the rotor suspension principle resulted
comparatively high nacelle or tower head masses for the ENERCON WTs until
the recent past. The top marked, until the introduction of the V164-
8.0 MW
Offshore WT from Vestas, the uprated E-
126 (7.58MW) EP8 from 2010 was
the most powerful onshore WT ever. This type had a nacelle weight includ-
ing the hub of 364 tons; with blades, the tower head mass was ~650 tons. Its
“grandfather” model the E-
66 had 67 tons for the nacelle and ~100 tons for
the total head mass. Besides other ENERCON thus underlined the prejudice,
which is still rampant in some cases, that DD WTs generally have a very high
nacelle mass, especially for high-
rated power WT. The generator of the E126
alone already weighed ~220 tons, but the high mass in total to large extent
resulted from the special, necessary rigid construction of this Enercon turbine
series. Until the EP4 series from 2014 with the type E-
126 and E-
141 remained
true to the “old” ENERCON design principles, so the E-
141 still had a tower
head mass of ~490 tons. The 2018 less powerful but very innovative, optimized](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-300-320.jpg)
![274 Wind turbine system design
5.4.3
General Electric wind energy (GE)
The history of WTs from GE started with the companies Tacke Windtechnik
(Germany) and Zond (United States) which were bought by Enron in 1997. After
the Enron bankruptcy in 2002, GE took over their wind division. GE Renewables
was and is particularly well represented on the onshore WT world market, the rather
changeable offshore division only gain traction with the takeover of Alstom’s energy
division in 2015. However, in terms of market share, GE occupies third place behind
Vestas and Siemens Gamesa among the western manufacturers. Especially in the
onshore sector, GE was influential with its turbines on the way that the standard
WT developed into a commodity with the largest possible quantities, standardized
production, and low prices. GE’s 1.5-
MW WT already reached the limit of 10,000
manufactured and installed systems in 2008. A total of ~16,500 GE 1.5sle turbines
were installed. Since the takeover of Enron’s wind division and from a historical
perspective, GE has rather relied on reputed robust DT concepts, which means
non-
integrated designs with separated DT parts (no dual functionalities of parts)
as far as possible (refer to Figure 5.7). The 3 MW platform is still being produced
using a four-
point main shaft suspension in principle; however, in the case of the
3 MW platform, it is not designed with two completely separate bearing housings
but integrated into a cast structure at the nacelle frame. The gearbox then hangs on
the comparatively short main shaft, supported by two torque arms. The design is
relatively compact compared to the standard four-
point bearing design, but not a
dedicated lightweight construction due to the massive nacelle support structure. In
contrast, the older 2 MW GE platform, still a high seller in specific markets, applies
a three-
point suspension for the rotor shaft. A three-
stage gearbox with torque arms
and a high-
speed DFIG system is the standard configuration for the other DT parts
of this smaller platform, too (refer to Figure 5.8). Since the introduction of variable-
speed turbine control, GE has relied almost exclusively on double-
fed asynchronous
machines with partial converters as a particularly cost-
effective solution for the elec-
tric DT. The WTs of the former 1.x MW range (especially the 1.5 MW types) also
used the classic three-
point suspension for the rotor shaft. After gearbox damage
increasingly occurred in the field at that time, comprehensive test and validation pro-
grams started at the NREL with regard to DT dynamics and reliable main bearings
configurations [20, 42], which continue till now. The development of the very suc-
cessful 2 MW platform has its roots in the design of the 1.5 MW (1.5i) in 1996. The
1.5sle with its large numbers was presented in 2004, followed by variants up to 1.85
MW and with rotor diameters up to 100 m until 2013, in comparison to the Western
OEM competition, comparatively small systems (rotor sizes) with lower capacity
(rated power). During the development of the 2 MW platform and the introduction
of the 2.5 MW DT, high-
speed PM generators with a full power converter were used
for the first time. But already with the introduction of the GE 2.5-
120 variant, GE
returned to the internally valued, above all cost-
effective, DFIG technology. The
2 MW and 3 MW platforms are primarily aimed at emerging markets and mass
markets, where robust systems with uncritical specific wind farm site area to power
density appear to be proving their worth. GE now develops on the basis of the 2 MW](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-303-320.jpg)
![Drivetrain concepts and developments 279
5.5
Offshore wind turbines and drivetrain developments
Due to the subsequent chronological development of offshore WTs, the diversity
of their DT concepts and variants used over time is somewhat less manifold than
those for onshore turbines, but nevertheless impressive. Even if the current market
shares of the two main DT concept speak rather clearly in favor of the DD concept,
the high-
power Hybrid-
Drive for offshore applications with two to three planetary
stages, which was mainly pushed and built by Vestas, has won a respectable posi-
tion. For OWT, the competition for the best DT concept remains open, too, whereby
a disappearance of the DD technology in this market segment is unimaginable from
today’s perspective. As in the onshore sector, some OEMs, including pioneers in the
offshore sector (e.g., BARD,Areva, Senvion/former REpower, Sinovel) disappeared
for different reasons from the market after some initial success. Other well-
known
turbine OEMs decided at a very early stage not to get involved in the “adventure”
offshore wind energy, such as ENERCON or Nordex; some appeared undecided,
such as GE, where the time phases of commitment and abandonment alternated.
This draws a quite realistic picture of the offshore wind industry as a modern adven-
ture for manufacturers, suppliers, and operators. Little or poorly usable experience
from the offshore and maritime sector paired with major technical challenges (sea-
bed foundations, very large WTs, complex electrical shore connection, complex
logistics for construction and maintenance, and limited accessibility) and the most
adverse environmental conditions (sun, salt, ice, ocean currents, and maritime veg-
etation cover) standing for a few challenges, which can be somehow compared to
other great technical “adventures” like the moon landing. Therefore, all pioneers in
this field deserve a high degree of respect for the achievements, which are already
taken for granted today in some cases.
Strictly speaking, the era of offshore wind energy started in 1991 with the con-
struction of 11 turbines of the 450 kW class (hub height 35 m, rotor diameter 35 m)
in the first offshore wind farm Vindeby in Denmark; 25 years of experience from
operating the systems in shallow water (water depth below 4 m, distance from shore
2 km) under rough offshore conditions is the reward for this Danish pioneering work
when the farm was decommissioned in 2017, due to economic reasons. Important
findings were missing in other countries, e.g., during the construction of the first
German offshore wind farm, indeed with a completely different turbine size, water
depth, and distance from the coast, when they started. The contracted OEM for the
wind farm Vindeby at that time was Bonus Energy. They supplied a modified ver-
sion of their standard Bonus 450 kW WT for the project. The turbines were modi-
fied for offshore use by sealing the towers, controlling the humidity inside with air
conditioning to extend the life of the machinery, and hardening the gearboxes using
heavy-
duty gearbox concepts with an input planetary stage.
The first industrial phase of offshore wind energy started in the year 2000
with the first noteworthy, comparatively xx MW near-
shore farm installations in
the North Sea (DK/Middelgrunden, UK, Sweden) consisting of smaller multi-
megawatt turbines ([43], Table 5.3). Viewed from the history of development, the](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-308-320.jpg)
![Drivetrain concepts and developments 281
Nevertheless, the need for dedicated offshore turbine development was quite
obvious, and the era of such turbines, for the second phase of offshore wind farms
(Table 5.4), started with Enrons 3.6 MW (later GE 3.6s) turbine that was one of the
first offshore-
dedicated machines. Originally designed with a 100 m rotor, this was
replaced by a 104 m rotor after GE bought Enron’s assets in April 2002. The tur-
bine incorporated a conventional non-
integrated high-
speed geared DT with DFIG
and 4-
point main shaft suspension (refer to Figure 5.7). As an OWT, this machine
already had innovative equipment features that are customary and necessary to this
day in order to take into account typical offshore requirements (rough environmental
conditions, on-
site logistics). Some of them we see nowadays within some of the
latest OWT developments again, e.g., a container solution beneath the nacelle and
behind the tower where the electrical/electronic equipment gets a protected environ-
ment, for main component exchange a 40-
ton foldable portal crane in and a helicop-
ter platform installed above the nacelle. Seven units were produced for operation at
the Arklow Bank project of Ireland in 2003. Plans for a mass- and cost-
optimized
GE 3.6sl successor with an enlarged 111-
m rotor were dropped, and GE rather
stayed away from the offshore sector, except for a short intermezzo with the upcom-
ing DD technology also for offshore application in the early 2010s. At that time, GE
canceled its development of a 4.1 MW offshore WT with 113 m rotor and a non-
integrated DD (PMSG) concept with double suspension main shaft support [refer to
Figure 5.20 (top)]. The only prototype built, owned by Goteborg Energi, was erected
in Goteborg in 2011. Then again GE disappeared from the offshore wind segment
for years until its acquisition of Alstom in 2015. From then on, GE began to catch up
offshore technology, but more details on that later. First, we want to follow some of
the historical timelines for offshore turbine and especially DT development.
The second innovative turbine development with a really ground-
breaking DT
concept was the V90-
3.0 MW from Vestas. It was not originally designed for off-
shore purpose only but used for it and with some design features that enabled the
advance into the 5+MW offshore turbine class and at the same time pioneering for
the later development lines of high and fully integrated DTs. This early and cou-
rageous design owns the Krone for the best (lowest) weight-
to-
power ratio (tons/
MW) of a nacelle with DT at that time and even considered scaling rules even some
Table 5.4 Typical OWTs of the second phase of commercial offshore wind farms
Turbine type Rated
power
(MW)
Rotor
diameter
(m)
Rated
wind
speed
(m/s)
Rated
rotor
speed
(rpm)
Tower head mass
(tons) (nacelle +
rotor)
Tip
speed
(m/s)
GE 3.6s (sl)
- version II
3.6 104
(111)
14
(15.5)
13,6 185 + 83 73
(79)
V90-3 MW 3.0 90 15 16,1 70 + 41 76
SWT 3.6-107
SWT 3.6-120
3.6 107
(120)
13.5
(12.5)
13 125 + 95
(125 + 100)
73
(81)](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-310-320.jpg)
![Drivetrain concepts and developments 283
most powerful offshore turbine (REpower/later Senvion 5M-
126 and Multibrid/later
Areva M500-
116) available on the market at that time, which heralded the era of the
first 5-
MW OWTs (Table 5.5). With the Areva M5000, medium-
voltage technology
with IGCTs for a full power converter system was used in WTs for the first time.
In terms of converter reliability, this innovation has so far shown noticeable advan-
tages over common, conventional low-
voltage installations based on low-
voltage
IGBT (1 200 or 1 700 V types) modules. The big OEMs Vestas, Siemens, and GE
remain a power class below with their 2–3.6 MW turbines during this period.
The BARD Offshore 1 Park was Germany’s first commercial offshore wind
farm when it went into operation in 2013. This 400 MW wind farm is located 100
km off the northwest German coast in the North Sea and consists of 80 BARD
5.0-
122 newly developed WTs. The REpower/Senvion 5M-
126/6.xM-
152 utilized a
robust, conventional (non-
integrated) four-
point suspension system with high-
speed
three-
stage gearbox and DFIG system. Roughly speaking, REpower was supplying
a well-
engineered but not highly innovative offshore turbine. Accordingly, specific
performance data such as the specific weight-
to-
power ratio (tons/MW) and the spe-
cific rotor-
swept area performance (W/m2
) are relatively modest in direct compari-
son to more innovative concepts from other OEMs. Also, the BARD OWT followed
a rather conservative turbine concept, partially documented by the high tower head
masses, too, but showed some innovations within the DT. The conventional high-
speed DT was combined with a separated single-
bearing (OTRB and moment bear-
ing, respectively) solution with a greatly shortened main shaft (refer to Figure 5.13)
and thus a comparatively small overall length. A special feature here is the separate,
comparatively soft support of the gearbox via torque arms and elastomeric hydraulic
supports, similar to “classic” three-
point suspension solutions, which usually apply
double raw SRB for the upwind bearing without capabilities to carry moments in the
rotor plane. Measurements on the turbines in the field [44] documented still rather
high dynamically displacements of the entire gearbox during operation, taking into
account the actually bending resistant main bearing. The generator, a DFIG system,
Table 5.5
Era of first 5+MW OWT class introducing the third phase of OWT
development
Turbine type Rated
power
(MW)
Rotor
diameter
(m)
Rated
wind
speed
(m/s)
Rated
rotor
speed
(rpm)
Tower head mass
(tons) (nacelle) +
rotor/hub + blades
Tip
speed
(m/s)
REpower 5M
6.2M-126
6.2M-152
5.08
6.15
6.15
126
126
152
13
14.5
11.5
12.1
12.1
11.1
(290) + 120
(325) + 134.5
(325) + 156.5
80
80
88
Areva
M5000-116
M5000-135
5.0
5.0
116
135
12.5
11.4
14.8
13.4
(200) + 111.5
(235) + 140
90
95
BARD 5.0 5.0 122 12.5 15.7 (280) + 70 + 3*28.5 100](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-312-320.jpg)
![Drivetrain concepts and developments 285
The generator with improved closed-
loop internal air cooling operates at a slightly
higher rated speed, because the tip speed increases roundabout by 10% compared to
its predecessors with a 154 m rotor.
At Siemens Gamesa, offshore turbines with gearboxes [the so-
called internally
(older) Siemens G4 platform] do not seem to matter anymore in the future. The top
of development there was marked by the SWT-
4.0-
130 (former 3.6 MW design).
The usage of gear technology at SGRE nowadays is focused on onshore WT on the
basis of Gamesa 3.x, 4.x, and 5.x platforms. On the other hand, gearless onshore sys-
tems, originally based on the Siemens D3 platform (4.2 m outer rotor PMSG) with
capacities from 3.3 to 4.7 MW and rotor diameters of 120–130 m, were only built in
small numbers until around 2017, seem to be no further under ongoing development.
Following the purchase from Ecotècnia by Alstom and later on the acquisition
of Alstom by GE, the production for the 6MW Haliade OWT started in 2016 at the
St. Nazaire factory. For its DT, the Haliade featured an inner rotor DD PMSG con-
cept with a very high mechanical robustness (based on Alstoms PureTorqueR
design
principles, refer to Figure 5.24). The prototype started an extended test period in
spring 2016 at Østerild Wind Turbine Test Field.
At that time, the most powerful turbine under test (Prototype from 2014 to 2016
at Østerild Wind Turbine Test Field) was MHI Vestas V164-
7.0 MW, with a power
Table 5.6
Era of the latest (fourth phase) super class OWT—development line
of SGRE
Turbine type Rated
power
(MW)
Rotor
diameter
(m)
Rated
wind
speed
(m/s)
Rated
rotor
speed
(rpm)
Tower head mass
(tons) (nacelle +
hub) + blades
Tip
speed
(m/s)
SWT-6.0-154
DD
6.0 154 13.0 11.0 (275) + 3*25 89
SWT-7.0-154
DD
7.0 154 13.5 12.0 (285) + 3*25 96
SG-8.0-167
DD
8.0 167 12.0 12.0 ~(285) + 3*30 104
AD8-180 8.0 180 15.0 8.5 ~(356+ 107) +
3*34
80
SG-10.0-193
DD
10.0 193 ? ? ~(400) + 3*40 ?
SG-11.0-200
DD
11.0 200 ? ? ~550*
?
SG-14.0-222
DD
14.0 222 ? ? ~600*
?
SG-14.0-236
DD
14.0–15.0 236 ? ? ~650+ ?
“?” means, no reliable public information available, “~”means, slightly more, “*” assessment by the
author, no public data available.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-314-320.jpg)
![Drivetrain concepts and developments 289
uprated 14-
MW rated Haliade-
X offers 405 W/m2
, and Siemens Gamesa’s upcoming
SG 14-
222 DD in 14-
MW “standard” mode 362 W/m2,
respectively, 320 W/m2
for the
latest announced SG 14-
236 DD. In line with wind industry practice for onshore and
offshore, newly introduced platforms always have future scalability in their DNA. The
initial V164-
7.0MW (331 W/m2
) platform was followed by a V164-
8.0 MW prototype
and commercial model (379 W/m2
), then again uprated to the V164-
9.5 MW (450 W/
m2
), and finally to a V164-
10.0 MW (473 W/m2
). Perhaps further scaling of the cur-
rent platforms and OWT flagships along the same evolutionary development pathway
could therefore in future result in uprated outcomes toward 20 MW or site-
specific
extremely low specific power ratings of 250–300 W/m2
(refer to MingYang latest
MySE16.0-
242 OWT, 16 MW power rating using a low-
integrated Hybrid-
Drive with
bearing unit and a three-
planetary gear stage gearbox, and to Goldwind GW12-
242
with 261 W/m2
, also applying a low-
integrated Hybrid-
Drive PMSG DT). So, this Top
OEM (No. 4 worldwide) takes the step and seems to leave the proven former path as
a DD specialist, exactly the other way around Siemens did it in the past. This shows
once again very clearly, nothing is fixed, and the race for the best DT concept in the
wind industry remains open.
5.6
Outlook and potential development trends
A look back at the technical developments and the global growth of WT installations
shows, on the one hand, clear, consistent long-
term trends with regard to various
aspects [3, 5, 45]. This applies to the growth of onshore and offshore WTs, which
has already been mentioned several times in the chapter, in terms of the rotor swept
area and the output capacity (rated power). In 2022, the current prototypes that are
planned for the upcoming wind farms will have 14–15 MW for offshore and almost
6–7 MW for onshore applications. In retrospect, this growth of turbine sizes did not
always run at the same speed, but generally always with considerable dynamics and
with clearly positive gradients. However, the OEM’s commitment to take technical
risks for the development or adaptation of the turbine key systems (rotor, DT, and
tower) varied greatly during this period from the 1980s to the 2020s (i.e., within 40
years of development history). Roughly speaking, cycling periods of high and low
risk developments are clearly visible, sometimes temporal shifted between markets,
regions, and OEMs. Especially in recent years, the gradient of growth for turbine
rotor swept area and capacity has again increased significantly. A clear end even a
leveling off is not yet observable. But the last few years seem to have been shaped
more by evolutionary developments than by groundbreaking developments, even if
the turbine sizes are really remarkable. Further optimizations were carried out, espe-
cially for the DT, and existing design margins were exploited. This approach is also
documented by the performance indicators, which do not indicate any technological
leaps for the latest developments of OEMs.
On the other hand, the lack of such disruptive innovations can be clearly rec-
ognized. But what would be such a disruptive technological leap? For example, the
introduction of HTS [46] technology is suitable for series production, especially](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-318-320.jpg)
![290 Wind turbine system design
in the area of DD, in order to significantly reduce the specific installation space
for a higher capacity and the dead weight of the generators compared to the cur-
rent state. In recent years, the company Envision in cooperation with partners has
developed a prototype-
ready, scaled 3 MW generator for field testing in a WT.
The first test runs at Fraunhofer-
IWES on the 10 MW generator/nacelle test bench
DyNaLab in Bremerhaven (Germany) were quite promising. HTS technology for
WTs [46–50], which currently remains limited to DC sub-
applications within the
generators (means DC excitation of the SG) and offers the technological poten-
tial to increase the value of the air-
gap flux density (currently with state-
of-
the-
art
PM technology limited to ~1 T) to significantly higher values clearly above 2 T.
In order to achieve this, the entire generator must be designed accordingly (e.g.,
materials and magnetic flux paths within the generator). The constant power require-
ment for the cooling capacity of considerably above 100 kW for very large systems
15 MW remains disadvantageous. To date, the construction and weight advantage
has been decompensated by the additional systems required and the expensive spe-
cial materials, which are additionally not easy to handle within generator produc-
tion process. The cooling system for the HTS has to run for the stand-
by function
of the WT, which does not negligibly reduce the overall efficiency, flexibility, and
availability of operation. Finally, nothing or very little is known about the techni-
cal reliability of the systems (cryogenic systems in offshore use, rotary transmitter
for cooling medium, HTS materials in dynamic operation, etc.). So, the hurdles to
introduce new technology are always quite high.
A related question is whether the current growth will continue in this form. The
main argument of the proponents for even bigger turbines is the lower number of
required foundations and associated internal inner-
array cabling. As well as lower
service costs due to the smaller number of turbines (fewer personnel resources, etc.)
should enable decreasing LCoE, as far as the theory goes. The increasing complex-
ity and border lining utilization of material and physical principles within the largest
turbines speak against this. So, components become sometimes specifically more
expensive; thus, a 5 MW gearbox has lower specific costs than a high-
end 20 MW
gearbox; the same applies to generators, pitch systems, etc. Summing up some key
factors from the view of the author that could enable further substantial growth of
turbine size and power rating:
•
• new materials and designs for generator air-
gap flux densities above 2 T
•
• internal operating temperature ranges of 200°C and above for electrical compo-
nents and power electronics (permanently and with high reliability)
•
• new HTS materials for higher temperatures and at lower prices, combined with
a good processibility characteristics (e.g., less brittle)
•
• efficient cooling for higher current density in generators and converters/cooling
in general in combination with sealed systems (protection against environmen-
tal influences)
•
• new material surfaces (processes and coatings) supporting significant higher
wear resistance and surface pressure (e.g., for more than 25-
year bearing life-
time design and clearly exceeding dynamic stress limits of 1 650 MPa as well](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-319-320.jpg)
![Drivetrain concepts and developments 291
as extreme stress limits of 4 000 MPa [6, 51]). Metal-
composite materials with
higher high-
cycle fatigue capabilities for dynamic loading (e.g., main shaft,
support structures, hub, and gearbox housing)
•
• fatigue strength of connection elements/systems (gluing, screwing, shrinking,
etc.), more reliably quality (process), and monitoring technics
•
• new materials/coatings with high strength and surface wear-
out resistance for
mixed friction operation
•
• further introduction of journal bearing (hydro-
static, -dynamic, combined) for
further critical application in the WT (e.g., for main shaft suspension, yaw-,
pitch-bearings)
•
• wireless energy supply and DD technology for blade-
pitch systems
•
• continuous, in-
situ residual lifetime assessment of individual DT/WT compo-
nents—call it “real” condition monitoring of individual components
Furthermore, there are reasonable assumptions for even broader application of
wind energy utilization in the future. This means some more specific applications,
e.g., in the offshore sector with dedicated hydrogen production in off-
grid operation
directly in the wind farm or the requirements for new operational regimes for WT,
e.g., to compensate for the lack of dynamic operation capability (in terms of degra-
dation) of electrolyzers or for the direct use as mechanical/hydraulically drives for
heat pumps applications. So, at the end, the share of WTs with similar designs to
current turbines dedicated to low wind speed applications could increase, including
special operational strategies, just to achieve the highest capacity factors for specific
applications.
Currently, there seem to be no materials (which have already left the laboratory
stage) available to enable further middle-
term uninterrupted growth of the turbines. In
the very personal opinion of the author, simple upscaling the currently existing and now
rather outbid technologies and platforms of the 15 MW offshore class and 7–8 MW
onshore class makes less sense if the LoCE shall be further reduced according to the
general growth and scaling rules (refer to section 5.3.4). Maybe if also in future even
less sites are available, and the overall installed capacity has to be maximized under this
boundary condition. The not new but interesting idea of Multi-
Rotor-
WT [52] should
not be underestimated for application with special requirements for highly concentrated
power per footprint. The same is for the idea of splitting the rotor power internal into
smaller portions, which are easier to handle with more standardized components (e.g.,
multi-
generator concepts, refer to Chapter 6).
However, the scaling idea could also push a reverse approach, which has so far
been little discussed until now. This is about downscaling, i.e., scaling down the
already optimized technology of the super class turbines in order to use their tech-
nological potential, e.g., in terms of costs not only CAPEX (lightweight design) but
also OPEX (reliability) for special applications, e.g., in the onshore area and markets
or the already mentioned multi-
rotor concepts. This could end up with much smaller
high-
end turbines combining maximum efficiency, reliability, and minimized envi-
ronmental emissions (noise, shadow, and hazards potential). Just as a compromise
in order to take into account the general problem of acceptance through a special](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-320-320.jpg)
![292 Wind turbine system design
design and layout and still generating at an attractive LCoE, especially in densely
populated areas of industrialized nations, i.e., a certain departure from commodity
(low-
cost) WTs toward a technological even more optimized, smarter, renewable
power plant, at least for specific global regions.
Of course, further growth of standard turbines will certainly remain attractive for
other markets (mass markets or markets with other boundary conditions). Future Mega
WT with outputs 20 MW, on the other hand, could remain the clearly favored direc-
tion of developments for floating [53, 54] and autonomously swimming WT (“free
floater”). Here an optimized availability and reliability as well as the ratio between the
mass of the supporting and floating structure and the tower head mass may be more
relevant.
References
[1] Voith Turbo Wind GmbH Co. KG Windstrom einzigartig erzeugen. Die
Windrive-Technologie. technical report; 2018.
[2] Schulze T. Ganzheitliche Dynamische Antriebsstrangbetrachtung von Wind-
energieanlagen, dissertation. Göttingen: Sierke Verlag; 2007.
[3] Rohrig K., Berkhout V., Callies D, et al. ‘Powering the 21st century by wind
energy—options, facts, figures’. Applied Physics Reviews. 2019, vol. 6(3), p.
031303.
[4] Karimi H.R. Structural control and fault detection of wind turbine
systems[online]. IET Energy Engineering, Band 117. ISBN-
10:1785613944;
2018 Aug 31. Available from https://digital-library.theiet.org/content/books/
po/pbpo117e
[5] Wenske J. Direct Drives and Drive-
train Development Trends [Technical
Report]. Fraunhofer; 2012 Apr.
[6] Chovan C., Fierro A.J. Improving Bearing Life in Wind Turbine Main Shafts
and Gearboxes [technical artical, The Timken Company; 2021]. technical
report, 2012.
[7] Bergua R., Jove J., Echarte J. ‘Pure torque drivetrain design: a proven solution
for increasing the wind turbine reliability’. Brazil Windpower Conference;
technical report, 2014.
[8] Guo Y., Bergua R., van Dam J., Jove J., Campbell J. ‘Improving wind turbine
drivetrain designs to minimize the impacts of non-
torque loads’ [online]’.
Wind Energy. 2015, vol. 18(12), pp. 2199–2222. Available from http://doi.
wiley.com/10.1002/we.v18.12
[9] Zhang H., Ortiz de Luna R., Pilas M., Wenske J. ‘A study of mechanical
torque measurement on the wind turbine drivetrain - ways and feasibilities’.
Wind Energy. 2018, vol. 21(12), pp. 1406–1422.
[10] Fanzhong M., Wenske J., Gambier A. ‘Wind turbine loads reduction using
feedforward feedback collective pitch control based on the estimated effec-
tive wind speed’ [online]’. American Control Conference (ACC); Boston,](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-321-320.jpg)
![Drivetrain concepts and developments 293
MA, 2018. Available from https://ieeexplore.ieee.org/xpl/mostRecentIssue.
jsp?punumber=7518121
[11] Srikanth P., Sekhar A.S. ‘Wind turbine drive train dynamic characterization
using vibration and torque signals’. Mechanism and Machine Theory. 2020,
vol. 98, pp. 2–20.
[12] Bartschat A., Morisse M., Mertens A., Wenske J. ‘Analysis of dynamic inter-
actions between different drivetrain components with a detailed wind turbine
model’. Journal of Physics. 2016, vol. 753, p. 082022.
[13] Wenske J., Beckert U. ‘Sensorless control for active damping of torsional
vibrations in wind turbine drivetrains with doubly-
fed induction generator’.
Renewable Energy and Power Quality Journal. 2016, pp. 242–248.
[14] Chen L., Xu H., Wenske J. ‘Active damping of torsional vibrations in the
drive train of a DFIG wind turbine’. Renewable Energy and Power Quality
Journal. 2014, pp. 270–275.
[15] Xu H., Chen L., Wenske J. ‘Active damping control of DFIG wind tur-
bines during fault ride through [online]’. 3rd International Conference on
Electric Power and Energy Conversion Systems (EPECS); Istanbul, Turkey,
2013. Available from https://ieeexplore.ieee.org/xpl/mostRecentIssue.
jsp?punumber=6704878
[16] IEC standard 61400-1:2019-02. Wind energy generation systems – part 1:
design requirements. Geneva, Switzerland: International Electrotechnical
Commission; 2018.
[17] Röder J., Jacobs G., Duda T., Bosse D., Herzog F. ‘Investigation of dynamic
loads in wind turbine drive trains due to grid and power converter faults’.
Energies. 2016, vol. 14(24), p. 8542.
[18] Helsen J., GuoY., Keller J. ‘Gearbox high‐speed‐stage bearing slip induced by
electric excitation in a test facility [online]’. Wind Energy. 2018, vol. 21(11),
pp. 1191–201. Available from https://onlinelibrary.wiley.com/toc/10991824/
21/11
[19] Gallego-
Calderon J., Natarajan A. ‘Assessment of wind turbine drive-
train
fatigue loads under torsional excitation’. Engineering Structures. 2015, vol.
103, pp. 189–202.
[20] Guo Y., Keller J. ‘Combined effects of gravity, bending moment, bearing
clearance, and input torque on wind turbine planetary gear load sharing’.
Conference Paper NREL/CP-
5000-
55968; Dearborn, MI, 2012.
[21] Veritas D.N. Guidelines for design of wind turbines. 2nd Edition. Denmark:
Wind Energy Department, Risø National Laboratory. ISBN 87-
550-
2870-
5;
2002.
[22] Helsen J., Peeters P., Vanslambrouck K., Vanhollebeke F., Desmet W. ‘The
dynamic behavior induced by different wind turbine gearbox suspension
methods assessed by means of the flexible multibody technique’. Renewable
Energy. 2014, vol. 69, pp. 336–346.
[23] Jenkins N., Burton T., Bossanyi E., Sharpe D., Graham M. Wind Energy
Handbook. 3rd ed. Hoboken, NJ: Wiley; 2014 Sep. Available from https://
onlinelibrary.wiley.com/doi/book/10.1002/9781119451143](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-322-320.jpg)
![294 Wind turbine system design
[24] Gasch R., Twele J. (eds.) Windkraftanlagen. Grundlagen, Entwurf, Planung
und Betrieb. 7. Aufl. Wiesbaden: Vieweg + Teubner Verlag; 2011.
[25] Manwell J.F., McGowan J.G., Rogers A.L. Wind Energy Explained. 2nd
Edition. West Sussex: Wiley; 2011. Available from https://onlinelibrary.wiley.
com/doi/book/10.1002/9781119994367
[26] Hau E. ‘Windkraftanlagen, Grundlagen, Technik, Einsatz, Wirtschaftlichkeit.’
4 Auflage. Heidelberg: Springer; 2008.
[27] Hau E., von Renouard H. Wind Turbines, Fundamentals, Application,
Economics. 2nd edition. Berlin, Heidelberg: Springer. ISBN-
10
3-
540-
24240-
6; 2006. Available from http://link.springer.com/10.1007/
3-540-29284-5
[28] Tan J., Zhu C., Song C., Han H., Li Y. ‘Effects of flexibility and suspension
configuration of main shaft on dynamic characteristics of wind turbine driv-
etrain’. Chinese Journal of Mechanical Engineering. 2019, vol. 32(1), p. 36.
[29] Li Y., Zhu C., Chen X., Tan J. ‘Fatigue reliability analysis of wind turbine
drivetrain considering strength degradation and load sharing using survival
signature and FTA’. Energies. 2019, vol. 13(8), p. 2108.
[30] Kirsch J., Kyling H. ‘Optimized cast components in the drive train of wind
turbines and inner ring creep in the main bearing seat’. Forschung Im
Ingenieurwesen. 2021, vol. 85(2), pp. 199–210.
[31] Bork C.-P. ‘Leitfaden für eine betriebsfestigkeitsrechnung’ in Gudehus H.,
Zenner H. (eds.). Materials and corrosion. Vol. 47. 3. Auflage. Düsseldorf:
S., Verlag Stahleisen mbH; 1995. p. 112.
[32] ForschungskuratoriumM. (FKM), Rennert R., KulligE., VormwaldM. (eds.)
“Analytical strength assessment of components in mechanical engineering:
made of steel, cast iron and aluminium materials” [VDMAVerlag]. Frankfurt;
2021. Available from https://d-nb.info/1238617476
[33] Pyttel B., Varfolomeyev I., Luke M., Berger C., Siegele D. ‘Welding in the
world’ in “Fracture mechanics proof of strength for engineering compo-
nents” — overview and extension topics. Vol. 51; 2007. pp. 5.
[34] McEvily A., Suresh S. Fatigue of Materials. Cambridge: Cambridge
University Press. 1992. Available from http://doi.wiley.com/10.1002/adma.
v5:4
[35] Barla S. ‘Global wind turbine technology trends 2021 – market report’ [on-
line]’. Wood Mackenzie. 2021. Available from woodmackenzie.com
[36] Gille D. ‘Schneller schrumpfen. Einige Turbinenbauer erschließen mit weni-
gen Handgriffen andere Leistungsklassen. Eine Wende im Verständnis der
Plattformstrategie’. ERNEUERBARE ENERGIEN. 2011.
[37] Nam J.S., Park Y.J., Kim J.K., Han J.W., Nam Y.Y., Lee G.H. ‘Application
of similarity theory to load capacity of gearboxes’. Journal of Mechanical
Science and Technology. 2010, vol. 28(8), pp. 3033–3040.
[38] Vries E. de WindTech: MHI Vestas goes global with 9.5MW typhoon-
class
V174. Denmark: WindPower Monthly; 2019.
[39] Guo Y., Parsons T., King R., Dykes K., Veers P. An Analytical Formulation
for Sizing and Estimation the Dimensions and Weight of Wind Turbine Hub](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-323-320.jpg)
![Drivetrain concepts and developments 295
and Drivetrain Components [Technical Report]. NREL/TP-
5000-
63008;
2015 Jun.
[40] Klinger F. UPSCAL Beschreibung von Gesetzmäßigkeiten des Größenwachs-
tums von Windenergieanlagen und Lösungsansätze für die sich daraus erge-
benden Problemstellungen, Forschungsbericht. Hochschule für Technik und
Wirtschaft des Saarlandes, University of Applied Sciences; 2002.
[41] Molly J.P. ‘Technical trends in wind turbine development’ [online]’. DEWI
Magazine. 2002. Available from www.dewi.de
[42] Guo Y., Keller J. ‘Investigation of high-
speed shaft bearing loads in wind tur-
bine gearboxes through dynamometer testing’ [online]’. Wind Energy. 2018,
vol. 21(2), pp. 139–50. Available from http://doi.wiley.com/10.1002/we.v21.2
[43] Zhang Z., Matveev A., Ovrebo S., Nilssen R., Nysveen A. ‘State of the art in
generator technology for offshore wind energy conversion systems, 2011’.
IEEE International Electric Machines Drives Conference (IEMDC),
Conference Paper; Niagara Falls, Ontario, Canada, 2011.
[44] Gellermann T. Investigation into the dynamic behaviour of drive trains
in wind energy converters. Bremen: VDI Tagung, Schwingungen von
Windenergieanlagen; 2011.
[45] Vries E. de ‘Top turbines of 2020-
spotlight on new hardware in our annual
awards for the best technology’. WindPower Monthly. 2021, vol. 37(1).
[46] Tsukamoto O. ‘Present status and future trends of RD for HTS rotational
machines in Japan’. Physica C. 2014, vol. 504, pp. 106–110.
[47] Abrahamsen B., Mijatovic N., Seiler E, et al. ‘Superconducting wind turbine
generators, IOP PUBLISHING, supercond’. SciTechnol. 2010, vol. 23(8pp),
034019.
[48] Ohsaki H., Queval L., Terao Y. ‘Design and characteristic analysis of 10 MW
class superconducting wind turbine generators with different types of stator
and rotor configurations’. 4th International Conference on Clean Electrical
Power (ICCEP 2013); Alghero, Italy, 2013.
[49] Liang Y., Rotaru M.D., Sykulski J.K. ‘Electromagnetic simulations of a fully
superconducting 10-
MW-
class wind turbine generator’. IEEE Transactions
on Applied Superconductivity. 2019, vol. 23(6), pp. 46–50.
[50] Le T.D., Kim J.H., Kim D.J., Boo C.J., Kim H.M. ‘Status of the technology
development of large scale HTS generators for wind turbine’. Progress in
Superconductivity and Cryogenics. 2015, vol. 17(2), pp. 18–24.
[51] SKF Group Bearing damage and failure analysis. technical report: PUB BU/
I3 14219/2 EN; 2017 Jun.
[52] Giger U., Kleinhansl S., Schulte H. ‘Design study of multi-
rotor and multi-
generator wind turbine with lattice tower—a mechatronic approach [online]’.
Applied Sciences. 2021, vol. 11(22), p. 11043. Available from https://doi.org/
10.3390/app112211043
[53] Moghadam F.K., Nejad A.R. ‘Evaluation of PMSG‐based drivetrain technol-
ogies for 10‐MW floating offshore wind turbines: pros and cons in a life cycle
perspective’ [online]’. Wind Energy. 2020, vol. 23(7), pp. 1542–63. Available
from https://onlinelibrary.wiley.com/toc/10991824/23/7](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-324-320.jpg)
![296 Wind turbine system design
[54] Nejad A.R., Torsvik J. ‘Drivetrains on floating offshore wind turbines: les-
sons learned over the last 10 years’ [online]’. Forschung Im Ingenieurwesen.
2021, vol. 85(2), pp. 335–343. Available from https://doi.org/10.1007/
s10010-021-00469-8](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-325-320.jpg)
![Gearbox concepts and design 301
Very early Kleinhenz (1937) introduced plans for wind power systems,
which should be able to handle power up to 20 MW [1–3]. In cooperation with
Maschinenfabrik Augsburg-
Nürnberg (MAN) in Gustavsburg, Kleinhenz was able
to further develop the concept of his large-
scale wind power plant in 1938, result-
ing in the MAN-
Kleinhenz project, the technical features of which still appear very
modern today. Namely, a three- or four-
blade rotor with a diameter of up to 130 m
was to sit on a braced tubular steel tower with a hub height of 250 m. The approach
of the gearbox was very interesting. The power distribution was followed by means
of helical gears and four generators. However, the concept was never built.
The very high torque of almost 20 000 kNm at the input shaft of the gigantic
wind turbines would have required a PGT of almost a 4 m large ring gear with a
module of 35 mm (with eight planets), but this was not manufacturable at that time.
Epicyclic gearing was freely used during the course of the 19th century in a wide
variety of industrial applications but did not keep place with the demand for large
powers, because improvements in the technique of cutting external gears outstripped
developments in the production of internal gears. The ingenious gear designer W.G.
Stoeckicht was working on large planetary gearboxes at the same time in 1937 and
awarded a Diploma of Honour (Diplome D’Honneur) at the Paris World Exhibition
for a planetary gear in a 1 400 hp Diesel Locomotive, exhibited by the German
Railways.
Shortly afterward followed a speed reducer for an underwater application of
5 000 hp power. We are talking about a weight of only 1 000 kg and a ring gear
of 0.9 m diameter at relatively high speeds from 3 370 to 580 rpm [4]. However,
such fast speeds at the input shaft for a wind generator do not seem to be appli-
cable. Toward the end of this chapter, let us recall these marginal conditions. The
Stoeckicht design appeared in the wind world later nevertheless in the 1980s. With
very compact gears and little use of materials, he pursed precisely the requirements
that are necessary for a nacelle high above the ground.
An early example with parallel shaft gears can be found in the two wind tur-
bines WE 10/G6 and StGW-
34 from Allgaier Werke, a manufacturer from Germany
and designed by Prof. Hütter. The StGW-
34 rated power was 100 kW. The turbine
started working at a wind speed of 3.7 m/s. The wind turbine was equipped with a
Figure 6.2 Development of turbine size](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-330-320.jpg)
![302 Wind turbine system design
total of two rotor blades, and the diameter was 34 m at a maximum speed of 36 rpm.
The smaller rated power with 6 kW was supplied by the Allgaier WE 10/G6, with
a 78 m² rotor area.
The early gearing concepts, based on parallel shaft gears, are too large and too
heavy as the power increases. Light and compact also apply to wind turbines, as the
weights in a nacelle must be kept to a minimum. It is therefore significant that the
gearboxes that were redesigned in the aircraft industry in the 1950s were often used
as a model for wind energy. The Ivchenko AI-
20 is a Soviet turboprop engine devel-
oped by the Ivchenko design bureau in the 1950s. The compactness of the engine is
achieved by splitting the torque between two gear stages, with the two-
stage plan-
etary gear having a differential stage, as shown in Figure 6.3.
In the 50s of the previous century, a split epicyclic gearbox approach is known
from France, in which two individual stages were spatially separated and connected
with a long coupling. The nominal apparent power of 800 kVA or 650 kW. The pro-
peller was connected to the alternator by a coaxial mechanical linkage involving two
sets of epicyclical spur gears. In order to make the gearboxes favourable as a sin-
gle stage, they were designed as epicyclical gearboxes (Figure 6.4). The turbine was
installed in the field of Institute Aérotechnique from Saint-
Cyr L'Ecole Nogent-
le-
Roi
[5]. The rotation speed of the rotor was 47 rpm. The alternator rotation speed was at
1 000 rpm (invariable as regulated by the 50 Hz frequency of the EDF grid).
An interesting example was built in Holland in the 1980s at STORK-
FDO
(rated power: 300 kW, NEWECS 25) Holland. The planetary gear, flanged to the
double-
bearing main shaft, shows a very modern and up-
to-
date drive technology
(Figure 6.5). The engagement in a ring gear of the high-
speed shaft is chosen quite
favourably in terms of epicyclic technology, which employed an annular gear, with
the intention of taking advantage of that design’s ability to engage more teeth than
Figure 6.3 Planetary coupling gears from aviation 1950s](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-331-320.jpg)
![Gearbox concepts and design 303
the normal spur gear, and therefore reduce the load and wear factor on the individual
teeth. There is a concave contact pressure. And the ring gear serves as a very good
bucket wheel for the lubricating and cooling oil.
The most exciting drivetrain development story in the wind industry in the last
40 years is firmly associated with the 3 MW wind turbine LS-
1 in Burgar Hill,
England, in 1987. 3 MW British Aerospace/GEC/Taylor Woodrow turbine LS-
1
is a production of Wind Energy Group (WEG) Ltd., a manufacturer from the UK
(Figure 6.6). This manufacturer had been in business since 1978. However, WEG
Ltd. has not been in business since 1998. The manufacturer was also acquired by
NEG Micon A/S. The irony of history, the much smiled at Danes with their small
wind turbines 100 kW, which were quickly repaired over the weekend and thus
achieved a fabulous availability, took over the highly convinced megawatt pioneer
wind turbine manufacturer from England.
Frequently cited and reported, this experimental system had been ground-
breaking in the powertrain. Ray Hicks from Wales (GB) was the designer of the
Figure 6.4 L'Eolienne de Nogent Le Roi (France) 1955–1966
Figure 6.5 AII-
PGT gearbox [6]; A external meshing, I internal meshing](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-332-320.jpg)
![306 Wind turbine system design
Somewhat later, an interesting 3 MW design was added by the same gearbox
manufacturer. The version of TGW shown here in the WTS 3 MW wind turbine
was built in a collaboration between Hamilton Standard from the USA and Wind
Turbine Systems Corp. (Swedyards) of Sweden. The overall gear ratio was 1:60.
The large input moments were worked through by a special trick. It was decided to
split the first stage. A common dual-
bearing sun shaft summed the half-
stages and
fed the total torque via the coupled planet carrier into the second planet stage. As
with the 2 MW Tjeaborg machine, the housing was welded. Remarkable was the
planet bearing, this time with the outer rings inserted in the planet.
The MAAG gearbox design in 2003 for the power class looked different from the
state of the art at that time (Figure 6.9). A compound two-
carrier PGT was selected
for the 1.3 MW power rating. In a joint partnership, the bearing company Timken
[7] stood aside and used the newly created Integrated Flexible Pin in both epicyclic
stages. A pseudo-
PGT so-
called ‘Star epicyclic stage’ combines with a differential
Figure 6.8
TGW Thyssen Getriebe- und Kupplungswerke GmbH, Herne,
Germany
Figure 6.9 MAAG DPPV-
7-
79, turbine type: Nordex N-
60](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-335-320.jpg)
![Gearbox concepts and design 307
stage, by a so-
called ‘floating member’, which is a total support (bearing) free cou-
pling shaft. The second stage is formed as a reaction plate, a planet carrier equipped
with up to eight planets, presenting a innovative concept. Two gearboxes were built.
In Orkney (Kirkwall district), the first gearbox is still running in an N60 Nordex
turbine, after almost 20 years.
MAAG has chosen the concept that was already successfully used in 1987 in
the LS1. Of course, it was proposed, designed, and implemented with the help of
Ray Hicks with a very high-
quality standard as in a Swiss watch. The inventor of
the elastic equalizing pin (flexible pin) already had consulting mandates at MAAG
in the 1970s. For a first time, in 1983, MAAG and Hicks worked together in wind
project for a Dutch general contractor FDO for turbines rated at 1 000–1 500 kW
power (NEWECS 45).
Presented at the same time as the MAAG solution in 2003 and also comparable,
Bosch Rexroth also uses a stage as a differential stage in the third stage (1c) in
Figure 6.10. In the same figure, only three planets per stage were deliberately drawn.
For a long time, good load balancing in a rotary stage was considered acceptable
only with three planets. The manufacturer Bosch Rexroth, later taken over by ZF,
built these gearboxes very successfully. The concept is explained in more detail in a
disclosure document DE19963597A1 from 2007.
The colour illustration from the company brochure and an abstract from the
disclosure document DE19963597A1 are listed one after the other. Wolf [9] intro-
duced symbols in 1958 that represent an epicyclic gear unit by a circle with three
lines going outward (the three connecting shafts). These symbols, types and uses are
defined in more detail, e.g., in the VDI 672 guideline. It is not difficult to see how
the layman is easily overwhelmed by Wolf’s scheme. Because also the calculation
of Willis does not contribute necessarily to the dismantling of the question marks, at
that time, Prof. Kiril Arnaudov from Bulgaria [10] developed a simple and practical
teaching method for his students.
A history worth mentioning was the PSC 1002 gearbox from the Jahnel-
Kestermann company in Bochum. The power was 750 kW, and about 570 units
Figure 6.10 GPV gearbox principal Bosch Rexroth [8]](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-336-320.jpg)
![308 Wind turbine system design
were built, mainly for the NEC Micon turbine NM48. The basic design of the
PSC1002 wind turbine gearbox is characterized by the use of a planetary stage on
the input side and two helical gear stages on the output side. The entire gear unit is
connected to the rotor and the machine carrier via a so-
called three-
point support.
This design is used in numerous wind turbines up to approximately 2.5 MW and
has proven itself in compliance with certain narrow bearing clearances in the planet
carrier.
The technical innovations in the PSC1002 wind turbine gearbox are primarily to
be found in the planetary stage bearing concept of the sun gear shaft and were pat-
ented by Jahnel-
Kestermann. On the generator side, the bearing is accommodated in
the housing as standard. The sun shaft was newly supported in the planet carrier on
the rotor side. This mounting was made in a comparatively thin-
walled sleeve on the
planet carrier, which protruded towards the sun wheel. The required adjustment flex-
ibility is achieved exclusively by the clearance of the radial roller bearing, which is
matched to this, and by the deformability of the planet carrier support, which specifi-
cally encompasses the radial roller bearing. As a further advantage, this bearing con-
cept offers the possibility of integrating the wheel of the first spur gear stage on the sun
gear shaft. This eliminates the need for an additional coupling between the planetary
and helical gear stages, thus enabling a shorter gear unit design. The shorter gear unit
design also results in a smaller gear unit housing and thus in weight and cost savings.
The American laboratory NREL has consulted this gearbox PSC 1002 for
extensive investigations (GRC) and modified it several times. Probably, the best-
documented wind turbine gearbox is publicly available through a great number of
measurements and reports [11]. No organization has contributed as much to the bet-
ter understanding of the kinematic relationships in a wind turbine gearbox as NREL
located at the foot of the Rocky Mountains near Boulder, Colorado. Hidden here is
great will for wind power to become better and more reliable. Due to gearbox prob-
lems in the US turbines, even these machines did not reach the required lifetime. At
first, general turbine problems were suspected, not manufacturer-
dependent ones.
The basic quality deficiency was not suspected as the main cause. Even compliance
with the state of the art was not sufficient according to the initial findings, because
problems nevertheless occurred. Furthermore, the bearings were mainly suspected
as the big evil, less the gear teeth. The first culprit was the bearing failures and only
secondarily the teeth, so seen as a secondary cause. And the failures in the 500–700
kW wind turbines with gears continued in the megawatt class. For these reasons, a
large investigation program with its own large test rig and sophisticated measure-
ment technology was started.
In the year 2009, still another built and also patented solution of the company
GGS is presented (Figure 6.11). To avoid indefinitely loaded roller bearings, the
sun of the first stage was mounted on a plain bearing mandrel. On the one hand,
the coupling shaft is guided in this way, and on the other hand, the second stage is
supplied with the necessary lubricating and cooling oil. The solution was exhibited
at the Hanover Fair in the same year. The concept was installed in various licensed
machines in the 2.5 MW power class.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-337-320.jpg)
![Gearbox concepts and design 309
6.3.1
Hybrid systems
Hybrid systems are a middle path between the conventional solution with three gear
stages in the megawatt range and direct drive solutions, which usually require a
generator with a relatively large diameter. The goal is a simpler and more reliable
gearbox with a generator of comparable size, resulting in a dimensionally balanced
and compact powertrain.
Around the turn of the millennium, the medium-
speed concept gained a lot of
attention. Based on the many gearbox failures at the fast output stage, it was learned
that the omission of this HSS stage should result in a considerable increase in avail-
ability. At the same time, permanently excited generators with 150–300 rpm were
developed and directly coupled to the two first gear stages. The first known produc-
tion turbine in the megawatt class was the WinWind WWD-
1 wind turbine from
Finland. The hybrid drivetrain of the WWD-
1 wind turbine consists of a single-
stage planetary gear and a low-
speed synchronous generator. This Multibrid®
con-
cept combines the reliability of a direct drive and the compactness of a gear system.
This period also witnessed the installation of the Falcon 1.25 MW wind turbine in
Nordenham (Germany), which was commissioned in December 2009. The com-
pany GGS from Andermatt [12] developed and built the new patent drivetrain for it
(Figure 6.12).
A compound two-
carrier PGT unit was inserted into the main shaft and sur-
rounded by two main taper roller bearings. The driveline consisted of two stages,
each with seven and five planets. The tooth forces were transmitted on flexible pins,
which ensured perfect load balancing. The gear solution incorporated in the main
shaft is called an integrated tubular gear system (ITGS).
Integrated systems eliminated misalignment between the rotor main shaft
and the gearbox/generator. The same gearbox lubricant was used to lubricate the
enclosed rotor shaft bearings, eliminating the need for grease-
lubricated main bear-
ings. In an integrated system, the seals must be carefully designed because of the
possibility of gearbox oil entering the generator or vice versa and wear debris or
other contaminants entering the gearbox.
Figure 6.11 Compound two-
carrier AI-
PGT; source: GGS](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-338-320.jpg)
![310 Wind turbine system design
Today’s modern drivetrains with gearboxes of the class up to 15 MW mostly
belong to this category and are specially developed combinations of the main shaft,
main bearing, coupling, gearbox and medium generator.
6.3.2
Exceptional developments in the drivetrain
Essentially, these are ways to implement a variable speed in the gearbox that allows
a synchronous generator with sinusoidal power generation to be connected directly
to the output, eliminating the need for an electric converter. We remember that the
LS1 in Orkney demonstrated this in 1987 successfully.
WinDrive from Voith Craislheim was essentially a mechanical solution for
variable-
speed operations, based on a torque converter combined with a planetary
gear based on the Stoeckicht principle. As a fluid machine, the torque converter is
well matched to the wind turbine rotor, and via the fluid in the converter, the system
decouples the input and output shafts by absorbing the peaks in input torque and
providing vibration damping. With the WinDrive solution, the added mechanical
complexity and cost of the transmission system were offset by the elimination of the
cost, mass, and losses of an electric converter. The damping and compliance inher-
ent in the hydrodynamic coupling ensure that a synchronous generator can be used.
Voith technology had long been established in industrial drives, but the wind energy
application presented new challenges, particularly in terms of service life and effi-
ciency, which Voith addressed but ultimately failed to bring to market.
The counterpart to the hydrodynamic solution is the electromechanical solution
of the LS1 drivetrain at first and after also SET company [13]. Again, preferably
externally excited medium-
voltage synchronous generators directly connected to
the mains were to be used. To compensate for the disadvantage of the fixed speed
of synchronous generators directly connected to the mains, a differential PGT was
placed in front of the generator. This system from SET [13] was also difficult to
establish itself on the market. The complexity of controlling the superimposed gear-
box was a major challenge for the turbine manufacturer.
Figure 6.12
1.25-
MW Falcon wind turbine with PMG system (2010); source:
GGS](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-339-320.jpg)
![Gearbox concepts and design 311
6.3.3
A Swiss geared wind turbine
For years, the Swiss landscape has prevented the development of an independent
wind industry, because the claim to protect the landscape has priority over the issue
of domestic renewable energy production. Wind project durations of 20 years are
not uncommon in Switzerland. In this difficult environment, the company GGS from
Andermatt is trying to develop and build its own wind turbine for the demanding
Alpine region.
The distributed generation drivetrain (DGD) drive opens up a very broad solu-
tion (Figure 6.13). The DGD drive reduces the gear load by consistently utilizing
the formula, power P equals speed times torque, and applying it several times at the
gearbox output. At very high speeds, over 5 000 rpm at the generator shafts, up to
12 generators work together in a distribution gear to deliver the incoming power
package from the wind rotor. The two input stages correspond to the template from
the LS1 and the ITGS patent of the company GGS from 2011. It is probably the
most compact power drive in the wind world. Enormous advantages result from
maintenance. A generator–inverter unit weighs only 140 kg. In case of failure or
malfunction, these lightweight units can be replaced within a very short time. The
basic idea was supported in Switzerland, and a report on it was published [14]. Due
to the ultra-
short overall length, the entire drivetrain can be tilted downward by 90°
— just right for installation and maintenance in the Swiss mountains.
6.3.4
State of the art
Exemplary, for the current state of the art in wind power gearboxes to see, shows a
whole gearbox series (Figure 6.14) on the website of the manufacturer Flender (former
Figure 6.13 Tilt turbine Altanus and right, general view of the ground test](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-340-320.jpg)
![312 Wind turbine system design
Winergy). Of course, one learns little about the internal structure of the gearboxes. But,
to achieve more torque per kilogram of gear weight (up to 200 Nm/kg), as mentioned
on the website, between four and seven planets per stage will have to be installed. This
is not new knowledge, let us remember again the LS1 machine. The input torque of
913 kNm was successfully shared among 10 planets. The weight of the gear parts was
4.6 tons, resulting in 198 Nm/kg of torque per kilogram of gear weight, and this was
already in 1987.
The ‘modern’ gearbox concepts actually all go back to existing and proven
gearboxes from earlier wind turbines. For the large power range, these planetary
gearboxes have become established. The transmission of high specific torques/gear-
weight numbers of up to 200 Nm/kg is now state of the art.
Thus, we have arrived in the middle of the subject area. We remember hav-
ing read at the beginning of the subchapter that from the analysis of the existing,
the evaluation of all experiences, the verification of the theoretical knowledge, the
search for new solutions, the testing of new ideas, the product of the future results
with an inventive spirit. To understand the actual gear design, a little more we need
to look at some theories.
6.4
Basic gear tooth design
The calculation and dimensioning of gearings are complex and fill entire books.
It cannot be described in detail in this short publication. For further information,
please refer to the relevant standards and technical literature [15, 16].
The gear wheel is not a spontaneous invention; it is rather a phenomenon (simi-
lar to the wheel) and has evolved over two millennia. It is still evolving today and
is certainly one of the products that can be called ‘high tech’. The famous Basel
mathematician Leonard Euler already recognized the advantages of involute gearing
in 1762. For a uniform and shock-
free transmission of the tooth forces in an engage-
ment of two gears, he referred to the rolling curve ‘circular involute’ as a suitable
shape for the tooth flanks.
The gear mechanism has historically been the most effective and efficient mech-
anism for coupling machines with different optimum speeds. In its simplest form, a
Figure 6.14 Gearboxes for modern wind turbines; © Flender GmbH, 2022](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-341-320.jpg)
![Gearbox concepts and design 313
fixed ratio gearbox consists of a pinion with a smaller number of teeth meshing with
a wheel with a larger number of teeth, whose respective axes are parallel. The ratio
of the number of teeth, diameters, torques of the driven to the driving gears (pinion
and wheel) to each other is equal to the gear ratio. For example, a wheel with 80
teeth drives a pinion with 20 teeth at four times the speed. Thus, with the help of the
number of teeth, the rotation can be reduced or increased accordingly. This rather
simple principle is used in almost all mechanical drives and plays a particularly
important role in our lives. Mechanical clocks may be mentioned as examples.
The total gear ratio of several parallel gears is obtained by multiplying the indi-
vidual gear ratios. Intermediate gears (R gear) only change the direction of rotation.
During tooth meshing, the point of contact of the tooth flank moves along a straight
line, and a sliding and rolling movement takes place between the tooth flanks them-
selves (refer to Figures 6.15 and 6.16). The transmission of torque in tooth meshing
Figure 6.15 Simple wheel and pinion
Figure 6.16 Base tangent contact path [17]](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-342-320.jpg)
![Gearbox concepts and design 315
Without modification, the part of the involute starting directly at the base circle
is used, with modification, a part of the involute further away is selected. This is the
great secret that MAAG accidentally discovered by a mistake in the manufacture
of a gear. Through long trial and error, he subsequently obtained a new strong gear
system that made the teeth at the root stronger. Even today, these addendum modi-
fications are important in choosing the right roll and glide ratios in each tooth mesh.
The calculation of the pressure distribution is performed for each support point
using the formulas according to Hertz [18]. Based on the local line loads, a pressure
distribution is calculated for each load level on the contact lines of the considered
engagement position. Since each discrete individual force Fi from the force vector Fi
acts on a small section Δl of the contact line, it can be assumed for simplicity that the
force is distributed uniformly along the partial section Δl of the contact line. Along
the meshing section, the existing local equivalent radii of curvature ρi
of the tooth
flanks are determined at the local support points of the contact lines in the mesh.
The relative radius is the product of the respective tangent lengths at the points of
contact divided by their sum, i.e., the constant length of the common tangent. For
a given common tangent length, the product of the respective pinion and wheel
tangent lengths would be maximum if they were equal. Of course, this would only
be the case if the pinion and wheel are of the same size. It follows that the relative
radius of curvature is minimum at the lowest point of contact between the gear tip
diameter and the pinion root. However, this lies in the area of the double-
tooth con-
tact, so that the selected load point for calculating the highest surface tension is at
the lowest point of the single-
tooth contact on the pinion flank.
In order to illustrate the essential design criteria for gear teeth, the gears and
their possible damage patterns should be studied first. Classical tooth damage can be
roughly divided into three groups, tooth fracture, pitting, and scoring. Tooth fracture
usually occurs near the base, caused by excessive bending stress. Pitting appears
as more or less extensive material chipping on the load-
bearing tooth flank. The
cause here is excessive pressure on the contact line of the two tooth flanks. This is
associated with excessive stresses immediately below the surface, which causes the
material on the tooth flank to literally flake off. The third cause is called scuffing
and occurs when there is insufficient oil film between the tooth flanks. The oil film
breaks down, and direct metallic tooth contact occurs. This in turn leads to micro-
welding of the material, which is immediately torn out, marking the typical seizure
at the head and root. As a designer, I have a choice of materials and heat treatments
that determine the allowable loads in tooth contact (Figure 6.18). The following is an
excerpt from the old DIN 3990 [4], the great role model of ISO 6336 [19].
The design of a gear must now be optimized in such a way that the desired
safety results with regard to the three damage phenomena. The effects caused by
the three damaging factors often behave in opposite directions when the essential
parameters of a gear are changed, so that an optimum compromise must always be
aimed for (Figure 6.19).
The surface pressure is the criterion that determines the volume of the pitch
cylinder of a gear pair, i.e., the square of the respective diameter multiplied by the
width of the tooth surface. The compressive stress generated by the normal force](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-344-320.jpg)
![318 Wind turbine system design
planetary gearboxes with more than five planets in wind power gearboxes and still
punished them with penalty factors for uneven load sharing. Thus, the weight advan-
tage of many planets was always immediately destroyed.
6.4.1
PGT planetary stage in detail
The planetary gearboxes (PGT) play an important role in wind turbine gearboxes.
However, thanks to their numerous possibilities, these gearboxes cause difficulties
in theory, calculation and design [9, 20–22]. Characteristics are three central shafts,
which together distinguish this compact gear design to the ideal speed or torque con-
verter (Figure 6.21). Thus, the power to be transmitted is distributed over several tooth
meshes. This advantage comes into its own where very high torques have to be trans-
mitted at medium and low speeds. Ideally, the largest possible number of planets is
used in one stage. This makes full use of the compactness. Large overall gear ratios are
then preferably realized in successively compound multi-
carrier PGTs. In each case,
two of the three central shafts are coupled together. There are three possibilities, full
input power, power sharing, or circulating power. In favourable arrangements, the cou-
pling shafts can carry only a partial power, in unfavourable cases, however, a multiple
of the input power as internal power. The theoretical relationships for power flow and
speed ratios have been compiled by Prof. Dr.-Ing. Kirill Arnaudov in a very practical
system [10].
But that is only the theoretical side of the story. In practice, the transmission sys-
tem must be designed to withstand the many loads, including the chosen bearings.
Table 6.1 Post calculation with KMAAG
of historical gear stages
Year 1986 1987 1998
Type/name 1250
DK
LS1
GB
PEAC 4440
Country D
Stage # and type 2 PU* 1 PU 2 PF†
1 PF 1 PU
P, nominal power kW 2205 2205 2252 996 1660
n, wind rotor RPM 21.9 110.8 34.0 607.8 19.0
Mt
, input torque kNm 960 190 912 51 834
Stage ratio - 5.053 5.727 2.767 4.481 5.647
kγ
, mesh load factor 1.15 1.00 1.03 1.03 1.00
b, tooth width mm 250 130 250 240 330
kappa b/d' - 0.81 0.48 0.44 0.88 1.17
m, module mm 16 12 13 10 16
KMAAG
N/mm² 15.02 7.27 4.19 4.61 11.7
σFMAAG
N/mm² 112 68 55 51 83
Gear rim—outside-Ø mm 1500 1440 1760 1380 1520
n, sun RPM 58.2 281.1 109.3 337 48.4
L10h
, planet h 50 367 101 348 2 186 2 435 31 762
Total weight stage kg 2 372 905 3 032 1 572 2 430
*PU annulus fix
†
PF carrier fix](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-347-320.jpg)
![322 Wind turbine system design
derived in each case via the retained planet carrier of the auxiliary stage. This leads
to generally thicker planetary pins in this auxiliary stage, because it has to act for the
whole compound multi-
carrier PGT.
The example in Figure 6.23 will briefly show exactly how much material can be
saved with the right combination of coupling gears and optimum number of planets
in a stage. Three possibilities of a gearbox design with a ratio of i = 40 were com-
pared at the end with the resulting weight. The variation of the number of planets
and the number of stages allows for an increasingly smaller weight. The interesting
thing is that the gears are always designed according to the same safety. This contra-
dicts the opinion of many designers that the translation of many planets in one stage
is a disadvantage. Only the consistent use of the available space with the largest pos-
sible number of planets results in a minimum gear weight. Of course, the stage with
the lowest speed but the highest torque determines the overall diameter. The savings
in gear weight with the same safety factors are considered.
6.4.5
The problem of load distribution and its control
In epicyclic gear units with several planetary gears, the applied load is not distrib-
uted quite evenly among the individual power branches. The unavoidable manu-
facturing deviations within the prescribed tolerances are responsible for this. This
is taken into account in the calculation by the mesh load factor Kγ
[1]. There is a
large number of patented compensation devices. Some of these compensators oper-
ate on a kinematic principle. The other part of the balancers uses the compliance of
certain gear elements for load balancing. The importance of some of these factors
was fully realized by the famous Lanchesters automotive from UK, who referred to
this question in their paper in 1924 [23] on epicyclic gears. They had also grasped
one of the constructions when they stated: The success of any mechanism, however,
well-
conceived, depends finally upon its correctness in detail.
Figure 6.23 Weight reduction due to more planets and stages](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-351-320.jpg)
![Gearbox concepts and design 325
point D. Immediately before this tooth constellation, the entire load is transmitted
by the second tooth pair, the driving gear moves the amount along the line of action.
In order to prevent a shock load, the gear is eased back an amount of tip relief Cαa
.
Modification methods are made mostly in house and belong to the best-
kept
secrets of a company. Simplified lead deformation values can be roughly estimated
from the MAAG Gear book [24]. Each company uses its own calculation method
and implements this in drawing templates. For example, the complex determination
can be read in detail in the work of Reference [25]. From standard ISO 21771-
1,
the correct designations exchange data information can be taken. Especially when
a gearbox fails, these values are often not exchanged. Unfortunately, this causes a
great loss of confidence in the industry. However, the prerequisite remains, to cor-
rectly study and interpret the tooth contact patterns. Gear design is still handcrafted
and requires a large backpack of different experiences.
When determining the microgeometry, it is important to keep an overview.
Especially in wind turbines, stationary gearboxes with one load step are not found.
On a tall slender tower, it vibrates constantly, depending on the tower head mass
even more and now the housing must be considered very flexible and the loads all
constantly changing. This big challenge needs a lot of calculation and test work and
the following support over 20 years. Unfortunately, this is forgotten, although every
gearbox should be designed according to machine guidelines 2006/42/EG [26] and
must be maintained for the entire life of the gearbox, signed by the manufacturer, not
only when it has banged. Therefore, test rigs play a very important role. The calcu-
lated tooth load pattern can be easily checked for plausibility by stepwise increasing
torque load. Of course, the observation of the gearbox up on the tower must con-
tinue, as just mentioned.
If, for example, a gearing correction is calculated for 100% nominal load and
the gear unit is then loaded with a load spectrum of 50–200% of the nominal load,
Figure 6.26
Load variation and profile diagrams for involute spur gears with
tip relief](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-354-320.jpg)
![326 Wind turbine system design
then a different load distribution is produced for each load. As the load increases, the
contact pattern of the gearing increases until, at 100% nominal load, there is a load
distribution that is constant across the width, which generates a face load factor KHβ
close to one for this load level. As the load continues to increase, the contact pat-
tern shifts, for example, to the left side of the gearing, resulting in damaging excess
loads. To illustrate this, Figure 6.27 shows the displacement of the contact pattern
for a load spectrum of 50–200 % of the nominal load.
Noise optimization is also the goal of corrected tooth interventions. However,
the wrongly determined and ground microgeometries are not always the only cul-
prit. Also, helical gears do not always solve the problems. The wrong number of
teeth can generate excitations in a gear stage and amplify them in the turbine through
natural frequencies. Torsional vibration analyses can help here in the design stage.
In planetary stages in particular, the choice of the number of teeth is of decisive
importance. Reference is made here to the relevant technical literature [27].
6.4.8
Absolute, coupling, and relative (rolling) power
When examining the question of power in PGTs, the notions derived from the super-
position ‘Swamp method’ are all useful. Two partial movements (rotations) transmit
their partial powers, e.g., coupling power Pcoup
(NKZ in Figure 6.28), which transmits
by coupling movement when PGT rotates as a coupling (as a whole) without relative
rotation of gears to the carrier. Consequently, this power is assumed to be transmit-
ted without internal losses. Relative (rolling) power Prel
(NBZ in Figure 6.28) or the
power in the mesh that is transmitted by relative movement of the gears with respect
to the carrier. This is the power in PGT after inversion (‘pseudo-
PGT’with fixed car-
rier, we call it also ‘PF’) that causes the meshing losses. These losses (lost power PѰ
)
are considered by the basic loss factor Ѱo
, respectively, and by the basic efficiency
ηo
. Of the above follows: PA
= Pcoup
+ Prel
.
6.5 Bearings
The most commonly used bearing in the wind industry is the rolling bearing. These
bearings are selected for their low friction and high load capacity. The reason for
this is that roller bearings have a higher capacity than larger bearings and are less
expensive. Tapered roller bearings are generally used for combined loads, both radi-
ally and axially.
Figure 6.27 Load distribution over the tooth face (width)](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-355-320.jpg)
![Gearbox concepts and design 327
In general, the design criteria for such bearings lead to a finite life, which
takes into account the total number of hours at different loads [29]. The heaviest
loads are in high-
torque, low-
speed planetary stages and, in particular, the plan-
etary spindles, which must support the sum of the tooth loads from the planetary
meshes with the sun and the ring. The most successful arrangement is a pair of
preloaded tapered roller bearings, which ensure that there is no risk of skidding
under light loads.
Tapered roller bearings are used extensively as mechanical components in most
auto-
moving machinery and must withstand time-
varying loads. The influence of
the preload required for this type of bearing to prevent pitting and fatigue problems
is well known. The aim is to achieve the most even distribution of contact pressure
in the inner and outer raceways. To maximize the available space for a better bearing
capacity design between the bore and the spindle, especially at low ring/sun ratios, it
is helpful to select a fine-
tuned module to increase the root diameter that allows the
planet bore to enlarge. It also helps to omit roller outer rings from bearings and inte-
grate the raceways into the planetary bores. Timken has gone a step further and also
integrated the inner races into the planetary spindle and used full-
roller, preloaded
tapered rollers. All planetary bearings, as well as all other lower-
loaded, higher-
speed bearings in the secondary trains, require a pressurized lubricant supply. No
bearing should be subjected to misalignment, and self-
aligning bearings should be
avoided. They cannot be effectively preloaded because they have a clearance that
can cause skidding at low loads.
The selection and arrangement of bearings are determined based on industry
experience and the recommendations of IEC 61400-
4. The fatigue strength of the
Figure 6.28
Power split and Pcoup,
Prel
according to Reference [28] in compound
two-
carrier three-
shaft PTGs (Diff and Star stage only). Numbers
are from the example in Figure 6.15.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-356-320.jpg)
![Gearbox concepts and design 335
the conspicuous gears. In the past, spherical roller bearings were dispensed with
entirely, paired tapered roller bearings were installed in their place (Figure 6.34),
and additional lubricating oil pumps were installed. It is the transient loads (every-
thing that does not run ‘normally’ is labelled transient) that are responsible for
difficult lubrication conditions in cold, slow-
running wind turbines. In addition,
there were major failures in maintaining the quality of the gearing parts. In the
event of problems with the gearbox, rotor shaft, or rotor bearings, the rotor blades
must usually be removed and the generator, gearbox, and rotor shaft removed as a
unit. The total weight of the assembly must be within the lifting capacity of nor-
mally available cranes and ground conditions. For large WTGs, this can be a major
obstacle to maintenance. For example, repair and maintenance of components in
WTGs with integrated gearbox systems may require removal and replacement of
the rotor blades, which can be significantly delayed and cause considerable down-
time in high winds.
6.12
Standards for load gear units in the drivetrain
Proven national and international standards exist to verify the load-
bearing capac-
ity of the individual components of a gear unit. In the 1980s, systematic damage
occurred to gearboxes in the first major wind farms in the United States, although
there was sufficient dimensioning of the individual components in accordance with
the relevant standards. For this reason, standardized regulations were required to
prove the service life specifically for gearboxes used in WTGs.
The first ‘regulation’ to appear in the United States in 1997 was the AGMA/
AWEA 921-
A97 information sheet [30]. Since information sheets have no legal
relevance, it was decided in 1999 to revise the information sheet and convert it into
Figure 6.34 Modified HSS shaft](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-364-320.jpg)
![336 Wind turbine system design
an American standard. Thus, the first standard published in 2004 was the American
standard ANSI/AGMA/AWEA 6006-
A03 [31]. At present, IEC 61400-
4 ‘Design
requirements for gearboxes’ [32] Edition 2 is being prepared at the international
level in a joint IEC/ISO working group.
The load capacity of helical gears is verified on the basis of ISO 6336 [19].
However, the increased incidence of gear damage, particularly in the early days
of wind turbines, has shown that there is too much room for interpretation in the
application of ISO 6336 or the previous versions for many factors for gears in wind
turbines. For this reason, specific provisions for the application of ISO 6336 have
been included in the wind power standards to narrow this scope for interpretation.
The failure rate of gearboxes that meet the boundary conditions of AGMA 6006, for
example, is significantly lower than that of gearboxes designed in accordance with
ISO 6336 using the scope for interpretation.
The load-
carrying capacity of bearings is generally verified by means of the
‘basic dynamic load rating’ and ‘modified rating life’ approaches described in ISO
281 [33]. The increased incidence of bearing damage, particularly in recent years,
has shown that these methods alone are by no means suitable for life verification and
can only be used for prediction purposes. In the newer wind power standards, there-
fore, the ‘extended contact analysis’ according to ISO 281 Annex B or according to
the procedures of the bearing manufacturers is provided as a service life verification
taking into account the maximum contact pressure. Furthermore, the wind power
standards contain tables for bearing selection, since many of the bearing failures
listed above were due to incorrect bearing selection.
A junior gear designer started 30 years ago with the DIN 3990 [4], lists with
characteristic values, or vice versa with a gear manufacturer his specific basic train-
ing. At the moment, a multitude of gear software is available that a young engineer
has to master. In addition, they are crushed by an enormous burden of standards in
the wind sector. Hand on the heart, who reads and understands standards? Everyone
adorns themselves to know standards, although certain standards are sometimes
contradictory in themselves. For this purpose, a simple understandable gear design
was deliberately pointed out in this chapter.
In the future, the gearbox unit will no longer be considered in isolation but as
part of the entire drivetrain, whose operational safety can no longer be described
solely on the basis of the load capacity verifications of the individual components.
Operational safety will therefore increasingly be assessed on the basis of the results
of dynamic simulations of the entire powertrain. The approach to such simulations
is thus an important point in the further development of standards and guidelines in
the field of wind turbines.
6.13
Gearbox design methodology
For the 7.5-
MW offshore wind turbine of Fraunhofer Institute for Wind Energy
Systems IWES, the main gearbox shall be designed exemplarily with the KMAAG
cri-
terion, recorded and compared and checked after with ISO 6336 [19].](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-365-320.jpg)
![Gearbox concepts and design 337
The specifications of the wind turbine can be found in IWES Wind Turbine
IWT-
7.5-
164 Rev 4-
Wind Energy Report [34]. The geometric design of the gearbox
is based on the GGS in-
house method and international standard IEC 61400-
4 [32].
All gearbox components, i.e., gears, bearings and shafts, are calculated only with
rated mechanical power due to the limited scope in this book.
To determine the maximum safety, the approach to write a simple spreadsheet
that will do the calculations in a matter of seconds and check the layout at the same
time is the fastest way. The selection of the gearbox layout is based on the explana-
tions in this chapter. Experience shows that very few parameters are needed to start
the work.
A good and experienced gear engineer makes their own gears and filing the
teeth by hand. This is a somewhat old English method, but it is all the more valid
today, as software tools suggest how easy it is to construct a gearbox. Instead, with
this proven quick method, it is basically still highly contemporary. As a special fea-
ture, a gearbox layout version is selected with two output shafts. That is, two smaller
generators can be used, which split the input power of the rotor for a generator in
half.
Especially the geometrically necessary manufacturing data are not very interest-
ing at the beginning. Much is determined by the standards. The designer starts the
design with proven data, which he has collected from successful experience gearbox
designs. Load and operating conditions, such as PA
-
input power (A), TB
-working
load (B), T1
-
torque on the sun gear, n1-
rotation speed of sun gear, are the basic
data. For this, minimum geometrical data, such as m-module, aw
-operating centre
distance, z1
, z2
, and z3
—number of teeth, b1
, b2
, b3
—tooth face widths, are selected
and sketched. For the kinematic calculation in the GGS torque–speed spreadsheet,
practical equations are used. With seven formulas, the speeds are easy to calculate.
The balanced gearing in the individual stages with a very high number of planets
(up to 8) is the result of a long iteration process involving drawings and calcula-
tions. If too much stiffness is chosen in the flexible pin, a high uncertainty remains
with respect to the Kγ
mesh load factor. And the reasonable ratio value of the torque
per weight of the stage remains unused. Moreover, the planetary bearing decisively
determines the whole system from inside to outside. A balanced epicyclic stage
requires to consist of well-
balanced gearing, flexible pin and bearing load-
carrying
capacity. The design starts with a consideration of the gear forces in the PGT stage.
This load determines the size (flexible pin in this case) of the planetary stage. It
enables a reasonable choice of stiffness in the epicyclic system. It is flexible enough
to compensate for manufacturing errors of the carrier plate and provide an additional
reserve for load operation.
With the equations shown in Figure 6.35, the required input variables from the
specification are a great help to keep a tabular overview in a simple sketch and table
like in Figure 6.36. In this stage of the design process, calculation is still possible
without a gear program. Building up the sizes of the gear stages volumetrically
based on the permissible Hertzian pressure at the contact point is simple and fast.
The KMAAG
empirical values automatically include limits; with these values, the bend-
ing in the tooth root (tooth fracture), the pressure on the tooth flank (pitting), and](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-366-320.jpg)
![338 Wind turbine system design
via a simple control function also the flash temperature criterion for large modules
are controlled. In slowly, high-
torque-
loaded rotating planetary gears, the mesh con-
tact must be using the flash temperature method derived by Prof. Blok [35]. Later,
further optimization steps are pending, such as noise-
reduced gearings by the small
choice of the module and possibly smaller pressure angle (17.5°) for so-
called high
Figure 6.35 Kinematic relationships in GGS design calculation spreadsheet
Figure 6.36 Torque–speed spreadsheet GGS](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-367-320.jpg)
![Gearbox concepts and design 339
contact ratio gearings, which use transverse contact ratio above 2.0. The asymmetric
GGS flexible pin for PTGs with multi-
planet solutions requires less space but must
be carefully checked against the permissible stresses and stiffnesses with the bound-
ary conditions. From the many possibilities we have learned from past concepts, we
select something that allows minimum weight and maximum safety factors.
The determination of the torque ratio of the compound three-
stage PTG is done
with the torque method [10, 36].
The known and very useful Wolf’s symbol [9] is used, however, with a few
additions. The shafts with the aligned torques, i.e., the sun gear shaft 1 and the annu-
lus gear shaft 3 are marked with single lines (Figure 6.37). Their thickness is based
on the value of the torques, where T1
T3
. The carrier shaft S is marked with double
lines. The carrier S has the greatest, however, opposite torque, which equals that of
the other two torques in absolute terms:
TS =
T1 + T3
(6.1)
A torque ratio t is defined (formed) based on the aligned ideal torques T1
and T3
(without gear loss):
t =
T3
T1
= |
Z3
Z1
| +1
(6.2)
which proves to be very advantageous for the analysis of precisely intricately assem-
bled multi-
carrier planet gears.
By starting with a torque +1 (unity) with one of the sun gear shafts, which is,
however, not a condition and any positive or negative value can be assumed. In the
present case, we start with the output torque TB
of the sun shaft 7 (Figure 6.38) of the
third gear (III) unit, i.e., with the torque method [10, 36]:
TB T7 = +1 (6.3)
Proceed step by step using the individual torque ratios tI
, tII
and tIII
of the three gear
trains I, II and III. The sequence of torque calculation is marked by the green circled
numbers (Figure 6.38). The following must be observed during this procedure. The
two free coupling shafts, i.e., the ring gear shaft 3 with the sun gear shaft 4 and the
sun gear shaft 1 with carrier shaft SIII
, do not have an outward connection. Two
equal but opposite torques, i.e., with different algebraic signs, therefore act on their
ends. The drive input torque TA
and the reaction torque TC
of the connected linkage
shafts—i.e., the carrier shaft SI
with annulus gear shaft 6 and the carrier shaft SII
with
Figure 6.37 Elementary (one-
carrier) planetary gear](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-368-320.jpg)
![Gearbox concepts and design 341
For both cases, the following expressions result in the real torque T3
of ring gear 3:
T3 = oI tI T1 (6.6)
in the first case and
T3 =
tI T1
oI
(6.7)
in the second case.
The determination of the flow direction of the rolling power PW
is in principle
not always straightforward and easy to determine. In such cases, the practical trial
procedures according to Seeliger [21] can be used. In the present case, the determi-
nation of the flow directions of the three rolling powers PWI
, PWII
and PWIII
does not
cause any difficulties, since the carrier SII
is fixed and ring gearIII
9 is also fixed.
To determine the stand efficiencies ηoI
, ηoII
, ηoIII
of the individual PGT, the very
simple formula of Förster [37], which primarily takes the dominant impact of the
number of teeth into account, is used here for lack of space. This formula related to
the first gear unit is as follows:
oI = 1
0.15
1
z1
+
1
z2
+ 0.2
1
z2
1
ˇ
ˇz3
ˇ
ˇ
!#
(6.8)
Further partial gear losses are usually taken into account across the board with a
certain percentage. In the present case, the calculation of with the concrete number
of teeth yields the following numerical value for the individual stationary efficiency
of the three single-
carrier PGTs is:
oI = 98.7%; oII = 98.8%; oIII = 98.5% (6.9)
With the flow directions of the three rolling powers PWI
, PWII
and PWIII
shown, there is
the following relationship between the torques of sun gear shafts 1, 4 and 7 and the
corresponding ring gear shafts 3, 6 and 9:
Figure 6.39 Verification of the efficiency using real torque](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-370-320.jpg)
![Gearbox concepts and design 343
selection is made. For a long time, not all requirements complement each other and can
be processed. Many things simply contradict each other and have to be rejected in excep-
tion lists. In the end, the designer bears the responsibility and no certification company
participates in the certified damage. The design of gears is based on the ISO 6336 [19]
series of design codes. The material properties and heat treatment processes are defined
in ISO 6336 Part 5. The bending fatigue strength and the fatigue strength of the gear
surface are checked using the safety factors recommended in IEC 61400-
4 (Table 6.3).
Before the ISO recalculation of each individual tooth contact, the material char-
acteristics must be determined. Here we adhere to ISO 6336-
5.
Minimum quality requirements for gear components: Q5 external, Q7 internal.
All gears and torque transmitting shafts will be delivered with certification
ISO10474-
3.1C to be in compliance with ME-
quality per ISO6336-
5. The certifica-
tion includes the following documentation:
•
• surface hardness Eht550 and Eht400 determined on a representative test piece
(see ISO 6336-
5 for definition)
•
• surface roughness
•
• grinding temper etch inspection per ISO 14104
•
• magnetic particle inspection
•
• inspection of the microgeometry
ISO 6336 [19] (all parts) consists of international standards, technical specifi-
cations (TS) and technical reports (TR) under the general title: calculation of load
capacity of spur and helical gears.
International standards contain calculation methods that are based on widely
accepted practices and have been validated. TS contain calculation methods that are
still subject to further development. TR contain data that are informative, such as
example calculations. The formulas of the ISO 6336 series are intended to create a
uniformly acceptable method for calculating the load capacity of cylindrical gears
with straight or helical involutes. Several methods for calculating the load capacity
as well as for calculating various factors are permitted. The instructions in ISO 6336
are therefore complex, but also flexible. The formulas include the major factors cur-
rently known to affect gear damage covered by the ISO 6336 series. The formulas
are designed to allow the addition of new factors as new knowledge emerges in the
future.
Table 6.3
The architecture of the can be roughly sketched consisting of a bottom
sensor layer a middle network layer, and a top application layer
Material
Allowable stress for
surface
Allowable bending
stress
Sun, planet:18CrNiMo7-
6 +HH σHlim
= 1 500 N/mm², σFlim
= 500 N/mm²
Ring gear: 34CrNiMo6, nitrated σHlim
= 1 000 N/mm², σFlim
= 370 N/mm²](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-372-320.jpg)
![344 Wind turbine system design
In the above (Figure 6.40), the selected gearbox is shown schematically
and the rough dimensions of the gearings are determined by KMAAG
for all four
stages:
If this range is exceeded, the calculated results must be confirmed by experi-
ence. Design considerations to avoid fractures originating from stress increases in
the tooth flank, chipping at the tooth tip and fractures of the gear blank by the web
or hub, must be analysed using general methods of machine design.
To calculate the safety factors of the gear meshing, the load factors given in the
standard according to ISO 6336-
1 method C have been taken into account, i.e., very
conservative values (Table 6.4).
Figure 6.40 GGS design for the 7.5-
MW gearbox [38]
Table 6.4 K-
factors for ISO 6336 calculation proposed in GGS design
Use parameters applied by GGS Name Value
Application factor KA
for nominal torque KA
1.30
Dynamic factor KV
1.05
Transverse load factor (root stress) KFα
1.00
Transverse load factor (contact stress) KHα
1.00
Face load factor (root stress) KFß
1.15
Face load factor (contact stress) KHß
1.15
Mesh load factor (eight planets) KγGGS
1.23
Mesh load factor (six planets) KγGGS
1.16
Life factor at 10E10 cycles YNT/ZNT 0.85
Pitting safety factor SH2
1.56 (SH 1.25)
Bending safety factor SF 1.56](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-373-320.jpg)
![346 Wind turbine system design
6.14 Future prospects
Gearboxes will maintain an important role in the wind industry. In 2013, GGS pre-
sented a 15 MW mono off-
shore wind turbine in Munich, Germany [39, 40]. The
drivetrain with a novel gearbox system with high ratio 400 used a six-
way power
split at the output for six generators. Much different, but with a much lower cost, the
2021 idea comes to the fore. The power split of the newly proposed GGS multirotor
(MR) is already pulled out five times on the lattice tower and five relatively small
3 MW machines are attached to a large low-
cost lattice tower structure. As read at
the beginning of Kleinhenz 1937, the system of generation of electricity moves in
this order of magnitude 15 MW and consequently again with multiple generators.
This turbine proposal is a real alternative to the high-
cost increasingly large turbines.
The presented GGS monoturbine 15 MW, 2013, in Munich comes to an esti-
mated tower head mass of about 1 000 tons, a single drivetrain of an MR, however,
only 130 tons. Figure 6.41 compares these two different GGS systems. The off-
shore
15 MW machine nearly 10 years ago planned to improve serviceability with a spa-
cious nacelle design. The now envisioned MR machine of the same power size is
built with 5 x 3 MW machines and with 12 superfast generators in each nacelle.
The maintainability is again much easier for the total of 60 distributed superfast
Table 6.5
As a matter limitation of space, only results from single Diff_Stage I,
listed
Particulars of toothing Sun Planet Annulus
Power kW 7 500
Speed rpm 100.4 −96.9 −21.6
Centre distance mm 688
Number of teeth PF-type 59 55 −169
Pitch circle diameter dw mm 711.623 663.377 −2 038.377
Helix angle ° 0
Normal module mm 12.00
Normal pressure angle ° 20.00
Active face width mm 390 385 380
Addendum circle da mm 738.240 684.893 −2 012.026
Dedendum circle df mm 680.640 627.293 −2 066.026
Transverse contact ratio [eps_α] 1.73, 1.88
Safety factor for contact
stress at operating
pitch circle
[SH] 1.44, 1.52/2.54, 1.63
Backlash mm 0.55](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-375-320.jpg)
![348 Wind turbine system design
the other hand, must always be evaluated skeptically. Furthermore, test runs remain
necessary for practical control and verification. Unfortunately, modern simulation tools
are displacing the classic simple methods. But greatest care is required, beware of the
operator!
The most important literature references for the basics are listed in Table 6.6.
References
[1] [ANSI/AGMA 6123-
B06] Design Manual for Enclosed Epicyclic Gear
Drives.
[2] Hertz H.G.W., Gesammelte Werke B., Band I. ‘Miscellaneous’. [Verlag:
Leipzig] 1895.
[3] ‘IEC 61400-
4:2012(en) wind turbines — part 4: design requirements for
wind turbine gearboxes’ in [IEC 61400-
1] International Electrotechnical
Commission. Geneva, Switzerland; 2012.
[4] Stoeckicht W.G. ‘Some advantages of planetary gears’. The American Society
of Naval Engineers Inc. 1948. Available from https://doi.org/10.1111/j.1559-
3584.1948.tb02762.x
[5] Buckingham, E. Manual of gear design. Industrial Press; 1935.
[6] Available from http://eolienne.cavey.org/en/commentaires.php
[7] Förster H. ‘Zur Berechnung des Wirkungsgrades von Planetengetrieben,
Konstruktion’.1969, vol. 5, pp. 165–178.
[8] Berger G. Windenergieanlagen effizienter gemacht. H.6/2005 S.35-
37;
[9] Wolf A. Die Grundgesetze der Umlaufgetriebe. Germany: Friedr. Vieweg
Sohn, Braunschweig; 1958.
Table 6.6 Short recap of relevant literature for the basics of gear design
# Standards used Remark
11 MAAG Taschenbuch Ausgabe 1985 Rough design
6 ISO 6336: Calculation of load capacity
of spur and helical gears (2019) Part
1: Basic Principles, Introduction and
General Influence Factors (1996) Part
2: Calculation of Surface Durability
(Pitting) (1996) Part 3: Calculation of
Tooth Root Bending Strength (1996)
Part 5: Strength and Quality of Materials
(2003)
Single tooth contact…recalculation
1 ISO 9084 and AGMA 6123 Kγ values
36 IEC 61400–4 Safety Factors
ISO 1328–1 Accuracy Grade](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-377-320.jpg)
![Gearbox concepts and design 349
[10] Kril A., Dimitar Petkov K. ‘Planetary gear trains, theory, calculations and
design, load capacity and durability, manufacturing and control, application’.
CRC Press, Boca Raton, Florida. 2017.
[11] Barenhorst Fet al. ‘New drive train concept with multiple high speedgenera-
tor’. J. Phys.: Conf. Ser. 2016.
[12] Giger U., Fox G.P. (eds.) ‘Leistungsverzweigte Planetengetriebe in
Windenergieanlagen mit flexibler Planetenlagerung’.2003.
[13] Hehenberger G. Electro-
mechanical differential drives for wind energy con-
verters. SET Sustainable Energy Technologies GmbH;
[14] Lanchester F.W., Lanchester G.H. ‘DEWI-
magazin’.1923, p. 605.
[15] Merritt H.E. Gears, London: Pittman Sons. 1955.
[16] Shigley J.E. Mechanical engineering design. United states: McGraw-
Hill;
1963.
[17] Jura Ir.G.J., Rademakers Amdrijvingen B.V. Optmalisatie van overbrengin-
gen voor windturbines. Rotterdam: Overdruk uit Aandrijfetechniek; 1983.
[18] Available from https://
buch-
der-
synergie.
de. © 2007 - 2022 achmed A: W.
Khammas Lager
[19] ‘Deutsches Institut für Normung, Berlin: Germany’. Calculation of Load
Capacity of Cylindrical Gears’[DIN 3990:1987]. 1987.
[20] Muller H.W. Epicyclic drive trains. Detroit, MI: Wayne state University
Press; 1985.
[21] Seeliger K. ‘Das einfache Planetengetriebe. Antriebstechnik’. 1964, pp.
216–221.
[22] Berechnungsgrundlagen ‘Planetengetriebe-
Begriffe’. Symbole. 2012.
[23] 2011. ‘Bestimmung von Verzahnungskorrekturen und Lagerkräften in
Planetengetrieben für Lastkollektive’. [Dipl.-Ing. Mohamed Zeyed Sfar].
Dissertation, Bochum
[24] ‘MAAG-
Taschenbuch: Berechnung und Herstellung von Verzahnungen in
Theorie und Praxis — 2’. Erw. u. Erg. Aufl., Maag-
Zahnräder AG, Zürich.
1985.
[25] Guo Y., Keller J. Combined Effects of Gravity, Bending Moment, Bearing
Clearance, and Input Torque on Wind Turbine Planetary Gear Load Sharing.
Michigan: National Renewable Energy Laboratory, W. LaCava University
of Massachusetts, to be presented at the American Gear Manufacturers
Association (AGMA) Fall Technical Meeting Dearborn; 2012.
[26] ‘Richtlinie 2006/42/EG (Maschinenrichtlinie)’.2006.
[27] Hähnel T. Auslegung von Maschinenelementen dynamisch hochbelasteter
Antriebe mittels Messung und Simulation; Fakultät Maschinenwesen der
Technischen Universitäten Dresden, 2009.
[28] Giger U. Entwicklung und Erprobung eines neuartigen Antriebstranges für
die Kompaktwindturbine Falcon; DMK, Dresden, 2009.
[29] ‘Rolling bearings - dynamic load ratings and rating life. Geneva, Switzerland’.
[ISO 281:2007] International Organization for Standardization. 2007.
[30] Kleinhenz F. ‘Das Gross-
Windkraftwerk MAN-
Kleinhenz, Bericht Nr.6 der
RAW’. 1943.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-378-320.jpg)
![350 Wind turbine system design
[31] ‘Design and Specification of Gearboxes for Wind turbines. Alexandria,
Virgina’. [ANSI/AGMA/AWEA 6006-
A03] American Gear Manufacturers
Association. 2004.
[32] IWES wind turbine IWT-
7.5-
164 rev 4. Bremerhaven: Fraunhofer-
IWES.
2018. Available from https://doi.org/10.24406/IWES-N-518562
[33] Rolling bearings — dynamic load ratings and rating life [ISO 281:2007]
International Organization for Standardization; Geneva, Switzerland, 2007.
[34] Giger U., Arnaudov K. International Conference on Gears; Sofia, Bulgaria,
2013. pp. 101.
[35] Blok H. ‘Les températures de surface dans des conditions de graissage
sous pression extrème’ in Congr. mondial du petrole, 2me congr. paris
blitztemperatur;
[36] Arnaudov K., Giger U. ‘High efficiency high torque gearboxfor multi
megawatt wind turbines’. Presented at SCIENTIFIC PROCEEDINGS
VIII INTERNATIONAL CONGRESS MACHINES, TECHNOLОGIES,
MATERIALS; Bulgaria:, 2011.
[37] ‘All parts, calculation of capacity of spur and helical gears’. [ISO 6336:2019]
International Organization for Standardization, Geneva, Switzerland. 2019.
[38] Giger U. Transmission for a Wind Power Installation, Torque Supports, and Wind
Power Installation. [WO 2008/104258 A1].
[39] Available from https://www.aramis.admin.ch/Texte/?ProjectID=36954
[40] Giger U., Kleinhansl S., Schulte H. ‘Design study of multi-
rotor and multi-
generator wind turbine with lattice tower—a mechatronic approach’.Applied
Sciences. 2021, vol. 11(22), p. 11043.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-379-320.jpg)
![Hydraulic systems and lubrication systems 355
The directional valves in general can be distinguished between the directional
seated valve (Figure 7.6) and the directional spool valve.
The directional seated valve technology offers many advantages over direc-
tional spool valves. A spool valve is inherently prone to leakage. The keeping of a
particular position of the consumer over a long time must be realized with additional
check valves. Since oil cannot leak from directional seated valves in the blocked
position, there is no need to use additional check valves. Spool valves can also clog
if they remain in one switching position for a long time. The annular clearance of the
spool valve piston in the housing can be obstructed by dirt particles contained in the
oil. This leads to increased stick-
slip effects and greater hysteresis or – in the worst
case – avoid the switching back into the desired position by the included spring.
Secondary measures such as, e.g. dither frequencies superimposed on the control
signal or an optimized mechanical design can eliminate these disadvantages, but
on the other hand, they cause higher demands and additional effort and costs. Also,
a higher oil purity involves more expensive and more complex filtration measures.
The directional seated valve offers considerably higher switching reliability than the
directional spool valve since clogging with dirt particles circulating in the oil does
not occur. These are flushed out repeatedly when the valve opens.
Anyway, there are several requirements that the brakes must stay safely closed
after a shut-
off for a certain time period until maintenance can be done [1]. Also, for
this, leakage-
free seated valves in combination with an accumulator charging opera-
tion is the optimal choice.
Cartridge valves
The most common design of hydraulic valves is the cartridge design (Figure 7.7).
A cartridge valve is a valve on which the housing is in the form of a cartridge, where
the port openings are aligned with the ports in the cavity of the control block (sec-
tion 7.1.3 Manifold/control block) and which, therefore, can be assembled by simply
screwing it into the control block. Unfortunately, the cavities are different when
talking about different hydraulics manufacturers and so the valves, even when hav-
ing the same function, are not simply interchangeable. Nevertheless, with cartridge
valves the most compact and flexible control block hardware solutions can be build
up easily.
Figure 7.6 Directional seated valve, type NBVP 16 (HAWE Hydraulik SE)](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-384-320.jpg)
![Hydraulic systems and lubrication systems 375
7.2.4
How to engineer a hydraulic pitch system
The key components which need to be properly chosen for a hydraulic pitch system
are the cylinder(s), the pump/motor assembly and the tank size, the accumulator(s)
and the pitch-
valve.
7.2.4.1 Cylinders
A cylinder is characterized by three main dimensions. Diameter of the rod, the diam-
eter of the bore and the stroke.
The main input data from the wind turbine manufacturer are the forces or the
torques which are required to turn the blade against the wind over the blade angle. The
geometric design of the blade adjustment mechanism is needed to get the necessary
stroke of the cylinder. To get the diameters of the rod and the bore, we need the forces
and calculate the necessary pressurized area at the cylinder via the formula 7.1 [2]:
A =
F
p
(7.1)
Certain limits must not be exceeded regarding the system pressure which is related
to the pump design (see Table 7.1) and the limits of the other circuit components. A
reasonable maximal system pressure is 350 bar or less.
If the diameters and the stroke of the cylinder are fixed, a buckling calculation
based on the mechanical installation situation in the hub of the turbine has to be
performed and the dimensioning has to be approved.
7.2.4.2
Pump/motor assembly and tank size
As we have already seen during the dimensioning of the cylinder when fixing the
maximum working pressure of the pitch system, a proper choice of the components
always depends on the rest of the integrated components.
Figure 7.22
Functionality of differential circuity: left, pitch IN; middle, pitching
OUT; right, OUT at grid loss/emergency case (HAWE Hydraulik
SE)](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-404-320.jpg)
![376 Wind turbine system design
When now talking about a suitable motor/pump combination used in the pitch
control power pack in the nacelle, we additionally need to fix the size, respectively
the oil flow Q of the pump.
To do so, another input from the wind turbine manufacturer is required: the
dynamic of the cylinder. What is the maximal velocity v for moving the cylinder IN
and/or OUT? This value, together with the cylinder area A, gives us the necessary
oil flow of the pump as per formula 7.2 [2]:
Q =
v
A
(7.2)
It must be mentioned here that the pump usually is not chosen based on the maximal
flow which has to arrive at the cylinders as the accumulator normally supports the
pump. Therefore, the pump and the electric motor can be downsized.
By the following equation 7.3 [2], we can calculate the necessary power P of the
electric motor which is driving the pump:
P = Q
p
(7.3)
Here the factor p is the nominal pressure to be built up. A reasonable value for the
efficiency ŋ of a “state of the art” pump is 0.75–0.85.
The maximum delivered oil flow by the pump also gives us an appropriate tank
size T which is calculated via formula 7.4:
T = 3...5Q (7.4)
The resulting value is additionally strongly influenced by the heat balance of the
power pack and by the number and the size of the accumulator(s).
7.2.4.3 Pitch accumulators
The tasks of the accumulator(s) in a hydraulic pitch system are as follows:
•
• Ensure “back-
up” energy in a grid loss case to turn the blades in flag position
and bring the turbine to stop.
•
• Support the pump when flow peaks are necessary to fulfill reaction time to
adjust the blades quickly to changed wind conditions.
Table 7.1 Pump type overview
Pump design Maximum pressure (bar) Special feature
External gear pump 210 bar Low costs
Internal gear pump 250 bar Low noise level
Radial piston pump 700 bar Long lifetime
Axial piston pump 350 bar Variable flow](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-405-320.jpg)
![Hydraulic systems and lubrication systems 379
Usually, the control valves receive an input signal from the main wind turbine
controller via Profibus or CAN bus based on the monitoring of the generator output
and other monitored data. The valve flow and performance specifications have to
be matched to the system requirements to be compatible with the existing control
parameters and to co-
exist with the valves on the other axis.
The pitch control valve must be capable of withstanding ambient temperatures,
ranging from −30°C to 50–60°C in the turbine hub and should be designed to com-
ply with the standards for protection against dirt, dust, and moisture [3].
7.2.5
Outlook
The future of the design of hydraulic pitch controls will be driven by cost reductions
on the components side, optimization in manufacturing and production processes
and reduction of assembly effort. Some of their optimizations can be achieved by
the application of “modularization” approaches, breaking down the big system into
smaller sub-systems.
One example is shown below. Such a “pitch module” represents a complete unit
to control the setting the angle for one blade. It is mounted into the bearing between
the hub and the blade. The main power pack to supply the energy (pressure and flow)
or to charge the blade accumulators remains in the nacelle and the oil is transferred
to the hub via a rotary transmission unit.
Figure 7.25 Pitch valve (HAWE Hydraulik SE)](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-408-320.jpg)
![Hydraulic systems and lubrication systems 387
barely any lubricant comes to the bottles. This effect has to be taken into account for
the size of the collection bottles or containers.
To avoid, that the internal back pressure in the connection of the collecting
bottles becomes too high and the grease is pressed out over the sealing system, the
connection borehole at the bearing should be as big as possible (typically M16 × 1.5
mm) – of course without negatively affecting the bearing design – and angular fit-
tings, as well as a long tube from the outlet port at the bearing up to the bottle should
be avoided. Also, the bottle has to be equipped with a ventilation, so that the air
inside the bottle can exhaust when the grease comes to the bottle.
7.3.3
Simplified exemplary design of an automatic lubrication
system
In the following steps, a simplified design of an automatic lubrication system for a
yaw bearing raceway and gearing system is shown. Generally, the design is driven
by the following points:
•
• The necessary re-
lubrication quantities for raceway or teeth lubrication;
•
• Grease type for bearing raceway or bearing teeth;
•
• Maintenance interval, the tank of the lubrication pump should be refilled (half-
yearly, yearly, or for a longer period);
•
• Number of lubrication ports at the bearing/number of lubrication pinions.
With these points, the design of the lubrication system can also be adapted to the
other bearing components.
Following assumptions and yaw bearing specifications are made for the exem-
plary design (Table 7.3).
Table 7.3
Assumptions and boundary condition for the automatic lubrication
system design
Type Value
Type of bearing
Number of lubrication points
Size of lubrication points
Type of grease for raceway
Re-
lubrication quantity raceway*
Gear module
Teeth width
Type of grease for bearing teeth
Re-
lubrication quantity teeth*
Number of lubrication pinions
Maintenance interval
Yaw bearing and double-
row ball bearing
2 × 14, equally spaced
M10 × 1
NLGI class 1 (e.g. Klüberplex BEM 41-
141)
7 kg = 7.78 liter†
16 mm
160 mm
NLGI [4] class 1 (e.g. Klüberplex AG 11-
461)
~4 kg = 4.44 liter†
4
6 months‡
*Amount per year.
†
Liter quantity already calculated with grease density of ~0.9 g/cm³.
‡
+1 month tolerance.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-416-320.jpg)
![Hydraulic systems and lubrication systems 391
7.3.4.1
Final clarifications
7.3.4.1.1 Control software
Depending on using a built-
in controller in the pump, which is driven by a pump and
break interval activities, or if the pump is controlled via the turbine controller, the
programming has to be written, tested, and implemented.
7.3.4.1.2 Electrical connection
The electrical connection of the lubrication pump and the sensors (e.g. empty tank
alarm, cycle switch sensors) has to be clarified, so that it is in line with the electri-
cal setup, respectively with the supply voltage for those components in the turbine.
Also, the plugs, cable types and length must be chosen.
7.3.4.1.3 Mechanical connection
In general, most lubrication system suppliers can offer a mechanical adaption of
their lubrication system by using special designed brackets based on 3D model
design. Of course, those mechanical connection points could also be provided by
the turbine manufacturers themselves.
Particularly when connecting larger lubrication pumps to the hub, a detailed
analysis of the connection points must be taken into account, so that the tank body
gets no defects due to little deformations of the hub and/or the hub internal structure.
References
[1] IEC standard 61400-
1:2019-
02, Wind energy generation systems – part 1:
Design requirements. Geneva, Switzerland: International Electrotechnical
Commission; 2018.
[2] Gieck K., Reiner G. Technische Formelsammlung. 2005, vol. 31.
[3] ‘DIN EN 60529; VDE 0470-
1: 2014-
09, Schutzarten durch Gehäuse (ipcode)
(IEC 60529:1989 + A1:1999 + A2:2013)’. Deutsche Institut für Normung.
2019.
[4] ‘DIN 51818:1981-
12’. [NLGI grades] Lubricants; Consistency Classification
of Lubricating Greases. 1981.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-420-320.jpg)
![Cooling systems concepts and designs 395
In such a dry sump lube system the oil is not directly stored in the gearbox, it accu-
mulates below (the gearbox) in a separate steel reservoir.
The mechanical transmissions within the gearbox cause relevant power losses,
and thus most of these losses have to be transferred to and dissipated by the gear
oil. The typical maximum range of entire losses is between 2% and 3.5% of the
Figure 8.2
Winergy/Flender, gearbox for a GE 1.5 turbine, © Flender GmbH,
2022
Figure 8.3 IEC 61400-
4 standard for wet sump lube systems, refer to [1]](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-424-320.jpg)
![Cooling systems concepts and designs 397
The way this sub-
system works is as follows:
The flow path to the cooler is open all the time. The bypass way to the gearbox
will be continuously controlled by a temperature-
sensitive element by this thermal
valve (refer to Figure 8.3). At about 58°C to 60°C the way to the gearbox is com-
pletely closed. As a side note, the same technology is also used in automotive cool-
ing systems, but of course with different temperature settings.
8.2.1
Filtration
Besides cooling and lubrication, the filtration of the gear oil in wind turbine gear-
boxes is essential. The life time of the bearings today is not only calculated by the
load, but there is also an additional input by the cleanliness about the lube oil (e.g.,
according ISO 281, Figure 8.5). In addition to the lubrication, the oil also has the
task of keeping the particles in suspension and transporting them to the filter system.
These particles are collected in the lube filter and will be removed from the lube
system by the next filter element exchange (service operation).
The target for the general oil cleanliness is during operation −17/15, −/16/14
according to ISO 4406 and it is already known that with wire mesh filter technology
it is not possible to reach any defined ISO classes, because the filtration rate is not
fine enough. Initial contamination means the particles result from the production
and assembling process. So nearly all gearbox manufacturers have their own test
benches for flushing, cooling and filtration units with particle counting according to
the ISO code 4406. A sufficient filtration of the gear oil in wind turbine gearboxes
during operation is also essential for the overall DT reliability. As already men-
tioned, the lifetime of the gearbox bearings is not only calculated for the specific
design load cases (DLCs) but also their corresponding time shares in the anticipated
service life of the system, there is an additional input factor regarding cleanliness for
the lube oil (refer to [2, 3]).
In the beginning of wind turbine gearbox technologies the filtration rate had been
at about 50 microns up to 100 microns nominal, this was realized by wire mash filter
elements. The present design is about 10 microns and made of glass fiber [4, 5]. In lube
systems made by the HYDAC group there are two-
stage filter elements (Figure 8.6).
The two-
stage filter elements have the advantage that, if the bypass valve opens during
Figure 8.5 General definitions for cleanliness and bearing life time calculation](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-426-320.jpg)
![Cooling systems concepts and designs 415
References
[1] ‘Wind turbines – part 4: design requirements for wind turbine gearboxes’. [IEC
61400-
1] International Electrotechnical Commission, Geneva, Switzerland.
2012.
[2] 'Rolling bearings – dynamic load ratings and rating life’. [ISO 281] International
Organization for Standardization, Geneva, Switzerland. 2007.
[3] Dean J.A. Lange’s handbook of chemistry. New York: MacGraw-
Hill Inc.;
1990.
[4] “Hydraulic fluid power – filters – multi-
pass method for evaluating filtration
performance of a filter element”. [ISO 16889] International Organization for
Standardization, Geneva, Switzerland. 2008.
[5] ‘Hydraulic fluid power – filter elements – verification of material compatibil-
ity with fluids’. [ISO 2943] International Organization for Standardization,
Geneva, Switzerland. 1998.
Figure 8.27
Features of a compact, complete ‘smart’gearbox cooling system
(gearbox wet sump lube system) with sensor package, © Hydac
International GmbH, 2022](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-444-320.jpg)
![1
Fraunhofer Institute for Wind Energy Systems IWES (IWES), Bremerhaven, Germany
Chapter 9
Validation, verification, and full-
scale testing
Hans Kyling1
, Anna Wegner1
, Karsten Behnke1
,
Malo Rosemeier1
, and Alexandros Antoniou1
This chapter deals with the basic ideas behind, methodologies used and derived
activities to prove product compliance with stakeholder’s expectations.
9.1
Introduction
For a beginning, the terms used in the chapter title need some explanation. Dependent
on the field of expertise, the industry and the people involved you will get a broad
set of definitions for validation and verification. In this chapter, validation is defined
in accordance with guideline 2206 of Verein Deutscher Ingenieure (VDI) [1] and
cited from the Guide to the Project Management Body of Knowledge (PMBOK) [2]
as the assurance that a product, service, or system meets the needs of the customer
and other identified stakeholders. It often involves acceptance and suitability with
external customers. Whereas verification is often an internal process and understood
as the process to assess whether a product, service, or system complies with a regu-
lation, requirement, or specification.
9.2
Validation and verification strategy
A validation and verification process should accompany any professional product
development process. Some of the advantages coming along with a verified and
validated product are or can be:
•
• Less risk for the manufacturer and user
•
• Shorter time-
to-
market (less intensive field testing)
•
• Better conditions for product financing
•
• Better conditions for insurances linked to the product
•
• Higher product reliability and availability time](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-446-320.jpg)
![418 Wind turbine system design
•
• Create customer confidence
•
• Added value for product advertising
The validation and verification process and the derived activities require expen-
diture in the lower two digit per cent range of the costs for a new turbine develop-
ment. It is challenging to find a good trade-
off between gained benefits and their
related costs.
At this stage, the validation and verification strategy comes into play, which is
understood as the (company) specific approach to deliver the best trade-
off between
the advantages resulting from validation and verification activities and the linked
costs. It is specific to a company or even a smaller unit as the priorities of the exem-
plary advantages given above will vary. A broadly applicable mindset that can be
helpful in developing a validation and verification strategy is, e.g., the Golden Circle
[3]. Here, the starting question is why validation and verification activities should
be conducted and thus will help to prioritize expected outcomes. The next ques-
tion is how the strategy should be implemented. A proven four-
step procedure was
developed within the European Union (EU) funded project ReaLCoE [4] consisting
of following steps:
1. Initialization
2. Detailing
3. Evaluation and decision
4. Implementation
The first step of the procedure, Initialization, starts with the collection of infor-
mation on the to be developed wind turbine (WT) and on past experiences such as
design description, field experiences, and solved and open challenges. With this
knowledge, a qualitative risk evaluation can be implemented for all subsystems and
components. A list of all mandatory (“must”) validation and verification activities
is developed taking into account the certification needs. This list will be comple-
mented by a list of “can” activities that help to further mitigate the risk. By the end
of this phase, a complete and quantifiable list of validation and verification activities
based on the potential risks is defined, and a first review regarding the overall bud-
get, timing, and technical feasibility can be performed (see Figure 9.1).
The next step, Detailing, is an iterative process to elaborate verification and
validation specifications. These contain information as the validation and verifica-
tion aim (functional testing, end of line testing, field measurements, and simulation
report), needed input, requirements, relevant interfaces, and explicit success criteria.
At the same time, the viability of the approach is investigated. For this, aspects as
budget, time, technical feasibility, as well as, e.g., the availability of any involved
test rigs have to be elaborated.
In the third step, Evaluation and Decision, the list of validation and verification
needs is assessed, and the decision on which activities will be performed is taken.
Therefore, the identified risks and their proposed mitigation activities have to be](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-447-320.jpg)
![420 Wind turbine system design
9.3
Purpose of testing
The validation and verification activities defined in the section before can be, e.g.,
analytical proofs, virtual experiments (simulations), hardware tests, or a hybrid
form. In the following, we will deal only with real hardware tests. The specific aim
of conducting experiments using real hardware can be versatile ranging from simple
functional, over model validation and robustness to fatigue tests. The so-
called bath-
tub curve (Figure 9.3) is a representation to help understand the different reasons for
failure and thus test aims. Obviously, one major aim of product development should
be to minimize or at least manage actively the overall number of failures. The con-
tinuous line in Figure 9.3 represents the overall failure occurrence during a product’s
life and has the form of a bathtub. This line results from adding up the three main
types of failures. The early “infant mortality” or “teething failures” can be reduced,
e.g., by functional tests. To minimize the constant failures, e.g., model validation
is a very constructive test type. The failures due to wear out can be addressed by
robustness or fatigue testing.
Not only the expected failure mode and phase of the product lifetime for a
failure to occur are relevant information, when planning the required test activity,
but also the maturity of the product and its underlying technology. For this pur-
pose, the concept of the technology readiness level (TRL) was introduced in the
1970s at NASA [5]. The European Commission (EC) defines the TRLs, as given
in Table 9.1.
A product development based on a predecessor can use partially the results from
earlier validation and verification activity. That is why complex systems such as
airplanes and cars are usually developed in product families.
Figure 9.3
A typical bathtub curve (product failure rate vs. time) and its
composing portions (© Fraunhofer IWES/Kyling, source Wikipedia.
com)](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-449-320.jpg)
![Validation, verification, and full-
scale testing 421
9.4
Product development using the V-Model
The V-
Model as explained in Reference [1] and shown in Figure 9.4 can be used to
derive or to structure the validation and verification activities derived in a separate
procedure as introduced in section 9.2. On the left half of the V-
Model, the product
is decomposed from top-
level system requirements down to material-
level require-
ments. On the right half of the V-
Model, the design properties on different integra-
tion levels are assured by means of validation and verification and integrated into the
overall product property fulfillment. To keep it simple, Figure 9.4 shows only one
test activity on each integration level, but in general, there will be many activities
on each level.
Table 9.1 Definition of TRLs according to EC [6]
TRL Definition
1 Basic principles observed
2 Technology concept formulated
3 Experimental proof of concept
4 Technology validated in lab
5 Technology validated in relevant environment
6 Technology demonstrated in relevant environment
7 System prototype demonstration in operational environment
8 System complete and qualified
9 Actual system proven in operational environment
Figure 9.4
V-
Model with examples from application to a WT (© Fraunhofer
IWES/Kyling)](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-450-320.jpg)
![422 Wind turbine system design
The following chapters are structured and named in accordance with Figure 9.4.
Different exemplary validation and verification activities, devices under test or spec-
imen, and the involved test infrastructure are presented and discussed. If applicable,
the certification relevance of the presented activities is explained. Where possible
and in alignment with the other chapters, the Fraunhofer IWES Wind Turbine (IWT)
7.5 generic model [7] shall be used as an example.
9.5
Full-system testing
9.5.1
Certification measurements
The full system of a modern multi-
megawatt WT can only be tested in the field. This
test presents the final comprehensive check of the complete turbine, with the goal to
verify the design assessment and is an important step within the certification process
of the turbine.
For WTs, the certification measurements follow the standards of the series
61400 of the International Electrotechnical Commission (IEC). The individual
parts most relevant for the full turbine testing comprise IEC 61400-
11 “acoustic
noise measurement techniques” [8], IEC 61400-
12-
1 “power performance measure-
ments of electricity producing wind turbines” [9], IEC 61400-
21 “measurement and
assessment of electrical characteristics” [10], and IEC 61400-
13 “measurement of
mechanical loads” [11]. The latter one is the relevant standard for the structural test-
ing of the turbine. Details of this standard and the application on load validation are
presented in chapter 1 of this book.
Whereas for some components (e.g. a rotor blade test according to IEC 61400-
23 “full-
scale structural testing of rotor blades” [12]), a component test beforehand
is mandatory, for others (e.g. the nacelle as a whole) this is not the case. Testing
components both in field and on the test bench provide the additional opportunity to
validate test procedures. One example of the direct comparison of a nacelle test with
the field tests is given in Reference [13].
Compared with test bench applications, the field test poses additional challenges
on the measurements. First of them is the external environment with its variable and
non-
influenceable conditions. Other than during laboratory component tests, during
the field test, external conditions cannot be influenced but have to be assessed as
precise as possible in order to correctly interpret the data obtained during the field
test. Also, surrounding obstacles influence the measurements in the field.
Prior to any field tests a site evaluation must be performed to assess the site-
specific conditions. Details of the site evaluation are given in the standard IEC
61400-
12-
1 [9]. All obstacles, especially surrounding WTs, are assessed, whether
the wind conditions at the WT site or the position of the meteorological mast are
influenced by an individual obstacle or not. For certification measurements, all wind
directions, which are influenced by obstacles or other WTs, have to be identified.
Data acquired during times with wind direction originating from the disturbed sec-
tor will be excluded from the analysis to obtain an undisturbed measurement sector.](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-451-320.jpg)
![424 Wind turbine system design
beforehand. This method has the large drawback that the nacelle anemometers mea-
sure the disturbed wind field by the rotor.
Other environmental conditions that are typically quantified during a field test
are wind direction, precipitation, temperature, humidity, and air pressure. When
using a hub height mast, the latter three can be measured at hub height, otherwise
they are measured typically at 2 m or 10 m above ground.
The duration of the field tests also depends on the environmental conditions at
the site. To complete the certification measurement of a prototype, a capture matrix,
which is specified in the respective standards, has to be filled. To this end, each
dataset of a 10-
minute time interval is sorted according to the corresponding prevail-
ing wind speed into the respective wind speed interval (so-
called bins). The capture
matrix covers wind speed bins from cut-
in wind speed to rated wind speed. For load
measurements also, certain turbulence intensities are covered by the capture matrix.
This procedure ensures that the measurements cover the main operating range of
the WT.
Another challenge is the time synchronization. A measurement system on a
WT often consists of different sub-
systems, which comprise the data collection of
a certain part in the turbine, e.g., hub with blades, nacelle, and tower sub-
systems.
To ensure a correct analysis of the data, the time stamps of all sub-
systems have to
be synchronized to one common system or to each other. To this end, all data are
typically referenced to the same time stamp, which can be acquired using a Global
Positioning System (GPS) time stamp.
Besides prototype certification, a full-
system testing can be used to provide a
proof of the applicability of the test bench results to full-
turbine operation. In order
to use the results from the test bench, a link between test bench and field measure-
ments has to be established to identify comparable test situations. The main refer-
ence parameter of the field measurements is the incoming wind speed. However, the
wind speed is not available for test bench measurements. The electrical power would
be one approach to overcome this gap. Eustorgi et al. [13] used the rotor speed to
identify comparable datasets from the field test and the test bench experiments. All
approaches present a simplification and have limitations, e.g., neglect the wind pro-
file over the complete rotor. The use of full-
system testing for load model validation
is explained in Chapter 1.
9.5.2
Measurements on the yaw system
In addition to the aforementioned certification relevant works during the full-
system
testing in the field, a lot of further validation activity is typically conducted. As an
example, the often-
neglected yaw system will be discussed in the following lines
regarding field validation to give an impression of the versatile questions the engi-
neers seek to answer and the high level of detail hidden behind all components con-
stituting a WT. The yaw system is explained at length in Chapter 4. The components
of the yaw system are validated to a certain extent depending on the underlying strat-
egy of the WT manufacturer. Typical tests on the yaw bearing are proper functional-
ity of the sealing (also under harsh environmental conditions) and measurement of](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-453-320.jpg)
![Validation, verification, and full-
scale testing 427
have emerged. On the one hand, the parasitic loads can be transferred to the test
specimen by means of hydrostatic power actuators via a rotating disk located on the
drive shaft. Such systems are used, e.g., on the test rigs of the Center for Wind Power
Drives (CWD) in Aachen, Germany [14] or Clemson University in Charleston, USA
[15]. On the other hand, load application systems have been developed in which
hydraulic cylinders apply forces to a steel structure that is connected to the drive
shaft via a roller bearing. Such a load application system is used in the Dynamic
Nacelle Testing Laboratory (DyNaLab), the nacelle test rig of Fraunhofer IWES in
Bremerhaven (see Figure 9.8).
The following list gives a very brief summary of key features of the DyNaLab:
•
• Parasitic load application: application of up to 20 MNm bending moment and 2
MN thrust and shear forces,
•
• Nominal torque: 8.6 MNm (electrically excited synchronous motors in tandem
configuration) and overload torque: 13 MNm
•
• Artificial grid with 44 MVA installed inverter power
•
• Measurement system: more than 600 synchronous, high resolution and fre-
quency measuring channels.
Due to the missing rotor and tower in the laboratory, the nacelle has different
system characteristics on the test stand compared to the field. In order to simulate
real conditions (at least regarding the torsional DOF) in the laboratory, a so-
called
Figure 9.8 DyNaLab nacelle test bench (© IDOM)](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-456-320.jpg)
![428 Wind turbine system design
hardware-
in-
the-
loop operation mode was developed. A real-
time WT model is
simulating for any wind input the resulting load and motion on a defined interface
(e.g. hub flange). The test bench is used to simulate this input accordingly, and the
specimen including the turbine controller will react as in the field (e.g. adjustment
of the pitch angle or generator torque). This reaction is fed back to the real-
time WT
model and, thus, results in an updated input to the specimen from the test bench. The
development of such a real-
time capable WT specific model is complex and needs to
be planned for prior to testing.
A system test bench as explained above cannot only be used to test entire nacelles
but also drivetrains or direct drive generators as the EcoSwing high-
temperature super-
conductor generator [16] shown in Figure 9.9 as the requirements are very similar.
9.6.2
Test requirements
Once the turbine developer has taken the decision to test the nacelle as part of
their individual and project-
specific validation plan, a suited test bench needs to be
selected and its availability contractually secured. In order to do so, the major aims
of the test campaign need to be clarified (see section 9.3) as it influences strongly
the required functionalities and loads of the test bench. For the Fraunhofer IWT 7.5
generic WT, the extreme and damage equivalent loads (DEL) given in Table 9.2 are
assumed on the hub interface of the drivetrain.
Comparing this load set with the operation envelope of the DyNaLab nacelle test
bench leads to the conclusion that the extreme loads cannot be applied as needed.
Consequently, using this system test bench design extreme load tests cannot be a
part of the validation campaign. Whereas the calculated DEL are entirely covered.
Figure 9.9
EcoSwing HTS direct driven generator on DyNaLab nacelle test
bench (© Fraunhofer IWES)](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-457-320.jpg)
![432 Wind turbine system design
9.7
Sub-system testing
On the sub-
system level of the V-
Model (as shown in Figure 9.4), a practical very
relevant validation and verification activity is gearbox testing. In addition, valida-
tion examples of a brake system will be presented.
9.7.1
Gearbox
The international standard IEC 61400-
4 [17] gives guidance on a meaningful valida-
tion approach for WT gearboxes and states mandatory tests required for certifica-
tion. Based on the demands of the gearbox specification, the following stakeholders
are to be involved during the planning and conduction of the validation process: tur-
bine manufacturer, gearbox manufacturer, bearing manufacturer, lubricant supplier,
and certification body. Test criteria and requirements are developed during a design
failure mode and effect analysis and conclude in an overall test plan. This test plan
includes a mandatory unit prototype test of the gearbox, some tests of the gearbox
as an integrated part of the WT as well as requirements for serial production accept-
ance testing. The more of the identified validation activities can be performed on a
gearbox test bench under repeatable and reproducible conditions the better. Each
individual test requires a specific test description containing amongst other purposes
and objectives, acceptance/rejection criteria, environmental conditions, and a list of
physical quantities to be measured. The prototype test results are processed to define
the parameters used for series production acceptance tests. A list of a minimal scope
of testing is given in the standard and includes, e.g., low torque application until
oil cleanliness requirements are met, torque application in a minimum of four load
steps up to nominal torque preferably under rated speed, measurement of actual load
sharing for planetary or other split path gear meshes at each load step and heat runs
under nominal conditions to check for thermal stability.
Major WT gearbox manufacturers dispose of their own test rigs in order to
carry out development, certification-
relevant, and production-
related gearbox tests
internally [18]. These gearbox test benches are usually designed in a so-
called back-
to-
back configuration and feature only the rotational DOF as shown in Figures 9.13
and 9.14 in case of a mechanically closed loop setup.
The back-
to-
back configuration consists of two gearboxes mechanically con-
nected on the low-
speed side (LSS), a driving engine acting on the high-
speed side
(HSS), and in case of a mechanical power circulation, a torque-
tensioning system
might be included as well. The torque-
tensioning system is needed to simulate rated
torque and power conditions while the drive only has to supply the torque to surpass
losses.
Alternatively, the test setup shown in Figures 9.15 and 9.16 can be applied.
In this case, the second HSS shaft is connected to a generator. The power is circu-
lated electrically and consequently requires both machines to operate at the required
power range of the gearbox test, but again only power losses have to be fed to the
system under static operation. Both variants have in common that differing from the
WT field condition of the gearbox the LSS is not directly driven. The reason for this](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-461-320.jpg)
![Validation, verification, and full-
scale testing 437
torque into a gearbox or in the case of a direct driven WT, as the Fraunhofer IWT 7.5
generic, directly into the generator.
The main shaft was focused on two research projects of Fraunhofer IWES [19,
20]. The major aims were to validate existing models used for main shaft design
and during the certification process and compare component lifetime acquired based
on material with full-
scale tests. Consequently, with a validated design concept,
lightweight design toward material savings can be driven. The purpose in this case
is fatigue testing (as described in section 9.3). This leads directly to the question of
how to fatigue test the main shaft. When analyzing all design load cases and their
time distribution during a WT lifetime, it can be found (see Table 9.3) that the bend-
ing loads are dominant for the fatigue driven damage to the main shaft.
Table 9.3 Load DOF and corresponding WT life time damage sum
Load Damage sum [%]
FXR (vertical shear force) 0.0
FYR (lateral shear force) 0.0
FZR (thrust) 0.0
MXR (bending moment) 49.7
MYR (bending moment) 50.3
MZR (torque) 0.0
Figure 9.19
Plot of a test run under driving torque (© Fraunhofer IWES/
Kyling)](https://image.slidesharecdn.com/windturbinesystemdesignvolume1-240715221605-92add2d7/85/Wind-Turbine-System-Design_-Volume-1_Diseno-466-320.jpg)
Wind Turbine System Design_ Volume 1_Diseño
- 1. Wind Turbine System Design Volume 1: Nacelles, drivetrains and verification Edited by Jan Wenske Wenske Wind energy is a pillar of the strategy to mitigate greenhouse gas emissions and stave off catastrophic climate change, but the market is under tremendous pressure to reduce costs. This results in the need for optimising any new wind turbine to maximise the return on investment and keep the technology profitable and the sector thriving. Optimisation involves selecting the best component out of many, and then optimising the system as a whole. Key components are the nacelles and drivetrains, and the verification of their work as a system. Wind Turbine System Design: Volume 1: Nacelles, drivetrains and verification is a valuable reference for scientists, engineers and advanced students engaged in the design of wind turbines offering a systematic guide to these components. Chapters written by industry experts cover load calculation and validation, models and simulation, pitch and yaw system concepts and designs, drivetrain concepts and developments, gearboxes, hydraulic systems, lubrication, and validation. The book aims to enable readers to make informed and systematic choices in designing the best turbine for a given situation. About the Editor Jan Wenske is a professor at the University of Bremen and deputy director of the Fraunhofer- Institute for Wind Energy Systems (IWES), Germany Wind Turbine System Design Volume 1: Nacelles, drivetrains and verification Wind Turbine System Design Volume 1: Nacelles, drivetrains and verification Edited by The Institution of Engineering and Technology theiet.org 978-1-78561-856-7
- 2. Wind Turbine System Design IET ENERGY ENGINEERING SERIES 142A
- 3. Other volumes in this series: Volume 1 Power Circuit Breaker Theory and Design C.H. Flurscheim (Editor) Volume 4 Industrial Microwave Heating A.C. Metaxas and R.J. Meredith Volume 7 Insulators for High Voltages J.S.T. Looms Volume 8 Variable Frequency AC Motor Drive Systems D. Finney Volume 10 SF6 Switchgear H.M. Ryan and G.R. Jones Volume 11 Conduction and Induction Heating E.J. Davies Volume 13 Statistical Techniques for High Voltage Engineering W. Hauschild and W. Mosch Volume 14 Uninterruptible Power Supplies J. Platts and J.D. St Aubyn (Editors) Volume 15 Digital Protection for Power Systems A.T. Johns and S.K. Salman Volume 16 Electricity Economics and Planning T.W. Berrie Volume 18 Vacuum Switchgear A. Greenwood Volume 19 Electrical Safety: a guide to causes and prevention of hazards J. Maxwell Adams Volume 21 Electricity Distribution Network Design, 2nd Edition E. Lakervi and E.J. Holmes Volume 22 Artificial Intelligence Techniques in Power Systems K. Warwick, A.O. Ekwue and R. Aggarwal (Editors) Volume 24 Power System Commissioning and Maintenance Practice K. Harker Volume 25 Engineers’ Handbook of Industrial Microwave Heating R.J. Meredith Volume 26 Small Electric Motors H. Moczala et al. Volume 27 AC-DC Power System Analysis J. Arrillaga and B.C. Smith Volume 29 High Voltage Direct Current Transmission, 2nd Edition J. Arrillaga Volume 30 Flexible AC Transmission Systems (FACTS) Y-H. Song (Editor) Volume 31 Embedded generation N. Jenkins et al. Volume 32 High Voltage Engineering and Testing, 2nd Edition H.M. Ryan (Editor) Volume 33 Overvoltage Protection of Low-Voltage Systems, Revised Edition P. Hasse Volume 36 Voltage Quality in Electrical Power Systems J. Schlabbach et al. Volume 37 Electrical Steels for Rotating Machines P. Beckley Volume 38 The Electric Car: Development and future of battery, hybrid and fuel-cell cars M. Westbrook Volume 39 Power Systems Electromagnetic Transients Simulation J. Arrillaga and N. Watson Volume 40 Advances in High Voltage Engineering M. Haddad and D. Warne Volume 41 Electrical Operation of Electrostatic Precipitators K. Parker Volume 43 Thermal Power Plant Simulation and Control D. Flynn Volume 44 Economic Evaluation of Projects in the Electricity Supply Industry H. Khatib Volume 45 Propulsion Systems for Hybrid Vehicles J. Miller Volume 46 Distribution Switchgear S. Stewart Volume 47 Protection of Electricity Distribution Networks, 2nd Edition J. Gers and E. Holmes Volume 48 Wood Pole Overhead Lines B. Wareing Volume 49 Electric Fuses, 3rd Edition A. Wright and G. Newbery Volume 50 Wind Power Integration: Connection and system operational aspects B. Fox et al. Volume 51 Short Circuit Currents J. Schlabbach Volume 52 Nuclear Power J. Wood Volume 53 Condition Assessment of High Voltage Insulation in Power System Equipment R.E. James and Q. Su Volume 55 Local Energy: Distributed generation of heat and power J. Wood Volume 56 Condition Monitoring of Rotating Electrical Machines P. Tavner, L. Ran, J. Penman and H. Sedding Volume 57 The Control Techniques Drives and Controls Handbook, 2nd Edition B. Drury Volume 58 Lightning Protection V. Cooray (Editor) Volume 59 Ultracapacitor Applications J.M. Miller Volume 62 Lightning Electromagnetics V. Cooray Volume 63 Energy Storage for Power Systems, 2nd Edition A. Ter-Gazarian Volume 65 Protection of Electricity Distribution Networks, 3rd Edition J. Gers Volume 66 High Voltage Engineering Testing, 3rd Edition H. Ryan (Editor) Volume 67 Multicore Simulation of Power System Transients F.M. Uriate Volume 68 Distribution System Analysis and Automation J. Gers Volume 69 The Lightening Flash, 2nd Edition V. Cooray (Editor) Volume 70 Economic Evaluation of Projects in the Electricity Supply Industry, 3rd Edition H. Khatib Volume 72 Control Circuits in Power Electronics: Practical issues in design and implementation M. Castilla (Editor) Volume 73 Wide Area Monitoring, Protection and Control Systems: The enabler for Smarter Grids A. Vaccaro and A. Zobaa (Editors) Volume 74 Power Electronic Converters and Systems: Frontiers and applications A. M. Trzynadlowski (Editor) Volume 75 Power Distribution Automation B. Das (Editor) Volume 76 Power System Stability: Modelling, analysis and control A.A. Sallam and B. Om P. Malik Volume 78 Numerical Analysis of Power System Transients and Dynamics A. Ametani (Editor) Volume 79 Vehicle-to-Grid: Linking electric vehicles to the smart grid J. Lu and J. Hossain (Editors) Volume 81 Cyber-Physical-Social Systems and Constructs in Electric Power Engineering S. Suryanarayanan, R. Roche and T.M. Hansen (Editors) Volume 82 Periodic Control of Power Electronic Converters F. Blaabjerg, K.Zhou, D. Wang and Y. Yang
- 4. Volume 86 Advances in Power System Modelling, Control and Stability Analysis F. Milano (Editor) Volume 87 Cogeneration: Technologies, Optimisation and Implentation C. A. Frangopoulos (Editor) Volume 88 Smarter Energy: from Smart Metering to the Smart Grid H. Sun, N. Hatziargyriou, H. V. Poor, L. Carpanini and M. A. Sánchez Fornié (Editors) Volume 89 Hydrogen Production, Separation and Purification for Energy A. Basile, F. Dalena, J. Tong and T.N.Veziroğlu (Editors) Volume 90 Clean Energy Microgrids S. Obara and J. Morel (Editors) Volume 91 Fuzzy Logic Control in Energy Systems with Design Applications in Matlab/Simulink® İ. H. Altaş Volume 92 Power Quality in Future Electrical Power Systems A. F. Zobaa and S. H. E. A. Aleem (Editors) Volume 93 Cogeneration and District Energy Systems: Modelling, Analysis and Optimization M. A. Rosen and S. Koohi-Fayegh Volume 94 Introduction to the Smart Grid: Concepts, technologies and evolution S.K. Salman Volume 95 Communication, Control and Security Challenges for the Smart Grid S.M. Muyeen and S. Rahman (Editors) Volume 96 Industrial Power Systems with Distributed and Embedded Generation R Belu Volume 97 Synchronized Phasor Measurements for Smart Grids M.J.B. Reddy and D.K. Mohanta (Editors) Volume 98 Large Scale Grid Integration of Renewable Energy Sources A. Moreno-Munoz (Editor) Volume 100 Modeling and Dynamic Behaviour of Hydropower Plants N. Kishor and J. Fraile-Ardanuy (Editors) Volume 101 Methane and Hydrogen for Energy Storage R. Carriveau and D. S-K. Ting Volume 104 Power Transformer Condition Monitoring and Diagnosis A. Abu-Siada (Editor) Volume 106 Surface Passivation of Industrial Crystalline Silicon Solar Cells J. John (Editor) Volume 107 Bifacial Photovoltaics: Technology, applications and economics J. Libal and R. Kopecek (Editors) Volume 108 Fault Diagnosis of Induction Motors J. Faiz, V. Ghorbanian and G. Joksimovic ’ Volume 109 Cooling of Rotating Electrical Machines: Fundamentals, modelling, testing and design D. Staton, E. Chong, S. Pickering and A. Boglietti Volume 110 High Voltage Power Network Construction K. Harker Volume 111 Energy Storage at Different Voltage Levels: Technology, integration, and market aspects A.F. Zobaa, P.F. Ribeiro, S.H.A. Aleem and S.N. Afifi (Editors) Volume 112 Wireless Power Transfer: Theory, Technology and Application N.Shinohara Volume 114 Lightning-Induced Effects in Electrical and Telecommunication Systems Y. Baba and V. A. Rakov Volume 115 DC Distribution Systems and Microgrids T. Dragičevic ’, F.Blaabjerg and P. Wheeler Volume 116 Modelling and Simulation of HVDC Transmission M. Han (Editor) Volume 117 Structural Control and Fault Detection of Wind Turbine Systems H.R. Karimi Volume 119 Thermal Power Plant Control and Instrumentation: The control of boilers and HRSGs, 2nd Edition D. Lindsley, J. Grist and D. Parker Volume 120 Fault Diagnosis for Robust Inverter Power Drives A. Ginart (Editor) Volume 121 Monitoring and Control using Synchrophasors in Power Systems with Renewables I. Kamwa and C. Lu (Editors) Volume 123 Power Systems Electromagnetic Transients Simulation, 2nd Edition N. Watson and J. Arrillaga Volume 124 Power Market Transformation B. Murray Volume 125 Wind Energy Modeling and Simulation Volume 1: Atmosphere and plant P. Veers (Editor) Volume 126 Diagnosis and Fault Tolerance of Electrical Machines, Power Electronics and Drives A.J. M. Cardoso Volume 128 Characterization of Wide Bandgap Power Semiconductor Devices F. Wang, Z. Zhang and E.A. Jones Volume 129 Renewable Energy from the Oceans: From wave, tidal and gradient systems to offshore wind and solar D. Coiro and T. Sant (Editors) Volume 130 Wind and Solar Based Energy Systems for Communities R. Carriveau and D. S-K. Ting (Editors) Volume 131 Metaheuristic Optimization in Power Engineering J. Radosavljevic ’ Volume 132 Power Line Communication Systems for Smart Grids I.R.S Casella and A. Anpalagan Volume 134 Hydrogen Passivation and Laser Doping for Silicon Solar Cells B. Hallam and C. Chan (Editors) Volume 139 Variability, Scalability and Stability of Microgrids S. M. Muyeen, S. M. Islam and F. Blaabjerg (Editors) Volume 143 Medium Voltage DC System Architectures B. Grainger and R. D. Doncker (Editors) Volume 145 Condition Monitoring of Rotating Electrical Machines P. Tavner, L. Ran, C. Crabtree Volume 146 Energy Storage for Power Systems, 3rd Edition A.G. Ter-Gazarian Volume 147 Distribution Systems Analysis and Automation 2nd Edition J. Gers Volume 151 SiC Power Module Design: Performance, robustness and reliability A. Castellazzi and A. Irace (Editors) Volume 152 Power Electronic Devices: Applications, failure mechanisms and reliability F Iannuzzo (Editor) Volume 153 Signal Processing for Fault Detection and Diagnosis in Electric Machines and Systems M. Benbouzid (Editor) Volume 155 Energy Generation and Efficiency Technologies for Green Residential Buildings D. Ting and R. Carriveau (Editors) Volume 156 Lithium-ion Batteries Enabled by Silicon Anodes C. Ban and K. Xu (Editors) Volume 157 Electrical Steels, 2 Volumes A. Moses, K. Jenkins, Philip Anderson and H. Stanbury Volume 158 Advanced Dielectric Materials for Electrostatic Capacitors Q Li (Editor) Volume 159 Transforming the Grid Towards Fully Renewable Energy O. Probst, S. Castellanos and R. Palacios (Editors) Volume 160 Microgrids for Rural Areas: Research and case studies R.K. Chauhan, K. Chauhan and S.N. Singh (Editors) Volume 161 Artificial Intelligence for Smarter Power Systems: Fuzzy Logic and Neural Networks M. G. Simoes Volume 165 Digital Protection for Power Systems 2nd Edition Salman K Salman
- 5. Volume 166 Advanced Characterization of Thin Film Solar Cells N. Haegel and M Al-Jassim (Editors) Volume 167 Power Grids with Renewable Energy Storage, integration and digitalization A. A. Sallam and B. OM P. Malik Volume 169 Small Wind and Hydrokinetic Turbines P. Clausen, J. Whale and D. Wood (Editors) Volume 170 Reliability of Power Electronics Converters for Solar Photovoltaic Applications F. Blaabjerg, A.l Haque, H. Wang, Z. Abdin Jaffery and Y. Yang (Editors) Volume 171 Utility-scale Wind Turbines and Wind Farms A. Vasel-Be-Hagh and D. S.-K. Ting Volume 172 Lighting interaction with Power Systems, 2 volumes A. Piantini (Editor) Volume 174 Silicon Solar Cell Metallization and Module Technology T. Dullweber (Editor) Volume 180 Protection of Electricity Distribution Networks, 4th Edition J. Gers and E. Holmes Volume 181 Modelling and Simulation of Complex Power Systems A. Monti and A. Benigni Volume 182 Surge Protection for Low Voltage Systems A. Rousseau (Editor) Volume 184 Compressed Air Energy Storage: Types, systems and applications D. Ting and J. Stagner Volume 186 Synchronous Reluctance Machines: Analysis, optimization and applications N. Bianchi, C. Babetto and G. Bacco Volume 191 Electric Fuses: Fundamentals and new applications 4th Edition N. Nurse, A. Wright and P. G. Newbery Volume 193 Overhead Electric Power Lines: Theory and practice S. Chattopadhyay and A. Das Volume 194 Offshore Wind Power Reliability, availability and maintenance, 2nd edition P. Tavner Volume 196 Cyber Security for Microgrids S. Sahoo, F. Blaajberg and T. Dragicevic Volume 198 Battery Management Systems and Inductive Balancing A. Van den Bossche and A. Farzan Moghaddam Volume 199 Model Predictive Control for Microgrids: From power electronic converters to energy management J. Hu, J. M. Guerrero and S. Islam Volume 204 Electromagnetic Transients in Large HV Cable Networks: Modeling and calculations Ametani, Xue, Ohno and Khalilnezhad Volume 208 Nanogrids and Picogrids and their Integration with Electric Vehicles S. Chattopadhyay Volume 211 Blockchain Technology for Smart Grids: Implementation, management and security Gururaj H L, Ravi K V, F. Flammini, H. Lin, Goutham B, Sunil K. B R and C Sivapragash Volume 212 Battery State Estimation: Methods and Models S. Wang Volume 215 Industrial Demand Response: Methods, best practices, case studies, and applications H. H. Alhelou, A. Moreno-Muñoz and P. Siano (Editors) Volume 213 Wide Area Monitoring of Interconnected Power Systems 2nd Edition A. R. Messina Volume 217 Advances in Power System Modelling, Control and Stability Analysis 2nd Edition F. Milano (Editor) Volume 225 Fusion-Fission Hybrid Nuclear Reactors: For enhanced nuclear fuel utilization and radioactive waste reduction W. M. Stacey Volume 238 AI for Status Monitoring of Utility Scale Batteries Shunli Wang, Kailong Liu, Yujie Wang, Daniel-Ioan Stroe, Carlos Fernandez and Josep M. Guerrero Volume 905 Power system protection, 4 volumes
- 6. Wind Turbine System Design Volume 1: Nacelles, drivetrains and verification Edited by Jan Wenske The Institution of Engineering and Technology
- 7. Published by The Institution of Engineering and Technology, London, United Kingdom The Institution of Engineering and Technology is registered as a Charity in England & Wales (no. 211014) and Scotland (no. SC038698). © The Institution of Engineering and Technology 2022 First published 2022 This publication is copyright under the Berne Convention and the Universal Copyright Convention. All rights reserved. Apart from any fair dealing for the purposes of research or private study, or criticism or review, as permitted under the Copyright, Designs and Patents Act 1988, this publication may be reproduced, stored or transmitted, in any form or by any means, only with the prior permission in writing of the publishers, or in the case of reprographic reproduction in accordance with the terms of licences issued by the Copyright Licensing Agency. Enquiries concerning reproduction outside those terms should be sent to the publisher at the undermentioned address: The Institution of Engineering and Technology Futures Place Kings Way, Stevenage Herts, SG1 2UA, United Kingdom www.theiet.org While the authors and publisher believe that the information and guidance given in this work are correct, all parties must rely upon their own skill and judgement when making use of them. Neither the author nor publisher assumes any liability to anyone for any loss or damage caused by any error or omission in the work, whether such an error or omission is the result of negligence or any other cause. Any and all such liability is disclaimed. The moral rights of the author to be identified as author of this work have been asserted by him in accordance with the Copyright, Designs and Patents Act 1988. British Library Cataloguing in Publication Data A catalogue record for this product is available from the British Library ISBN 978-1-78561-856-7 (hardback) ISBN 978-1-78561-857-4 (PDF) Typeset in India by Exeter Premedia Services Private Limited Printed in the UK by CPI Group (UK) Ltd, Croydon Cover Image: Vinzo via Getty Images
- 8. Contents 1 Load calculation and load validation 1 Philipp Thomas, Mareike Leimeister, Anna Wegner, and Matthias L. Huhn 1.1 Design loads of wind turbines 3 1.1.1 Standard load calculation 4 1.1.2 Use cases and exemplary loads 11 1.2 Design load validation 16 1.2.1 Standard load measurements 16 1.2.2 Data evaluation process 21 1.2.3 Standard load validation 21 Acknowledgements25 References 25 2 Models and simulation 27 Paul Robert Feja, Mareike Leimeister, and Muhammad Omer Siddiqui 2.1 Introduction 27 2.1.1 Overview of modelling at different levels of fidelity 28 2.1.2 Requirements of standards for model fidelity 29 2.2 Modelling of environmental conditions 31 2.2.1 Modelling of wind conditions 32 2.2.2 Modelling of sea conditions 36 2.2.3 Modelling of soil conditions 39 2.3 Fully coupled wind turbine modelling 39 2.3.1 Aeroelasticity and standard tools 40 2.3.2 Aerodynamic models 40 2.3.3 Hydrodynamic models 53 2.3.4 Modelling of structural components 54 2.3.5 Modelling of other components 58 2.4 Detailed modelling of wind turbine drivetrains 58 2.4.1 General modelling approaches, methods and tools 59 2.4.2 Different approaches of modelling a wind turbine drivetrain 61 2.4.3 Modelling recommendations and best practices 66 2.5 Conclusion and summary 68 References 68 About the Editor xiii Prefacexv Abbreviations and Terminologies xxv
- 9. viii Wind turbine system design 3 Pitch system concepts and design 75 Karsten Behnke, Arne Bartschat, Eike Blechschmidt, Matthis Graßmann, Florian Schleich, Oliver Menck, and Heiko Jungermann 3.1 Blade bearing 78 3.1.1 Preliminary outer bearing design 79 3.1.2 Preliminary inner bearing design 84 3.1.3 Preliminary design of the bolted connections 89 3.1.4 FE blade bearing model 93 3.1.5 FE simulation of internal blade bearing loads 98 3.1.6 Calculation and dimensioning 101 3.1.7 Lubrication system 110 3.1.8 Coating 113 3.2 Pitch actuator 114 3.2.1 Electrical actuator 114 3.2.2 Operating conditions 118 3.2.3 Calculation and dimensioning 118 References 121 4 Yaw system concepts and designs 125 Christian Bulligk and Daniel von dem Berge 4.1 Fundamentals 125 4.1.1 Introduction 125 4.1.2 Wind direction and yaw misalignment 129 4.1.3 Typical key data 131 4.2 Design loads 133 4.2.1 Introduction 133 4.2.2 Yaw bearing loads 135 4.2.3 Yaw drivetrain aerodynamic loads 139 4.2.4 Loads acting on the yaw drivetrain 143 4.2.5 Modification of yaw drivetrain aerodynamic loads 146 4.2.6 Yaw slippage events during non-yawing operation 148 4.2.7 Overload events during yawing operation 150 4.2.8 Yaw start and stop events 152 4.3 System concepts and components 153 4.3.1 Differentiating features at system level 153 4.3.2 Yaw bearing 156 4.3.3 Yaw brake system 164 4.3.4 Yaw gearbox 167 4.3.5 Yaw motor and yaw motor brake 172 4.3.6 Auxiliary systems 176 4.3.7 Evaluation criteria 178 4.3.8 Common system concepts 180 4.4 System dimensioning and design aspects 181 4.4.1 Introduction and general requirements 182 4.4.2 Step 1: yaw system, holding torque and driving torque 185 4.4.3 Step 2a: yaw bearing, yaw brake and yaw drive 188
- 10. Contents ix 4.4.4 Step 2b: dimensioning of the yaw brake system 190 4.4.5 Step 2c: dimensioning of the yaw bearing 192 4.4.6 Step 2d: dimensioning of the yaw drive system 197 4.4.7 Step 3: auxiliary systems 202 4.4.8 Summary 202 References 205 5 Drivetrain concepts and developments 207 Jan Wenske 5.1 Fundamentals 207 5.2 Drivetrain concepts 210 5.2.1 Drivetrain diversification and classification 210 5.2.2 Drivetrain concepts and design principles 216 5.3 General design rules and procedures 237 5.3.1 Safety, protection, reliability and control 238 5.3.2 Loads and load cases 244 5.3.3 Loads analysis and strength verification 249 5.3.4 Modularization, standardization, and platform concepts 258 5.3.5 Scalability of designs and performance indicators 263 5.4 Onshore wind turbines and drivetrain developments 270 5.4.1 ENERCON 271 5.4.2 Nordex 272 5.4.3 General Electric wind energy (GE) 274 5.4.4 Vestas 275 5.4.5 Siemens Gamesa Renewable Energy 277 5.5 Offshore wind turbines and drivetrain developments 279 5.6 Outlook and potential development trends 289 References 292 6 Gearbox concepts and design 297 Urs Giger 6.1 Introduction 297 6.2 Challenge for load gearboxes in wind turbines 298 6.3 Historical drivetrains in wind turbines 300 6.3.1 Hybrid systems 309 6.3.2 Exceptional developments in the drivetrain 310 6.3.3 A Swiss geared wind turbine 311 6.3.4 State of the art 311 6.4 Basic gear tooth design 312 6.4.1 PGT planetary stage in detail 318 6.4.2 PGTs have a number of advantages and applications 319 6.4.3 Difficulties in using PGTs 320 6.4.4 Increasing the power sharing 320 6.4.5 The problem of load distribution and its control 322 6.4.6 The load-sharing measurement 323
- 11. x Wind turbine system design 6.4.7 Microgeometry 324 6.4.8 Absolute, coupling, and relative (rolling) power 326 6.5 Bearings 326 6.5.1 Bearing failure mechanisms 329 6.6 Coupling 329 6.7 Mechanical brakes 330 6.8 Lubrication system and its design principles 330 6.9 Bolted joints 332 6.10 Pitch tube 333 6.11 Repair work 334 6.12 Standards for load gear units in the drivetrain 335 6.13 Gearbox design methodology 336 6.13.1 Oil quantities and power losses 342 6.13.2 Calculation of gearing according to ISO 6336 standard (Part 1–6) 342 6.14 Future prospects 346 6.15 Conclusion 347 References 348 7 Hydraulic systems and lubrication systems 351 Andreas Nocker, Arved Hildebrandt, Christian Bulligk, and Daniel von dem Berge 7.1 Hydraulic systems 351 7.1.1 Main Components 352 7.1.2 Hydraulic auxiliaries 356 7.1.3 Manifold / control block 358 7.1.4 Centralized and decentralized systems 359 7.1.5 How to engineer a hydraulic power pack 360 7.2 Hydraulic pitch systems 365 7.2.1 History 365 7.2.2 Pitch control 365 7.2.3 Hydraulic pitch adjustment systems 370 7.2.4 How to engineer a hydraulic pitch system 375 7.2.5 Outlook 379 7.3 Automatic lubrication system for bearings 380 7.3.1 Fundamentals 380 7.3.2 Components of an automatic lubrication system 382 7.3.3 Simplified exemplary design of an automatic lubrication system387 7.3.4 Schematic overview and final clarifications 389 References 391 8 Cooling systems concepts and designs 393 Ernst-Wilhelm Langhoff 8.1 Introduction 393
- 12. Contents xi 8.2 Gearbox 394 8.2.1 Filtration 397 8.3 Generator 398 8.4 Main converter 400 8.5 Main transformer 403 8.6 Essential questions for cooling system design 404 8.7 Example – cooling design for IWT-7.5-164 variant 405 8.8 Experiences 412 References 415 9 Validation, verification, and full- scale testing 417 Hans Kyling, Anna Wegner, Karsten Behnke, Malo Rosemeier, and Alexandros Antoniou 9.1 Introduction 417 9.2 Validation and verification strategy 417 9.3 Purpose of testing 420 9.4 Product development using the V-Model 421 9.5 Full-system testing 422 9.5.1 Certification measurements 422 9.5.2 Measurements on the yaw system 424 9.6 Integration testing 426 9.6.1 System test benches 426 9.6.2 Test requirements 428 9.6.3 Projecting a nacelle test campaign 429 9.7 Sub-system testing 432 9.7.1 Gearbox 432 9.7.2 Brake system 435 9.8 Component testing 436 9.8.1 Main shaft 436 9.8.2 Pitch bearing 439 9.8.3 Rotor blade 442 9.9 Material testing 444 9.9.1 Leading edge protection 445 9.9.2 Polymer and composite testing 445 9.10 Outlook 446 References 447 10 Main shaft suspension system 451 Marc Reichhart, Tobias Baumgratz, and Clemens Brachmann 10.1 Introduction and bearing arrangement selection 451 10.1.1 Cylindrical roller bearings 452 10.1.2 Spherical roller bearings 452 10.1.3 Toroidal roller bearings 453 10.1.4 Tapered roller bearings 453 10.1.5 Moment bearings 453
- 13. xii Wind turbine system design 10.1.6 Bearing type and bearing arrangement selection 454 10.1.7 Bearing type selection in relation to the drivetrain concept 457 10.1.8 Influence of turbine size on rotor bearing size and type 459 10.2 General design and bearing calculation process 461 10.2.1 Drivetrain for calculation example 461 10.2.2 Calculations according to applicable standards and guidelines462 10.2.3 Rated life calculation 462 10.2.4 Contact stress 465 10.2.5 Static safety 467 10.2.6 Loads for rotor bearing calculation 468 10.2.7 Extreme loads for rotor bearing calculation 469 10.2.8 Fatigue load cases for bearing calculation 470 10.2.9 Bearing calculation models and software 473 10.2.10 Rigid calculation model 473 10.2.11 Calculation with the stiffness matrix 474 10.2.12 Calculation with non-linear stiffness (FE calculation) 476 10.3 Example for rotor bearing calculation 477 10.3.1 Influence of calculation model and boundary conditions 479 10.3.2 Definition and influence of bearing system preload 481 10.4 Reliability, failures and root causes 483 10.5 Development trends 486 References 487 Index489
- 14. About the Editor Jan Wenske is a professor at the University of Bremen and deputy director of the Fraunhofer-Institute for Wind Energy Systems (IWES), Germany. He received his PhD in 1999 at the Institute of Electrical Engineering at the TU Clausthal. Professional assignments from 2000-2010 included advanced development for elec- trical drives at the STILL GmbH and the Power Electronics Development at Jenoptik Defense Civil Systems. He works intensively with major industry players in wind energy technology.
- 15. This page intentionally left blank
- 16. Preface Synopsis The work in the field of wind energy use and the development of modern on- as well as offshore wind turbines has long since arrived in its own very special, complex and multidisciplinary research universe. Due to the enormous progress, in model- ling, simulation, experimental validation and measurements, it is now possible to model the subsystems of a wind power turbine in detail, to analyse it in the multi- domain realm. On the other hand, to devise a real turbine design based upon the model, study the construction about the highly complex interaction between the subsystems, and feed lessons learned back to the construction process. The turbine manufacturers have established and optimised their internal processes, however sometimes adjusted them primarily to formally meeting the requirements of the certifying authorities. Currently, at least onshore turbines as a still highly complex product are in permanently danger of becoming a commodity. Means a product, which perceived by the customers as nearly identical regardless of the manufacturer, with the price as the only easily distinguishable attribute, even though substantial differences in detail do exist. There is already an enormous pressure on manufactur- ers to reduce costs. Evidence for this is the rapid consolidation movement on the side of the manufacturers as well as of the equipment suppliers and introduction of new platforms, modifications and variants in always-shorter periods. For the wind industry, continuous innovation is essential to escape from this situation. Research institutions and universities try to support with specific expertise in various field of research from wind physics (explaining the character of wind with its turbulences as a complex, stochastic process and the atmosphere), about turbine technology and controls, aerodynamics, civil and ocean engineering, grid integration down to material science. The offshore industry has also developed rapidly and optimized all parts of the value chain within a few years, especially in the areas of logistics, founding and construction. Reducing costs to achieve even lower LCoE is still the main driver for technical developments. A few years ago, operators of new offshore wind farms still expected that the rated output of next-generation offshore wind turbines would almost double to around 15 MW within the next five years in order to be able to operate in a free electricity generation market without subsidies, and it really hap- pens. In the face of these rapid developments in the market industrial, application- oriented research on the one hand, and academic research in wind power on the other hand should be coupled. Appropriate specialist literature available to students, PhD
- 17. xvi Wind turbine system design candidates, industry experts and interested laypeople or carrier changes is always necessary and one key enabler for the motivation to dig deeper in the fantastic world of wind turbines. There is a lot of standard literature available for wind energy utili- zation and turbine technology. Mostly they aim at providing the readers with a first general introduction and broad overview about the topic of wind energy. Also in the specific areas of simulation, modelling, control, aerodynamics, hydro-and aero- elastic, fibre based materials, and structural dynamics in general a broad range of specialised literature is available as well. So why to write a small book series on turbine technology, particularly with a focus on power mechatronics of the drivetrain and the nacelle systems? Because the authors and the editor are of the opinion that there are still gaps in the literature for application oriented readers from engineering, mechatronics and electronics, i.e. written for novices, engineers, scientists and students engaged in applied research in wind power. With the goal to establish, the link between foundations and practical application under consideration of boundary conditions (such as construction related, system related, or economical), this book shall offer answers to questions like: • • How to configure a main bearing arrangement? • • Which loads are design relevant for different subsystems? • • Why especially the pitch system is such a critical turbine subsystem? • • What is essential for designing a gearbox for wind turbines? • • How to deal with the losses within the turbine? • • Which are boundary conditions to configure a Direct-Drive generator? • • What are the secrets of a proper DFIG system design? • • What is a modern, generic physical controller design for a wind turbine? • • What is the status of current CMS and SHM technologies and how to integrate them? • • Is an extensive test and validation program reasonable? • • Is the next challenge still the turbine or its grid integration? • • … The reader shall develop a sense for the concrete, practical design work. Of course, also this book cannot cover each design process to the minute detail. The goal is rather to describe the processes using examples, referencing, where applicable and necessary, to further literature or existing requirements and standards. Having read this book, the reader will be capable of understanding the designing of specific com- ponents and subsystems of modern wind turbines or to specifying them in more detail, and have a sound understanding of boundary conditions, dos and don’ts, system interaction and requirements in the design process of nacelle systems and drivetrains. Some chapters describe less the practical design process, but providing rather an application-oriented overview of the state of technology and research, such as CMS/SHM, controls and signals and drivetrains concepts. A small group of experts wrote therefore each, individual chapter. The teams are professionals from industry, experienced researcher or a mixed team always with a significant share of authors with industry expertise to ensure the practical relevance
- 18. Preface xvii of the contents. Each chapter will convey briefly the necessary basics, sometime a historical overview, examples for design processes, requirements, challenges, opti- misation potentials as well as future research issues. Readership, who might be interested Scientists and engineers interested or engaged in the design of wind turbines in general, and the involved drivetrain and mechatronics in particular. For students of the subjects of wind energy systems, in particular for those focusing on mechanical design, simulation, mechatronics or electrical engineering, this book series (Vol.1, Vol.2) hopefully is of special interest in terms of applied research and a deeper understanding of the entire wind turbine as a complex system. Especially in combi- nation with the established fundamental literature on wind energy systems but also other books of the IET wind energy series (e.g. Modeling and Simulation), this book shall be a practical addition. Thus, engineers and practitioners in wind power industry, for them to obtain a detailed overview of this topic area. In particular, when their day-to-day work involves adjacent subsystems, such as tower or blades, drivetrain specialists from other industries, open for inspiration and perspectives from wind industry; students specializing on different topics, in order to comprehend the requirements for turbine and subsystem designs. This series of books is only intended to provide a contribu- tion to effectively dealing with the necessary prerequisites, the complex challenges in the areas of multi-body simulation, FEM calculation or advanced fatigue strength calculation for machine components, control and generator design and power elec- tronics development for advanced wind turbines. Wind Turbine System Design and its authors Volume 1: Nacelles, drivetrains and verification The first Chapter deals with the condensed, fundamental topic of load calculation and validation of wind turbines, explaining where loads are coming from and how to calculate, briefly explain processes to calculate them respectively, for the entire tur- bine. Within this chapter also the generic 7.5MW wind turbine IWT-7.5-164 is intro- duced, which serves in the following chapters of the book as a source for design data that are sometimes required. This generic turbine is largely open to all technologies (in terms of offshore, onshore application, direct drive, gearbox, rotor shaft bearing, etc.) and serves as a kind of common thread for exemplary design cases, which may be supplemented by the teams of authors of the following chapters, according to their requirements. A team of experienced wind turbine modelers wrote this chapter under the lead of Dipl.-Ing. Philipp Thomas, who studied mechanical engineering at the Otto-von- Guericke University Magdeburg. Since 2012, he researches in the area of holistic modelling and load analysis of wind turbines and is one of the main programmers of the load analysis tool MoWiT. Today he heads the group global turbine dynamics
- 19. xviii Wind turbine system design at the Fraunhofer-Institute for Wind Energy Systems. Philipps had competent sup- port from his co-authors, who were; M.Sc. Matthias L. Huhn studied mechanical engineering at the Hamburg University of Technology, RWTH Aachen University, and EPFL in Lausanne. Since 2017, he is a research associate at Fraunhofer-IWES in the group for global turbine dynamics and works on modelling, verification, and validation of aeroelastic models of wind turbines. The latter is also the subject of his PhD he is working on. He is a member of the international committee Joint Working Forum on Model Validation of the IECRE; Dr. Anna Wegner studied Physics at the University of Heidelberg. She received her PhD at the University of Bremen. She worked as a postdoctoral research fellow both in Germany and in the U.S. Today she heads the Application Center for Wind Energy Field Measurements at the Fraunhofer-IWES; and Dr.Eng. Mareike Leimeister is an offshore wind engineer with special interest in floating offshore renewable energy systems. She holds a double master’s degree from TU Delft and NTNU and received her EngD from the University of Strathclyde in 2020. Since 2017, she is also a research associate at the Fraunhofer Institute for Wind Energy Systems, where she both coordinates and works on joint research pro- jects. Her expertise lies in global turbine dynamics, numerical modelling, simula- tion, and optimization as well as floating wind turbines. The authors of Chapter 2 dig much deeper into the topic of models and simulation with a specific focus on rotor and drivetrain modelling. The rotas have to be dis- cussed in detail, despite it is not part of nacelle or drivetrain, but the main source of drivetrain loading and with strong interaction in-between, of course not exclusively. M.Sc. Paul Robert Feja, who studied mechanical engineering at RWTHAachen with a focus on renewable energy, managed the team. In 2015, he joined Fraunhofer- IWES, where he worked on global wind turbine model development and simulation. He was responsible for the implementation of a real-time virtual rotor model for hardware-in-the-loop tests at the nacelle test bench DyNaLab. Since 2020, he is the group manager for test and method development, focusing on simulation of wind turbine drivetrains and test benches. Paul was supported from his co-author team, whose members are; M.Sc. Muhammad Omer Siddiqui, a mechanical engineer with a focus on modelling and simulation of mechanical systems. He has a master’s degree in Simulation Sciences from RWTH Aachen University. In 2018, he joined Fraunhofer IWES as a research associate where he is primarily involved in developing high fidel- ity simulation models of the test bench and nacelle drivetrains; and again Dr.Eng. Mareike Leimeister, who was already introduced in the remarks to Chapter 1. In Chapter 3 M.Sc. Karsten Behnke and his team from wind industry and IWES co-authors give an insight into the concepts and design of pitch systems which is still considered as one of the most failure prone subsystems within the entire turbine. Karsten Behnke studied mechanical engineering at the Otto von Guericke University Magdeburg and he wrote his master’s thesis on the topic of multibody simulation of rolling bearings. In 2017 he joined Fraunhofer-IWES as a research associate. Since
- 20. Preface xix 2018 he is part of the group slewing bearings, where he researches in the field of pitch bearing and gearbox bearing damages. The team of authors for chapter 3 is complemented by; M.Sc. Arne Bartschat, who is the manager of the group slewing bearings in the department validation and reliability at Fraunhofer IWES. He has professional expe- rience with blade bearing and reliability dedicated research and project management since 2014. His research interests include finite element modeling at various levels from single components to complex assemblies, design, execution and evaluation of blade bearing tests, SCADA analysis of 1000+ turbines, model development and load simulation of wind turbines; M.Sc. Matthis Graßmann studied mechanical engineering at the University Rostock with focus on simulation. In his master thesis, he created a fully para- metrized FE bearing model and run simulations considering realistic surrounding structures of an experimental test rig. He joined the Fraunhofer-Institute of Wind Energy Systems IWES as a research associate in 2019. There, he is part of the group Slewing Bearings and researches in the field of large slewing bearings in wind tur- bines. He is responsible for FE calculations of bearings and their surrounding struc- tures like test rigs and rotor hubs. M.Sc. Florian Schleich studied mechanical engineering at the Hamburg University of Technology. In 2017 he joined Fraunhofer-Institute for Wind Energy Systems to write his master thesis in the field of finite element modelling of blade bearings. Since 2018 he is working as a research associate in the department vali- dation and reliability. He has been working on several projects related with blade bearing simulation and testing. His focus is on the development of FE wind tur- bine rotor models and the simulation of blade bearings internal load distributions; furthermore, Dipl.-Ing. Eike Blechschmidt, who studied mechanical engineering with mechatronics as a major field of study. He wrote his master thesis on condition mon- itoring of slowly rotating bearings. Eike worked eight years for REpower/Senvion in different functions in the fields of condition monitoring, test validation and pitch yaw systems. Since 2021 he is working for Fraunhofer-IWES as a research associate with a focus on data analysis and artificial intelligence as well as on grease comparison tests.; and M.Sc. Oliver Menck studied mechanical engineering and mechatronics at the Hamburg University of Technology. He joined Fraunhofer-IWES as a research asso- ciate. Here he is involved in the planning and execution of tests, data analysis, and lifetime calculation of bearings in wind turbines. His interests include anything related to mechanical engineering, mechatronics and what else wind energy; and finally, from wind industry Dipl.-Ing. Heiko Jungermann, who has studied electrical engineer- ing at the “Fachhochschule Osnabrück” from 1996-2000. He started as an application engineer at Nidec SSB Windsystems from 2000 with the main tasks, design and layout of customized pitch systems for customers world-wide. In 2013 Heiko Jungermann became the head of the systems engineering at Nidec SSB Windsystems. Since 2020 he has the responsibility for the systems engineering and the electronic development team at Nidec SSB Windsystems in the role of Director ED.
- 21. xx Wind turbine system design Chapter 4 deals with another wind turbine subsystem, which applies large bearing devices, the yaw-system, usually only dealt with very briefly in the current literature but explained here in detail by professionals from wind industry. The yaw system author-team consists of; Dipl.-Ing. Christian Bulligk, who graduated in mechanical engineering at Dresden University of Technology in 2009. He has been working in the wind indus- try since 2010. For the wind turbine manufacturer REpower/Senvion, he developed mechanical components for pitch and yaw systems for turbines up to 6 MW. Since 2020, he has been lead engineer for pitch and yaw systems at bewind GmbH, an engineering office with extensive knowledge and experience in design, transport, installation, and operation of wind turbines; and his Co-author, Dipl.-Ing. Daniel von dem Berge, who studied mechanical engineering at the University of Applied Sciences Gelsenkirchen, department Bocholt. He started working in the wind indus- try in 2009 at the wind turbine manufacturer Kenersys, where he was responsible for pitch and yaw system, drivetrains and various auxiliary systems for 2 up to 2.5MW turbines. Since 2015, he has been working as engineer for mechanical systems and since 2019 as project manager for the maxcap project at engineering office windwise GmbH, an service provider that specializes in the wind industry - from the develop- ment and construction of multi-megawatt wind turbine generators to support with purchasing, quality assurance, project management and the technical management for wind-energy projects. Just explaining and discussing the astonishing range of drivetrain concepts of tur- bine manufactures and over time since the 1980s is the main content of Chapter 5. Development lines divided into on- and offshore application are discussed in detail as well as general aspects from literature, scaling laws and performance indicators. The author and editor of this book put in his experience from discussion with many wind energy experts to provide a broad overview of drivetrains in wind turbines. Prof. Dr.-Ing. Jan Wenske studied mechanical engineering at the Technical University of Clausthal with focus on high performance drives and power electron- ics. He received his PhD in 1999 at the Institute of Electrical Engineering at the TU Clausthal on the field of power electronic application for grid stabilization under high share of wind energy feed in. He worked another year as senior scientist and leader of research group distributed renewable energy systems. In 2000 he changed to industry, as project manager within the pre-development division for forklift truck drivetrains at the STILL GmbH. Subsequently he was in charge of the Department for Power Electronic Development at Jenoptik Defense Civil Systems from 2005 to 2010 with focus on high performance hybrid drivetrains, high-voltage vehicle power supplies and more electric aircraft projects. Since 2011 he has been deputy director of Fraunhofer IWES. 2013 he become Professor at the University of Bremen for Wind Energy Systems and is Chief Technology Officer (CTO) at the IWES. A true pioneer in the field of gearbox design for wind turbines describes in Chapter 6 very personally his experiences, old and new innovations and the example of a hands- on design process for the design of a gearbox for the 7.5MW IWT. He also presents
- 22. Preface xxi an outlook on a possible future with very high ratio, multi power split gears for future double digit rated power turbines. Dipl.-Ing. Urs Giger is a Senior Mechanical Engineer, holds a HTL Diploma in Mechanical Engineering from the FHNW School of Engineering and run his own company GGS in Andermatt. His most recent devel- opment work has focused to the design of Multi Rotor (MR) wind turbines. He holds three patents and has evolved the flexible pin for PTGs into a low-cost and effective element. His long-term collaboration with Ray Hicks † (Wales) and Kiril Arnaudov † (Sofia) has resulted in innovative drivetrains for the wind industry. He is an active member of the JWG 1 ISO TC 60 IEC TC 88 JWG GEARBOXES FOR WIND TURBINES, and active member in IEC 61400-8. He is representa- tive of Switzerland in the International Electrotechnical Commission TC88: WIND TURBINES and lives in Mühlau, Switzerland. An expert team from HAWE has compiled all relevant information regarding the hydraulic assistance systems inside the nacelle. Safe operation of the WT is not pos- sible without these systems. They control and supply centralized or decentralized hydraulic actuators for controlling the brakes, the rotor lock, the on-board crane, the nacelle-roof opener and quite often the entire pitch system of the turbine. System properties such as leakage free, reliability and safety are of essential importance. The authors are; Dipl.-Ing. Andreas Nocker who is with HAWE Hydraulik since 2000. More than ten years he worked as Product Manager. Since 2011 he is in charge as Key Market Manager for the application field Energy worldwide and especially for the use of hydraulics in wind turbines. After finishing his studies in 1991, he began his career with Bosch Rexroth, Lohr am Main/Germany working in the RD department for mobile hydraulics. At Oil Control Deutschland, Augsburg/Germany he was assistant of the head of technology from 1993 to 1999. He studied at the Technical University Munich/Germany and holds a degree (Dipl.-Ing. TU) as mechanical engineer. After studying mechanical engineering at the University of Applied Sciences in Kiel, Arved Hildebrandt directly started his professional career in the sector of wind turbines in 2009. In more than 10 years he designed various components and systems for different wind turbine manufacturers and engineering offices in Germany. In 2021 he has started in the technical sales team of HAWE SE and is responsible for wind turbine related products. Besides his engineering and sales activities, Arved Hildebrandt has a passion for innovation management and is part of several patent applications. Also in this chapter Daniel von dem Berge and Christian Bulligk provide detailed information and experiences, here about the central lubrications system for bearings in wind turbine, an important auxiliary system and essential for the overall reliability of the entire turbine. The importance of well-designed cooling systems within wind turbines is often underestimated. Cooling circuits which are at least temporarily to hot or cold cause significant trouble (power derating or insufficient coolant flow respectively). The cooling system is one key enabler for the performance and also efficiency of the overall drivetrain system. Gearbox, Generator and Converter always need sufficient
- 23. xxii Wind turbine system design cooling. Within the gearbox the oil additionally serves the lubrication and therefore reliability and mitigation wear-out. Combined systems are efficient but not easy to design. Ernst-W. Langhoff gives a deeper look in the secrets of the design of such systems. In closely cooperation with Urs Giger (Chapter 6) he designs and explains a suitable cooling and lubrications concepts for the 7.5MW geared drivetrain concept with power split and dry lube system, introduced by Urs Giger. Ernst-W. Langhoff is employed by the Hydac Group since May 1985, the first years as a sales engineer, later as a key account manager for wind energy Industry solutions for gearboxes and wind turbines. In technical cooperation together with the employer’s development department he developed the two-stage filter element for gearboxes, not only for WTB`s but also for industrial application. Beside the lubrication systems, he also designed water glycol systems for combined gear cooling circuits with converter as well as generator and in general further more special wind industry solutions. Now 68 years old and retired, he continues work with the Hydac Group, with passion. The Chapter 9 presents the recommended verification and validation process according to the V-model for complex product development, exemplary for the wind turbine, with the focus on rotor and drivetrain. The lead at the experienced team of authors had Dipl.-Ing. Hans Kyling, who studied aeronautical engineer- ing at the RWTH Aachen. For more than a decade he has worked in different roles in both numerical and experimental investigations of complete drivetrains as well as individual subsystems and components of wind turbines. Today he heads the department System Validation of Mechanical Drivetrains at the Fraunhofer Institute for Wind Energy Systems. Specific parts as Co-authors and specialists for test and validation took over M.Sc. Karsten Behnke and Dr. Anna Wegner, both already introduced above as well as M.Sc. Malo Rosemeier a mechanical engineer. Since 2013, he works as Research Associate at the Fraunhofer-IWES. In the Department of Rotor Blades, he is responsible for the applied research on rotor blade structures. His focus areas are among others the development of validation tests and structural analysis methods.; and Dr. Alexandros Antoniou, who is a PhD Mechanical Engineer with 22 years’ experience in design, manufacturing, and testing of composite materials and sub-structures for wind turbine rotor blades. Currently, he is heading the Group Modelling of Polymers and Composite Materials at Fraunhofer-Institute for Wind Energy Systems. Finally, as a real expert in bearing systems for wind turbine main suspension sys- tem Dipl.-Ing. Marc Reichert and his Co-authors M.Sc. Tobias Baumgratz and M.Sc. Clemens Brachmann both from Eolotec GmbH give a comprehensive insight in the corresponding design process, the requirements and challenges in Chapter 10. After graduating in mechanical engineering in 2005, Marc Reichhart started working in the application-engineering department of a bearing manufac- turer and later became the head of application engineering. In 2010, he moved to a wind turbine development company where he was able to expand his detailed knowledge of rolling bearings to the overall drivetrain system and the corresponding
- 24. Preface xxiii interactions. Finally, in 2012, he joined the newly founded company Eolotec GmbH. Since then, his tasks have included the new and further development of bearing and system calculation methods as well as the development of new measurement sys- tems for large size bearing arrangements in wind turbine drivetrains. In the recent years, it has become more and more his vision to bring together the extensive field experience regarding bearing damages, results from measurement campaigns and system calculation results, in order to gain a better understanding of the influence of load dynamics. This should help to avoid bearing failures in the future and thus to increase the reliability of large size bearing systems. The Co-author Tobias Baumgratz holds a master’s degree in mechanical engi- neering with focus on product development, he joined Eolotec GmbH as a working student in 2019. Through this occupation, he was able to build up initial knowledge in the field of rolling bearings for wind turbines and to extend this knowledge while writing his master thesis on the development of rolling bearing calculation models in FEM. After the master’s degree in mechanical engineering, he started to work as development engineer at eolotec GmbH, where Tobias Baumgratz have now acquired further expertise in the field of calculation methods for rolling bearings and structural components as well as in the design of drivetrain concepts. The second Co-author Clemens Brachmann gathered his first theoretical and practical experiences about laser additive manufacturing and about polymer pow- der deposition for laser sintering during his studies and interning in Taiwan and Germany. After his Master Thesis about the computational implementation of a heat conduction model, he joined eolotec in 2022 as a project engineer, now coordinating engineering services around roller bearings for wind turbines. Volume 2: Electrical Systems, Grid Integration, Control Monitoring The content of Vol.2 shall just explained briefly here. The chapters in more detail and the authors are described in the equivalent preface of Volume 2 of this book. In contrast to Vol.1, Volume 2 focuses on the content of the electrical drivetrain (generator, converter systems) of the wind turbine. In addition, the topics turbine control, bus systems and monitoring are discussed in detail. Another extensive focus are wind turbine HiL test systems, not exclusively but specifically for measuring and certifying their electrical properties, grid integration testing and model valida- tion. The Volume 2 concludes with chapters related to the topics advanced control for smarter turbines and wind farms as well as system integration in an anticipated, highly decentralized electric energy supply systems of the future (principles, mod- eling and grid-forming control). The Editor and the whole team of authors, which work all with great commitment and general passion for wind energy and wrote this book for whom interested, hope all readers enjoy reading and a successful future work in the fascinating world of wind energy systems. With best regards Jan Wenske (Ed.)
- 25. xxiv Wind turbine system design Acknowledgements On behalf of all authors of this book, the editor would like to thank the following companies for their kind support in the publication of this book. The information and images provided are of great value for understanding and explaining the com- plex areas of knowledge. bewind GmbH Bonfiglioli Deutschland GmbH DNV Denmark A/S Eolotec GmbH Federal-Mogul DEVA GmbH (a Tenneco Group Company) Flender GmbH Fraunhofer-IWES GGS Groeneveld-BEKA GmbH HAWE Hydraulik SE Hydac Group Kendrion INTORQ GmbH Liebherr-Components Biberach GmbH NIDEC SSB WIND SYSTEMS Svendborg Brakes A/S Trebu Technology B.V. windwise GmbH Please note that all images and tables marked accordingly are subject to copyright of the respective companies or institutions and have been reproduced exclusively for use in this book by individually permission.
- 26. Abbreviations and Terminologies 1D One-dimensional 3D Three-dimensional AC Alternating current ASME American Society of Mechanical Engineers B2B Back-to-back BEM Blade element momentum BTC Bend-twist coupling CAB Controlled atmosphere brazing CARB Toroidal roller bearing CCV Cold climate version CFD Computational fluid dynamics CMS Component mode synthesis COG Compact orbital gear CRB Cylindrical roller bearing CTOD Crack tip opening displacement CWD Center for Wind Power Drives DC Direct current DD Direct drive DEL Damage equivalent load DFIG Doubly-fed induction generator DFMEA Design failure mode and effect analysis DGD Distributed generation drivetrain DIN Deutsches Institut für Normung DLC Design load case DNV Det Norske Veritas DOFs Degree of freedom DRTRB Double-row tapered roller bearing DT Drivetrain DUT Device under test DyNaLab Dynamic Nacelle Testing Laboratory EC European Commission ECM Extreme current model EESG Electrically excited synchronous generator EFC Emergency feather command EP Extreme pressure ESS Extreme sea state ETM Extreme turbulence model
- 27. xxvi Wind turbine system design EU European Union EWH Extreme wave height EWM Extreme wind speed model F2F Face-to-face FDC Force-distributed constraints FE Finite element FEA Finite element analysis FEM Finite element method FFST Fatigue full-scale blade testing FMEA Failure mode and effect analysis FMECA Failure mode, effects, and criticality analysis FST Full-scale blade testing FTA Fault tree analysis GBTC Geometric bend-twist coupling GD Geared drivetrain GDW Generalised dynamic wake GEBT Geometrically exact beam theory GFRP Glass Fiber Reinforced Plastic GL Germanischer Lloyd GPS Global positioning system GRC Gearbox reliability collaborative GRP Glass reinforced polyester HAPT Highly Accelerated Pitch Bearing Test HCV Hot climate version HIL Hardware-in-the-loop HSS High-speed shaft HTS High-temperature superconductor IEC International Electrotechnical Commission IG Induction generator IPC Individual pitch control IR Inner ring ISO International Organization for Standardization ITGS Integrated tubular gear system IWES Institute for Wind Energy Systems JONSWAP Joint North Sea Wave Project LCC Life cycle cost LCoE Levelized cost of energy LDD Load duration distribution LEFM Linear-elastic fracture mechanics LEP Leading edge protection LES Large eddy simulation LiDAR Light detection and ranging LRD Load revolution distribution LSS Low-speed shaft LVRT Low voltage ride through MAN Maschinenfabrik Augsburg-Nürnberg
- 28. Abbreviations and terminologies xxvii MBS Multibody simulation MLC Measurement load case NCM Normal current model NCV Normal climate version NLGI National Lubricating Grease Institute NREL National Renewable Energy Laboratory NSS Normal sea state NTM Normal turbulence model NVH Noise, vibration and harshness NWH Normal wave height OEM Original equipment manufacturer OVRT Over voltage ride through PA Polyamide tube PAO Poly-alfa olefin PC Point contact PGT Planetary gear train PL Performance level PMBOK Project Management Body of Knowledge PMSG Permanent magnet synchronous generator PSF Partial safety factor RANS Reynolds-averaged Navier–Stokes RCF Rolling contact fatigue SBTC Structural bend-twist coupling SFST Static full-scale blade tests SG Synchronous generator SODAR Sound detection and ranging SPMT Self-propelled modular transporter SRB Spherical roller bearing SRP/CS Safety-related parts of control systems SSS Severe sea state SWH Severe wave height TANDEM Towards an Advanced Design of Large Monopiles TCO Total cost of ownership TI Turbulence intensity TR Technical reports TRB Tapered roller bearing TRL Technology readiness level TS Technical specifications UC Ultra-caps UMP Unbalanced magnetic pull VDI Verein Deutscher Ingenieure VV Verification and Validation WBS Work breakdown structure WP Work package WT Wind turbine YFM Yielding fracture mechanics
- 29. This page intentionally left blank
- 30. 1 Fraunhofer Institute for Wind Energy System, Großer Westring, Bremerhaven, Germany Chapter 1 Load calculation and load validation Philipp Thomas1 , Mareike Leimeister1 , Anna Wegner1 , and Matthias L. Huhn1 The design process of wind turbine (WT) generators is an iterative process. In the beginning, there are requirements regarding the electrical power or the specific power (i.e., power per swept area) for certain locations as well as the topology of the WTs. These requirements form the basis for an initial design of the rotor, which then pro- vides loads for the design/selection of the load- carrying components and drivetrain. This results in a design of the overall system, whose interaction is examined with numerical simulation tools, and requirements for the next iteration of the WT com- ponents are provided. The load assumptions for the individual components and the dynamics of the overall system are constantly being refined until the requirements for the system design are deemed to have been met. To ensure that the load assump- tions always contain the same operating states that are relevant for the lifetime of the WTs, regardless of the manufacturer, they are determined based on the specifica- tions of international standards. After completion of the numerical design process, the design loads and the system dynamics are verified by independent certification bodies before a prototype of the WT can be built. The numerical design loads must be validated on the prototype in the field. This proves that the real loads and dynam- ics are within the limits of the numerical design. At the same time, the quality of the numerical simulation tools used in the design process can be quantified. Typically, the design process of the entire system takes place at the manufac- turer of the WT. Specific components are purchased from specialized companies, such as bearings or gears. The supplier companies receive the necessary load pro- files for the component design from the manufacturers and return numerical model parameters to the manufacturer, which integrates these into the simulation of the overall system and checks the requirements for the system dynamics. If the compo- nent manufacturer also wants to verify the requirements for its design in the over- all system, he/she usually has to fall back on freely available simulation models from WTs. These so- called generic WTs exist for various power classes from a few hundred kilowatts to 15 megawatts and beyond. This means that component
- 31. 2 Wind turbine system design manufacturers are able to understand the load simulations in accordance with the requirements of international standards, independently of the WT manufacturer, and to generate relevant load information for the design process themselves. Although the parameters of generic WTs are never 100% consistent with the parameters of the manufacturers and generic loads are always subject to uncertainty compared to manufacturer loads, generic WTs offer indispensable added value for suppliers and research. After all, they offer the only way to determine and research the loads and dynamics of WTs independently of the manufacturer. Fraunhofer Institute for Wind Energy Systems IWES develops generic WTs itself and has already published a 7.5 MW design, the onshore WT IWES Wind Turbine IWT- 7.5- 164 [1]. The 7.5 MW WT has a rotor diameter of approximately 164 m, a hub height of 119.3 m (hybrid tower) and a total mass of 2004 t. The design combines a large onshore WT with relatively low specific power and a focus on detailed blade design with a load mitigation control strategy. Therefore, the WT was developed for coastal locations with high average wind speeds and turbulence inten- sity (IEC IA). The key figures of the WT can be found in Table 1.1. An excerpt of previous use cases shows the potential of generic WTs. The IWT- 7.5- 164 was designed in the Smart Blades project [2] and used to investigate various concepts for passive and active load reduction techniques. The loads generated were later used for the IWES Highly Accelerated Pitch Bearing Test (HAPT) test bench [3]. In the Towards an Advanced Design of Large Monopiles (TANDEM) research project [4], it was used to calculate loads for the design of an XL monopile, in the InterWiLa research project [5] for load simulations with new types of wind fields, and for the optimised [6] and automatic [7] upscaling of the IEA Wind Task 30 OC4 Table 1.1 Summary of IWT- 7.5- 164 main properties Description Value Unit Rated electrical power 7542 kW Rotor diameter 163.96 m Hub radius 2 m Arc length of the blade 79.98 m Hub height 119.3 m Cut- in wind speed 3 m/s Rated wind speed 11.7 m/s Cut- out wind speed 25 m/s Rated tip speed 85.53 m/s Cut- in rotor speed 5 rpm Rated rotor speed 10 rpm Rated tip speed ratio 7.31 – Blade cone angle 2 degrees Shaft tilt angle 5 degrees Mass of rotor- nacelle assembly 536.78 t Single blade mass 30.93 t Specific power per rotor swept area 360 W/m²
- 32. Load calculation and load validation 3 semi- submersible [8] and OC3 spar floater [9], respectively, from a 5 MW to the 7.5 MW WT. This chapter deals with the requirements from the following standards for deter- mining the loads of WTs: • • IEC 61400- 1 design requirements • • IEC 61400- 3 design requirements for fixed offshore WTs • • IEC 61400- 13 measurement of mechanical loads Section 1.1 first introduces the topic of load calculation and then goes into load determination requirements for onshore and offshore WTs. Topics such as taking environmental conditions into account when determining the loads by simulation and the procedure for determining the design loads are considered. At the end of the section, two use cases for load calculation are presented. Section 1.2 deals with the validation of the load calculation models. After the general procedure for model validation is explained, the acquisition of measurement data is presented. Finally, the specific procedure for validating the load calculation models is outlined. 1.1 Design loads of wind turbines The design loads are determined in accordance with the requirements of interna- tional standards. IEC 61400- 1 [10] for onshore WTs and 61400- 3 [11] for offshore WTs are mentioned here as examples. The standards define requirements regarding the consideration of environmental conditions and how these are to be considered when designing specific components of the WT. The aim is to consider all operating and extreme load conditions that are likely to occur during the service life of the WT and their worst combinations when determining the design loads. The standards contain requirements for the simulative determination of the design loads. These requirements are sometimes very specific and sometimes leave a lot of room for interpretation. Detailed simulation scenarios are prescribed, the so- called design load cases (DLCs), but the requirements for the numerical simula- tion tools are only outlined schematically. With the so- called fully coupled simula- tion tools, the DLCs specified in the standards can be implemented numerically and converted into time series. The time series contain the dynamic reaction loads and deformations for a period of typically 10 minutes. Load spectra are extracted from the time series and converted into component- specific operating loads for a service life of 20 or more years as well as extreme loads. Furthermore, the simulation pro- vides the natural frequencies of the individual components and the coupled natural frequencies of the overall WT system as well as the specific characteristic curves, such as power and thrust depending on the wind speed. The results of the load simulation are checked by measurements on the WT in the field to show their validity. The load calculation and load validation are two essential steps in the type certification process, which is essential for the commercial sale of the WT.
- 33. 4 Wind turbine system design 1.1.1 Standard load calculation International standards define different requirements for determining the design loads for the design and certification of onshore and offshore WTs. The require- ments for onshore locations are taken here as an example from IEC 61400- 1 and reproduced to the extent necessary for understanding. The standard defines essential design requirements to ensure the structural integrity of WTs. Its purpose is to pro- vide adequate protection against damage from all hazards during the designed life- time. Requirements for the numerical simulation models, the consideration of dif- ferent environmental conditions, the type and implementation of simulations and the processing and evaluation of the simulation results are specified. In addition, IEC 61400- 1 describes the use of an aeroelastic dynamics model to predict design loads. Such a model is intended to be used to determine loads over a range of wind speeds using turbulence and other wind conditions. All relevant combinations of external conditions and design situations shall be analysed. The standard also defines a set of simulations to be performed. These DLCs (cf. Table 1.2) can be seen as the mini- mum requirement for the scope of the simulations to be carried out and the operating situations to be considered. For onshore WTs, the DLC simulations differ mainly in terms of the operat- ing and wind conditions and the way they are evaluated. The operating conditions include all states that are expected during the lifetime of the WT. This includes normal operation, start- up and shut- down, standstill as well as various error states. Each operating state is intended to simulate either normal or extreme operating con- ditions, or sometimes a combination of these. This is also reflected in the type of evaluation, where a distinction is made between fatigue loads (F) and ultimate loads (U). Furthermore, suitable partial safety factors (PSFs) are assigned to normal (N) and abnormal (A) operating conditions and the resulting favourable or unfavourable loads in the evaluation. Deterministic or stochastic models are used to represent the wind conditions. While onshore the most relevant environmental condition is the wind, there are also waves and currents for offshore locations. 1.1.1.1 Wind conditions Since the locations of WTs and the wind conditions prevailing there always dif- fer, the wind conditions have been categorised and WT classes introduced (cf. Table 1.3). This means that a WT can be assigned to an annual average wind speed Vave , a reference wind speed Vref (usually 5Vave , except for areas subject to tropical cyclones ( Vref,T )) averaged over 10 minutes, and a reference wind turbulence Iref , for which the WT is designed. The standard differentiates between the four wind speed classes I, II, III and S. Class I is for high wind speed locations typically found offshore. A class III WT would be pre- destined for a weak wind site. In addition, a distinction is made between four turbulence classes A+, A, B and C. Turbulence is the temporally and spatially varying fluctuation in wind speed, which is caused by meteorological interactions and the interaction of the wind with obstacles. These include the influence of the wind by buildings, such as other WTs,
- 34. Table 1.2 Onshore DLC according to IEC 61400- 1 and MLC according to IEC 61400- 13 Design situation DLC MLC Wind condition Other condition/objectives/remarks Type of analysis PSF 1) Power production 1.1 NTM vin vhub vout For extrapolation of extreme events U N 1.2 1.1 NTM vin vhub vout WT is running and connected to the grid F * 1.3 ETM vin vhub vout U N 1.4 ECD vhub = vr +/- 2 m/s and vr U N 1.5 EWS vin vhub vout U N 3.1 vin vhub vout Normal operation below rated wind speed and above rated wind speed with relatively steady rotational speed; Objectives: Blade, tower and drivetrain frequencies 2) Power production plus occurrence of fault 2.1 NTM vin vhub vout Control system fault or loss of electrical network U N 2.2 NTM vin vhub vout Protection system or preceding internal electrical fault U A 2.3 EOG vhub = vr ± 2 m/s and vout External or internal electrical fault, including loss of electrical network U A 2.4 2.4 NTM vin vhub vout Control, protection, or electrical system faults, including loss of electrical network F * 3) Start up 3.1 2.1 NWP vin vhub vout vin and vr + 2 m/s F * 3.2 EOG vhub = vin , vr ± 2 m/s and vout U N 3.3 EDC vhub = vin , vr ± 2 m/s and vout U N 4) Normal shutdown 4.1 2.2 NWP vin vhub vout vin , vr and vr + 2 m/s F * 4.2 EOG vhub = vr ± 2 m/s and vout U N 5) Emergency shutdown/stop 5.1 2.3 NTM vhub = vr ± 2 m/s and vout Rated power U N 3.3 vhub vr Emergency stop from rated power; Objectives: Blade, tower and drivetrain frequencies (Continues)
- 35. Design situation DLC MLC Wind condition Other condition/objectives/remarks Type of analysis PSF 6) Parked (standing still or idling) 6.1 EWM 50- year recurrence period U N 6.2 EWM 50- year recurrence period Loss of electrical network connection U A 6.3 EWM 1- year recurrence period Extreme yaw misalignment U N 6.4 1.2 NTM vhub 0.7 vref Rotor either at standstill or idling F * 3.2 High wind speed (high enough to get sufficient excitation, this will be WT specific) WT is parked (standstill or idling); Objectives: Blade and tower frequencies 7) Parked and fault conditions 7.1 EWM 1- year recurrence period U A 8) Transport, assembly, maintenance and repair 8.1 8.2 NTM vmaint to be stated by the manufacturer EWM 1- year recurrence period U U A A 9) Yaw start/stop 3.4 Low wind speed With an instrumented blade in a horizontal position, the blade gets excited by starting and stopping the nacelle yaw rotation. Test shall be conducted with blades in normal operating position (targeting the flatwise frequencies) and with blades feathered (targeting the edgewise frequencies); Objectives: Blade frequencies Table 1.2 Continued
- 36. Load calculation and load validation 7 and by the terrain, such as hills and forests. A typical low turbulence class C site would be offshore, and an A+ site with very high turbulence would more likely be onshore in complex terrain. The combination of wind and turbulence class results in the WT class, such as IC or IIIA+. Theoretically, all WTs could be designed for an IA+ location, which would also make them suitable for all other locations. However, this would mean that the majority of the WTs (if not all of them) would be oversized, which is uneconomical for the manufacturers and would unnecessarily drive up the costs of the WTs. The future location of a WT is therefore already of great relevance in the design phase. The class S WTs are intended for very specific locations for which the manufacturer must determine the wind and turbulence parameters himself. Further parameters are derived from the parameters Vave , Vref and Iref , which are required to describe the specific wind conditions for the simulation of each DLC. A distinction is made between stochastic and deterministic wind conditions. While the stochastic wind conditions attempt to examine the inflow occurring during everyday operation and its influence on the loads and dynamics of the WTs as realistically as possible, the deterministic wind conditions are intended to depict special operating conditions. For the stochastic wind conditions, turbulence models are used to generate the random fluctuations in wind speed in the time domain. A distinction is made between normal (normal turbulence model, NTM) and extreme (extreme turbulence model, ETM) turbulence intensity (TI). For the simulation of the DLCs with stochastic wind conditions, wind files are generated during the preparation of the simulation. The turbulence models provide the necessary parameters, namely the mean wind speed and TI at hub height over a period of 10 minutes. For NTM, for example, the stan- dard deviation is determined by the following equation: = Iref 0.75Vhub + 5.6 m/s (1.1) The standard deviation is the TItimes the mean wind speed Vhub . Depending on the average wind speed and reference TI, the associated TI for each DLC simulation can be determined (cf. Figure 1.1). To avoid having to simulate any number of wind speeds in the operating range of the WTs, wind speed ranges are defined, and suitable wind files are generated for each range. A common subdivision would be to start a new range every 2 m/s. Each Table 1.3 Basic parameters for WT classes according to IEC 61400- 1 WT class Unit I II III S Vave m/s 10 8.5 7.5 Designer specific values Vref m/s 50 42.5 37.5 Vref,T m/s 57 57 57 A+ Iref – 0.18 A – 0.16 B – 0.14 C – 0.12
- 37. 8 Wind turbine system design range represents an average wind speed and TI. Common software tools for calcu- lating stochastic wind fields use pseudo- random numbers to determine the temporal fluctuations in wind speed. The sequence of these random numbers is determined by a starting value (seed) and is thus made reproducible. So that not all stochastic DLCs use the same stochastic fluctuation, each seed is only used once in a load simulation. To ensure the statistical significance of the stochastic DLC simulations, each wind speed range is simulated with several seeds. At least six seeds must be used in DLC 1.2, which corresponds to a total of 60 minutes of turbulent simulation per average wind speed. IEC 61400- 1 specifies 12 seeds for DLCs 2.1, 2.2 and 5.1, and even 15 seeds for DLC 1.1. For the implementation of DLC 1.2, which is intended to map normal power production, there are at least 72 simulations each with a length of 600 s (with an operating range of 3–25 m/s with a range width of 2 m/s and 6 seeds each) and even more if yaw errors are considered. The stochastic DLCs represent both normal and extreme operating situations of WTs and are evaluated according to fatigue or extreme loads, depending on the DLC. The deterministic wind conditions consist of laminar inflow, which is partially overlaid with specific wind speed profiles. Since there are no stochastic fluctua- tions and therefore no turbulence, these DLCs are only suitable for investigating specific wind and operating conditions and their impact on the load and the dynamic behaviour of the WT. (Extreme) gusts, changes in wind direction, wind shear and combinations thereof are simulated. In addition, the effect of extreme wind speeds (extreme wind speed model, EWM) is examined, which can be implemented sto- chastically or deterministically. This should take into account particularly strong storms, which only occur every year or once in 50 years. Figure 1.1 TI for the normal turbulence model
- 38. Load calculation and load validation 9 Each DLC uses horizontal and vertical oblique flow. Oblique flow is the devia- tion of the main wind direction from the rotor plane. Horizontal oblique flow is taken into account differently for operating load cases and standstill load cases. While the maximum dead range of the yaw control, in which the yaw control is not yet active (e.g., ±10 degrees), is to be examined in the case of operating load cases, special attention is paid to the failure of the yaw control in the standstill load cases, with angles of up to ±180 degrees. Here, DLC 6.2 in particular leads to numerical instabilities with many load simulation tools at a 90- degree oblique flow. Vertical oblique flow takes into account the effect of terrain on the wind direction. It is rec- ommended to turn the main wind direction downwards by 8 degrees against the horizon for onshore load cases. Furthermore, the wind speed, which changes with altitude, must be taken into account. In IEC 61400- 1, an exponential shear model is specified with an exponent of 0.2 or 0.11 for onshore load cases, depending on the wind model. In addition to specifications on wind conditions, the standard contains other requirements that must be taken into account. The natural frequencies for the tower, drivetrain and rotor must be determined. If coupled system frequencies are in the range of up to the sixth harmonic of the speed (6 P), further analysis must be carried out to check for any resonance points and, if necessary, countermeasures must be taken. If there is a risk of earthquakes at the potential installation site of the WT, the load simulations must also be adapted accordingly. The same applies to regions where ice formation on the rotor blades is to be expected. 1.1.1.2 Waves and current conditions In the case of offshore WTs, also sea conditions need to be taken into account in addition to wind conditions. The underlying standard IEC 61400- 3 [11], hence, is mainly based on IEC 61400- 1 and extends it by offshore- specific design require- ments. Thus, in addition to the wind and operating conditions detailed in 0, different waves, the directionality between wind and waves, the presence and type of sea cur- rents as well as the water level need to be considered and are included in the offshore DLC specification. The waves occurring offshore are irregular and stochastic. They can therefore be represented by a wave spectrum, which is based on a significant wave height, peak spectral period and the mean direction of the waves. Similar to the wind conditions, it is differentiated between normal (normal sea state, NSS), severe (severe sea state, SSS) and extreme (extreme sea state, ESS) conditions with the corresponding nor- mal (NWH), severe (SWH) and extreme (EWH) wave heights, and associated peak spectral periods. While the SSS is only used in combination with NTM to represent the severe conditions at a wave- dominated site, the ESS in conjunction with EWM reflects the extreme environmental condition that has a recurrence period of one year or 50 years. Sea currents are only taken into account in DLCs for ultimate strength analyses and not for fatigue analyses. Depending on the type of environmental and site condi- tion, it is differentiated between normal (normal current model, NCM) and extreme
- 39. 10 Wind turbine system design (extreme current model, ECM) currents. Both types contain currents that are gen- erated by the wind and, hence, only reach down to a limited depth (mostly 20 m) below the water surface. Depending on the location of the offshore WT, there might also be surf currents to be considered due to breaking waves occurring close to the coast. Only in the ECM, subsurface currents need to be additionally considered, which may source from storm surge or tides prevailing at the offshore site. With the consideration of both wind, waves and currents in the DLCs, the direc- tionality of the environmental factors may have a significant impact on the resulting loads on the offshore system. For ultimate and, hence, worst- case scenarios, wind and waves may be considered as codirectional and unidirectional. In some fatigue DLCs, also multidirectionality is considered for aligned wind and waves. The mis- alignment of wind and waves along with their multidirectionality mainly needs to be taken into account in parked DLCs. A separate third combination of the directionali- ties of wind and waves with the current direction is not required. For the currents, the direction of the main source of the specific subtype is applied, i.e., the near- surface wind- generated currents follow the wind direction and the subsurface currents are codirectional with the waves, while the direction of the breaking wave induced surf currents is, due to the nature of this current type, parallel to the coastline. 1.1.1.3 Fatigue and extreme loads The fatigue and extreme loads are required when designing the individual compo- nents of the WT. The simulations to determine the loads use aeroelastic models. This allows the complex, nonlinear interaction of the WT to be examined in the time domain, and the transient loads and deformations that occur in the process to be determined. The models account for gravitational, inertial, aerodynamic, actua- tion and other relevant loads as required by the IEC 61400- 1. More specifically, it requires the following to be taken into account: • • The influence on the wind field by the WTs • • The influence of the three- dimensional (3D) flow on the aerodynamic properties of the rotor blades • • Transient aerodynamic effects • • Structural dynamics and the coupling of vibrational modes • • Aeroelastic effects • • The interaction of the control system with the WT. In addition to the requirements for the capabilities of the load model, the standard also requires subsequent validation of the aeroelastic simulation model through measurements that should be carried out in accordance with the require- ments of IEC 61400- 13. This is to ensure that the simulated loads and dynamics reflect reality. The extreme loads are determined from the results of all extreme load DLCs at different positions of each component of the WT. The evaluation is carried out for all degrees of freedom and determines the largest load in terms of absolute value. PSFs
- 40. Load calculation and load validation 11 are used to take into account the uncertainties in loads, analysis methods and the importance of components with regard to the consequences of failure. The design load consists of the PSF and the characteristic load. The characteristic load is deter- mined from the simulated loads and, if necessary, applied with a specific safety fac- tor per component, which should take into account the effect of a failure, depending on the requirements of the respective DLC. The standard defines minimum PSF values, the use of which also requires a validated load model. The extreme loads are required for the design of the respective component and indicate the load level to be endured. Therefore, extreme loads are also referred to as design driving loads. The fatigue loads are a measure of the damage caused by the cyclic load changes during operation of the corresponding DLC. The limit state is reached when the component reaches damage 1 at the combination of a load level, a number of cycles and an oscillation frequency. The number of cycles depends on the fatigue strength range of the Wöhler curve of the respective material of the component, and the oscil- lation frequency on the desired service life of the WT, e.g. 20 years. In the standard, the use of Miner’s rule is recommended, which uses a rainflow counting method. The Weibull distribution of the wind speeds measured at the site is used to take into account the occurrences of the various mean wind speeds during the lifetime of the WT. This allows the loads simulated in the DLC to be extrapolated for the desired lifetime of the WT. In addition to determining the loads, the load model is also used to investigate the deformation behaviour. For this purpose, the DLCs are evaluated for critical deformations, such as a sufficiently large distance between the rotor blade and the tower at each operating point. 1.1.2 Use cases and exemplary loads The generic WT IWT-7.5-164 has already been used many times to determine design loads. These two use cases are presented below as examples: • • Rotor blade design • • Monopile design The right- hand Cartesian coordinate systems of the following two examples are shown in Figure 1.2 and are defined as follows. The tower coordinate system origin is located at the point where the tower bottom horizontal plane crosses the vertical centreline of the tower. The xt - axis points downwind with respect to the main wind direction, the yt - axis points to the side, and the zt - axis points vertically upwards. The blade root coordinate system is located at the centre of the blade root and rotates with the rotor. It is tilted with the main shaft tilt angle, coned with the rotor cone angle and pitched with the blade pitch angle. This means that, in the case of zero pitch angle, the xb - axis points downwind, the yb - axis is parallel to the rotor plane and points against the rotational direction, and the pitch zb - axis points radially outwards.
- 41. 12 Wind turbine system design 1.1.2.1 Rotor blade design During the design process of the IWT- 7.5- 164 rotor blades, a selection of the five DLCs 1.1, 1.2, 2.3, 6.1 and 6.2 were simulated in order to determine the loads rel- evant to the design of the rotor blades. This selection of DLCs was then used to eval- uate different blade design approaches. For this purpose, the DLCs were repeated several times and the results were compared. The loads along the complete rotor blade are required for the rotor blade design. For the sake of clarity, only the results at the blade root are shown here. Fatigue results in Table 1.4 are presented in terms of damage equivalent loads (DELs) cal- culated for all force and moment components: Fx , Fy , Fz , Mx , Myand Mz , with the following settings: • • 20 years WT lifetime • • 107 load cycles • • Wöhler slope exponent m of 4, 8, 10 and 14 for blade outputs Table 1.4 Fatigue loads at the blade root of the IWT- 7.5- 164 m Fx [kN] Fy [kN] Fz [kN] Mx [kNm] My [kNm] Mz [kNm] 4 261.5 549.9 557.3 11879.3 10469.2 304.9 8 252.6 420.6 464.5 9229.3 10623.8 409.0 10 261.0 399.4 461.1 8844.9 11193.0 458.9 14 281.2 377.7 471.1 8540.6 12456.6 535.4 Figure 1.2 Coordinate systems at tower and blade root
- 42. Load calculation and load validation 13 Ultimate loads are presented in terms of the minimum and maximum load com- ponents located on the main diagonal of an ultimate load table with the correspond- ing (same time step) load components listed in the rows. Each ultimate load table also lists DLC names (with wsp: wind speed in m/s, yaw: yaw misalignment in degrees, seed: wind seed), at which the ultimate loads occurred, with the corre- sponding PSFs. For reasons of space, only the moments in the three spatial direc- tions x , yand zare shown in Table 1.5, and the forces are left out. With the help of the load calculation, variables influencing the loads can also be examined. The IWT- 7.5- 164 reference blade design was modified in two ways to account for bend- twist coupling (BTC). BTC couples the blade bending to the blade torsion and automatically twists the rotor blade out of the wind at high bending. This reduces the aerodynamic torque with large blade deflection and thus the loads. Two BTC approaches were investigated: structural (SBTC) and geometric (GBTC) bend-twist coupling. Table 1.6 compares the relative change in fatigue loads at the blade root with the reference design, as an example for S- N slope m = 8 . The three different blade design methods affect the loads differently. With these exemplary numbers, GBTC seems the most promising for load reduction. Only the torsional loads Mzincrease with BTC, which is to be expected. In addition to evaluating different design approaches for rotor blades, the blade root loads of the IWT- 7.5- 164 offer other practical uses. For example, they can be used for the design of WT manufacturer- independent test stands for which no other loads are available. The IWT- 7.5- 164 blade root loads were used to design the IWES Table 1.5 Extreme loads at the blade root of the IWT- 7.5- 164 DLC Mx [kNm] My [kNm] Mz [kNm] PSF [-] Mx max DLC11_wsp25yaw0seed6 16656.7 −9789.1 −7.9 1.35 Mx min DLC11_wsp25yaw-8seed5 −14939.6 −6242.1 −199.9 1.35 My max DLC62_wsp50yaw300seed2 −1968.8 30363.4 728.5 1.10 My min DLC11_wsp11yaw-8seed6 8983.3 −40819.8 −52.5 1.35 Mz max DLC11_wsp13yaw-8seed2 −21486.5 −756.4 1231.7 1.35 Mz min DLC62_wsp50yaw180seed4 −15477.1 −1359.5 −1194.0 1.10 Table 1.6 Comparison of fatigue loads for different blade design approaches at the blade root of the IWT- 7.5- 164 Fx Fy Fz Mx My Mz Reference 100.00% 100.00% 100.00% 100.00% 100.00% 100.00% GBTC 95.59% 99.95% 99.21% 100.04% 90.45% 104.69% SBTC 99.58% 101.35% 101.34% 101.89% 97.78% 112.78%
- 43. 14 Wind turbine system design HAPT test stand [3], where endurance tests for large bearings are carried out. The fatigue and extreme loads served as the basis for defining the capacities of the test bench, with additional safety factors being taken into account so that the test bench is also suitable for testing future bearing generations. In addition, the IWT- 7.5- 164 loads were used for the planning and execution of the pitch bearing test program. 1.1.2.2 Monopile design The design process of bottom- fixed offshore foundation structures usually consists of the following three sequential steps: 1. Provision of representative cutting loads 2. Design of the monopile 3. Optimization of the overall system In the example described below, the DLCs 1.1, 1.2, 3.1, 4.1, 6.1, 6.2 and 6.3 were simulated to determine the representative cutting loads at the transition piece of the tower. The environmental conditions were chosen for a representative North Sea location. The fatigue and extreme loads at the transition piece of the tower for six degrees of freedom of the three spatial directions x , yand zas well as the resultant loads in the x − y – horizontal plane are shown in Table 1.7. The fatigue loads were calculated for 107cycles in 20 years lifetime and for the Wöhler slope exponents mof 4, 5 and 6. The extreme loads at the transition piece are shown in Table 1.8 for DLC 6.2. For reasons of space, only the moments in the three spatial directions x , yand z and the resulting moment xyare shown, and the forces are left out. The values of the extreme loads can be found on the main diagonal (shaded grey in Table 1.8). The absolute largest and smallest values are given as well as the subcase of the DLC in which the values occurred (with u: wind speed in m/s, y: yaw misalignment in degrees, ww: wind- wave misalignment in degrees, s: seed). In addition to the extreme values, the other loads that occurred in the simulation at the time of the extreme load are also entered. In addition to the loads, the PSF is also listed. Table 1.7 Fatigue loads at the transition piece of the IWT- 7.5- 164 in the tower coordinate system m Fx [kN] Fy [kN] Fxy [kN] Fz [kN] Mx [kNm] My [kNm] Mxy [kNm] Mz [kNm] 4 408.5 873.7 583.0 1819.6 23583.4 18758.7 20532.9 20237.1 5 468.6 975.6 600.7 2113.7 29210.8 23378.8 22087.3 22992.4 6 523.5 1086.4 643.3 2353.9 37778.3 28556.1 25083.4 25550.5
- 44. Table 1.8 Extreme loads at the transition piece of the IWT- 7.5- 164 in the tower coordinate system Load case Mx [Nm] My [Nm] Mxy [Nm] Mz [Nm] PSF [-] Mx max DLC62_u37y30ww30s4 1.10E+08 1.29E+07 1.11E+08 3.42E+06 1.10 Mx min DLC62_u37y240ww30s6 −1.10E+08 −2.97E+07 1.14E+08 −2.53E+06 1.10 My max DLC62_u37y0ww0s1 2.00E+07 7.52E+07 7.78E+07 −5.86E+05 1.10 My min DLC62_u37y180ww0s2 1.02E+07 −9.53E+07 9.59E+07 1.15E+06 1.10 Mxy max DLC62_u37y120ww30s3 8.30E+07 −8.25E+07 1.17E+08 2.78E+06 1.10 Mxy min DLC62_u37y0ww-30s1 −2.90E+04 1.79E+04 3.41E+04 −1.91E+05 1.10 Mz max DLC62_u37y30ww-30s4 4.24E+07 −1.00E+07 4.36E+07 5.80E+06 1.10 Mz min DLC62_u37y330ww-30s1 −7.03E+07 4.96E+06 7.05E+07 −6.79E+06 1.10
- 45. 16 Wind turbine system design 1.2 Design load validation Within the numerical design process, the design loads and system dynamics are calculated and form the basis for the manufacturing process of the turbine. Before being built into a serial product, the results from the numerical design process must be validated on the prototype in the field. This provides proof that the real loads and dynamics are within the limits of the numerical design. Based on the numerical design, a prototype of the WT is built, on which the type certification measurements are carried out. The type certification measurements for load validation are specified in the standard IEC 61400- 13 ‘Measurement of mechanical loads’ [12], as part of the IEC 61400 series, that comprises all IEC stan- dards relevant for WTs. In Chapter 10, details of further IEC standards relevant to the certification process of WTs can be found. The standard of the American Society of Mechanical Engineers (ASME) for Verification and Validation in Computational Solid Mechanics ASME VV 10 defines validation as the ‘process of determining the degree to which the model is an accurate representation of corresponding physical experiments from the perspective of the indented uses of the model’ [13, p. 3]. Accordingly, the case of intended use for which the model is to be validated must be defined. However, the area for which the model can be considered valid after successful validation is not defined by the intended use domain but by the validation experiments that span the validation space (see Figure 2.3- 2 in ASME VV 10 [13, p. 5]). A general process for verification and validation is presented in the standard ASME VV 10 [13] and illustrated in Figure 1.3* . This process should be carried out in parallel with the model development. However, the different steps of verification of aeroelastic models, such as code veri- fication and calculation verification, are not part of this book. This chapter focuses on validation. First, the load measurements to receive the experimental outputs from the physical experimentation branch are described. Then, the simulation results of the modelling and simulation branch are described. 1.2.1 Standard load measurements The measurements of mechanical loads in the design validation follow the standard IEC 61400- 13. The quantities to be measured can be divided into three different groups: • • Load quantities • • Meteorological quantities • • Operational quantities * Figure 3.3- 1 reprinted from ASME VV 10- 2019, by permission of The American Society of Me- chanical Engineers. All rights reserved.
- 46. Load calculation and load validation 17 Meteorological quantities, specified in the standard IEC 61400- 12- 1 [14], com- prise wind speed, wind direction, TI, air density and wind shear for the lower rotor half as mandatory parameters. For the load quantities, the minimum instrumentation according to the standard is given in Table 1.9. Operational parameters comprise electrical power, rotor or generator speed, yaw misalignment, rotor azimuth angle, pitch position and speed, brake status and WT status. The status values are usually taken from the SCADA system of the Figure 1.3 Verification and validation process [13, p. 9]
- 47. 18 Wind turbine system design WT. In the following, the implementation of the corresponding measuring points is described. 1.2.1.1 Blade bending moment For the measurement of the bending moments, classical electrical strain gauges are usually used. Recently, fibre- optic sensors have also been applied more and more frequently. Especially when measuring the bending moments of the rotor blades, the advantage of fibre- optic sensors is relevant, as no additional lightning protection has to be considered with this type of sensor. Table 1.9 Minimum instrumentation for mechanical load measurements according to IEC 61400- 13 Load quantity WT 1500 kW or rotor diameter 75 m WT 1500 kW and rotor diameter 75 m Bending moment blade root flatwise 1 blade mandatory, 2 blades recommended 2 blades mandatory, 3 blades recommended Bending moment blade root edgewise 1 blade mandatory, 2 blades recommended 2 blades mandatory, 3 blades recommended Rotor tilt moment Mandatory Mandatory Rotor yaw moment Mandatory Mandatory Rotor torque Mandatory Mandatory Bending moment tower base (normal) Mandatory Mandatory Bending moment tower base (lateral) Mandatory Mandatory Bending moment tower mid (normal) Recommended Bending moment tower mid (lateral) Recommended Bending moment tower top (normal) Mandatory Bending moment tower top (lateral) Mandatory Bending moment distribution blade flatwise 2 blades mandatory, 3 blades recommended Bending moment distribution blade edgewise 2 blades mandatory, 3 blades recommended Blade torsional frequency and damping Recommended Pitch actuation loads 1 blade mandatory Tower top acceleration (normal) Mandatory, when used for controller feedback Tower top acceleration (lateral) Mandatory, when used for controller feedback Tower torque Mandatory
- 48. Load calculation and load validation 19 The strain gauges in the blade root are installed in the cylindrical part of the rotor blade. Here, two pairs of strain gauges are installed perpendicular to each other so that each sensor of a pair is in 180° opposite position (Figure 1.4). The best posi- tion to place the strain sensors can be determined gravitationally. To do this, the blade is brought into the feather position, and the lowest point in the rotor blade circumference is determined with the help of a cylindrical body. The other three positions can be determined by measuring or using the blade bolts as a reference. Following the standard, the measurement requirements are different for turbines with a rotor diameter of more than 75 m and a rated power of more than 1500 W compared to smaller turbines. Almost all modern WTs fall into the category of larger turbines. Therewith, the measurement of the bending moment distribution in at least two rotor blades is required. Especially here, the advantages of fibre- optic strain sensors come into play to reduce the risk of lightning strikes. To realize the bending moment distribution, typically one additional set of strain sensors is installed further inside the blade. At this position, another challenge is posed in the installation of the sensors. Further into the blade, it is not as straightforward to identify the most suitable position for the strain sensor. Whereas for the leading edge the position is clear, the opposite position at the trailing edge does not allow the installation of the sensor due to the tapered shape of the blade. Here, the position on the pressure or suction side of the blade has to be chosen. For the flatwise bending moment, the best position is in between the webs (the middle position displayed on the right in Figure 1.4). However, depending on the construction of the blade, this position might be difficult to access. Then, the strain gauges are applied on one side of the web on the pressure side and on the other side of the web on the suction side. This alignment is also used for blades with only one web. Typically, signals from the sensors are transmitted to a control cabinet located in the hub of the turbine. When installing the cables, the additional cableways due to pitching have to be considered. This is typically addressed by applying an additional cable loop that expands during pitching. As the measured quantity for both fibre- optic and conventional strain gauges is strain, both systems have to be calibrated to convert strain into the correspond- ing bending moment. The most accurate way to calibrate the strain gauges is the Figure 1.4 Installation of strain sensors in the rotor blade in the blade root (left) and at a larger rotor radius (right)
- 49. 20 Wind turbine system design application of a defined force on the rotor blade. This type of calibration can be carried out well on a test stand, but on a real WT in the field, it involves a large technical and logistical effort and is therefore usually not applied. Here, the mass and the centre of gravity of the rotor blade are used for calibration. By turning the blade into a horizontal position on either side of the turbine, the relation between the applied moment of gravity and the output of the strain gauge reveals the calibration factors for the bending moment at the blade root. To reduce external forces other than the gravitational force during the calibration process, aerodynamic loads on the blade must be minimized. Therefore, during the calibration, the mean wind speed must be below 5 m/s. Alternatively, a data set from the operating turbine, which fulfils the conditions of the calibration, can be extracted and used for the calibration. An additional calibration at the end of the measurement campaign allows for drift correction. 1.2.1.2 Tower moments The measurement of tower moments typically is realized using electrical strain gauges, as most towers are made of armoured concrete and lightning protection is therefore not relevant. Both for larger and smaller turbines, the measurement of the bending moments at the tower bottom is mandatory for normal and lateral direc- tions. For larger turbines, an additional set of sensors is required at the tower top. A third set of sensors is recommended in the middle of the tower. As in the blade root, a set of four strain gauges are installed in pairs in lateral or normal direction. The sensors should be installed not too close to the turbine door and to tower flanges in order to avoid any effects from these elements. For the tower moments, an analytical calibration can be applied. To this end, the mass, the overhang moment and centre of gravity of the nacelle as well as the Young’s modulus are needed as input from the turbine manufacturer. For the cal- culation, additionally, the geometry of the tower is measured. The wall thickness can be quantified using an ultrasonic gauge. To find the offset of the calibration, the nacelle has to be yawed several times over 360°. To minimize aerodynamic effects, the wind speed should be below 5 m/s. 1.2.1.3 Main shaft The measurement of the moments on the main shaft poses an additional challenge to the measurement system. The sensors installed on the shaft are in the rotating system of the turbine, whereas the data acquisition system is typically installed in the non- rotating system of the turbine. The rotational forces pose a higher demand for the stability of the sensor. This is particularly important for the fast- rotating high- speed shaft, which is not part of the certification process. However, also for the low- speed shaft, the rotation of the shaft has to be considered. Besides the rotational forces, data and energy transfer from the rotating to the non- rotating part of the turbine poses additional challenges to the measurement of the main shaft, if a slip ring cannot be used. Here, battery packs mounted on the shaft can serve as a solution.
- 50. Load calculation and load validation 21 1.2.2 Data evaluation process During the measurement campaign, regular plausibility checks are applied to the data. The end of the measurement campaign is reached with the completion of the capture matrix. After a final plausibility check, the data evaluation process starts. The final documentation is implemented by the report, which end is defined by the IEC 61400- 13. Additionally, special measurement load cases (MLCs) have to be ful- filled. Some of these MLCs correspond to specific DLCs. An overview of all MLCs and DLCs is given in Table 1.2. Before evaluation, certain data are rejected from the data set. This includes, e.g., wind directions outside the valid sector, defined by the site evaluation (for details, see Chapter 10) or spikes in the data sets. 1.2.3 Standard load validation Aeroelastic models, like all models, are subject to assumptions about the state and physics they represent. In order to have confidence in the results of the models, they must be validated. The standard IEC 61400- 1 specifies that load calculations must be based on validated methods and approved codes. The aeroelastic simulation model that is used for the specific design calculations must be subsequently vali- dated by measurements on a dynamically and structurally similar turbine. However, they may differ in detail, e.g., in alternative tower designs [10, p. 40]. 1.2.3.1 General information To validate models, output variables are generally compared between the model and the real system to match the response of the model and the real system and to assess whether the model validly describes the real system. This requires the model input to represent the system and its excitation as accurately as possible. However, all model input data are subject to uncertainties. The model input can be divided into system data and surroundings data (cf. Figure 1.5). System data include geometry, initial conditions and physical modelling parameters. Surrounding data include boundary Figure 1.5 Source of uncertainty, adapted from Reference [15]
- 51. 22 Wind turbine system design conditions and system excitation. For WTs, the system excitation is derived from the boundary conditions. For aeroelastic models, geometry includes the geometry of the individual com- ponents: blade and tower as well as the drivetrain, which represents the geometric connection between blade and tower. The blade geometry includes chord length, thickness, twist and pre- bend along the blade. Physical modelling parameters include density, elastic modulus and lift coefficients. The real prevailing wind cannot be measured at every position at every time. Often, only the measurements at one to five measuring points of a permanently installed met mast are available. Therefore, in the validation of aeroelastic models, especially the input of the system excitation is subject to uncertainty and poses a great challenge. 1.2.3.2 Procedure for the validation of load calculation models A procedure to perform the validation of the load model based on measurements is provided in the informative Annex E.1 of IEC 61400- 13 [12, p. 81]. This pro- cedure comprises 10 steps and is reprinted here with the kind permission of the International Electrotechnical Commission (IEC)† . “1) Set validation requirements and acceptance criteria. Specify the conditions for the tests and the acceptance criteria for the results. The acceptance criteria are the maximum allowable differences between the measured loads and the simulated loads for equivalent wind condi- tions. The acceptance criteria may be different for each load component or test condition. 2) Specify the measurement setup which is needed to measure the desired quantities. The starting point is of course the sensors specified in this stand- ard [(i.e., IEC 61400- 13)] but the specific turbine design could require other sensors or measurement techniques. Also special attention should be given to the inflow conditions where additional sensors could be needed to get a suitable understanding of the wind. 3) Assure the model is representative of the real turbine to be tested (main structural data, natural frequencies, controller settings, sensor positions etc.). 4) Rerun model for test turbine based on site specific inflow conditions to ensure the structural integrity of the turbine during the test campaign to be done afterwards. 5) Perform the load measurements campaign. Collect data about turbine performances, loads and inflow conditions during the tests. † IEC 61400- 13 ed.1.0 “Copyright © 2015 IEC Geneva, Switzerland. www.iec.ch”
- 52. Load calculation and load validation 23 6) It is recommended to make a small comparison between measurements and simulations early in the campaign to identify possible errors in the measurements or the model. 7) Create a database for model validation by filtering the measured data. Normally a capture matrix is already defined by the measurement Institute, further filtering can be required to: reduce the number [of] data for a certain wind speed, trending in the wind speed, yaw activity, data quality issues etc. 8) Reproduce test conditions in simulations, by rerunning the model for test turbine with the inflow conditions measured during the tests. Synthetic wind time series can be created to match measured time history at some grid points. 9) Compare simulated loads and measured loads. This comparison can be done based on statistical values, frequency spectral density functions, point by point time series, etc. (see Clause E.2 [in IEC 61400- 14]). Besides load magnitudes, it is recommended to compare also other magnitudes as power, rotor speed, pitch activity, control variables, etc. 10) If there are differences between simulations and measurements look into potential reasons: a) External conditions not captured in the model: low level jets, wind shear, veer, TI, etc. b) Other external conditions; controller settings, location of measured loads vs. model, etc. c) Measurement issues: calibrations mistakes, spikes, cross talks, etc. d) Model issues: profile data, masses, stiffness’s, etc.” Remarks on the proposed procedure are given below: Step 1 states that the validation requirements and acceptance criteria should be defined. Requirements for the acceptance criteria, which describe the maximum deviation between measured and simulated loads, are not given in this standard. In step 2, the measurement setup has to be defined. In addition to the study of the turbine design in documents, a first inspection of the WT and the site is required. For example, the presence of electric or hydraulic pitch requires different sensors, or the unavailability of a slip ring or electrical power supply requires solutions for the data and power transfer from the rotating to the non- rotation system. To assure that the model is representative of the real turbine to be tested, as stated in step 3, the explanations from section 3.1 structural properties of sub- components in the publication by Huhn and Popko [16, p. 3] could be used. Step 4 is especially important if installed measuring equipment can lead to mass and/or aerodynamic imbalance. The latter can be caused, e.g. by pressure sensors installed outside the blade. In step 5, the measurement campaign is conducted. The duration of the cam- paign depends on the environmental conditions. In addition to the different load cases, a so- called capture matrix has to be filled during normal power production. The measured data are categorized into wind speed intervals (wind bins) and TI
- 53. 24 Wind turbine system design intervals (TI bins). For each of these bins, a certain number of time intervals of 10 minutes have to be collected to fill the capture matrix and thus complete the mea- surement campaign. A comparison between measurement and simulation at an early stage of the val- idation campaign, as mentioned in step 6, helps to identify possible differences, e.g., in the coordinate systems of measurement sensors and sensors from the simulation. While adjustments and corrections (including renewed simulations) can usually be carried out without much effort on the simulation side, this is usually not possible afterwards for the real measurement. Filtering the measurement data is an important step in validation. In addition to the aspects mentioned in step 7, it is essential to filter by wind direction. If wind measurements from a met mast are used, wind directions where the met mast is in the wake of the turbine must be excluded. Furthermore, depending on the intended use case, other wind directions should be excluded if necessary. If the focus of the model validation is on the power curve, e.g. only the measurement sectors accord- ing to section 6.3.3 and Annex A from IEC 61400- 12- 1 [14] should be considered. If measured time series of the wind are used as input variables for the creation of the wind fields, a so- called narrow sector, as used by Dimitrov et al. [17], is advan- tageous. The fact that the met mast is located exactly upwind of the WT increases the significance of the wind measured at the met mast in relation to the wind expe- rienced by the WT. In addition, special operating conditions that lead to a power reduction or lower rotor speed must be filtered out. The following are examples: • • Noise reduced mode at night • • Safe mode of turbine control that could occur in conjunction with warnings that may be triggered by high wind direction change or temperature warnings, among other things • • Regulation from grid operator • • A shutdown can be triggered at certain times and seasons when bats fly near the WT. Step 8 mentions the reproduction of the test conditions. As mentioned in section 1.2.3.1, this is a particular challenge and key role in the validation of aeroelastic WT models. The use of synthetic wind fields that satisfy the measured time series at some grid points is a helpful element and especially necessary when the evaluation is performed by point- by- point comparison. The tools TurbSim [18] and PyConTurb [19] are suitable for the generation of these synthetic wind fields, which fulfil the time series at the measured points. The wind profile can be measured using vertical profiler light detection and ranging (LiDAR) systems. The mean values (over the duration of the wind field) of the wind speeds at the different heights can then addi- tionally be used as input for the creation of the synthetic wind field. Step 9 deals with the comparison between simulation and measurement results and mentions statistical values, frequency spectral density functions and point- by- point time series as comparison methods. These are explained in more detail in Annex E.2 of IEC 61400- 13 [12, pp. 82- 85]. It should be considered that, on the one hand, the
- 54. Load calculation and load validation 25 comparison utilizing statistical values leads to a certain loss of information because the simulation and measurement results of 10 minutes are condensed to one single value. On the other hand, when comparing using the point- by- point method, it must be ensured that the synthetic wind field also reflects the measurements over time (cf. step 8). Furthermore, it must be taken into account that the point- by- point comparison is made utilizing plots and thus always represents a subjective observation. In addi- tion, the runtime length between the wind measuring point (e.g. met mast) and the WT must be taken into account. For each individual 10- minute measurement, this can be determined, e.g., using a cross- correlation function. For this purpose, on the one hand, the measured wind speed signal at the wind measuring point and, on the other hand, the signal of the electrical power can be used as an indicator for the wind speed in the rotor plane (neglecting the rotor inertia). If synthetic wind fields that satisfy the time series at the measured points are used, each measurement is reproduced by a specific simulation. Pairs of measurement and simulation are created. In addition to the comparison meth- ods presented in IEC 61400- 13, a linear regression can be performed from these simu- lation and measurement pairs and used as an indicator of the model’s quality [20, 21]. Acknowledgements “The author thanks the International Electrotechnical Commission (IEC) for per- mission to reproduce Information from its International Standards. All such extracts are copyright of IEC, Geneva, Switzerland. All rights reserved. Further information on the IEC is available from www. iec. ch. IEC has no responsibility for the placement and context in which the extracts and contents are reproduced by the author, nor is IEC in any way responsible for the other content or accuracy therein.” References [1] Popko W., Thomas P., Sevinc A, et al. IWES wind turbine IWT- 7.5- 164. rev 4 [online]. 2018. Available from https://doi.org/10.24406/IWES-N-518562 [2] Teßmer J., Daniele E., Balzani C, et al. ‘Schlussbericht SB: smart blades: 01.12.2012- 30.04.2016’. 2016. Available from 10.2314/GBV:871472740 [3] Stammler M. ‘Endurance test strategies for pitch bearings of wind turbines’ [Dissertation]. Fraunhofer IRB- Verlag, Gottfried Wilhelm Leibniz Universität Hannover [4] Leimeister M., Spill S., Dose B, et al. ‘TANDEM Schlussbericht: towards an advanced design of large monopiles’. [in de] Research Gate. 2019. Available from 10.2314/KXP:1678117404 [5] Thomas P., Wächter M., Antoniou A. ‘InterWiLa Schlussbericht: berechnung und validierung von lasten an windenergieanlagen aufgrund intermittenter windfelder’. [in de] 2018. Available from 10.2314/GBV:1028484186 [6] Leimeister M., Bachynski E.E., Muskulus M., Thomas P. ‘Rational upscal- ing of a semi- submersible floating platform supporting a wind turbine’.
- 55. 26 Wind turbine system design Energy Procedia. 2016, vol. 94(1), pp. 434–442. Available from 10.1016/j. egypro.2016.09.212 [7] Leimeister M., Kolios A., Collu M., Thomas P. ‘Larger MW- class floater designs without upscaling?: a direct optimization approach’. ASME 2019 38th International Conference on Ocean, Offshore and Arctic Engineering; Glasgow, Scotland, 2016. pp. 06092019. [8] Robertson A., Jonkman J., Masciola M, et al. Definition of the semisubmers- ible floating system for phase II of OC4 [online] [technical report NREL/ TP- 5000- 60601]. 2014. Available from https://www.nrel.gov/docs/fy14osti/ 60601.pdf [Accessed 2 Nov 2022]. [9] Jonkman J. Definition of the floating system for phase IV of OC3 [online] [Technical Report NREL/TP- 500- 47535]. 2010. Available from https://www. nrel.gov/docs/fy10osti/47535.pdf [Accessed 2 Nov 2022]. [10] IEC ‘Wind turbines- part 1: design requirements’. [IEC 61400- 1] International Electrotechnical Commission, Geneva. 2019. [11] IEC ‘Wind turbines- part 3: design requirements for offshore wind turbines’. [IEC 61400-3] International Electrotechnical Commission, Geneva. 2009. [12] IEC ‘Wind turbines- part 13: measurement of mechanical loads’. [IEC 61400- 13] International Electrotechnical Commission, Geneva. 2015. [13] ASME Standard for verification and validation in computational solid me- chanics: ASME. VV 10. New York, NY: American Society of Mechanical Engineers; 2019. [14] IEC ‘Wind turbines- part 12- 1: power performance measurements of electrici- ty producing wind turbines’. [IEC 61400-12-1] International Electrotechnical Commission, Geneva, Switzerland. 2017. [15] Oberkampf W.L., Roy C.J. Verification and validation in scientific computing. Cambridge: Cambridge University Press; 2010. [16] Huhn M.L., Popko W. ‘Best practice for verification of wind turbine nu- merical models’. Journal of Physics: Conference Series. 2020, vol. 1618(5), p. 052026. Available from 10.1088/17426596/1618/5/052026 [17] Dimitrov N., Borraccino A., Peña A., Natarajan A., Mann J. ‘Wind turbine load validation using lidar‐based wind retrievals [online]’. Wind Energy. 2019, vol. 22(11), pp. 1512–1533. Available from 10.1002/WE.2385 [18] Jonkman B. ‘TurbSim user’s guide v2.00.00: draft version’. 2016. [19] Rinker J.M. ‘PyConTurb: an open- source constrained turbulence generator’. Journal of Physics: Conference Series. 2018, vol. 1037, p. 062032. Available from 10.1088/1742-6596/1037/6/06203 [20] Huhn M.L., Gómez- Mejía A.F. ‘Aeroelastic model validation with 8 MW field measurements: influence of constrained turbulence with focus on power performance’. Journal of Physics: Conference Series. 2022, vol. 2265(3), p. 032058. Available from 10.1088/1742-6596/2265/3/032058 [21] Conti D., Dimitrov N., Peña A. ‘Aeroelastic load validation in wake con- ditions using nacelle- mounted lidar measurements’. Wind Energy Science. 2020, vol. 5(3), pp. 1129–1154. Available from 10.5194/wes-5-1129-2020
- 56. 1 Fraunhofer IWES, Fraunhofer Institute for Wind Energy Systems, Bremerhaven, Germany Chapter 2 Models and simulation Paul Robert Feja1 , Mareike Leimeister1 , and Muhammad Omer Siddiqui1 During the design phase of wind turbine (WT) systems, simulation models are used in order to design and optimise the WT system behaviour and its components. Since (offshore) WTs are complex systems that interact with various environmental condi- tions and other technical systems, it is a considerable challenge to develop simula- tion models in such a way that all relevant influences are reproduced properly. Overall, the content of this chapter comprises approaches for modelling and simulation of WT systems, including hands- on experience in their applications and suitability. Due to the huge diversity of types and fidelity levels of simula- tion models, as introduced in section 2.1, the subsequent elaborations are limited in scope and mainly focus on aerodynamic and mechanical models of WT sys- tems and their components. Since the mechanical behaviour of a WT is not only driven by the structure itself but also strongly affected by environmental influ- ences and resulting loads on the system, the modelling of environmental condi- tions is addressed first (section 2.2). The subsequent description of WT system and component modelling starts from a medium- fidelity level in the general and fully coupled modelling of WTs (section 2.3), including aero-, hydro- and struc- tural dynamics, and pointing out the modelling of structural components while just touching upon the modelling of other components. The final focus, then, lies on drivetrain models (section 2.4), comprising details on modelling approaches and best practices. A short summary and conclusions are provided at the very end (section 2.5). 2.1 Introduction Various approaches may be followed, and one can choose from a large number of types of simulation models of different levels of fidelity. Despite this diversity, stand- ards and guidelines specify requirements for the fidelity of WT simulation models
- 57. 28 Wind turbine system design and provide recommendations for them. Both topics are introduced in the following (sections 2.1.1 and 2.1.2, respectively). 2.1.1 Overview of modelling at different levels of fidelity Considering a WT as an aggregation of several technical subsystems, such as rotor blades, hub, drivetrain (i.e. shaft, bearings and gearbox), generator/inverter system, tower, foundation, etc., various mathematical descriptions of the underlying physics can be taken into account. It is the engineer’s challenge to select, according to the current state of the design phase and intended analyses, the proper tools of an ade- quate level of fidelity to cost- efficiently achieve reliable and helpful results from the utilised simulation tools. For complete system- level modelling of such a complex system – which is a WT, namely dealing with different physical domains, such as structure dynamics, aerodynamics, fluid–structure interaction, composites, dynam- ics of machine elements, hydraulic systems, control systems and electric power sys- tems – a multi- physical approach is required. However, based on the focus area and quantities of interest, WT system and component modelling are usually broken down into single physics or bi- physical domains. In terms of structural analyses and to assess the structural integrity of the WT system, aeroelastic and computational fluid dynamics (CFD) tools are utilised to estimate the aerodynamic loads that are endured by the WT structure. In such a scenario, the internal components of the WT system, such as the drivetrain, electrical systems and hydraulics, are disregarded for analysis. For the analysis of hydraulic circuit performance, only the fluid power circuit is modelled. On the other hand, the dynamic models of the mechanical drivetrain are developed using multibody simulation (MBS) methods. In the classical sense, this approach uses rigid bodies with lumped masses to represent component inertias and force ele- ments for the component stiffness. A flexible MBS approach introduces fully flex- ible bodies, which has become a more common practice for modelling drivetrains in recent years. In principle, the level of fidelity increases as the design process progresses. This means that in the early design phase, i.e. for the conceptual design, rough analyses that, e.g. mainly focus on the global system response and natural fre- quencies (for which reason these early stage analyses may be just performed in the frequency domain) are sufficient and also most appropriate when considering the large number of simulations that need to be performed when still investigating different basic design concepts and solutions. However, as soon as the conceptual design is laid down, the detailed design stage follows, which requires in- depth analyses that are no longer only at the system but also at the component level and include fully coupled approaches to consider interactions between the single com- ponents. Such high fidelity simulations are typically more time- consuming than the low- fidelity ones; however, they provide more accurate results in turn, since even complex physical phenomena, such as non- linearities, are captured in more detail by the underlying high- fidelity simulation models and commonly utilised time domain analyses.
- 58. Models and simulation 29 2.1.2 Requirements of standards for model fidelity Due to the various levels of detail of the models that are capable of representing the behaviour of WTs, national and international standards and guidelines have been agreed on. These standards and guidelines provide a reference for the selection of valid simulation models for a given task. They ensure that simulation results are within a reasonable uncertainty band, regardless of the WT model or site analysed, and they ensure comparability between different simulation tools by defining stand- ardised models for certain tasks. A crucial step in the design of a WT system is the iterative load simulation process in order to estimate the extreme and fatigue loads of the individual compo- nents (cf. Chapter 1). Since these simulation results are of high importance for both the design and the certification process of the WT as a whole as well as its subsys- tems and subcomponents, there are detailed requirements specified on simulation models to be used for load determination and certification, mainly provided in the International Electrotechnical Commission (IEC) standards. This chapter will focus mainly on models described in the following references: • • IEC 61400- 1 [1] • • IEC 61400- 3 [2] • • IEC 61400- 4 [3] • • DNVGL-ST-0437 [4] The first two elements, as well as the last one on this list, focus on the model require- ments for global load analysis of onshore and (bottom- fixed) offshore WTs. These docu- ments give advice on how aeroelastic simulation tools should be implemented in order to estimate the global WT loads used for design (cf. section 2.1.2.1). The third item, however, focuses on the design of WT gearboxes (cf. section 2.1.2.2), therefore putting a focus on higher dynamics and more complex mechanical models in comparison to global aeroelastic simulation models. These gearbox models require a higher level of detail with respect to mechanical components, so typically other and more specialised MBS software is used. 2.1.2.1 Requirements for global load simulation The international standard IEC 61400- 1 [1] defines the requirements for load simulation of (onshore) WTs and specifies wind conditions and design situations by defining load cases. Furthermore, for determining WT design loads, the use of dynamic aeroelastic simulation models is required by the standard, and it is described what is sufficient to meet these requirements. Suitable simulation codes should be validated subsequently by measurements and must take into account at least the following types of loads: • • gravitational and inertial loads • • aerodynamic loads • • actuation loads, mainly caused by the WT controller
- 59. 30 Wind turbine system design • • other loads, e.g. wake loads, impact and ice loads as well as vortex- induced tower vibrations More specifically, IEC 61400- 1 lists relevant effects that should be taken into account by using appropriate models, which will be described in this chapter. Among the most relevant ones are wind field perturbations due to the WT (e.g. tower shadow or wake- induced velocities), three- dimensional (3D) flow phenomena, such as tip and hub losses as well as 3D stall, unsteady aerodynamic effects and the realis- tic representation of the influence of the control system behaviour on the WT. These requirements lead to the application of the fully coupled aero- servo- elastic simula- tion tools that are typically used for load calculation in the WT design. With respect to modelling the environmental conditions causing loading of the considered WT during design load calculations, the IEC 61400- 1 standard mentions two turbulence models for representation of the wind inflow: the Mann uniform shear model and the Kaimal spectral and exponential coherence model. Icing of the rotor blades shall be accounted for by applying profile coefficient modifications to the airfoil characteristics as well as adding additional mass to the rotor blades. Therefore, only model input parameter corrections, and not the implementation of specific models, are required to consider rotor blade icing during load simulations. Adding up to IEC 61400- 1, the international standard IEC 61400- 3 [2] focuses on the additional assessment requirements to be considered for (bottom- fixed) off- shore WTs. Since the models used for representing the WT and the aerodynamic wind loads are the same as for onshore WTs, this standard focuses mainly on the influence of specific offshore loads, such as sea loads and sea ice loads. In particular, the following additional loads have to be considered and therefore require specific models in load simulation to capture aero- hydro- servo- elastic dynamics: • • wave loads, considering different sea states as well as breaking waves • • sea currents • • differences in water level (e.g. due to tides) • • sea ice • • marine growth • • seabed movement, including scour All loads shall be calculated utilising full dynamic simulations and the appropri- ate modelling approaches. The IEC standard points out the relevance of correct rep- resentation of various sources of system damping, such as aerodynamic damping, hydrodynamic damping, structural damping or soil energy dissipation. Further guid- ance on the calculation of hydrodynamic loads is given by referring to the Morison equation as a standard method as well as the MacCamy- Fuchs approach to account for diffraction effects. The most relevant models for the above- mentioned loads will be presented in this chapter. Besides the IEC standards, Det Norske Veritas (DNV), formerly DNV GL, provides supplementary standards and recommended practices, such as the stan- dard DNVGL-ST-0437 Loads and site conditions for wind turbines [4], which is
- 60. Models and simulation 31 applicable to onshore and offshore WTs. This standard defines normal and extreme turbulence models, which differ from the models given in IEC 61400- 1 [2] to account for offshore wind conditions. Furthermore, DNVGL- ST- 0437 gives addi- tional advice on the estimation of environmental conditions. With respect to the aeroelastic WT simulation model, it points out the importance of non- linear effects to be accounted for where important (e.g. for soil–structure interaction), while linear elastic theory is considered to be applicable for the structural dynamics, e.g. for blade and tower/support structure elasticity. Some more specific requirements on model fidelity, such as consideration of bearing friction moments, elastic machinery mountings (if relevant) or rotor mass eccentricity, are mentioned as well. 2.1.2.2 Requirements for gearbox and drivetrain simulation Detailed knowledge of the drivetrain dynamics is of major importance for the evalu- ation of the design loads, and dynamic analysis of the drivetrain forms a mandatory part of the certification of WTs. The primary aim of drivetrain dynamic models, apart from component load calculation, is to investigate and avoid the occurrence of possible resonances in the drivetrain during normal operation [5]. As drivetrain models in aeroelastic codes are usually simplified with very few degrees of freedom (DOFs) for the global WT analysis, the dynamic analysis of the drivetrain is mostly performed in a separate analysis with a more detailed drivetrain model. While the WT structural analysis methods have been well established over the past decades, the methods for drivetrain dynamic analysis have still not achieved the same level of maturity in terms of standardisation. The standard practices for designing WT gearbox systems follow the Germanischer Lloyd (GL) guidelines [6, 7], AGMA 6006 [8] and IEC 61400- 4 [3]. For standard practices of modelling the drivetrain for dynamic analysis, very limited material is available. To tackle this issue, GL has set the minimum modelling requirements for drivetrain dynamic analysis, which are summarised in Table 2.1. These requirements emphasise the use of multi- DOF models. However, the GL guidelines also offer the possibility to use the classic torsional drivetrain models in combination with complementary mea- surements. These measurements can be carried out either at a gearbox test bench or taken from drivetrain system testing. The reason for including these measurements is to verify the modelling assumptions and to identify any missing eigenfrequencies and mode shapes not covered by the simulation model. 2.2 Modelling of environmental conditions WTs have to face various environmental impacts due to site- specific wind and soil conditions as well as, in the case of offshore systems, sea and marine conditions. An overview of environmental conditions for the example of a bottom- fixed offshore WT system is presented in Figure 2.1. The commonality of these environmental factors is their stochastic nature. Thus, e.g. wind, waves and currents are never constant but fluctuate. Similarly, the soil characteristics change over depth and time depending on further impacts, such as
- 61. 32 Wind turbine system design installation method, decommissioning or scour. Icing of blades or sea ice impact only occurs in some climates at certain times of the year and is subject to random- ness and the characteristics of nature. And also the marine growth on an offshore WT structure is ever changing due to the living organisms, and its development depends on the sea conditions and structural characteristics. Overall, the environmental conditions are dependent on the specific site of the WT, its surroundings, the elevation or depth, the time and season and further com- plex correlations. Due to the random and stochastic nature of the environmental conditions and their dependency factors, there is therefore always a certain degree of uncertainty prevailing when trying to model environmental impacts and condi- tions. In the following, different methods for representing environmental conditions in numerical models are presented, grouped into wind (section 2.2.1), sea (section 2.2.2) and soil (section 2.2.3) conditions. 2.2.1 Modelling of wind conditions For WT systems, the wind is actually the resource for energy production. Due to its fluctuating character, the wind can be best described by a wind speed Table 2.1 Minimum requirements for modelling drivetrain components for dynamic analysis according to GL guideline [6, 7] Component Minimum model Minimum degrees of freedom Hub Rigid body Torsional, axial, bending Main shaft Two rigid bodies (flexible recommended) Torsional, axial, bending Low-speed shaft coupling Rigid body Torsional, axial, bending Gearbox housing Rigid body (flexible recommended) Torsional, axial, bending Planet carrier Rigid body (flexible recommended) Torsional, axial, bending Gearbox shafts Three rigid bodies (flexible recommended) Torsional, axial, bending Gearbox gears Rigid bodies Translational, axial, bending Gearbox support Spring-damper element Translational Brake disc Rigid body Torsional, axial, bending High-speed shaft coupling Three rigid bodies (flexible recommended) Torsional, axial, bending Generator rotor Rigid body Torsional, axial, bending Generator housing Rigid body Torsional, translational Generator legs Spring-damper element Translational Main frame Rigid body (flexible recommended) In compliance with model of component Bearings Spring-damper element Full stiffness matrix recommended
- 62. Models and simulation 33 time series. Most commonly, met- masts and light detection and ranging (Lidar) systems are used to record such site- specific wind speed time series at different heights and over the years. Additionally, the variable direction of the fluctuat- ing wind speed is captured by measuring the three components (e.g. horizontal x-component, horizontal y- component and vertical z- component) of the wind speed vector separately. To reflect the characteristics of real wind speed time series and naturally occurring wind- related environmental conditions in numeri- cal models, different aspects need to be included, which are addressed in the following. 2.2.1.1 Wind turbulence and power spectral densities The characteristics of a turbulent wind speed time series can be expressed by the corresponding power spectral density. Traditional wind speed spectra con- tain most of the energy at very low frequencies. Thus, the shape of a wind speed spectrum resembles the curve of a power function with a negative exponent. This is also reflected in the equations of wind spectral models, which commonly depend on the frequency f , the wind speed Vhubat hub height zhuband the turbu- lence intensity, mostly represented by the standard deviation of the wind speed. Based on the example of the Kaimal model and considering only the longitudinal component of the wind velocity (corresponding to index 1), the power spectral density function S1 f is presented in (2.1) [1]. Currents and tides Scour Marine growth Soil mechanics (Extreme) waves Sea ice impact Tidal depth variantions Wake Icing Turbulent wind and gusts Figure 2.1 Environmental conditions at an offshore WT system ©Fraunhofer IWES
- 63. 34 Wind turbine system design S1 f = 0.052 1 ƒ1 Vhub 2 3 f 5 3 with ƒ1 = 8 : 0.7zhub for zhub 60 m 42 m for zhub 60 m (2.1) While the Kaimal model [9] and the Mann model [10, 11] are mentioned in the stan- dard IEC 61400- 1 [1], tools for generating turbulent wind speed time series, such as TurbSim by the National Renewable Energy Laboratory (NREL) [12], support, among others, the Kaimal model and von Kármán model [13]. Beyond this, special wind spectral models like the Frøya model [14] are recommended and supported for representing extreme wind conditions, such as hurricane winds [12, 15, 16]. As mentioned before and apparent from (2.1), the turbulence intensity feeds into the power spectral density function via the corresponding standard deviation. While the turbulence intensity I , which is derived from the standard deviation following (2.2), decreases with the wind speed, the standard deviation of the turbulent wind speed increases with the wind speed. Depending on the condition considered (distin- guishing between normal and extreme turbulence models), the turbulence standard deviation is computed differently. I1 = 1 Vhub (2.2) 2.2.1.2 Wind speed distributions in space and time The wind speed distribution in space is more commonly referred to as wind shear. Due to the non- slip condition and in dependence on the roughness of the environment (i.e. the roughness of the ground or sea surface if offshore), which is influenced by surround- ing objects (e.g. the landscape, trees, buildings) as well, the wind speed increases from low altitudes to higher ones and follows a logarithmic profile. This sheared profile can be expressed by a power law, setting the wind speeds ( V ) and the associated heights (z) in relation to each other, as given in (2.3). The power law exponent ˛depends on the roughness as well as on the conditions considered, i.e. normal or extreme. Thus, values of 0.2 and 0.14 are recommended for normal wind conditions onshore and offshore, respectively, while a value of 0.11 should be taken for extreme wind conditions [1, 2]. V z1 = z1 z2 ˛ V z2 (2.3) The distribution of the wind speed in time is the probability distribution of dis- tinct wind speeds or wind speed bins. Commonly, the distribution is derived for a specific site based on 10 min average wind speeds from measurement data over at least a year or averaged over several years. If measurement data is not available or a distribution function should be fitted to the measurements, a Weibull or Rayleigh distribution is often used.
- 64. Models and simulation 35 2.2.1.3 Extreme wind and gust models Apart from the extreme turbulence model, there are other extreme events that need to be reflected in the numerical models. Among them is the extreme wind speed model, which aims at representing extreme wind speeds that are expected to occur with a certain return frequency. Based on the design standards for WTs, return peri- ods of 1 year and 50 years are considered. However, due to climate change and more extreme environmental conditions, it is expected that even more extreme cases would need to be taken into account when designing WT systems. Special consid- eration must be given as well to regional extreme wind conditions like hurricanes, typhoons or cyclones [15]. Other extreme wind conditions are, e.g. extreme operating gusts, meaning a sudden and significant increase in wind speed for a very short period of time, or extreme direction changes. These events could, of course, also happen simultane- ously. Furthermore, the wind shear might be extreme as well and have an impact on the loading over the rotor plane. Another extreme and special condition can be the icing of blades. This is not directly related to a wind or turbulence model; however, it needs to be considered in the corresponding aerodynamic load calculations due to the affected shape of the blades, as well as in the loads acting on the blades and entire WT due to gravity and inertia. 2.2.1.4 Tower shadow and wakes The ambient wind flow gets disturbed by the presence of a WT. The two main aspects that need to be addressed at this point are the tower shadow and wake effects. While the wake of a WT mainly affects other WTs that are located in the lee behind it, the tower shadow effect has a direct impact on the WT itself. The tower influences the undisturbed ambient wind flow in such a way that the wind speed is reduced at that point of time when a blade passes the tower. This affects the power output and is visible in the 3P oscillations of the power output, considering a com- mon three- bladed WT. The tower shadow can be mainly modelled based on the flow around a cylindrical structure and potential flow theory. The wake of a WT, on the other hand, plays a key role in the wind farm’s aero- dynamics. A WT that operates in the wake of another WT no longer experiences an undisturbed inflow. The wake behind a WT is characterised by increased turbu- lence intensity and reduced wind speed. These aspects mainly affect the loads on the WT operating in a wake as well as the performance and power production. Due to these wake effects, WTs are placed at certain distances to each other in a wind farm, and the park layout is designed such that the main wind flow directions are incorporated as well. To take the wake effects, including their amplification when considering several rows of WTs, into account in the numerical modelling of a wind farm, different models can be applied, which can be grouped into analytical or semi- analytical models, such as the N.O. Jensen model [17], the Sten Frandsen model [18, 19] or dynamic wake meandering models [20–22] and CFD- type models, such as
- 65. 36 Wind turbine system design linearised Navier–Stokes, parabolised Navier–Stokes, Reynolds- averaged Navier– Stokes (RANS), detached eddy simulations or large eddy simulation (LES). 2.2.2 Modelling of sea conditions In the case of offshore WTs, additional environmental impacts have to be taken into account. These may be sourced from waves, currents, sea ice and marine growth. 2.2.2.1 Wave models Similar to the wind, waves are also stochastic. Thus, to represent the irregular char- acteristics of the time series for the wave elevation, the power spectral density is uti- lised. There are two most commonly followed approaches: the Pierson–Moskowitz spectrum SPMand the Joint North Sea Wave Project (JONSWAP) spectrum SJ[2]. While the Pierson–Moskowitz spectrum, as presented in (2.4), represents a fully developed sea, an extension and modification are established within JONSWAP to represent a developing sea. Thus, additionally, the normalising factor C (2.5) and the peak enhancement factor ˛ , which both depend on the peak- shape parameter (2.6), are comprised in (2.7) for the JONSWAP spectrum* . This spectrum model, however, is limited to only a low amount of swell being present and is mainly repre- sentative of the North Sea and rather shallow waters. SPM f = 0.3125 H2 s f 4 p f5 exp 1.25 fp f 4 # (2.4) C = 1 0.287 ln (2.5) = 8̂ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ :̂ 5 for Tp p Hs 3.6 exp 5.75 1.15 Tp p Hs for 3.6 Tp p Hs 5 1 for Tp p Hs 5 (2.6) SJ f = C SPM f ˛ with ˛ = exp f fp 2 22f 2 p # with = 8 : 0.07 for f fp 0.09 for f fp (2.7) In any of these spectrum models, the irregular wave elevation time series is specified by the significant wave height Hsand peak spectral period Tp(or the corresponding * The JONSWAP spectrum can be transferred into the Pierson- Moskowitz spectrum if the peak- shape parameter is set equal to 1.
- 66. Models and simulation 37 peak frequency fp ). Depending on the specific type considered – distinguishing between normal, severe or extreme sea states – the significant wave height and peak spectral period are defined, so that, e.g. a 50- year event can be represented in the numerical models as well. Specific adjustments and enhancements may be required to account for and represent other extreme conditions, such as hurricanes and the associated wave fields generated [15]. Furthermore, the wave direction is essential for describing the sea state, for which reason a directional wave spectrum, including some directional spreading function, is often utilised. Apart from the power spectral density- based approach, the irregular wave ele- vation time series can also be seen as the superimposition of a set of regular wave elevation time series of different amplitudes (i.e. wave heights), frequencies (i.e. wave frequencies which are related to the wavelengths and wave numbers as well) and phase angles. Depending on the specific site and wave conditions, i.e. the water depth, wave height and wave period, different wave theories are applicable for deter- mining the wave kinematics [23]: • • Linear Airy wave theory, whose approximation – namely that the waves are linear- harmonic – is valid only for deep and intermediate waters under the con- dition that the waves are not very steep; • • Stokes wave theories, which can be of different orders – depending on the num- ber of additional harmonic waves added to the basic linear theory to include corrections for non- linear effects – but again are only applicable in deep and intermediate waters, while they can represent steep waves and can, in the case of deep waters and high- order Stokes waves, even be applied up to the breaking wave limit; • • Stream function theory of Dean, which has the same application range as Stokes wave theories and also considers a set of harmonic waves, but simultaneously rather than sequentially as in Stokes’ theory; • • Cnoidal theory, which accounts for non- linear effects due to the limited depth in shallow waters and, hence, is applicable in shallow waters as well as to steeper waves in intermediate waters; or • • Solitary wave theory, which describes a soliton and represents in shallow and also intermediate waters the very steep waves and waves that are about to break. The maximum wave height for a certain wave period and at a specific water depth is limited. At that limit, the wave is breaking. Since wave kinematics just up to the still water level can be obtained following the linear wave theory, commonly, stretching methods are applied supplementary to con- sidering the actual water level elevation. Such wave stretching methods may follow an extrapolation, a vertical stretching or apply the stretching method by Wheeler [24]. 2.2.2.2 Current types and models Sea currents can be classified by the depth range where they exist as well as the source causing the current.
- 67. 38 Wind turbine system design While currents that are generated by the wind are just felt to a limited depth below the water surface and are, hence, called near- surface currents, currents that stem from storm surge, tides as well as density, temperature, pressure or salinity gra- dients are present over the entire water depth and are, hence, called sub- surface cur- rents. Towards the seabed, however, the current velocity decreases due to friction. In the preceding classification according to the respective depth range of occur- rence, wind, storms, tides or any gradients are already mentioned as sources for cur- rents. Beyond that, another source that causes a third current type has to be listed: breaking waves. These wave- induced surf currents are only present close to the shore. Additional sources of currents may be specific local topographies or estuaries as well as Coriolis forces or other causes leading to ocean circulations on a large scale or eddy currents. Depending on the current type, the corresponding current velocity may be mod- elled as a constant, as a uniform flow in the horizontal direction for a certain depth range below the water surface or as a velocity distribution over the depth following, e.g. a linear or power function. The overall current impact may then be reflected by the superimposition of the different single current models. 2.2.2.3 Modelling of sea ice In cold regions or winter periods, sea ice may be present and, hence, act as another environmental impact on an offshore WT.As for the other environmental conditions, the occurrence and characteristics of sea ice highly depend on the site, time and environment. Thus, it is always best if some real data exists that can be taken as a basis for modelling sea ice. One characteristic of sea ice is its thickness hice . After the frost season, this can be approximated as a value in metres according to (2.8), with the 1- day mean tem- perature mean . hice = 0.032 p 0.9Kmax 50 with Kmax = X days ˇ ˇmean day ˇ ˇ for mean 0ı C (2.8) Further characteristics of sea ice are, among others, its crushing strength, bending strength and velocity of motion. In general, it is recommended to derive sea ice mod- els and the behaviour of sea ice impacting the offshore structure based on ice model tests and detailed site assessments [2]. 2.2.2.4 Modelling of marine growth Due to the presence of living organisms in the oceans, these may settle on parts of the offshore structure, either those that are fully submerged or those that are just periodically under water or in the splash zone. The amount, type and distribution of living organisms on the support structure highly depend on the location of the offshore WT.
- 68. Models and simulation 39 Marine growth is commonly not represented by means of a separate model but taken into account by adjusting certain parameters for the depths where marine growth occurs. This implies larger outer dimensions of the affected structural parts and – depending on the type of living organisms – modified surface properties, which both are especially relevant for the hydrodynamic load calculation. Furthermore, the additional mass due to the marine growth needs to be modelled as distributed mass over the affected part of the support structure using an appropriate density (depend- ing on the type of living organisms). In this way, the impact of marine growth on the total system mass, its frequencies and dynamic response can be reflected. 2.2.3 Modelling of soil conditions For bottom- fixed (onshore or offshore) WTs, the soil is of substantial relevance for the stability of the entire system. The soil properties are highly site- specific and even vary across the depth due to different soil layers. Some main soil characteristics, such as stiffness, damping and shear strength, can be deduced from the complex geophysics; however, geotechnical investigations and cone penetration tests should preferably be conducted at the site of interest. While the soil–structure interaction can simply be modelled as a clamped beam connection of the foundation to the ground and, hence, represents just a rigid foundation, more detailed and advanced soil models may be built based on the site- specific soil parameters. Thus, the infor- mation on the soil characteristics may be fed into a stiffness matrix to represent the soil and soil–structure interaction. An apparent fixity method may be utilised to reflect the rather flexible characteristics of the foundation. Even the p–y approach can be followed in numerical modelling by the implementation of non- linear springs as well as dampers, distributed along the depth of the structure in the soil. In the case of offshore WT systems, additional wave–soil–structure interactions have to be taken into account. These include scour, which might occur locally or also globally, as well as movements of the seabed, such as soil settlement or mud- slides. Such effects may be reflected by adjustment of the soil characteristics and the embedded length of the structure, which may be implemented in the numerical soil models in a time- dependent manner, allowing for consideration of any scour protec- tive measures during maintenance and repair work offshore. Other soil mechanics that may occur both on- and offshore are earthquakes. The impact of earthquakes on WT structures is mostly represented by the resulting soil acceleration on the ground. This may be derived from time series or a response spectrum [1]. 2.3 Fully coupled wind turbine modelling To determine the load assumptions for WT system design, fully coupled simula- tions of the WT are required by the standards, as described above in section 2.1.2. In the following section, an introduction to the most commonly used models for global load calculation will be given. Since WTs are dominated by the aerodynamic
- 69. 40 Wind turbine system design behaviour of their rotor blades, emphasis will be put on the description of the aero- dynamic models. 2.3.1 Aeroelasticity and standard tools Over the years, several codes have been developed for the purpose of aeroelastic WT simulation. While some of them originate from other multi- purpose dynamic simulation software, most codes commonly used by both industry and academia have been specifically developed for WT simulations. Among the most commonly used and mentioned codes are Bladed [25], HAWC2 [26], Flex [27] as well as the free and open- source software OpenFAST [28]. These dedicated aero- servo- elastic tools are used in industry, but some WT original equipment manufacturers (OEMs) have also developed their own tailor- made in- house software. Such in- house devel- opments allow WT manufacturers to put a specific focus on real turbine challenges that may arise during the design or validation stages and can be directly adapted in code development. These aeroelastic codes couple the calculation of aerodynamics with a struc- tural solver in order to be able to determine the material loads at various locations of the WT. The (time- varying) loads from an aerodynamic model are applied to the structural model to calculate its dynamic response. The deflections of the structural components are in turn provided to the aerodynamic (and – in the case of offshore WTs – also hydrodynamic) models to update the respective loads. For design load estimation, WTs are modelled as aeroelastic multibody systems. 2.3.2 Aerodynamic models Modern WTs use aerodynamically tailored rotor blades to transform the kinetic energy of the wind into (rotational) mechanical energy that can in turn be used to generate electricity. This section will describe the fundamentals of aerodynamics and efficiency of horizontal axis WTs. The classical aerodynamic analysis for WTs was developed at the beginning of the 20th century and is based on momentum theory as well as blade element theory. Combining these two, creating the so- called blade element momentum (BEM) theory, allows the analysis of an idealised rotor. Several additional correction models have been developed and implemented in order to account for more realistic simulation results, as computation capabilities and WT size have increased over the years. This section will focus on an introduction to the BEM theory, since it is the most common modelling approach to determine the aerodynamics of a WT rotor for design load calculation. 2.3.2.1 Momentum theory The power conversion from the wind’s kinetic energy to mechanical energy of a horizontal axis WT can be described using fundamental physical equations. First, the WT will be described as an ideal non- rotating actuator disc. In a second step, this simplified one- dimensional (1D) momentum theory will be extended to account for rotational effects.
- 70. Models and simulation 41 Actuator disc model: 1D momentum theory Considering an area A , the perpendicular material flow P mpassing this area can be described as follows: P m = v A (2.9) where is the air density, and vis the (local) wind speed. Basic consideration of the kinetic energy per time gives the theoretical maximum of the power that could be extracted from this flow stream: Pwind = 1 2 v3A (2.10) However, not all the energy of this flow can be extracted from the wind. Since the kinetic power of the wind depends on the wind speed, any power extraction by a WT in the considered rotor plane and its conversion to mechanical energy will result in a wind speed deficit behind the rotor, since the total mass flow does not change. Furthermore, any reduction in the wind speed leads to an increase in the area passed by the air, meaning that the flow stream widens after the rotor as it slows down. Assuming stationary, incompressible and frictionless flow, the extracted power can be described by the difference between the undisturbed flow’s power before the rotor plane and its power after the rotor plane, with V0being the undisturbed wind speed upstream of the rotor and 1the wind speed in the wake, as shown in Figure 2.2. Considering conservation of mass, i.e. P m0 = P m1 = P m , this difference can be expressed as Pex = 1 2 V3 0A0 1 2 v3 1A1 = 1 2 P m V2 0 v2 1 (2.11) Even though this equation might indicate that power extraction could be maximised for v1 = 0 , i.e. reducing the flow velocity to zero, this solution is physically impos- sible, since it would require the mass flow to either vanish or accumulate in the rotor plane (which is clearly not the case). The wind speed deficit behind the rotor must result from a thrust force F act- ing on the fluid, which is opposite to the main wind direction. This force can be described by the change of the fluid momentum as well as by a local pressure drop pR = p+ R p R over the rotor plane: F = P m V0 v1 = vAR V0 v1 = pR AR (2.12) where ARis the (known) rotor area, p+ Rand p R are the local fluid pressures right before and behind the rotor plane and vis the wind speed in the rotor plane, as indicated in Figure 2.3. Due to the energy extraction, Bernoulli’s equation cannot be applied over the rotor plane, but it is possible to formulate it twice: from the undisturbed inflow with wind speed V0and stream tube cross section A0to right in front of the rotor, as well as a second time from right behind the rotor to the free wake flow with reduced wind speed v1 , as shown in Figure 2.3. This results in the following equations, assuming that the free flow in the wake has reached ambient pressure p0again:
- 71. 42 Wind turbine system design p0 + 1 2 V2 0 = p+ R + 1 2 v2 p R + 1 2 v2 = p0 + 1 2 v2 1 Combining these equations gives an expression for the pressure drop over the rotor plane: pR = p+ R p R = 1 2 V2 0 v2 1 (2.13) Together with (2.12), it is possible to derive an equation for the wind speed at the rotor plane, which is exactly the mean of the inflow and wake wind speeds: v = V0 + v1 2 (2.14) This surprisingly simple finding allows for further consideration of the optimal power output from such an idealised wind stream tube. Defining the power coef- ficient Cpas the ratio of the extracted energy (2.11) and the available wind power (2.10) in a stream tube of cross section ARas Figure 2.2 Stream flow at a WT according to momentum theory ©Fraunhofer IWES
- 72. Models and simulation 43 Cp = Pex 1 2 V 3 0 AR (2.15) allows to obtain a non- dimensional value for the power output to be further anal- ysed. Next, the axial induction factor acan be defined, which describes the wind speed reduction as a factor of the undisturbed inflow velocity: v = (1 a) V0 (2.16) Similarly, with (2.14), the wake wind speed can be described as v1 = (1 2a) V0 (2.17) Rearranging and inserting (2.11), (2.16) and (2.17) into (2.15) eventually lead to a formulation of Cpas a function of the axial induction factor, which is depicted in Figure 2.4: Cp = 4a 1 a 2 (2.18) This function obviously has a maximum, which can be determined by finding the derivative and setting it to zero, at a = 1/3 . This means that an optimal WT rotor will reduce the undisturbed inflow wind speed by one- third until it arrives at the rotor plane, v = 2/3 V0 . This in turn, as can be seen from (2.17), will result in a Figure 2.3 Pressures and thrust force at the WT as assumed by momentum theory ©Fraunhofer IWES
- 73. 44 Wind turbine system design total far wake wind speed deficit of two- thirds, i.e. v1 = 1/3 V0 . Furthermore, the power coefficient has a theoretical maximum of Cp,max = 16/27 0.593 . This means that no WT can extract more than approximately 60 per cent of the wind’s theoretical kinetic energy. This important relationship is known as the Betz limit, named after the physicist Albert Betz [29]. Although this finding is crucial for understanding WT aerodynamics in general, the above-mentioned theory does not consider the WT itself as a mechanical system but rather as an idealised extrac- tion of power from the wind. To describe the influence and behaviour of the rotor and its blades in more detail, further considerations need to be taken into account, as will be explained in section 2.3.2.2. Momentum theory with rotational effects The idealised actuator disc model does not expect any rotor rotation and the cor- responding torque, which acts both on the rotor as well as on the fluid flow. This torque, as a reaction, will impose an additional rotational movement on the wake flow, which is also present in the rotor plane, reducing the usable power that can be extracted from the wind. First of all, the momentum theory is expanded to describe this phenomenon. Therefore, an angular or tangential induction factor Figure 2.4 Power coefficient as a function of the axial induction factor according to momentum theory
- 74. Models and simulation 45 a0can be defined to quantify the induced rotational movement of the flow com- pared to the non- rotating inflow wind upstream of the WT. It is assumed that the introduced 1D momentum theory is valid for an annular ring with thickness dr and local radius rof the rotor plane, as depicted in Figure 2.5, while the induc- tion factors and pressure change with the local radius. The rotation of the rotor at angular velocity Rinduces a tangential velocity component vtto the air, which is opposite to the rotor blades’ rotation and varies with the local radius: vt = r a0 (2.19) Similar as for the axial induction factor, the tangential induction factor is the mean between the tangential component of the undisturbed inflow of V0,t = 0and the wake value of v1,t = 2 r a0 (2.20) The overall rotor area can be discretised into several annular stream tubes. These stream tubes are assumed not to influence each other, i.e. there is no radial flow from one tube to another, and no further forces are acting between the individual annular tubes. Furthermore, there is no variation of the forces with respect to the azimuth angle, which means that a rotor with infinite blades is assumed. With these assump- tions, from the momentum theory an equation for the local thrust force on this annu- lar ring of cross section ARing = 2r drcan be formulated, similar to (2.12): dF = (V0 v1) d P m = (V0 v1)v2r dr (2.21) With the axial induction factor, this can further be written as dF = 4a(1 a)V2 0r dr (2.22) Taking into account the rotational interaction between the rotor and the fluid flow, an equation for the torque on the annular element dM can be established in the same way and by considering the induced tangential velocity component in the wake v1,t : r Figure 2.5 Annular stream tube cross section in the rotor plane ©Fraunhofer IWES
- 75. 46 Wind turbine system design dM = rv1,td P m = v1,tv2r2dr (2.23) Again, with the tangential induction factor, this can be formulated as dM = 4a0(1 a)V0r3dr (2.24) 2.3.2.2 Blade element momentum theory Based on the described momentum theory, which focuses on the wind resource as an idealised fluid stream, an additional theory was developed by Glauert in the 1930s that takes the influence of the rotor blade into account [30] and extends the 1D momentum theory with respect to the local forces that act on the blades. This theory is best known as the BEM theory and is the classical formulation for WT aerodynamics. Figure 2.6 shows the local cross section of a rotor blade located in a stream tube at radius r . The blade has the chord length cas well as local pitch angle ‚, and the rotor rotates in such a way that the depicted blade cross section moves from right to left. The blade cross section experiences a lift force FLand a drag force FD , combining to the total resulting force R . This force can be expressed with a normal component FNand a tangential component FT , of which the latter one drives the rotor rotation. All experienced velocities are seen from an observer moving with the Figure 2.6 Local relative velocities and forces at an airfoil cross section ©Fraunhofer IWES
- 76. Models and simulation 47 blade. The axial induction reduces the local wind speed in the main flow direction, as indicated in Figure 2.6. The total relative inflow velocity Vrelis the vector sum of this normal wind component and the tangential blade’s velocity relative to the fluid velocity in the rotor plane. This relative tangential velocity in turn is the sum of the blade’s velocity due to the rotor rotation rand the tangential induced fluid velocity vt (cf. (2.19)). These two tangential components act in the same direction, since the induced fluid velocity is a reaction to the rotor rotation. The total angle of the rela- tive wind ˆis the sum of the local pitch angle and the angle of attack, ˆ = ‚ + ˛ (2.25) and can be calculated by geometrical considerations based on the following math- ematical relationship: tan ‚ = (1 a)V0 (1 + a0)r (2.26) At the blade section, the local lift and drag forces FLand FD , respectively, acting on the blade can be determined by making use of the airfoil’s lift and drag coef- ficients Cland Cd , respectively. The values for the lift and drag coefficients can be determined experimentally or by means of simulation and are usually available in tabulated form for different airfoil geometries as a function of the local angle of attack ˛for different Reynolds numbers. These coefficients describe the relationship between the experienced lift and drag forces dFLand dFD , respectively, relative to the local dynamic pressure due to the relative wind speed: Cl = dFL 1 2 V2 relc dr (2.27) Cd = dFD 1 2 V2 relc dr (2.28) With this data, the resulting local force Racting on the blade can be calculated as a vector sum of lift and drag forces. From Figure 2.6, it is visible that lift and drag forces are defined with respect to the relative flow direction. With simple geomet- ric relations, the force components normal and tangential to the rotor plane can be found: dFN = dFL cos ˆ + dFD sin ˆ (2.29) dFT = dFL sin ˆ dFD cos ˆ (2.30) In the same way, lift and drag coefficients can be combined to compute the resulting normal and tangential force coefficients Cnand Ct , respectively. This, however, will not be shown here. The real rotor has a finite number of rotor blades B. The above-mentioned forces can be combined, so that the total thrust force and the torque resulting from the tangential forces in the annular stream tube can be written as dF = B dFN (2.31)
- 77. 48 Wind turbine system design dM = B r dFT (2.32) Now, these forces, which were derived from the local velocities at the blade, describe the same total loads on the annular stream tube as (2.22) and (2.24), which were derived from the momentum theory. With some algebraic manipulations, combin- ing (2.22) and (2.31), a formulation for the axial induction factor can be established: a = 1 4 sin2 ˆ rCn + 1 (2.33) Here, rdescribes the solidity of the local annular stream tube, which is the fraction of the annular area that is covered by the rotor blades of chord length c : r = cB 2r (2.34) In the same way, the torque equations, i.e. (2.24) and (2.32), for the stream tube can be combined to find an equation for the tangential induction factor: a0 = 1 4 sin ˆ cos ˆ rCt 1 (2.35) Equations (2.33) and (2.35) are the classical formulation of the induction factors for the BEM. With these formulations, the aerodynamics of a WT rotor can be com- puted. Since the radial discretisation of the stream tubes assumes the stream tubes to be independent from each other, each tube can be solved individually. Typically, an algorithm that solves for the induction factors aand a0 is developed, as shown below: 1. Initialise induction factors aand a0 , e.g. by taking the values from the previous iteration step or assuming a = 1/3and a0 = 0as an initial guess. 2. Calculate the local angles ˆ and ˛using (2.26) and (2.25). 3. Determine Cnand Ctfrom tabulated data for Cland Cdfor the calculated ˛. 4. Calculate updated values aand a0with (2.33) and (2.35) and compare them with the initial guess. This algorithm scheme to solve the simple BEM is iterated until the predefined tolerance criteria for the induction factors are met. Instead of solving for the induc- tion factors, an equivalent algorithm can be formulated to solve for Cland a . To speed up simulation time, the number of iterations can be limited, which, however, might result in not fully converged solutions in some cases. A modified solution method for the BEM equations was suggested by Andrew Ning, which reduces the two- dimensional to a 1D problem [31]. Instead of iterating for the induction fac- tors, the problem is re- formulated to solve for the local inflow angle ˆ , which can increase the robustness of the solution approach as well as reduce the computation time.
- 78. Models and simulation 49 2.3.2.3 Correction models to the blade element momentum theory The described BEM solution of the aerodynamics of a WT rotor is a simplified approach that neglects several physical effects that occur in reality. In the simplified momentum theory, e.g. the induction factor cannot exceed 0.5since the wake wind speed would become negative, as can be seen from (2.17). In reality, induction fac- tors can become larger than this theoretical boundary. Furthermore, there are several more dynamic effects that are not accounted for by the classical static BEM theory. Tip loss correction factor The aerodynamic shape of a rotor blade creates a pressure difference between the pressure and suction side of the blade, which results in the total acting force at the blade and is exploited to create a torque at the rotor. At the tip of the blade, this pressure difference results in air flowing around the tip from the pres- sure to the suction side, resulting in a radial velocity and ultimately reduced lift and therefore power losses. Since this effect is particularly noticeable with fewer rotor blades, a correction for the momentum theory’s assumption of infinite rotor blades was suggested by Ludwig Prandtl, as described in Reference 30. The tip loss correction factor Ftlis introduced to account for the additional losses at the blade tip: Ftl = 2 arccos exp (R r)B 2r sin ˆ (2.36) Here, R and rare the total and local rotor radii, respectively, Bis the number of blades (as used previously) and ˆis the angle between the rotor plane and the rela- tive velocity. The correction factor Ftlhas a value between 0and 1 . Since it reduces the forces estimated from momentum theory, it has to be applied as a factor to (2.22) and (2.24), resulting as an additional factor in the first term of the denominator in (2.33) and (2.35) (see also (2.38) below as an example). Blade root losses In aeroelastic simulations of WTs, blade root losses are less important compared to tip losses, since relative flow velocities at the blade root are much lower compared to the blade tip. However, it is not uncommon to apply Prandtl’s tip loss correction factor in a similar way to account for blade root losses, since the circulation is zero there as it is at the blade tip. The correction can be written as Frl = 2 arccos exp (r r0)B 2r0 sin ˆ (2.37) In this equation, r0is the distance along the blade to start considering the root loss, which is usually the hub radius. The correction factors for tip loss and root loss can be combined by multiplication for one single loss factor to be used in the BEM algorithm, i.e. F = Ftl Frl .
- 79. 50 Wind turbine system design Turbulent wake state As mentioned above in section 2.3.2.3, momentum theory, and therefore also classical BEM theory, is only valid for axial induction factors smaller than 0.5, since this is the limit for the wake flow field to have a positive velocity (cf. (2.17)). In reality, however, the axial induction factor can exceed this theoretical limit since the flow field can become more complex compared to what the simplified theory predicts. For such higher induc- tion factors, momentum theory suggests that the rotor thrust has a maximum for a = 0.5 (cf. (2.22)) and approaches zero for a ! 1 , while in reality, the wake can turn into a turbulent state and the thrust can increase above the theoretical maximum for a = 0.5 . To describe this turbulent wake state, a correction was originally proposed by Glauert, establishing an empirical relationship between the non- dimensional thrust coefficient CTand the axial induction factor a[30]. The detailed description will not be shown here, but using this empirical relationship, a correction for the axial induc- tion factor can be applied if its value exceeds a certain limit ac , which is in this case ac = 1/3 . A solution was given by Hansen [32], which also includes Prandtl’s tip loss factor F . For small axial inductions a ac , the standard BEM formulation is valid: a = 1 4F sin2 ˆ rCn + 1 (2.38) For larger induction factors, a ac , the wake state can become turbulent and the following solution can be applied: a = 1 2 2 + (1 2ac)K q 2 + (1 2ac)K 2 + 4(Ka2 c 1) (2.39) where: K = 4F sin2 ˆ rCn (2.40) Dynamic inflow The described aerodynamic theory so far assumes quasi- static behaviour of the flow, i.e. that the wake is in equilibrium with the aerodynamic loads acting on the rotor. If there is a change in the rotor loads, e.g. due to a change in blade pitch angle or wind speed, the wake will not instantaneously be in equilibrium according to momentum theory. Instead, a dynamic transient to a new steady state will occur since the flow cannot respond quickly enough. This effect is known as dynamic inflow or dynamic wake and can be modelled in different ways. A simple and straight- forward dynamic inflow model was proposed by Øye [33], which acts as two filters connected in series for the induced velocities and can be applied for both the axial and tangential induced velocities. The dynamic inflow model consists of two first- order differential equations, taking into account time derivatives of the induced velocities: Wint + 1 dWint dt = Wqs + 0.6 1 dWqs dt (2.41)
- 80. Models and simulation 51 W + 2 dW dt = Wint (2.42) In this model, Wqsis the quasi- static value of the induced velocities (both axial and tangential) coming from the BEM algorithm, Wintis an intermediate value for the filter and W is the final value, including the dynamic inflow correction. The time constants 1and 2are fitted to: 1 = 1.1 1 1.3a R V0 (2.43) 2 = 0.39 0.26 r R 2 1 (2.44) where ais the axial induction factor, Ris the rotor radius, V0is the undisturbed inflow velocity and ris the local radius of the considered rotor annulus. When cal- culating 1 , the induction factor shall be limited to a 0.5 . Dynamic stall Another aerodynamic effect occurring at WT blades, which is not considered in the static BEM theory, is the dynamic separation and re- attachment of the flow at the airfoil, known as dynamic stall. The local angle of attack at an airfoil varies continu- ously, e.g. due to wind turbulence and shear, tower shadow or yaw misalignment. The static airfoil polar curves for Cland Cdconsider stalling of the blade at a fixed static angle of attack. In reality, the flow separation is a dynamic phenomenon that does not happen immediately but has a time delay, comparable with the dynamic inflow effect. The timescale for the phenomenon of flow separation, however, is usually faster than for dynamic inflow since it can be estimated with the order of the time it takes the local wind speed to pass the blade chord. Dynamic stall occurs when the angle of attack changes rapidly and delays the onset of stall as well as the re- attachment of the flow. This trailing edge flow separation can be modelled with a separation function, as suggested by Øye [34]. The model interpolates linearly between the lift coefficients for the attached and the fully separated flows, such that the steady lift coefficient is continuously restored. This interpolated lift coefficient can be described as Cl = fs Cl,inv(˛) + (1 fs) Cl,sep (2.45) where Cl,inv(˛)is the (idealised) lift coefficient for inviscid flow without separa- tion, Cl,sepis the lift coefficient for fully separated flow and fsis the factor which describes the degree of stall from 0 (fully separated) to 1 (no stall), which is given in dfs dt = f st s fs st (2.46) Here, fst s is a value of the stall factor chosen in such a way that the static airfoil behaviour is obtained when inserted into (2.45), while st is a time constant. A description of how to estimate the parameters in these equations can be found, e.g., in References 35, 36 and is not given here. The application of a dynamic stall model in aeroelastic simulations has been observed to be important for the stability of the
- 81. 52 Wind turbine system design turbine, since otherwise blade vibrations might be calculated that do not occur in reality [34]. Another important dynamic stall model often used in WT simulation is the so- called Beddoes- Leishman model [35, 37]. This model includes more physical effects of the dynamic stall phenomenon that are neglected by the Øye dynamic stall model. It is, however, beyond the scope of this chapter to explain this model in more detail here. Further blade element momentum theory correction models Besides the models described above, there are further correction models which account for specific effects that are not represented in the classical BEM theory, typically with respect to unsteady aerodynamics [38]. One example is a correction for the case that the WT rotor is not aligned perfectly with the main wind direc- tion. In such a yawed or tilted skewed inflow operating point, the induced velocities vary with the azimuthal rotor angle since the blade tip will point either upstream or downstream during one rotation, so that the blade will be deeper or less deep into the wake. An overview of different implementations of such a skewed inflow correction can be found in Reference 39. Another correction with respect to the aerodynamics of a WT is the consider- ation of the tower passage effect when a blade passes the tower. The reduced wind velocity can be modelled based on potential flow theory, as mentioned in section 2.2.1.4. Furthermore, a BEM correction model has been suggested to account for radial flow [40]. 2.3.2.4 More complex aerodynamic models The focus of the previous section was the introduction of the BEM theory and its most important correction models, which is a comparably fast and robust low- fidelity method to solve the aerodynamics of a WT. Since several hundred or even up to thousands of single simulations need to be conducted for WT design load calcula- tions, such efficient engineering models are widely used and therefore explained in detail above. Besides the BEM theory, the more general formulation of the so- called generalised dynamic wake (GDW) theory is utilised in some simulation codes. It is based on the potential flow theory and accounts for unsteady and 3D effects while avoiding the requirement for an iterative solution. More information on the model- ling of GDW theory can be found in Reference 41. There are, however, several other ways which enable a more accurate estimation of WT rotor aerodynamics. Given the current trend of ever- increasing WT and rotor sizes, longer and more flexible rotor blades might require improved simulation methods in the future. Aerodynamic models that exceed BEM capabilities can usually be assigned to either the category of vortex wake and actuator type models or to CFD models. Compared to BEM models, vortex wake and actuator type models resolve the full wake instead of just the rotor plane and are thus more computationally expensive [42]. For certain applications, however, the increasing computational power of mod- ern computers allows the utilisation of vortex wake models even for WT design load
- 82. Models and simulation 53 calculations, which can lead to significantly reduced rotor fatigue loads compared to classical BEM codes [43]. Vortex wake and actuator type models can be seen as a mid- fidelity approach to modelling rotor aerodynamics. Most details in aerodynamic simulations can be expected from high- fidelity CFD codes. These codes typically solve the 3D Navier–Stokes equations and pro- vide a very detailed solution of the flow field around the blades as well as in the WT wake. A challenge is the modelling of turbulence effects, which is either extremely computationally expensive in direct numerical simulations or relies on turbulence model assumptions to reduce the computational cost. In so- called LES, only larger scale motions are considered, while small eddies rely on specific models. A different approach is the RANS equations, where time- averaged Navier–Stokes equations are solved, while different models have been proposed to capture the turbulence effects. The simulation of WT aerodynamics with CFD codes of all types requires large computational resources and is therefore not suitable for design load estimation. 2.3.3 Hydrodynamic models The most traditional method for determining hydrodynamic loads on a structure is the Morison equation [44]. Based on a semi- empirical approach, the horizon- tal component of the hydrodynamic load @Fon a section of a vertical cylindrical structure (with length @land diameter D ) is determined based on an inertial and a drag component with corresponding coefficients CMand CD , respectively, and depending on the horizontal orbital velocity component Uand corresponding water particle acceleration P Uas well as the water density water , as given in (2.47) [45]. The values for the inertial and drag coefficients need to be determined empirically and depend on the flow and surface conditions. Thus, the drag coefficient may range from 0.6 for large Reynolds numbers (105 ) to about 1.2 for smaller Reynolds num- bers (105 ), while a value of 2.0 may be considered for the inertial coefficient for small Keulegan–Carpenter numbers (10) and a value of 1.5 for larger Keulegan– Carpenter numbers ( 10) [46]. @F = waterCM D2 4 P U@l + 1 2 waterCDDU ˇ ˇU ˇ ˇ @l (2.47) The Morison equation, however, assumes that the structure is hydrodynamically transparent, meaning that the structure experiences a wave impact, whereas the wave itself is not altered due to the presence of the structure. This assumption is valid for non- moving structures that exhibit a very small diameter in relation to the wave length. With a more advanced approach based on the potential flow theory [46], some of these limitations can be overcome. According to this theory, the total velocity potential is a summation of the potentials of the incident waves, the diffracted waves as well as the radiated waves. The hydrodynamic load on an offshore structure is then derived from the pressure distribution, which is itself determined based on the velocity potential. In numerical modelling, the potential flow theory may be realised by using the boundary element method. An alternative method, which is also based on the potential flow theory, is the MacCamy–Fuchs approach [47]. This is based on
- 83. 54 Wind turbine system design the Morison equation and extends it: The inertial coefficient is determined in depen- dence on the structural diameter and the wave number, and the phase shift between the incident wave and the resulting force is taken into account. Thus, it is no longer assumed that the wave is not affected by the structure, but radiation and diffraction effects are now accounted for. While waves feed into both inertial and drag forces, currents – if assumed to be constant – only lead to an additional drag force resulting from the current velocity. Additional hydrodynamic impacts might be impulse loads due to horizontal or ver- tical wave slamming and slapping, breaking waves close to or at the structure and wave run- up. Tailored approaches are required to capture these complex hydrody- namic phenomena [2, 15, 16]. Another alternative is CFD, by means of which a much higher level of fidelity in modelling the hydrodynamic effects is achieved. CFD approaches are based on the Navier–Stokes equations and can represent even complex non- linear flow phenom- ena. The underlying equations are the momentum equation (2.48) and the continuum equation (2.49) for incompressible flows and Newtonian fluids, with the flow veloc- ity vector U , the Nabla operator r , the time t , the pressure p , the Laplace operator , the dynamic viscosity and the volume and gravitational force vector F. These Navier–Stokes equations are numerically approximated by means of CFD methods. water @U @t + U r U = rp + U + F (2.48) r U = 0 (2.49) 2.3.4 Modelling of structural components Depending on the purpose of the analysis, different modelling approaches might be used for representing the structural parts of a WT. In general, the mathematical model should be as simple as possible and only as complex as necessary to account for the relevant physical effects while being computationally efficient. Depending on the purpose of the model, a simple 1D mass spring system might be sufficient to compute the elastic response, while for other purposes, an in- depth stress analysis of, e.g. the blade root trailing edge might be necessary, requiring a detailed descrip- tion of the local geometry and material distribution as well as an exact knowledge of the aerodynamic loading scenario. Due to the subject area of this book on WT system design and therefore the calculation of design loads, this section will focus on the description of models that are used for aeroelastic simulation. More complex finite element method (FEM) models are often used in structural analysis of blades, tower and other mechanical components (see also section 2.4.1.1), but it is beyond the scope of this section to explain this method here in more detail. In fully coupled aeroelastic WT simulations for design load calculation, the drivetrain is most often modelled as a torsional spring–damper system with addi- tional rotational inertia for the generator, low- speed shaft (LSS) and high- speed shaft (HSS). Some implementations additionally consider bending of the main shaft as well. For direct- drive WTs, it is also common to disable any torsional
- 84. Models and simulation 55 drivetrain dynamics and just connect a rotational generator inertia rigidly to the rotor hub. Different modelling approaches of varying complexity for WT drivetrains are described in more detail in section 2.4. In the following, however, this section focuses on suitable simulation models for rotor blades and tower since these are the most important structural components of a WT. 2.3.4.1 Blade and tower modelling as beams The main structural components that make up a WT are the rotor blades and the tower. These components not only mainly define the optical appearance of a WT, but they are also the most important structural components with respect to creating a model of an (onshore) WT. WT blades and towers are both tall and slender struc- tures that have one dimension significantly larger than the other two, so physically it is a good assumption to be considered as beams. While the tower is mostly a sym- metrical steel tube, the blade shape and, therefore, the blades’ bending behaviour are more complex compared to a tower, but the beam- like shape is comparable. To combine the structural WT component models for time domain simulations, typi- cally a multibody approach is chosen, where different bodies, which might be rigid or flexible, are connected to each other with certain kinematic joints or links (see also section 2.4.1.2). During operation of a WT, the rotor blades experience deflections in flapwise (out- of- plane) as well as in edgewise (in- plane) direction. Furthermore, especially the large and flexible blades of modern WTs experience considerable elastic tor- sional deflections as well as rigid body blade pitch rotations around their main axis. Similarly, WT towers experience large fore–aft bending moments as well as side– side deflections and torsion due to elastic yaw rotations of the rotor- nacelle assem- bly. To allow for these types of loading in simulation models, WT blades and towers are usually represented as a 1D beam model with beam elements. For more detailed design simulations, 3D FEM shell element models are often used, which allow for local stress determination. However, this section will focus on the beam modelling approach in the following, since this simplified approximation is computationally faster and therefore suitable for design load calculation. Commonly used beam models are the Euler–Bernoulli beam model, as described, e.g. in References 48, 49, and the Timoshenko beam model [50]. Both models account for axial and bending loads as well as torsion of the beam. While the former neglects shear deformation and is therefore not suitable for shorter beams, the Timoshenko beam model includes the deformation of the local cross section due to shear in its mathematical formulation. This, however, makes it slightly more computationally expensive, so that the simpler Euler–Bernoulli beam model has been the standard beam model in aeroelastic load simulation in the past [51]. This section will focus on blade modelling since the blades usually require more detailed models compared to the tower. However, the same model may be used to model the tower as well. To apply the beam models, a discretisation method is required. Most often, one of the following methods is used [52]:
- 85. 56 Wind turbine system design • • Modal reduction method: In a preprocessing step, the flexible eigenmode shapes of the beam are calculated based on a static FEM implementation. During the dynamic simulation runtime, the beam deflection is approximated by a linear combination of the calculated mode shapes. This method solves computation- ally efficiently but may be less accurate for larger blades with higher deflections due to its linearity and the limited number of DOFs. • • Multibody dynamic approach: Several (usually flexible) beam elements are interconnected by kinematic joints or force elements. This method allows for increased DOFs to be considered in the model. • • 1D FEM approach: The beam model is represented as a number of finite ele- ments (FEs) connected by nodes. It requires higher computational effort but provides a good approximation for the deformation of the structural component. As the implementation of a dynamic structural model for aeroelastics can vary significantly with respect to mathematical implementation as well as computational complexity, only a brief overview will be presented in the following. Most beam models rely on the assumption of small deflections, which may not be valid for large and increasingly flexible rotor blades. Therefore, more complex theories, such as the geometrically exact beam theory (GEBT), have been developed that account for the geometric non- linearities occurring at large deflections [53]. For further implemen- tation guidance, the reader is encouraged to have a deeper look into some example implementations of structural models for aeroelastic simulation [54–56]. 2.3.4.2 Parametrisation of beam models The parametrisation of a blade (or tower) beam model requires information on the stiffness and mass distribution along spanwise stations of the beam, which are usu- ally an input to the aeroelastic simulation model and can be computed with specific preprocessors or blade/tower design tools from the respective geometry and design. The mass distribution may also contain additional masses not originating from the blade structure itself, such as blade ice accretion. The structural properties of a spe- cific cross section of the rotor blade can be described by a 6 6mass matrix and stiffness matrix, respectively. The stiffness matrix Kdescribes the relation between the vector of forces and moments Fand the vector of elastic deflections and rota- tions uat the cross section: F = Ku (2.50) The stiffness matrix contains information on axial stiffness EA , shear stiffness kGA , bending stiffness EIand torsional stiffness GJ . Typically, some of the DOFs show a structural coupling, so that the stiffness matrix can contain off- diagonal terms, e.g., for structural bend–twist coupling. The mass matrix Mdescribes the inertial behaviour of the cross section. It relates the vector of forces and moments Fto the linear and angular accelerations R u , as given in (2.51). F = MR u (2.51)
- 86. Models and simulation 57 Figure 2.7 shows a cross section of a blade and the most important geometrical properties. The elastic centre is the point at which a normal force pointing in the main beam direction does not lead to additional bending of the beam. Analogously, the shear centre is the point at which a transversal force does not result in torsional deflection but in pure bending. Pure bending about an applied bending moment will occur if the bending moment is aligned with one of the principal axes. 2.3.4.3 Equation of motion The generalised equation of motion of a discretised mechanical system can be writ- ten as follows [54]: MR q + CP q + Kq = Fg (2.52) Here, M, Cand Kare the mass matrix, damping matrix and stiffness matrix, respec- tively, while Fgis the generalised vector of external forces acting on the system. If the generalised loads are known, e.g. from solving the aerodynamic problem, (2.52) can be solved for the system accelerations R q . This set of generalised coordinates q corresponds to the DOFs of the system. By representing the deflection shapes as linear combinations of the sys- tem’s eigenmodes, the number of DOFs can be reduced to increase the com- putational speed of the simulation. This is a common approach in aeroelastic codes, where typically the first 6–12 eigenmodes of the blades are used and 4–10 eigenmodes of the tower. However, these numbers should only be seen as a rule of thumb, since they depend strongly on the modelled system and the simulation purpose. For such a modal approach, each generalised coordinate qi corresponds to a (modal) deflection shape ui . The Craig–Bampton method [57] is often used to identify the eigenmodes and corresponding mode shapes of the system. The total deflection state can be described by superposition of the selected number of considered mode shapes N : u(q) =q1u1 + q2u2 + : : : + qNuN (2.53) Figure 2.7 Structural cross section of a blade ©Fraunhofer IWES
- 87. 58 Wind turbine system design Besides the described modal reduced approach, some codes use a full (1D) FEM representation of the blade (and tower) beams. As explained above, this method increases the accuracy of the model as well as computational costs and is becom- ing more and more the state of the art for design load calculations of modern WTs [51]. The derivation of the dynamic equations for such a model is summarised in Reference 36. More details on the non- linear equations of motion and the corre- sponding implementation as a state- space model can also be found in Reference 58. An implementation of a multibody discretisation is shown in Reference 59. 2.3.5 Modelling of other components Besides the already discussed component models, the most visible parts of a WT, which have not been discussed yet, are the hub and the nacelle system. In aeroelastic simulations used for design load calculation, these parts are usually assumed to be rigid. Therefore, lumped masses are used to model the inertia of the hub, including pitch system and blade bearings, as well as for the nacelle system with the masses of main shaft supports, gearbox housing, generator sta- tor and the electrical installations in the nacelle. The drag of the rotor hub and the nacelle is typically accounted for by apply- ing simple drag coefficients. The same holds for the tower drag, which includes the BEM induction factors as described in section 2.3.2.2. The electrical system of a WT is usually not considered in aeroelastic simula- tions. The influence of grid faults, such as sudden power losses, however, can be a relevant source for fatigue loads on the rotor- nacelle assembly. To account for such grid faults, a sudden drop of the generator torque can be applied in the aeroelastic simulation model. For short- time grid losses, so- called fault ride through events, the WT does not shut down during the generator torque loss, so that high loads can occur. Depending on the type of the electrical configuration of the WT, this can rep- resent the mechanical excitation of the system in an adequate way to reproduce the structural loads acting on blades, hub, tower and foundation. 2.4 Detailed modelling of wind turbine drivetrains The drivetrain of a WT is located inside the nacelle and consists of all components from the main bearing to the electrical generator and power conversion system. It is responsible for converting rotor kinetic energy into electrical power. Modern WTs are designed to have a minimum operational life of 20 years with high reliability. Unfortunately, this has not been the case, as studies have revealed high rates of failure in WTs, with drivetrain- related failures causing the longest downtimes [60]. Furthermore, due to the complexity of WTs disassembling procedures and, in many cases, remote locations with difficult access, such frequent failures often result in costly repairs. As a result, understanding the causes of such failures and reducing their occurrence to improve system reliability have been the most important areas of research in recent years.
- 88. Models and simulation 59 NRELconducted extensive case studies in the Gearbox Reliability Collaborative (GRC) project to determine the root causes of drivetrain- related failures. One of the major causes of premature drivetrain failures, according to these studies, is a lack of a system- level approach and incorrect estimation of drivetrain loads during the design process [61, 62]. Limiting uncertainties in drivetrain component load estima- tions during the design phase can aid in the development of reliable designs. Hence, it is critical for design engineers to have simulation models that can provide realistic load estimates. The following sections discuss the modelling methods for WT drive- trains that are widely used in research and the wind industry. 2.4.1 General modelling approaches, methods and tools Before going into details, it is worth mentioning the general goals or requirements that determine the modelling approach or method. The selected modelling approach might depend on the following criteria: • • Analysis scope – – Material-level analysis – – Component-level analysis – – System-level analysis • • Analysis types – – Static vs dynamic – – Linear vs non- linear – – Explicit vs implicit • • Physical domain – – Mechanical – – Electrical – – Hydraulic – – Thermal – – Fluid • • Analysis goals – – Load response – – Durability study – – Noise, vibration and harshness (NVH) – – Control – – Condition monitoring • • Coupling type – – Fully coupled analysis – – Decoupled analysis – – Partially coupled analysis Each of the aforementioned criteria requires specific modelling methods and has various fields of application and research domains. For the purpose of WT drivetrain design,a system level approach with sufficient modelling depthis necessary for complete
- 89. 60 Wind turbine system design depiction of the important drivetrain dynamics and load response. The available meth- ods for modelling WT drivetrain can be classified into the following categories: • • Finite element method • • Multibody simulation • • Bond graph methods • • Block modelling 2.4.1.1 Finite element method The FEM involves discretisation of the physical domain into a finite set of nodes and elements [63]. FEM models can have DOFs varying between hundreds and several millions depending on the model size and complexity. Some of the commer- cially available general purpose FEM software packages include Abaqus Unified FEA, Ansys® , Altair HyperWorks® and COMSOL Multiphysics® . The FEM model- ling method is suitable for detailed analysis of stresses, strains, contact and damage for systems undergoing small deformations and motions. It is the primary method for detailed component- level analysis of the drivetrain subsystems. However, FEM models require long computation times and are not suitable for performing system- level dynamic simulations of the entire drivetrain system. 2.4.1.2 Multibody simulation In MBS, the system is modelled as a combination of rigid bodies that interact with their surrounding via constraints and force elements. Each rigid body is defined as a lumped mass having a maximum of six DOFs. The flexibility of the individual components is modelled as spring elements with estimated stiffness of the components. The final system results in a set of ordinary differential equations. MBS is a widely practiced method for dynamic simulation of mechanical drivetrains as it can accurately replicate system dynamics and load effects while requiring little computational resources. The standard rigid body approach in MBS can be further enhanced with flexible bodies using modal reduction techniques. This optimally blends the advantages of the FEM and MBS approaches. The flexible MBS is currently the most popular method for modelling WT drivetrain systems. The commercially available MBS softwares like Adams™, Simpack and RecurDyn include the flexible MBS approach. 2.4.1.3 Bond graph methods Bond graphs are a graphical modelling method for dynamic physical systems. Bond graph models consist of elements (representing certain physics) connected together via bonds (representing power flow between the elements). It features a uniform notation for all types of physical systems. A major advantage of bond graph methods is that it is domain neutral, and system- level modelling of multi- physical domains can be seamlessly performed. Softwares like 20- sim, SYMBOLS 2000 and CAMPG provide bond graph modelling features. Bond graph techniques have been used for modelling WT drivetrains [64, 65], but their application in WT drivetrains is very limited as compared to MBS method.
- 90. Models and simulation 61 2.4.1.4 Block models Another method for simulating and analysing multi- domain dynamic systems is block modelling. Each block represents a model component with mathematical equations describing the system’s dynamic behaviour. Commercially available softwares with block modelling features include Simulink® , Dymola® , and 20- sim. Simulink® offers built- in blocks for WT drivetrain and component models [66, 67]. Block modelling methods are useful for developing real- time system models, con- trollers and hardware-in-the-loop applications. 2.4.2 Different approaches of modelling a wind turbine drivetrain WT drivetrains can be classified into two broad categories: geared drivetrains and direct drives. The geared drivetrains have a gearbox that increases the rotational speed of the rotor, allowing high- speed generators to be used. In direct drives, the generator is directly connected to the rotor shaft and runs at the same speed as the rotor. As a consequence, direct drives typically require larger generators with more pole pairs as compared to geared drivetrains. Although recent years are showing an increasing number of direct- drive WTs in offshore applications, geared drivetrains still account for the majority of WTs deployed worldwide. Based on the generator speed range, geared drivetrains are further divided into two categories: medium- speed geared and high- speed geared. The operating speeds of medium- speed geared drivetrain generators are around 90–500 rpm, whereas high- speed geared drivetrain generators operate around 1500–1800 rpm [68]. The high- speed geared drivetrains allow for smaller size generators and are popular in Generator Brake disc Gearbox Main bearing Main sha Torque arms Generator sha Generator foongs High speed sha Bedplate High speed coupling Figure 2.8 Main features of a typical three- point suspension drivetrain ©Fraunhofer IWES
- 91. 62 Wind turbine system design most of the onshore applications. The geared drivetrains commonly have either a three- point suspension topology or a four- point suspension topology. The following sections describe the different approaches for modelling a geared drivetrain system. A drivetrain with three- point suspension topology is considered, as the majority of geared WTs have this topology. Figure 2.8 shows the main fea- tures of a three- point suspension drivetrain. The main bearing and the gearbox torque arms transfer non- torque loads from the rotor to the bedplate. The generator is mounted on the bedplate by special foot- ings that provide the necessary damping. The high- speed shaft of the gearbox is connected to the generator shaft by a coupling. This ensures that only torque is transferred to the generator by allowing radial and angular misalignments. There is also a mechanical break system on the high- speed side. 2.4.2.1 Torsional model In a torsional drivetrain model, all bodies have only one DOF around the rotational axis. The remaining five DOFs are fixed, and hence, excluded from the system equations of motion. All bodies have rotational inertia, and the flexibility between them is modelled by a torsional spring that represents the torsional stiffness of the components. Such models can be used to describe the system torsional eigenfre- quencies and can provide insights for the dynamic analysis of the torque in the drivetrain. For the analysis of the global WT dynamics, the drivetrain model is often reduced to a two- mass model in the global WT model (see Figure 2.9). This is the simplest form of the drivetrain model and includes the first torsional mode of the drivetrain. The coupled system of the two masses involves only two DOFs resulting in the equation of torque of the following form: T1 = T2 = keq(2 1) (2.54) where T1and T2are the torques acting on the two masses, 1and 2are the angular displacements of the two masses and keqis the equivalent stiffness of the shafts. The non- zero undamped eigenfrequency of the two- mass drivetrain model is defined as fn = 1 2 r keq(JRotor+n2JGen) JRotorn2JGen (2.55) Figure 2.9 Drivetrain modelled as a two- mass spring system ©Fraunhofer IWES
- 92. Models and simulation 63 where fnis the undamped eigenfrequency, JRotoris the inertia of the rotor, JGenis the inertia of the generator and nis the gear ratio. Multi- mass torsional models are an extension of the two- mass model, as each component has one DOF in these models, allowing for a more detailed analysis of the drivetrain. Figure 2.10 represents a case with the drivetrain modelled as a multi- mass spring system. For the undamped free vibration case, the resulting equation of motion follows the following form: J R + K = 0 (2.56) where J , K and represent the system inertia matrix, total stiffness matrix and the vector of rotational displacements, respectively. The eigenfrequencies (! ) of the system are obtained by solving the following eigenvalue problem: !2 = eig(J1K) (2.57) Torsional models are mostly used in aeroelastic codes during the initial development phase of a new WT, when a global analysis of the WT is carried out for all relevant load cases to estimate the drivetrain design loads. These estimated loads provide a starting point for designing the internal components of the drivetrain, such as the main shaft and gearbox [69]. Although torsional models are useful for estimating external loads on the drive- train and avoiding natural frequency excitation of external WT components, they do not capture the details of the complete drivetrain dynamics and internal excitation sources. More details on the limitations of these models have been addressed in References 70–72 with suggestions for using higher- fidelity drivetrain models to overcome these issues. 2.4.2.2 Rigid multibody model with six degrees of freedom The next level of modelling fidelity involves rigid bodies (having lumped masses and six DOFs) that are connected by joints and flexible elements (see Figure 2.11). The bearings are modelled as spring elements representing the estimated stiff- ness of the roller elements. The stiffness values are estimated using an analytical Figure 2.10 Drivetrain modelled as a multi- mass spring system ©Fraunhofer IWES
- 93. 64 Wind turbine system design approach or FEM. The resulting spring element is defined by a 6 6matrix defining stiffness in three translational and three rotational DOFs. The simplest approach would involve a linear stiffness model of the bearings without any cross- coupling behaviour. This leads to a diagonal stiffness matrix having zero off- diagonal ele- ments. Such a model can represent the system dynamics in all six DOFs while main- taining low computational requirements. A more advanced bearing model would include a fully populated stiffness matrix with cross- coupled terms and non- linear stiffnesses addressing the bearing clearance. The gears are modelled as rigid bodies which, in simplest form, perform the speed and torque transformation according to the gear ratio. A more detailed model would include the teeth contact force, gear out- of- plane motion and back- lash. The teeth contact force is usually defined as a function of contact stiffness. The contact stiffness can be a constant or a variable function. The shafts can be modelled as discrete rigid bodies with stiffness or an Euler–Bernoulli beam ele- ment. Such a model can represent the dynamics of the drivetrain in all six DOFs in great detail. However, the rigid body assumptions for gearbox components and bedplate leave out several important dynamic aspects that can be very critical for simulating the true dynamic response of the drivetrain. Furthermore, housing deformations under high loads are not taken into account, which can influence gear and bearing loads. Figure 2.11 Drivetrain MBS model with six DOFs ©Fraunhofer IWES
- 94. Models and simulation 65 2.4.2.3 Mixed rigid flexible multibody model with six degrees of freedom A more realistic model of the drivetrain system takes component flexibility into consideration. The components can be modelled as flexible bodies using FEM. However, this results in models with very large number of DOFs that are not suit- able for dynamic drivetrain simulation. This issue is overcome by implementing modal reduction techniques via component mode synthesis (CMS). This method involves modal reduction of FEM models into flexible bodies that have the same dynamic behaviour as their parent FEM models but with significantly less DOFs. The Craig–Bampton method [57] is the most widely used CMS method for gen- erating flexible bodies in MBS. This method performs modal reduction based on the fixed interface normal modes and the constraint modes. As a result, the overall structure dynamics is a linear combination of these fixed interface normal modes and the constraint modes. To add modal reduced flexible bodies in MBS, the CMS procedure is first applied to the meshed models in the FEM software. In this procedure, an appro- priate set of component modes is selected, and interface nodes are defined at locations where the flexible body can interact with the external environment. These nodes are constrained to slave nodes on the surfaces of the body, where the load is distributed (see Figure 2.12). After the modal reduction procedure, the flexible body is imported in MBS, where it has connections with other bod- ies, force and constraint elements. The imported flexible body allows for lin- ear elastic deformation as well as large rigid body motion. The resulting MBS model leads to a very realistic representation of the drivetrain system in terms of deformations. FE models Reduced flexible bodies Interface definion Modal reducon Nodal coordinates Modal coordinates Slave node Kinemac constraint Interface node Figure 2.12 Modal reduction procedure for flexible body generation ©Fraunhofer IWES
- 95. 66 Wind turbine system design 2.4.2.4 Multi-physical high-fidelity model Including high level of modelling details can lead to very complex drivetrain models. Such high- fidelity models can simulate detailed contacts and flexibili- ties of machine elements. Multi- physical models can be coupled with the MBS model via co- simulation. The resulting system model can be used, e.g. to simulate electromagnetism of the generator, the control systems, the acoustic emissions, lubrication analysis and thermal analysis. Figure 2.13 depicts the possibilities to conduct multi- physical simulations for a drivetrain. This type of advanced modelling approach incorporates system complexity as much as is technologically possible. Of course, such models require very long computation times. These types of models usually find applications in research and development. Figure 2.14 illustrates the different modelling approaches in order of increasing modelling depth. 2.4.3 Modelling recommendations and best practices In the early years of modelling WT drivetrains, a purely torsional system with lumped masses and stiffness elements was a common practice in industry and a basic requirement from design standards. However, such a model lacked the ability to describe loads and dynamics at the component level [73]. With the Figure 2.13 Multi- physical high- fidelity model of the drivetrain ©Fraunhofer IWES
- 96. Models and simulation 67 increasing issues of WT system reliability, numerous research projects have been conducted to evaluate the existing design practices and determine the optimum method for dynamic modelling of drivetrains. According to the case studies con- ducted by NREL in the GRC project, a purely torsional model is insufficient to adequately describe gearbox loads. Based on their findings, new recommendations for drivetrain modelling fidelity have been proposed [74]. These recommendations suggest using MBS model for the mechanical drivetrain with multi- DOFs and a mix of flexible and rigid bodies to represent the internal drivetrain components. Table 2.2 lists the recommended modelling fidelity based on the findings of the GRC project. Following these recommendations has shown good correlation with experimental results. Computaon me Model complexity Very short Very long Lowest Highest Torsional model Rigid MBS model Mixed rigid-flexible MBS model Mul physical / high fidelity model Figure 2.14 Comparison of different modelling methods for WT drivetrain ©Fraunhofer IWES Table 2.2 Minimum model fidelity for drivetrain dynamics by NREL [74] Component Recommended approach Required DOFs Hub Rigid body, lumped mass N/A Main shaft Flexible, FE beams From convergence study Main bearing Stiffness matrices 5 (exclude rotation) Gearbox housing Flexible, condensed FEs From convergence study Planet carrier Flexible, condensed FEs From convergence study Gearbox shafts Rigid shafts N/A Gearbox support Stiffness matrices 6 Gears Rigid body and contact stiffness 6 Gearbox bearings Stiffness matrices 5 (exclude rotation) Spline Stiffness matrices 2 (tilting) Bedplate Rigid body or condensed FEs N/A Generator coupling Stiffness matrices 5 (exclude rotation)
- 97. 68 Wind turbine system design 2.5 Conclusion and summary WT system design strongly depends on the capability of simulation models to accu- rately predict the loads that a WT will experience during its lifetime. Within this chapter, the most relevant modelling approaches for all the different aspects that are required for WT design load estimation are introduced and summarised. Suitable models for the environmental conditions are necessary for predicting the loads and operating conditions the WT system will have to withstand. However, the stochastic nature of environmental conditions makes it difficult to always cor- rectly represent these in numerical approaches. Therefore, accurate and reliable measurement data is highly supportive for realistic representation during simula- tions. Various models for environmental conditions and corresponding impacts are available with different levels of detail and fidelity, which always have to be selected according to the point of interest of the corresponding analysis. Furthermore, the modelling approaches should be tailored to the applicable specific conditions, such as the determined installation site or the WT type. To calculate the aerodynamics of a WT, the BEM theory can be combined with a multibody- based structural representation of the WT to perform design load calcula- tions given the environmental conditions. Existing aeroelastic codes have generally achieved very good levels of validation, but some challenges remain to be solved, such as the accurate prediction of torsional deformations of rotor blades in certain load simulations. Therefore, as computational power and capacity become afford- able, traditional modelling approaches may be supplanted by more complex models in the near future, even for design load simulations. These advanced models can include non- linear FEM beam models for the structural components as well as vor- tex wake or actuator type models for the aerodynamics. WT drivetrains can be modelled using a wide range of approaches with various levels of detail. A balance between the required dynamic details, the correspond- ing modelling effort and the resulting computation time are needed. Sufficient accuracy in drivetrain dynamics can be achieved by following the NREL recom- mendations, as this modelling approach has shown good validation with experi- mental results. However, integration of high- fidelity drivetrain models into load calculation models will still remain computationally too expensive for the near future, so this fully coupled approach will only be used for some specifically selected load cases. References [1] International Electrotechnical Commission. ‘Wind energy generation systems–Part1:Designrequirements’.4thed.No.IEC61400- 1inInternational Standard; Geneva, Switzerland, International Electrotechnical Commission, 2019. [2] International Electrotechnical Commission. ‘Wind energy generation systems – Part 3- 1: Design requirements for fixed offshore wind turbines’.
- 98. Models and simulation 69 1st ed. No. IEC 61400- 3- 1 in International Standard; Geneva, Switzerland, International Electrotechnical Commission, 2019. [3] International Electrotechnical Commission. ‘Wind turbines – Part 4: Design requirements for wind turbine gearboxes’. 1st ed. No. IEC 61400- 4 in International Standard; International Electrotechnical Commission, 2012. [4] DNV GL Loads and site conditions for wind turbines DNVGL- ST- 0437. 2016. Available from https://www.dnv.com/energy/standards-guidelines/dnv-st-0437- loads-and-site-conditions-for-wind-turbines.html [5] Poeppel T, Bockstedte A. ‘Certification requirements for the drive train dy- namics of wind turbines’. Conference for Wind Power Drives (CWD); 2015. pp. 85–97. [6] GL Renewables Certification. GL Rules and Guidelines - IV Industrial Services - Part 1 - Guideline for the Certification of Wind Turbines. Hamburg, Germany: Germanischer Lloyd Industrial Services GmbH; 2010. [7] GL Renewables Certification. GL Rules and Guidelines - IV Industrial Services - Part 2 - Guideline for the Certification of Offshore Wind Turbines. Hamburg, Germany: Germanischer Lloyd Industrial Services GmbH; 2012. [8] American Gear ManufacturersAssociation AGMA 6006 - Standard for Design and Specifications of Gearboxes for Wind Turbines. AGMA; 2003. [9] Kaimal J.C., Wyngaard J.C., Izumi Y., Coté O.R. ‘Spectral Characteristics of Surface-Layer Turbulence’. Quarterly Journal of the Royal Meteorological Society. 1972, vol. 98(417), pp. 563–589. Available from http://doi.wiley. com/10.1002/qj.v98:417 [10] Mann J. ‘Wind field simulation’. Probabilistic Engineering Mechanics. 1998, vol. 13(4), pp. 269–282. [11] Mann J. ‘The spatial structure of neutral atmospheric surface- layer turbu- lence’. Journal of Fluid Mechanics. 1994, vol. 273, pp. 141–168. [12] Jonkman B.J. TurbSim user’s guide v2.00.00: DRAFT VERSION [online]. National Renewable Energy Laboratory; 2016. Available from https://www. nrel.gov/wind/nwtc/assets/downloads/TurbSim/TurbSim_v2.00.pdf [13] von Kármán T. ‘Progress in the Statistical Theory of Turbulence’. Proceedings of the National Academy of Sciences of the United States of America. 1948, vol. 34(11), pp. 530–539. [14] Andersen O.J., Løvseth J. ‘The Frøya database and maritime boundary layer wind description’. Marine Structures. 2006, vol. 19(2–3), pp. 173–192. [15] American Bureau of Shipping. Design Standards for Offshore Wind Farms: Final Report: Contract M10PC00105.ABS. 2011.Available from https://www. bsee. gov/sites/bsee.gov/files/tap-technical-assessment-program/670aa.pdf [16] DNV GL ‘Environmental conditions and environmental loads’. August 2017 ed DNVGL- RP- C205. DNV GL AS; 2017. [17] Jensen N.O. A note on wind generator interaction. Risø-M: Risø National Laboratory; 1983.
- 99. 70 Wind turbine system design [18] Frandsen S., Barthelmie R., Pryor S, et al. ‘Analytical modelling of wind speed deficit in large offshore wind farms’. Wind Energy. 2006, vol. 9(1–2), pp. 39–53. [19] Rathmann O., Barthelmie R., Frandsen S. ‘Turbine wake model for wind resource software’. EWEC 2006 Wind Energy Conference and Exhibition; European Wind Energy Association (EWEA), 2006. pp. 1–12. [20] Larsen G.C., Madsen H.A., Thomsen K., Larsen T.J. ‘Wake meandering: a pragmatic approach’. Wind Energy. 2008, vol. 11(4), pp. 377–395. Available from http://doi.wiley.com/10.1002/we.v11%3A4 [21] Madsen H.A., Larsen G.C., Larsen T.J., Troldborg N., Mikkelsen R. ‘Calibration and Validation of the Dynamic Wake Meandering Model for implementation in anAeroelastic Code’. Journal of Solar Energy Engineering. 2010, vol. 132(4), 041014. [22] Larsen T.J., Madsen H.A., Larsen G.C., Hansen K.S. ‘Validation of the dy- namic wake meander model for loads and power production in the Egmond aan Zee wind farm’. Wind Energy. 2013, vol. 16(4), pp. 605–624. [23] Méhauté B. An Introduction to Hydrodynamics and Water Waves. Berlin Heidelberg: Springer; 1976. [24] Wheeler J.D. ‘Method for Calculating Forces Produced by Irregular Waves’. J Pet Technol. 1970, vol. 22(3), pp. 359–367. [25] Det Norske Veritas.Wind turbine design software - Bladed. 2022.Available from https://www.dnv.com/services/wind-turbine-design-software-bladed-3775 [26] DTU wind energy. HAWC2. 2021. Available from https://www.hawc2.dk/ [27] Øye S. ‘FLEX4: Simulation of Wind Turbine Dynamics’. Proceedings of 28th IEA Meeting of Experts Concerning State of the Art of Aeroelastic Codes for Wind Turbine Calculations; Lyngby, Denmark: International Energy Agency, 1996. pp. 71–76. [28] National Renewable Energy Laboratory OpenFAST Documentation. 2022. Available from https://openfast.readthedocs.io/en/main/ [29] Betz A. ‘Das Maximum der theoretisch möglichen Ausnützung des Windes durch Windmotoren’. Zeitschrift für das gesamte Turbinenwesen. 1920, vol. 26, pp. 307–309. [30] Glauert H. ‘Airplane propellers’ in Durand W.F. (ed.). Aerodynamic Theory. Berlin, Heidelberg: Springer; 1935. pp. 169–360. [31] Ning S.A. ‘A simple solution method for the blade element momentum equa- tions with guaranteed convergence’. Wind Energy. 2013, pp. 1327–1345. [32] Hansen M.O.L. Aerodynamics of wind turbines. 2nd ed. London, UK: Earthscan; 2008. [33] Snel H., Schepers J.G. ‘Joint investigation of dynamic inflow effects and implementation of an engineering method’. [ECN- C–94- 107. ECN] 1995. Available from https://publications.ecn.nl/E/1995/ECN-C--94-107 [34] Øye S. ‘Dynamic stall simulated as time lag of separation’. Proceedings of the 4th IEA Symposium on the Aerodynamics of Wind Turbines; Rome, Italy, 1991. [35] Hansen M.H., Gaunaa M., Madsen H.A. ‘A Beddoes- Leishman type dy- namic stall model in state- space and indicial formulations’ [online]. No.
- 100. Models and simulation 71 Risø-R- 1354(EN) in Risø R, Report, Risø National Laboratory. 2004. Available from http://iis-03.risoe.dk/netahtml/risoe/publ_uk.htm [36] Hansen M.O.L, Sørensen J.N, Voutsinas S, Sørensen N, Madsen H.A. ‘State of the art in wind turbine aerodynamics and aeroelasticity’. Progress in Aerospace Sciences. 2006, vol. 42(4), pp. 285–330. [37] Leishman J.G, Beddoes T.S. ‘A Semi-Empirical Model for Dynamic Stall’. Journal of the American Helicopter Society. 1989, vol. 34(3), pp. 3–17. [38] Leishman J.G. ‘Challenges in modelling the unsteady aerodynamics of wind turbines’. [online] Wind Energy. 2002, vol. 5(2–3), pp. 85–132. Available from http://doi.wiley.com/10.1002/we.v5:2/3 [39] Schepers J.G., Snel H. ‘Dynamic inflow. Yawed conditions and partial span pitch control’. ECN-C-95-056. Netherlands Energy Research Foundation. 1995. [40] Lanzafame R., Messina M. ‘BEM theory: How to take into account the radial flow inside of a 1- D numerical code’. Renewable Energy. 2012, vol. 39(1), pp. 440–446. [41] Peters D.A, He C.J. ‘Correlation of Measured Induced Velocities with a Finite–State Wake Model’. Journal of the American Helicopter Society. 1991, vol. 36(3), pp. 59–70. [42] Sørensen J.N, Mikkelsen R.F, Henningson D.S, Ivanell S, Sarmast S, Andersen S.J. ‘Simulation of wind turbine wakes using the actuator line tech- nique’. Philosophical Transactions. Series A, Mathematical, Physical, and Engineering Sciences. 2015, vol. 373(2035), 20140071. [43] Boorsma K., Wenz F., Lindenburg K., Aman M., Kloosterman M. ‘Validation and accommodation of vortex wake codes for wind turbine design load calcu- lations’. Wind Energy Science. 2020, vol. 5(2), pp. 699–719. [44] Morison J.R., Johnson J.W., Schaaf S.A. ‘The Force Exerted by Surface Waves on Piles’. J Pet Technol. 1950, vol. 2(5), pp. 149–154. [45] Sumer B.M., Fredsøe J. ‘Hydrodynamics Around Cylindrical Strucures’ in Advanced series on ocean engineering. Revised Edition vol. 26. Singapore: World Scientific Publishing Co Pte Ltd; 2006 Sep. [46] Clauss G., Lehmann E., Östergaard C. Offshore Structures. London: Springer London; 1992. Available from http://link.springer.com/10.1007/ 978-1-4471-3193-9 [47] MacCamy R.C., Fuchs R.A. ‘Wave Forces on Piles: A Diffraction Theory’. Contribution from the Scripps Institution of Oceanography. U.S. Beach Erosion Board. 1954. Available from https://books.google.de/books?id= M0K5GQAACAAJ [48] Benaroya H., Nagurka M.L., Han S.M. Mechanical vibration: Analysis, un- certainties, and control. 4th edn. Boca Raton and London and New York: CRC Press Taylor Francis Group; 2018. Available from http://www.crc- press.com/p/book/9781498752947 [49] Bauchau O.A., Craig J.I. Structural analysis: With applications to aerospace structures. Vol. 163. of Solid mechanics and its applications. Dordrecht: Springer; 2009.
- 101. 72 Wind turbine system design [50] Timoshenko S.P. ‘LXVI. On the correction for shear of the differential equa- tion for transverse vibrations of prismatic bars’. The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science. 1921, vol. 41(245), pp. 744–746. [51] Lamei A, Hayatdavoodi M. ‘On motion analysis and elastic response of floating offshore wind turbines’. Journal of Ocean Engineering and Marine Energy. 2020, vol. 6(1), pp. 71–90. [52] Wang L., Liu X., Kolios A. ‘State of the art in the aeroelasticity of wind turbine blades: Aeroelastic modelling’. Renewable and Sustainable Energy Reviews. 2016, vol. 64, pp. 195–210. [53] Hodges D.H. ‘Geometrically Exact, Intrinsic Theory for Dynamics of Curved andTwistedAnisotropicBeams’.AIAAJournal.2003,vol.41(6),pp.1131–1137. [54] Shabana A.A. Dynamics of Multibody Systems. Cambridge University Press; 2009. [55] Kim T., Hansen A.M., Branner K. ‘Development of an anisotropic beam finite element for composite wind turbine blades in multibody system’. Renewable Energy. 2013, vol. 59, pp. 172–183. [56] Wang Q., Sprague M.A., Jonkman J., Johnson N., Jonkman B. ‘BeamDyn: a high‐fidelity wind turbine blade solver in the FAST modular framework’. Wind Energy. 2017, vol. 20(8), pp. 1439–1462. [57] Mapa L. de P.P., das Neves F. de A., Guimarães G.P. ‘Dynamic Substructuring by the Craig–Bampton Method Applied to Frames’. Journal of Vibration Engineering Technologies. 2021, vol. 9(2), pp. 257–266. [58] Veers P. Wind energy modeling and simulation. vol. 125 of IET Energy engineering series. London: The Institution of Engineering and Technology; 2019. Available from https://digital-library.theiet.org/content/books/po/ pbpo125f [59] Mo W., Li D., Wang X., Zhong C. ‘Aeroelastic coupling analysis of the flex- ible blade of a wind turbine’. Energy. 2015, vol. 89, pp. 1001–1009. [60] Faulstich S., Hahn B., Tavner P.J. ‘Wind turbine downtime and its impor- tance for offshore deployment’. Wind Energy [Online]. 2011, vol. 14(3), pp. 327–337. Available from http://doi.wiley.com/10.1002/we.v14.3 [61] Keller J., Wallen R. Gearbox Reliability Collaborative Phase 3 Gearbox 3 Test Report. No. NREL/TP- 5000- 67612 Technical Report. National Renewable Energy Laboratory; 2017. [62] Musial W., Butterfield S., McNiff B. ‘Improving Wind Turbine Gearbox Reliability NREL/CP- 500- 41548’. National Renewable Energy Laboratory, 2007. [63] Szabó B., Babuška I. Introduction to Finite Element Analysis: Formulation, Verification and Validation. John Wiley Sons; 2011. [64] Sanchez R., Medina A. ‘Wind turbine model simulation: a bond graph ap- proach’. Simulation Modelling Practice and Theory. 2014, vol. 41, pp. 28–45. [65] Torsvik J. Dynamic Analysis in Design and Operation of Large Floating Offshore Wind Turbine Drivetrains [Doctoral thesis]. NTNU; 2020.
- 102. Models and simulation 73 [66] Apostoaia C., Scutaru G. ‘A Dynamic Model of a Wind Turbine System’. In: The 10th International Conference OPTIM’06; Transilvania University Press, 2006. pp. 261–266. [67] Benbouzid M.E.H., Diallo D. ‘Development of a Matlab/Simulink®-based wind turbine prototyping software through undergraduate student projects’. ComputerApplications in Engineering Education. 2012, vol. 20(1), pp. 78–87. Available from http://doi.wiley.com/10.1002/cae.v20.1 [68] Manwell J.F., McGowan J.G., Rogers A.L. Wind energy explained: Theory, design and application. 2nd ed. Chichester: John Wiley; 2009 Mar. Available from https://onlinelibrary.wiley.com/doi/book/10.1002/ 9781119994367 [69] Vanhollebeke F. Dynamic Analysis of a Wind Turbine Gearbox Towards Prediction of Mechanical Tonalities [Ph.D. Thesis]. Belgium: Katholieke Universiteit Leuven; 2015. [70] Peeters J., Vandepitte D., Sas P. ‘Flexible multibody model of a three- stage planetary gearbox in a wind turbine’. Proceedings of ISMA2004; Katholieke Universiteit Leuven, 2004. pp. 3923–3942. [71] Peeters J. Simulation of dynamic drive train loads in a wind turbine [Ph.D. Thesis]. Belgium: Katholieke Universiteit Leuven; 2006. [72] Peeters J., Goris S., Vanhollebeke F, et al. ‘A need for advanced and validated multibody models as a basis for more accurate dynamic load prediction in multi- megawatt wind turbine gearboxes’. International Conference on Noise and Vibration Engineering; 2008. pp. 2097–2112. [73] Helsen J., Vanhollebeke F., Marrant B., Vandepitte D., Desmet W. ‘Multibody modelling of varying complexity for modal behaviour analysis of wind tur- bine gearboxes’. Renewable Energy. 2011, vol. 36(11), pp. 3098–3113. [74] Guo Y., Keller J., La Cava W, et al. ‘Recommendations on Model Fidelity for Wind Turbine Gearbox Simulations NREL/CP- 5000- 63444’. National Renewable Energy Laboratory, 2015.
- 103. This page intentionally left blank
- 104. 1 Fraunhofer Institute for Wind Energy Systems, Bremerhaven, Germany 2 Nidec SSB Wind Systems GmbH, Salzbergen, Germany Chapter 3 Pitch system concepts and design Karsten Behnke1 , Arne Bartschat1 , Eike Blechschmidt1 , Matthis Graßmann1 , Florian Schleich1 , Oliver Menck1 , and Heiko Jungermann2 The pitch system allows for turning the rotor blades of the wind turbine about their longitudinal axes. This turning movement is commonly called pitching and con- trols the aerodynamic loads on the blades and thus the power output, the rotational speed and structural loads of the turbine. Feathering of the blades means that the blades are turned to reduce the lift force and hence the torque. It leads to a stop of the turbine. Hence, the pitch system is one of the safety systems of a wind turbine and thus always designed in a redundant manner. Design guidelines like the one from Det Norske Veritas (DNV) make a redundant pitch system a mandatory pre- requisite for a turbine certification [1]. Redundancy means that at least two of the blades have independently operating pitch systems. Common wind turbines have three blades. This leaves the designer the choice to have two pitch systems, one for two and another one for one blade, or three pitch system, each for one blade. In terms of design effort, procurement and operational costs, three identical pitch systems are better, which makes them the common choice in today’s commercial wind turbines. The pitch system of one blade consists of the pitch bearing (Section 3.1) and the pitch actuator (Section 3.1 and Chapter 8). In addition, a lubrication system (Section 7.3), sensors for position and more, a power supply, backup power, connec- tion to the main turbine controller, and a local backup controller complete the pitch system. Interfaces of the blade bearing towards the blade and hub are usually bolted connections. The design of a pitch system is a complex task since many aspects have an influence on it. Furthermore, the pitch system design affects the turbine design. The blade and hub are directly connected to the blade bearing. Their design
- 105. 76 Wind turbine system design Table 3.1 Design steps for a pitch system Step Inputs Objectives 3.1.1 Preliminary outer bearing design • Blade inner and outer diameter • Blade bolt size number and bolt circle diameter • Decide if the blade is mounted to the inner or outer ring • Define blade bearing outer dimensions • Choose actuator concept: Hydraulic, electric geared, electric belt 3.1.2 Preliminary inner bearing design • Blade root loads: moments and forces • Bearing dimensions • Choose rolling element type and number of rows • Calculate the number of rolling elements and their size • Choose spacer or cage as separator • Determine hardening, hardening gaps and fill plug position 3.1.3 Preliminary design of the bolted connections • Bearing dimensions from previous sections • Bolt circle diameter and number of bolts from blade and or hub, if available • Blade root loads: moments and forces • Information about tightening processes, tools and available installation space • Calculate loads, which are acting on the bolts • Choose a bolt size as and number for both hub and blade flange 3.1.4 FE blade bearing model • Bearing data from the first steps, which are: dimensions, bearing type, number of rolling elements and bolts • Create a Finite element model of the blade bearing • Verify the plausibility of the model • Calculate load angle and contact forces 3.1.5 FE simulation of internal blade bearing loads • FE bearing model from previous section • FE model of the interface parts • Mount FE bearing model into a wind turbine rotor star model • Check the plausibility of the load application • Find possibilities to reduce computational time • Calculate load angle and contact forces (Continues)
- 106. Pitch system concepts and design 77 must therefore be considered for the pitch system, and vice versa, the pitch system design must be considered in the blade and hub design. Hence, a pitch system design cannot happen individually. Technical feasibility, experiences in the company and costs are main drivers for any decision. However, to find a start, Table 3.1 gives an overview of possible steps for such a project. The table starts with the blade bearing design. It connects blade and hub and defines con- sequently their dimensions. It must cope with the acting loads and ensure the ability to pitch. In addition, it determines possible other choices for the remain- ing components of the pitch system, like the actuator and auxiliary systems. The following sections give detailed information for each step of the pitch system design. These steps should not be understood as a gradual process. Calculations or other decisions could lead to changes in a previous design stage. It is possible that many iterations are mandatory. The individual steps will include examples of possible design choices for the IWT7.5- 164, a 7.5 MW reference wind turbine designed by Fraunhofer IWES [2]. The turbine is introduced in Chapter 1. Step Inputs Objectives 3.1.6 Calculation and dimensioning • FE model from the previous sections • Load time series containing: pitch angle, moments and forces at the blade root, and extreme loads • Perform static evaluation: use the maximum load situations for overload, truncation and core crushing check • Perform fatigue evaluation: calculate rolling contact, structural and bolt fatigue lifetime 3.1.7 Lubrication • Bearing dimensions and type • Operating conditions • Choose grease type • Design grease in- and outlets and sealing 3.1.8 Coating • Specific site conditions • Turbine location • Define the type of coating and layer thickness to protect against corrosion 3.2.1 Electrical actuator • Bearing interfaces and position of the blade (inner or outer ring) • Acting wind loads, friction torque and inertia of the blade and other rotating parts • Calculate the torque, which is needed to pitch the blade • Choose drive size • Choose kind of motor and back- up power • Determine back-up capacity for emergency drive Table 3.1 Continued
- 107. 78 Wind turbine system design 3.1 Blade bearing Blade bearings connect to the rotor blade and the rotor hub of a wind turbine by means of bolted connections. Their rings tend to be made of steel, with 42CrMo4- type steels being the most common choice of material. Rolling elements be likely made from 100Cr6. The rotor blade is mostly made of glass- fibre- reinforced plastic (GFRP). Its weight and the acting wind loads cause the loads on the blade bear- ing. The blade works as a huge lever, hence the dominant load on a blade bearing is a bending moment. The bearing also sees loads in every other degree of free- dom, however, their influence pales in comparison to that of the resulting bending moment caused by the tilting of the blade. Thus, the moments and forces at the blade root are input values for the bearing design. All of them vary according to the wind conditions. Values for said loads typically stem from the aero- elastic simulations (cf. Chapter 1). These can have the form of time series (see, e.g., Reference [3]) or postprocessed results like extreme loads as defined in References [4] and [5], bin counts, damage equivalent loads, rain flow counts for the last, see Reference [6]. A further example is a load revolution distribution (LRD) of a rolling contact fatigue (RCF). An aero- elastic simulation time series of IWT7.5- 164 reference turbine is avail- able online and can be downloaded under [3]. It can be used to understand the loads that are acting on the blade bearing and the movements the bearing does. As mentioned before, the blade bearing design process is iterative as it influ- ences and is influenced by the other systems such as controller and blade design. A heavy use of individual pitch control, e.g., reduces turbine loads significantly and therefore lowers costs in structural components. This on the other hand low- ers the lifetime of the blade bearing with respect to RCF, which can lead to changes in rolling body diameter or even bearing size. Especially the latter is commonly not wanted as it would influence the design of the blade root. A lower use of individual pitch control leads to higher loads on the structural components especially on the blade that would need more material to withstand. The higher masses increase, in turn, the bending moment acting on the blade bearings and could lead to higher RCF again. The aim during the design phase is to optimise for overall system costs. Figure 3.1 shows the iterative bearing design process. In the first iteration, it will be necessary to make some assumptions until the bearing completely fulfils the requirements. The design starts with the bearing geometry, which is described in Sections 3.1.1–3.1.3. The geometrical properties are necessary to create a FE bearing model to calculate the internal loads of the bearing (cf. Section 3.1.4). The rotor star simulates the load distribution of the bearing for realistic surrounding structures and turbine loads (cf. Section 3.1.5). The cal- culated bearing loads are used for the static evaluation and RCF calculation in Section 3.1.6. The calculation results may require adaptions in the dimensions of the bearing. When the geometry of the bearing changes, a new FE model needs to be created to run new simulations and recalculate the static evaluation and
- 108. Pitch system concepts and design 79 RCF. This repeats until the final geometry of the blade bearing. The lubrication (Section 3.1.7) and coating (Section 3.1.8) are not included in this loop, as they can be designed afterwards. 3.1.1 Preliminary outer bearing design This section explains how to get a first draft of the outer dimensions of the bearing. Thereby it presumes input values, most importantly the blade diameter and the num- ber of bolts. Starting with these values, it is possible to derivate the blade bearings main dimensions. Section 3.1.2 gives more and more details, like the bearing type and other bearing characteristics. The preliminary bearing draft is mandatory for the following sections, which include FE calculations. The inner and outer diameters of the bearing are determined by two design choices: the blade root diameter and the blade- mounted rotating ring of the bearing, which can be the inner or outer ring. The blade root diameter is part of the blade design. As the blade is one of the most important parts of the turbine and any design changes trickle down to all other major components, its properties are usually not influenced by design needs of any interface parts. As such, the pitch system designer usually takes the blade root diameter and the blade root bolt circle as granted and uses them for the blade- side ring of the bearing. However, as modern turbine sizes and loads increase constantly, the interface between blade root and blade bearing becomes even more critical. Therefore, it is possible that the development of the blade root and the blade bearing must be done in conjunction to achieve a reliable and, probably most important, an economical design. The final design should con- sider the loads but also the stiffness properties of the blade root, the hub and the bearings itself. Figure 3.1 Bearing design process © Fraunhofer IWES
- 109. 80 Wind turbine system design The rotor blade may be mounted either on the inner or outer ring of the bear- ing. Each choice has advantages and disadvantages. Both solutions can be found in commercial turbines. For example, Table 3.2 summarises the arguments. Vensys and Senvion turbines have the blade mounted to the outer ring, whereas Siemens, Vestas, Nordex and others have it on the inner ring (IR). Furthermore, the individual rolling element loads will be different for both designs. If the blade design and other boundary conditions stay the same but the hub size increases, the IR blade- mounted design can have more rolling elements due to the larger diameter and hence a better distribution of the loads. Additionally, the increased diameter leads to lower rolling element forces if the moment stays the same. Besides these, many other aspects influence the decision as well. These could be costs of the surrounding parts, assembly processes, maintenance accessibility and the technical feasibility. This decision is a major step, which again will influence the further design. For the IWT7.5- 164, the outer blade root diameter is 4650 mm, the bolt circle diameter is 4500 mm and the inner diameter is 4350 mm. There are 120 bolts of size M42 (cf. Section 3.1.3). The blade is mounted to the IR. With this size of the blade root, it is considered a major advantage to have the pitch drives inside of the hub. This allows to omit additional housings and makes maintenance easier. The holes on the IR are through holes, and their diameter is chosen to be 45 mm to accommodate the tolerance of both hole circles. The wall thickness of the blade root is 150 mm. The bolt hole centres are in the middle, 75 mm away from each outer side. The blade bearing can be slightly less Table 3.2 Advantages of inner ring and outer ring blade mount Inner ring Outer ring • A stiffener plate can be mounted between the inner ring and the blade without any additional parts. This provides additional structural stiffness to the bearing. For an outering mount, these parts are necessary to adapt the bolt heads heights of the inner ring. • In case of a ring crack, it is less likely that the blade falls down. • During turbine assembly, blade bolts can be tightened from within the hub, which is easier than on the outside. • Pitch drives are mounted in the hub, which makes additional housings unnecessary and allows for easier maintenance. • The hub can be smaller, lighter and stiffer. • It is possible to use a belt drive mechanism that is less prone to wear than gear drives or hydraulic pitch cylinders and works without lubricants. • The design of the connection of the lubrication system towards the blade bearing is easier.
- 110. Pitch system concepts and design 81 thick due to the higher load capacity of steel in comparison to GFRP of the blade. At this point, the value for the inner diameter of the bearing is set to 4380 mm. Between blade and bearing, an additional plate adds stiffness to avoid ovalisation of the bear- ing rings. This plate is the interface between blade and bearing. Figure 3.2 visualises the design state. The next missing diameter is the centre diameter, more commonly referred to as pitch diameter. Together with the rolling element size, it will determine the distance between the raceways and the minimum distance between the bore holes. For obvious reasons, a clash with the bore holes should be avoided. It is likely that every bearing manufacture has its own methods to determine the distance between both elements. Another missing part so far is the rolling ele- ment size, which will be discussed in the following section. Therefore, the pitch diameter will be preliminary, and a check afterwards is necessary. However, as a first draft, this value is set to 4690 mm for the IWT7.5- 164 pitch bearing. It is also the starting point for the fixed ring. Here, the bore holes for the bolted con- nection to the hub are the design driver. As a first step, the number and size of the bolts can be the same as for the blade side since the loads are similar. Thus, the outer ring will also have 120 bore holes with a diameter of 45 mm for M42 bolts. Detailed calculation of the bolted connection will be part of the follow- ing sections. Again, a clash between raceway and bore holes is not acceptable; therefore, the width of both rings can be the same. The outer diameter is 5000 mm, and the bolt circle diameter is 4880 mm. Figure 3.3 gives an update to Figure 3.2 and shows these values. The height is the final missing information for the bearing’s outer dimensions. Once again, it is dependent on the rolling element size, the cage or spacer design (cf. Section 3.1.2), the sealings, and their interface towards the other ring. The height Figure 3.2 IWT7.5- 164 pitch bearing design Step 1: inner bolt circle and blade mount ring decision © Fraunhofer IWES
- 111. 82 Wind turbine system design must be determined once these values are known. The IWT7.5- 164 blade bearing rings have a height of 284 mm. Both rings have the same height. However, the flanges of the rings are at different levels, which leads to a height of 294 mm. This gap simplifies the tightening process of the bolts and avoids clashes once the bearing is mounted. Figure 3.4 shows the outer dimensions. The choice of whether the blade is mounted to the inner or outer ring also deter- mines the pitch actuator and its interface. However, the interface of the bearing towards the actuator influences the design of the bearing, e.g., additional interface plates for a hydraulic system require more space. The IWT7.5- 164 pitch system is an electrically powered gear drive. Hence, the bearing IR needs gear teeth. Figure 3.5 shows the gear teeth. Instead of using an electric pitch system, it is also possible to use a hydraulic pitch system. One advantage of this design choice in terms of the bearing design is that there is no need for a gearing at the IR. This will lower the manufacturing costs of the bearings and probably also the weight as there is less material needed. Besides these advantages, the pitch system must include adapters to connect a hydraulic actuator to the IR. It is easier that the hydraulic actuator is connected to the IR. Although it is not generally required for the turbine operation, a hydraulic pitch system will most likely not be able to perform large angles or full rotations due to its kinematics based on an eccentric connection and limited strokes of the hydraulic actuators. Figure 3.3 IWT7.5- 164 pitch bearing design Step 2: outer bolt circle and hub mount ring decision © Fraunhofer IWES
- 112. Pitch system concepts and design 83 Figure 3.4 IWT7.5- 164 pitch bearing design Step 3: bearing height © Fraunhofer IWES Figure 3.5 Toothing of the IWT7.5- 164 blade bearing © Fraunhofer IWES
- 113. 84 Wind turbine system design 3.1.2 Preliminary inner bearing design With the external loads and the main dimensions, the next choices in the design are about the internal layout of the bearing. This includes the type of rolling bodies, their number and dimensions and the number of rows. These choices are influenced by whether a cage or spacers are used to keep the rolling bodies in equidistant positions, as a cage has different space requirements than spacers. The main distinction between the two rolling body types is the type of con- tact and the contact angle. A ball has a point contact (PC) where a roller has a line contact. Rollers typically have a lower contact pressure compared to balls. It is likely that the rolling contact lifetime is longer with lower contact pressure, although the number of rolling elements and hence the number of load cycles due to over rolling play a role as well. The fatigue lifetime will be calculated in Section 3.1.6. In comparable space, roller bearings tend to have higher RCF lifetimes. The prevalent types of bearings used as blade bearings are currently four- point bearings, which use balls. A second type is the three- row roller bearing. Four- point bearings can have two rows if one is not sufficient for the acting loads. Three- row roller bearings have two axial rows with thick rollers to withstand axial loads, where each of the two rows carries axial loads in one respective direction. The third row in a three- row roller bearing is a radial one with smaller rollers that merely must withstand significant smaller radial loads. Other possible but less common designs include, e.g., crossed- roller bearings or three- row bearings with two axial ball rows (‘T-Solid’). In a typical three- row roller bearing, the contact angle for the axial rows is at a constant 90°, whereas in a four- point ball bearing, it varies under load from an initial angle. The initial angle is often 45° but others are possible as well. If the contact angle increases further a specific limit, the raceway edges suffer from plastic deformation or cracking due to high stresses because of contact pressure ellipse truncation. In this case, a roller bearing could be a possible solution. A contact angle diverging from 90° creates additional radial forces that result in higher ring defor- mation and lower ring fatigue lifetime. If both considerations above have satisfactory results for both roller types, then the price is the next decision criterion. A four- point ball blade bearing is cheaper than a three- row roller bearing because it has only two rings that have to be manu- factured instead of three. Manufacturing an additional ring significantly increases the costs due to additional handling and machining costs. In the beginning, a rough calculation gives an estimation for the largest contact force in the bearing. The contact force is necessary to calculate the contact pressure, which depends on the rolling element type and size as well as the raceway geometry. The contact pressure is the key value and the base for most of the calculations in Section 3.1.6. It is important to know that the surrounding structure influences the load distribution in the bearing. To gain realistic contact forces, a FE calculation is mandatory (cf. Section 3.1.4). Nonetheless, a FE calculation requires an input. So, in the first stage, the rolling element type, number of rows and elements, and others are
- 114. Pitch system concepts and design 85 stated as in the following steps. Later it can be necessary to change these variables depending on the results in Section 3.1.6. The dominant load of a pitch bearing is the resulting moment (Mres ). As 3.1 shows, it is the combination of the edge- and flapwise bending moment. Mres = q Mx 2 + My 2 (3.1) Radial (Fr ) and axial forces (Fax ) will influence the rolling body loads as well, but play a minor role. Figure 3.6 shows a typical load distribution. Due to the bending moment, there are two opposite regions where the loads are transferred. One side is called traction side, the other, compression side. Between the peaks, the individual rolling element loads are lower. Under the assumption of stiff rings and interfaces, the bearing rows (m) share the load equally and the highest contact force (Qmax ) can be calculated as 3.2 shows. Qmax = 2 Fr m z cos(˛) + Fax m z sin(˛) + 2 Mres m r sin(˛) z (3.2) Here, z is the number of rolling elements per row, ˛is the nominal contact angle of a ball bearing (0° for pure radial and 90° for pure axial contact) and r is the pitch radius. This equation is based on DG03 [7], but the number of rows was added. It assumes of rolling elements with linear stiffnesses and stiff surrounding struc- tures; however, balls and rollers are non- linear in their deformation behaviour and the surrounding structures behave elastically. A real load distribution results into Figure 3.6 Characteristic load distribution of a double- row four-point contact ball bearing © Fraunhofer IWES
- 115. 86 Wind turbine system design higher maximum forces. They are typically 10−25% above those from 3.2 and can be even larger, depending on the specific deformation behaviour of the load case. For first approximations, a corrective factor of 1.1−1.25 applied to Qmax may thus be useful. For 3.2, m defines the number of rows that would carry loads under a pure axial force. In double- direction two- row roller bearings, only one row carries an axial load in a given direction, and therefore m should be chosen to be 1 in this case. For a double- row four- point bearing, on the other hand, two rows carry an axial load, and consequently, m should be 2. The rolling element loads are then used in Hertzian calculations for contact pressure of a single contact. The Hertzian calculations cannot be solved analytically. Hence, equations for an approximation were well established. For example, Houpert published a method. The maximum pressure (Pmax ) follows according to 3.3, see Reference [8]. The Hertzian contact ellipse radii (a and b) are required inputs for the equation. Here, another iterative process comes into play. The contact pressure decreases with larger rolling elements since the contact area is larger. But if one uses larger rolling elements, less of them fit into the available space on one raceway, and hence each ball or roller has to carry a higher load. Hence, an optimum will be a trade- off. Pmax = 1.5 Qmax a b (3.3) According to Reference [9], maximum permissible contact stresses are 4.2 GPa for ball bearings and 4.0 GPa for roller bearings. However, since these pressures are related to extreme wind conditions, typical contact pressures during normal opera- tion of the turbine are lower. Instead of 4.2 GPa for a ball contact or 4.0 GPa for a line contact, it is recommended to aim at 2.5 GPa for a PC and 2.0 GPa for a line contact. The experience shows that the results from fatigue or core crushing calcula- tion then finally determine which loads are permissible during normal operation. In addition, the initial contact angle of a four- point bearing increases with higher loads. A higher load could lead to contact ellipse truncation at the edge of the raceway before the maximum permissible contact stress is reached. Truncation should be avoided as well. The IWT7.5- 164 blade bearing design is a double- row four-point contact ball bearing. The pitch diameter and the rolling element diameter determine the possible number of rolling elements. It is necessary to have about 1–2 cm space between each rolling element for a cage or spacer. A possible rolling element diameter for this turbine class is 80 mm. In the following steps, it will be necessary to evaluate this first approach. The bearing has 147 balls per row with a diameter of 80 mm each. The number of rolling elements can be derived with the pitch diameter and the roll- ing element size. The IEC 61400 defines different design load cases (DLCs) [4]. For the following calculations, the maximum load (Mres_max = 30.3 MNm) in DLC 2.2 is used. DLC 2 are power production load cases with an occurrence of a fault. For this DLC, a partial
- 116. Pitch system concepts and design 87 safety factor for loads (γf = 1.1) must be considered [4]. According to 3.2, it leads to a contact force of 138 kN. Qmax = 2 7158 kN 2 147 cos 45 + 1245 kN 2 147 sin 45 + 2 30.3 MNm 2 147 2.345 m sin 45 ! 1.1 = 138kN (3.4) This value again is the input to calculate the Hertzian parameters. Due to the simplifications of Equation 3.2, a safety factor of 1.25 is added, which leads to Qmax = 172 kN. With Equation 3.3, the bearing has a maximum contact pressure of about 3.0 GPa. Table 3.3 gives the missing values to calculate the contact pressure. It also introduces the ball grove conformity. It describes the ratio of the radii from the rolling element and the raceway. The last one is slightly larger; typically, values for blade bearings are 0.52–0.53. The conformity influences the contact size and hence the contact pressure. With a higher conformity, the contact pressure decreases, but on the other hand, a higher frictional work and a higher friction are consequences. Other values from Table 3.3 stem from litera- ture values for steel. The calculated contact pressure is lower than the permissible. Typical loads during normal operation behaviour are even lower and lead to lower contact pressures. The bending moment during power production (DLC1.1) reaches val- ues from about 22 MNm, which again lead to contact pressures in the range of 2.5 GPa. To sum up, a rolling element diameter of 80 mm has satisfactory results in terms of ultimate loads and seems to fulfil the recommendations, explained before. The choice between spacers or a cage is the next topic. Both separate the rolling elements and keep them equidistant. Table 3.4 lists advantages and disadvantages for both types. The IWT7.5- 164 has a steel cage. It is made of S235 steel. Figure 3.7 shows an exemplary cage. The raceways of the bearing are inductively hardened to increase the resilience of the raceway. The hardening process induces currents that heat the raceways locally and thus harden the bearing steel. Figure 3.8 shows two case- hardened race- ways of a four- point bearing. As the magnetic field is not homogeneous, it is almost impossible to harden one whole raceway without superposing the starting point. Table 3.3 Bearing properties for Hertzian calculation Value Unit Conformity inner ring 0.53 Conformity outer ring 0.53 Hertzian ellipse radius a 2.05 mm Hertzian ellipse radius b 13.00 mm Young’s modulus E 210000.00 N/mm² Poisson’s ratio ν 0.30
- 117. 88 Wind turbine system design Therefore, it is quite common to leave space between the beginning and the end of the hardened raceway. This part of the raceway is called hardening gap. To prevent overloading of the non- hardened zone of the raceway, a relief grinding leads to a reduction of the forces on the rolling elements. The orientation of the so- called soft spot is chosen so that the lowest rolling element loads act upon it. Figure 3.7 Steel cage © Fraunhofer IWES Table 3.4 Pros and cons for cages and spacers Pros Cons Cage • Rolling elements are kept at a constant distance therefore no ball jams can occur. • Needs more space in axial direction so there is less space for the raceway, which leads to earlier truncation of the pressure ellipse at lower loads. Spacer • More space for raceways in axial direction and therefore higher loads until truncation of pressure ellipse are possible. • Spacers just push the next rolling element. Different rolling element speeds can lead to a ball jam and hence an increased friction torque. Segmented cages (primarily in roller bearings) • Rollers are kept at a constant distance and cannot jam. • Gaps between segments are possible.
- 118. Pitch system concepts and design 89 3.1.3 Preliminary design of the bolted connections This section will give a rough description for a preliminary and first design of the bolted connections between blade, bearing and hub.Adetailed calculation of the bolted connection has to be performed according to applicable standards like the DIN VDI 2230 in any case [10]. Usually, the blade bearing is connected to the blade root and to the hub flange by using bolted connections. There are several requirements that must be fulfilled with these connections. The most important requirement is to ensure a permanent and safe connection, which can withstand the dynamic and ultimate loads of the complete service life of the turbine. Besides this, the bolted connection allows to pre- load the rolling elements for specific bearing designs. These kinds of bearings have separated rings such as three- row roller blade bearings. In general, there are two flanges with bolted connections, which differ slightly in terms of demands on the design: the connection of blade bearing ring and blade root and the connection between blade bearing and hub flange. Typically, the connection of blade root and blade bearing can be characterised as follows: • • through holes on the bearing ring • • connection of different materials such as steel (bearing or stiffener plate) and fibre- reinforced plastics (blade root) with different stiffness properties • • possibility of having multiple joints within the clamp length due to additional parts such as stiffener plates, ring extenders or others • • long clamp length due to the bearing geometry and special requirements of the blade root especially for T- Bolt connections due to bearing stresses in the radial bore holes • • specific friction coefficients in the joint due to different materials • • additional sealing material in the contacts to prevent water from entering espe- cially in harsh environments Figure 3.8 Case- hardened raceway © Fraunhofer IWES
- 119. 90 Wind turbine system design The connection of blade bearing and hub flange can be considered to have the following characteristics: • • through- holes or blind- holes on the bearing ring • • shorter clamp length • • similar or comparable material properties of both connecting parts • • possibility of having multiple joints within the clamp length due to additional parts such pitch system carriers, ring extenders or others As the design of the blade root usually includes a detailed consideration of the bolted connection due to the specific requirements of the lightweight construction, the bearing design could be adapted to the needs of the blade root in terms of number and size of bolts. The following preliminary design for the outer ring is an iterative approach, which has to balance the requirements. If no information is available, the first step towards the definition of a preliminary design of the bolted connection should be the number of bolts. This number must be defined by the available space and the expected ultimate loads which have to be carried by the bearing flange. In case of the example in this section, the bolt circle diameter on the outer ring is 4880 mm (cf. Section 3.1.1). To define the number of bolts, the designer also must consider the tools needed to tighten the bolts. Usually, the bolted connections on a blade bearing are tightened with hydraulic torque wrenches or bolt tensioners, which need enough space to be applied to the bolts. In this example, 120 bolts (cf. Section 3.1.1) lead to a distance between the bolts of roughly 128 mm on the outer ring, which seems rea- sonable at first glance. The ultimate loads acting on the flange have to be transferred into an axial force carried by the bolts. Therefore, 3.2 is adapted slightly. The loads used for this calculation are based on DLC 2.2, according to the ultimate load analysis (cf. Section 3.1.2). According to IEC- 61400- 1 [4], different partial safety shall be con- sidered for the DLCs. In case of DLC 2.2, it is a safety factor of f = 1.1 . FB = Fax nB + 2 Mres r nB f = 0 B @ 1245 kN 120 + 2 q 29.265 MNm 2 + 7.827 MNm 2 2.44 m 120 1 C A1.1 = 239 kN (3.5) The clamping force needed to provide enough resistance against the opening of the contact can be estimated as follows. FK = s FB = 1.5 239 kN = 358.5 kN (3.6) The factor s is set to 1.5, which is typical for dynamic load situations. In general, the clamping force FK should be higher as the working load FB to ensure a positive residual clamping load. Both loads are used to estimate the needed stress cross sec- tion for the bolted connection. AS FB + FK Rp,0,2 kkA ˇ E fz lk (3.7) This estimation is based on Reference [11] and needs some more values. kA deter- mines a factor for the tightening method, kis a reduction factor based on the type of
- 120. Pitch system concepts and design 91 bolt, ˇa factor for the elastic resilience of the used bolt, Ethe Young’s modulus for steel, fza factor for the amount of embedding, lkthe clamping length and Rp,0,2the proof stress of the used bolt. For the preliminary design in this section, the following values are chosen according to Reference [11]: • • Rp,0,2 = 900 MPaaccording to bolts of type 10.9 • • kA = 1.2for hydraulic tensioning • • k = 1.15for setscrews • • fz = 0.011 mm • • ˇ = 1.1for set screws • • lk = 450 mmpreliminary assumption according to the height of the bearing ring of 284 mm (see Section 3.1.1), pitch carrier plates and the hub properties This calculation is based on some estimations. Although these estimations only aim towards a rough preliminary calculation of the bolted connection between the hub and the outer ring of the blade, the designer already needs at least some basic information regarding the major geometry. Based on the values estimated for this preliminary design, the needed stress cross section according to (3.8) is AS,min 239 kN + 358.5 kN 900 MPa 1.151.2 1.1 210 GPa 0.011 mm 450 mm = 924.3 mm2 (3.8) The stress cross section of the used bolt shall be equal or larger than the one esti- mated with (3.8). In this case, the next ISO metric screw size that has a slightly larger stress cross section is M42. The wrench size for bolt size M42 is 65 mm. Therefore, the mean distance between the bolts of 128 mm should be sufficient. However, a detailed analysis of possible collisions of the assembly must be performed according to the dimensions of the tools used for tightening. Therefore, the preliminary design of the bolted connection between outer ring of the bearing and the hub shall consist of 120 bolts of the size M42. For an estimation of the pretension, the designer needs to know the resilience of the bolt and the connected parts of the assembly. The val- ues used in this example lead to the following resilience of the M42 bolts. ıS = lk Es AS = 450 mm 210 GPa 1121 mm2 1.91e6 mm N (3.9) The resilience of the bearing outer ring, pitch carrier and the hub is estimated as follows. First the designer needs the substitutional area of the connected parts calculated with Asub = 4 d2 w d2 h + 8 dw DA dw 3 qlKdw D2 A + 1 2 1 (3.10) Using dw = 65 mm for bolt size M42, dh = 45 mm as a usual bore hole diameter for M42, the distance between the bolts Da = 128 mm and lk = 450 mm, the substitu- tional area is
- 121. 92 Wind turbine system design Asub = 4 65 mm 2 45 mm 2 + 8 65 mm 128 mm 65 mm 2 4 3 s 450 mm 65 mm 128 mm 2 + 1 !2 1 3 5 = 7996 mm2 (3.11) With the substitutional area, the resilience of the connected parts is calculated as ıT = lk ET Asub = 450 mm 210 GPa 7996 mm2 2.68e7 mm N (3.12) With the resilience of the screw and the connected parts, the load factor can be cal- culated with K = ıT ıT + ıS = 2.68e7 2.68e7 + 1.91e6 = 0.123 (3.13) Finally, the needed pretension of the bolts can be calculated with FV = fz K ıT + FK + 1 K FB = 0.011 0.123 2.68e7 + 358.5 kN + 1 0.123 239 kN = 573.2 kN . (3.14) This preliminary design of the bolted connection between blade bearing outer ring and the hub has to be verified in the detailed design process. Figure 3.9 shows exemplary three M42 bolts. For the very first detailed calculations using the finite element models described in the next sections, the preliminary number and size of bolts can be used. However, as the load situations and stiffness properties can be Figure 3.9 Outer ring bolts IWT7.5 © Fraunhofer IWES
- 122. Pitch system concepts and design 93 rather complex, a detailed analysis of the bolted connection should be performed towards the ultimate strength of the connection and the contacts and interfaces of the parts. Gaps and an opening of the contacts between the flanges should be avoided and shall be verified within the FE analysis. In addition, the fatigue strength of the connection has to be verified by more detailed calculations according to the stan- dards like DIN VDI 2230. 3.1.4 FE blade bearing model For the calculations in Section 3.1.6, the accurate contact force or pressure of the rolling elements in the bearing is essential. Because analytical calculation methods cannot consider the tilting of the bearing rings and complex load situations including the surrounding structures, finite element analysis (FEA) is the only means to deter- mine the load distribution realistically. FE models that contain the full bearing and potential surrounding structures like blade and hub are called global models in the following. Contact forces and angles as well as the deformation of the bearing rings are possible results of global bearing models. The contact pressure can be calculated with the contact force and the associated, analytically calculated, contact area. In this section, the preliminary bearing design from the previous sections is used to create the FE bearing model. After a plausibility check, the bearing is imple- mented in a full or one- third rotor star model in Section 3.1.5. There are many dif- ferent approaches to create a FE model of a blade bearing. One way is to create a detailed 3D model including all required geometrical properties like osculation and bearing preload. An FE software then meshes the model automatically. That would provide a bearing model with three- dimensional solid rolling bodies. Characteristic of any rolling bearing is that the rolling bodies roll in the raceways of the bearing rings. The FEM considers that with the definition of frictional contacts between the touching components. The blade bearing of the IWT7.5- 164 wind turbine is a double- row four- point contact ball bearing. That means, in initial state, every ball has four touching spots with the raceways. With 147 balls per row that sum up in a total number of 1,176 contact definitions. Furthermore, the calculation of cor- rect stresses in the material requires a sufficient fine mesh, especially in the contact areas. All in all, this approach leads to an FE bearing model with a high level of detail, many contact definitions and a very large number of nodes and elements. That results in a significant high computational effort and requires a lot of simula- tion time. In the design process, many simulations are necessary to investigate differ- ent load cases and related load distributions in the bearing. Long simulation run- times due to high number of nodes and elements are obstructive. Manufactures and researchers have developed and published different approaches to reduce the complexity of the bearing model by decreasing the number of defined contacts and elements. They have in common that they use non- linear spring elements to model the ball- raceway contacts instead of modelling three- dimensional solid balls. A force- deformation curve, calculated according to Reference [8], controls the behaviour of the spring elements. For a ball bearing, the following equation
- 123. 94 Wind turbine system design calculates the deformation ıfor a given ball force Q , ball diameter DW and oscu- lation s[8, 12]. ı = 8.97 104 1 s 0.1946 Q2/3 D1/3 W (3.15) The resulting force deformation behaviour of the blade bearing of the IWT7.5- 164 reference wind turbine is shown in Figure 3.10. Gao [13], Smolnicki [14] and Daidié [12] published different approaches to replace the solid ball with non- linear spring elements to represent the ball- raceway contact behaviour of four- point contact ball bearings. Each spring element repre- sents one force transmitting diagonal of the ball. Therefore, two springs model one ball with its four contact points with the raceways. The direction of the undeformed spring elements represents the initial contact angle. The initial contact angle is the angle between ball and raceway under unloaded conditions. For the blade bearing of the IWT7.5- 164 wind turbine, the initial contact angle is 45°. In [13], the springs are directly connected to the raceways. In [14], rigid beam elements connect the springs to the raceways. Starting and ending point of a spring are the centres of two opposite raceways. In [12], rigid shell elements, which roughly represent the size of the contact ellipse for a reference contact pressure, are located on the surface of the raceways. Rigid beam elements then connect these shell elements with the spring elements. Here, the springs are also placed between the centres of two oppo- site raceways. Figure 3.11 shows the arrangement of the spring (blue) and beam (orange) elements for a modelled ball in a four- point contact ball FE bearing. With this approach, a compression force on the ball leads to a tensile force in the spring. Figure 3.10 Force- deformation curve of the blade bearing of the IWT7.5- 164 reference wind turbine © Fraunhofer IWES
- 124. Pitch system concepts and design 95 A ball- raceway contact cannot transmit any tensile forces. According to that, the springs must not carry any compression forces. For the sake of completeness, Stammler [16] and Wang [17] published approaches to model roller bearings with non- linear spring elements. However, the focus in this sec- tion is a double- row four- point contact ball bearing that is used as blade bearing for the IWT7.5- 164 wind turbine and FE models of roller bearings are not considered further. The preliminary design is explained in Sections 3.1.1 and 3.1.2. Figure 3.12 visualises the FE bearing model that is used for the upcoming investigations in a cross- sectional view. A large rolling bearing has more important functional parts than the rolling ele- ments. For example, bolt holes, sealing surfaces, bore holes to allow lubrication, and for ball bearings fill plugs. However, not all parts are important to be consid- ered in a global FE bearing model to analyse the load distribution on the raceways. Bolt holes in the inner and outer ring allow to mount the bearing to its surrounding structures. Tightened bolts introduce additional stresses to the bearing rings and with that also influence the properties of the contacts between ball and raceway (e.g., contact pressure and ball force). To obtain realistic load distributions when the bearing is simulated with its surrounding structure, bolt holes and modelled bolts need to be considered. The other functional components do not greatly influence the load distribution and contact angle evolution of the bearing and can be left out for the FE bearing model. That decreases the complexity of the model and reduces the computational effort. Figure 3.11 Modelled ball with non- linear spring elements in four- point contact ball bearing [15]
- 125. 96 Wind turbine system design Another aspect that needs to be considered for creating FE bearing models is the geometrical details. A large bearing has some complex geometries to fulfil important tasks. There are grooves at the top and bottom of the rings to hold the sealing in posi- tion, and between the raceways to spread the lubricant in the bearing. Furthermore, there are additional bore holes for lifting accessories. All these details do not need to be included in the FE model. The osculation defines the deviation between the ball and raceway diameters. When the ball- raceway contact is modelled with non- linear springs, this parameter is included in the calculation of the force- deformation curve. It is not represented geometrically. Main results of a global FE bearing model with spring elements are the distri- bution of the contact forces and contact angles. The contact forces are input values for the static evaluation and RCF calculation in Section 3.1.6. However, in Section 3.1.1, a rough calculation to determine rolling element loads is presented, and the FEA gives a more accurate calculation for these. The contact forces are identical with the spring forces and can be directly obtained from the spring elements as an output parameter. For the contact angle, the position of each spring element, in an undeformed and deformed state, is compared. The differences between the two states of the springs describe the resulting contact angle. More detailed contact properties like contact pressure and shear stresses under the material surface require Figure 3.12 Cross- sectional view of a double- row four- point contact ball bearing FE model © Fraunhofer IWES
- 126. Pitch system concepts and design 97 a submodel. A submodel contains only a very small number of rolling elements but three- dimensional and solid modelled. For the IWT7.5- 164 ball bearing, a submodel can consider one pair of ball, one of every raceway and the belonging parts of the bearing rings or just one quarter of a ball and the raceway. Figure 3.13 shows such a submodel. In both cases, frictional contacts are defined to model the ball- raceway contacts. The results of the global model function as input parameters. With that, the submodel calculates stresses based on the acting contact forces and angles. After creating the FE bearing model, the plausibility of the model needs to be verified. In the first simulations with many restrictions on the possible bearing defor- mation, the results of the FE model can be compared to the analytical results of the Hertzian theory (cf. Section 3.1.2). The analytical results do not consider any tilting but only axial deformation of the bearing rings. When that behaviour is transferred to the FE model, the stiffness of the bearing model can be verified by loading the bearing with a pure axial force and comparing the resulting axial deformations of the model with the analytical ones. That also enables to evaluate whether the settings of the spring elements are correct. Only the diagonals of the ball- raceway contacts that Figure 3.13 FE submodel of the IWT7.5- 164 blade bearing containing two three- dimensional modelled balls © Fraunhofer IWES
- 127. 98 Wind turbine system design are compressed should transfer any load. Next, a bending moment is applied to the bearing. Unlike for the comparison with the Hertzian theory, the tiling of the bearing rings is not prohibited for these simulations. When a complex bearing deformation is permitted, the boundary conditions must not be directly applied to the bearing flanges. The clamping of the outer bearing flange and loading the inner flange leads to a very uneven load distribution of the raceways. Generic surrounding structures or simple steel rings that are mounted to the bearing flanges ensure a more realistic load distribution of the bearing model. Characteristic of a four- point contact ball bearing is a change of the loaded diagonals between traction and compression side (cf. Figure 3.6). The simulation of bending moments can verify the plausibility of the qualitative load distribution. At the traction and compression side are always two diagonals (one of each ball) loaded. Only in regions where the load changes the raceway, all four diagonals carry load. Typically, the maximum ball force at the compression side is higher than at the traction side. The qualitative progression of the contact angle should be similar to the load distribution. 3.1.5 FE simulation of internal blade bearing loads Once the bearing FE model is generated and the resulting contact forces and contact angles have been checked for plausibility, it must be mounted into a wind turbine rotor star model. A rotor star model containing the bearing’s surrounding structures enables the evaluation of realistic internal loads. When blade bearings become larger, the pitch diameter becomes disproportionately larger compared to the cross- sectional area of the bearing rings. Because of that, large bearings are structurally softer than small bearings and more sensitive towards the stiffness of their surround- ing structures. In turn, the resulting internal loads of a large blade bearing are sig- nificantly influenced by the stiffness of the rotor blade and the rotor hub [18]. As the blade flange and the rotor hub have inhomogeneous stiffnesses along their circum- ference, large blade bearings are exposed to complex load distributions. In addition, the overall stiffness is relatively low. Depending on the outer load, the load of every rolling element is different, and the specific load distribution needs to be considered in the bearing’s design process. It is best practice to model wind turbine rotor blades with shell elements to reduce computational time [19]. In order to connect such a model to the rest of the rotor star properly and to implement the opportunity to consider the bolted connection, a solid root has to be added to the blade model. The blade root, modelled with solid elements, should contain structural details like the metal inlets or T- Bolts. This allows the defini- tion of frictional contacts between the flange surfaces and the implementation of the bolted connection. Even if no sliding or separation of the flange surfaces occur, consid- ering the bolted connection can slightly change the bearings internal load distribution compared to a model with bonded contacts based on multipoint constraints. It is a suffi- cient way to model the bolts with beam elements that are connected to the surrounding structure with node couplings. The one- third IWT7.5- 164 rotor star model consists of the main components, mean- ing a one- third of the rotor hub, the blade bearing, the stiffener plate and the rotor blade.
- 128. Pitch system concepts and design 99 A part of it is exemplary shown in Figure 3.14. The outer ring of the bearing connects directly to the flange of the hub. The bearing’s IR connects to the blade with the stiff- ener plate in between. Using stiffener plates is a common way to reduce the ovalisation of the bearing, which is caused by the loads and blade design. Further details of the IWT7.5- 164 rotor star model can be taken from Reference [20]. In case it is foreseen to strengthen the connection of rotor blade and rotor hub with additional components like, e.g., ring extenders, those components need to be considered as well to end up with a realistic assembly situation for the blade bearing. Using only a one- third rotor star model greatly reduces the computational time as only one bearing model is implemented, which typically is the main driver for the com- putational effort. A cyclic constraint is applied to the cutting planes of the one- third model of the rotor hub. Doing so makes the model behave symmetrically, meaning that it is not possible to consider, that the three blades are loaded differently. The influence of this simplification on the bearing’s internal load distribution is depended on the asym- metry of the load acting on the entire rotor star. For load cases in which all blades are loaded in a similar way the effect of this simplification is negligible. However, for simu- lating the load distribution and contact angle variation in the blade bearing for the totality of all possible load combinations it is recommended to build up a full rotor star model containing the same components on all three flanges. A full rotor star model enables a correct consideration of the hub’s deformation behaviour for asymmetric load condi- tions that can affect the complex load distribution in the blade bearing at the flange of interest. In both cases, the downwind flange of the hub model is completely fixed and the rotation about the pitch axis is prevented by a very stiff torsional spring that is connected between bearing’s inner and outer ring. Fixing the rotation about the pitch axis is crucial as a rotation of inner and outer ring against each other would lead to incorrect results regarding postprocessed contact forces when using a simplified FE bearing model that bases on fixed spring elements, e.g., the Daidié approach shown in Section 3.1.4. Figure 3.14 Finite element model of 1/3 IWT7.5- 164 rotor star © Fraunhofer IWES
- 129. 100 Wind turbine system design To simulate the internal blade bearing loads, the loads acting on the rotor blade are applied to the blade model in the blade coordinate system [1]. For this purpose, two master nodes are created on the pitch axis of the rotor blade at two different positions as shown in Figure 3.15. Each master node connects with force- distributed constraints (FDC) to slave nodes that are located at the spar caps. No bending moments, only axial and radial forces are applied to the master nodes and the FDCs transmit the loads to the structure of the blade via the slave nodes. The position of the load application nodes should not be too close to the blade root flange as this can affect the structural deformation behaviour in that region, which could lead to unrealistic blade bearing loads. It is a good approach to use at least the first quarter of the rotor blade length in order to consider the characteristic rotor blade behaviour properly. For the exemplary one- third IWT7.5- 164 rotor star model, the z- positions for the load master nodes are L1 = 15 m and L2 = 20 m. By using two load application points, no fixed relation between bending moment and radial forces is maintained and, in turn, this configuration allows the generation of any desired load situation for the blade bearing even with fixed z- positions of the load master nodes. Based on the desired load situation at the blade root, the radial loads that need to be applied at the master nodes are calcu- lated as follows: Fx2 = My,root Fx,root L1 L2 L1 (3.16) Fx1 = Fx,root Fx2 (3.17) Fy2 = Mx,root + Fy, root L1 L1 L2 (3.18) Fy1 = Fy,root Fy2 (3.19) Figure 3.15 Load application at the blade of the 1/3 IWT7.5- 164 rotor star FE model © Fraunhofer IWES
- 130. Pitch system concepts and design 101 A first plausibility check of the load application can be done by using internal func- tions to sum the nodal force and moment contributions of the elements at the bearing flange surface. The summation point must be in the same position of the load refer- ence point. Further plausibility checks for the total deformation behaviour of the generated full rotor star model can be done by some synthetic load cases. Applying the same loads to three blades should result in the same deformation at all flanges. High loads (e.g., Mres,max ) only at one blade, while the others are not loaded, lead to an overall tilting of the entire rotor star model. After the plausibility check is done and it is ensured the rotor star model behaves realistically, further reduction techniques can be applied in order to obtain a more computationally efficient model. Superelement techniques based on condens- ing a selection of finite elements into one unique element, named superelement. Implementing the blade bearing between a blade- sided superelement and a hub- sided superelement allows a significant reduction in degrees of freedom and, in turn, computational time with only a very little loss of accuracy [21]. 3.1.6 Calculation and dimensioning At this step of the design process, it must be verified whether the bearing withstands the loads that it will see during the entirety of its lifespan. This consists, most impor- tantly, of a statical and a dynamical verification. The statical calculation aims to ensure that the bearing can survive a certain maximum load at least once, whereas the dynamical verification aims to ensure that the entirety of loads during its lifetime will not lead to fatigue failure of the raceways, rings or bolts. Bearing verifications of this kind are typically performed using ISO 76 [9] or ISO 281 [22]. These are intended for bearings placed in stiff surrounding condi- tions. Blade bearings are, however, typically surrounded by rather flexible struc- tures, which necessitates the FE models described in the previous section. The use of these FE models is highly recommended for the following calculations as well. Any formulae not including FE calculations risk significantly underestimating the actual loads that occur, both statically and dynamically. Static calculations The statical calculations aim to ensure that the blade bearing can withstand the maxi- mum load situations occurring for all operating conditions of the turbine. According to Reference [1], several DLCs shall be considered to verify the structural integrity of this component. The guideline provides a table with design situations and DLCs and specifies safety factors for the loads. The calculation starts with the analysis of the load time series for the DLCs in order to identify the maximum resulting bending moment, according to 3.1. For the IWT7.5-164, Mres,max occurs for DLC 2.2 (cf. Section 3.1.2). All load components of this load case are multiplied with the safety factor of 1.1 and afterwards simulated with the developed FE rotor star model (see Section 3.1.5). The highest contact force Qmax obtained with the global bearing FE model is used to calculate the highest resulting Hertzian pressure Pmax . It is important to consider that the highest resulting
- 131. 102 Wind turbine system design bending moment does not necessarily lead to the highest contact pressure between rolling element and raceway. As the structural behaviour of the rotor blade affects the bearings internal load distribution, a different load angle or a specific combina- tion of pitch angle and load angle can lead to higher rolling element forces and, in turn, slightly higher contact pressure. For that reason, the occurring load angle for the resulting bending moment needs to be calculated as well. In case the analysed load time series contain resulting bending moments slightly lower than Mres,max for significantly differing load angles, these load cases should also be simulated with the FE model. As stated in Section 3.1.1, according to Reference [9], the maximum permis- sible contact stresses shall be less than 4.2 GPa for ball bearings and 4.0 GPa for roller bearings. This check can be done with the rolling element forces, obtained from the global bearing model, which are used to calculate the maximum con- tact pressure and size of the contact area. For double- row four- point contact ball bearings the postprocessing of the resulting contact angles is required in order to check for possible contact ellipse truncation. With the maximum contact angle, it can be evaluated if contact ellipse truncation at the edge of the raceway will be a severe problem for the current blade bearing design. It is crucial to prevent the occurrence of contact ellipse truncation, as it goes along with an intensive increase of the contact pressure. However, not only high contact pressures but also high contact angle variations favour truncation. For that reason, it is not sufficient only to take a look at the highest contact forces but also to check for truncation in the region where the highest contact angle variations occur. In a double- row four- point contact ball bearing that is loaded with a bending moment, the contact angles on the traction side are usually higher than on the compression side of the bearing. In case the contact ellipse comes very close to the raceway edge, a further stress analysis with a detailed submodel (like exemplarily shown in Figure 3.13) should be carried out. Next step of the static calculations is the core crushing check according to Reference [7]. As large blade bearings are only case- hardened (see Figure 3.8), there is a rapid decrease of the hardness at the transition zone between case (race- way hardness) and core (ring material hardness). It is assumed that the core hard- ness starts at 110% of the case depth. For that reason, it has to be ensured that the resulting subsurface shear stresses do not reach down too deep into the material and exceed the yield stress in shear or the limit shear stress in fatigue of the core mate- rial. Therefore, the following equation has to be fulfilled: shear,actual shear,allowable 1 (3.20) The allowable shear stress can be calculated either by means of the material’s ulti- mate tensile strength and the given factor of 0.425 shear,allowable = 0.425 UTS (3.21) or with an available value for the core hardness and the listed correlation to the yield strength in shear (cf. Reference [6]). For the common material 42CrMo4, a typical
- 132. Pitch system concepts and design 103 value for the core hardness is around 25 HRC. The actual occurring subsurface shear stress at the transition zone of case hardness and core hardness can be calculated by the following equation: shear, actual = b P 1.8754 105 (3.22) This equation implies the semi- minor axis of the contact ellipse b, the cumu- lative measure of curvature † and an interpolated subsurface shear stress parameter , which depends, among others, from the hardening case depth z. For the IWT7.5- 164 blade bearing, the hardening case depth is z = 8 mm. The maximum contact force Qmax is postprocessed from the rotor star FE model (cf. Section 3.1.5) loaded with the Mres,max load case. On the inner ring raceway, the resulting contact pressure is slightly higher than on the outer ring raceway, which leads to two different values for the semi- minor axis of the contact ellipse b. Even though shear, actualis slightly higher on the inner ring, it can be more criti- cal if the calculated shear stress on the outer ring comes close to shear,allowable as the outer ring is more exposed to tensile stresses. Failure to satisfy this core crushing check requires a reduction in the maximum load or a proper increase in hardened case depth. Fatigue Any fatigue life calculation starts with the lifetime loads acting on the component in question. According to Reference [1], fatigue calculations of any turbine com- ponent include DLC 1.2, 1.7, 2.4, 3.1, 4.1, 6.4 and 8.3, weighed based on their share of turbine operational time. DLC 1.2 represents the normal operation of the turbine throughout its lifetime and is therefore undoubtedly the most important of these. Various safety factors can be applied to the calculation. IEC 61400 combines all uncertainties into one material factor γm and one load factor γf , (γf = 1 for fatigue states) and DNV GL [4] additionally uses a factor related to the consequences of failure γn , which are to be applied to the cyclic stress or strain in each fatigue cycle. DNV GL recommends safety factor γm as shown in Table 3.5. The choice of parameters therefore depends on the specifics of the turbine design, but in many cases, rolling contact fatigue will use a safety factor of 1.0, if the turbine can safely endure the rotational failure Table 3.5 Safety factors γm according to DNV GL [5] Component failure results in the destruction of the wind turbine or endangers people Component failure results in wind turbine failure or consequential damage Component failure results in interruption of operation 1.25 1.15 1.0
- 133. 104 Wind turbine system design of one bearing. Ring fatigue can cause severe subsequent damage and will therefore mostly require factor 1.25. Bolted connections will likely use γm = 1.15 because they can be inspected and detected before a complete failure of the connection. Rolling contact fatigue For the blade bearing calculation, DLCs without any bearing movement may be skipped as their effect on the lifetime calculation is assumed to be non- existent with all common calculation approaches. As written above, fatigue lifetime (also referred to as rating life) of bearings is typically performed according to ISO 281 [22]. For blade (and yaw) bearings, DNV GL [5] specifically recommends ‘the ISO 281:2007 rating life calculation shall be modified according to NREL Wind Turbine Design Guideline DG03, Section 4’. This is due to a number of details that arise with blade bearings that ISO 281 does not sufficiently factor in: in particular, blade bearings are primarily loaded with a bending moment, which ISO 281 does not account for, and they move in small oscil- lations rather than by rotating as assumed in ISO 281. The process illustrated in the following is inspired by, but not identical to, NREL DG03 [7] but leans more closely towards that described in Reference [20] which aims to improve some aspects of the former. There is not much available literature on the accuracy of fatigue lifetime cal- culations for large slewing bearings, but some published calculations [20, 23] sug- gest that the lifetime calculated will be lower than the actual usable lifetime of a blade bearing. At the time of this writing, fatigue lifetime calculation is an ongoing research topic. If readers have access to non- published lifetime data, it is recom- mended to compare this to the calculations performed in the following to be able to correct any inaccuracies contained therein. Particular attention should be paid to the fact that rating life is supposed to describe the point at which the first spall appears on the raceway; for large slewing bearings, operation may be possible far longer than this point. In this case, we will focus on one of the bearings only, as fatigue loads will not differ significantly between the three blade bearings typically used in modern wind turbines. Required values from the DLCs are Mx , My , Fx , Fy and Fz at the blade root and the respective multipliers of each DLC. Time series must be turned into a table of oscillations. Reference [6] describes the process of a range- pair count on a data set; for the following RCF calculations, a rain flow count may be used as well and is recommended by NREL DG03. Both approaches will result in similar outcomes. Figure 3.16 shows examples for a rain flow count and a range- pair count. Following the cycle count, a bin count akin to those shown in References [6, 20] is carried out. The IEC- 61400- 4 [24] recommends the usage of LRDs for gearbox bearings. The following approach is very similar, but not identical to that of using LRDs, and LRDs may be used instead of the following bin counting, too. A generalised mean of each load component M (i.e., both forces and moments) within each bin should be performed over all time steps t = 1 … T of the bin accord- ing to
- 134. Pitch system concepts and design 105 Mweighted = 0 B B @ T P t=1 Mp t T 1 C C A 1/p , (3.23) with exponent p = 3 for ball bearings and p = 10/3 for roller bearing as also used below for the lifetime calculation. Moreover, a mean oscillation frequency Nb , the mean oscillation angle θb and the time tb spent in each bin b will be required. These can be used to determine the accumulated movement lb = 4·Nb ·θb ·tb within the bin, measured in degrees. The above equation holds true for the oscillation angle θ defined as per Figure 3.17. The oscillation angle will later also be required to correct for the oscil- latory behaviour of the bearing. Within each bin, it is then possible to calculate the lifetime L10 , which gives the statistical point at which 10% of bearings are expected to have a first spall on the raceway, L10 = acorr Ca Pa p (3.24) where p = 3 and p = 10/3 for ball and roller bearings, respectively. The factor acorr can represent different corrective factors for oscillation, lubrication, etc., all of which are multiplied with each other. Ca refers to the dynamic axial load rating, which is defined according to ISO 281 [22]. It is a function of geometrical parameters of the bearing and can be determined by the manufacturer, irrespective of operating condi- tions, and it is hence constant over all bins. For ball bearings with ball diameters of above 25.4 mm and contact angles lower than 90°, it is defined as Ca = 3.647bm fc (cos ˛)0.7 tan ˛ Z2/3 D1.4 W (3.25) for other bearings refer to References [22] and [7]. The variable Z refers to the num- ber of rolling elements per row, and DW refers to the diameter of the rolling elements. The contact angle is α and equals 90 for pure axial loads, in which case a different equation for Ca should be used, (cf. Reference [22]). The factor bm = 1.3, and the number of rows is given by i. The value fc is either calculated according to Reference Figure 3.16 Rainflow count (left) vs. range- pair count (right) [15]
- 135. 106 Wind turbine system design [25] or interpolated. NREL DG03 [7] includes values for an interpolation of fcm = fc bm using an osculation of 0.53. The variable Pa in the equation above is the equivalent load. It is ideally cal- culated via ISO/TS 16281 [26] through the process described in Reference [20], Section 2.6.3. In that case, an equivalent load for each race, denoted Qei for races on the inner and Qee for ones on the outer ring, is, respectively, calculated using the contact forces Qj taken from finite element simulations. This is the most accurate procedure to obtain the lifetime L10 . Reference [7] proposes two simplified formulae that are analysed in Reference [20]. One of them, denoted NREL 2, is a simplified variation of ISO/TS 16281 and uses the force Qj on two contact Points A and B on a race. Pa = 1 ZNREL PZNREL j=1 QjA + QjB 3 1/3 ZNREL sin ˛ (3.26) where ZNREL = Z i. Alternatively, without using finite element simulations and con- sequently the most inaccurate, one may use NREL 1 for an estimate, which goes Pa = 0.75Fr + Fa + 2.5M dm (3.27) using an adjustment proposed by Reference [20]. Ideally, the factor 2.5 should be validated with FE simulations by comparing to ISO/TS 16281 or NREL 2. Figure 3.17 Angle definition used for θ © Fraunhofer IWES
- 136. Pitch system concepts and design 107 Pa must be determined individually for each bin. Ideally, this means run- ning finite element simulations for the weighted mean values in each bin. Practically, this may not be possible, depending on the number of bins used. In this case, one may simulate only a specified grid of points. Menck et al. [20] propose an approach for a regression analysis of contact forces; an interpola- tion between the simulated data points is possible, too. For simple calculations, usage of NREL 1 can remove the need for finite element simulations entirely or reduce its number significantly if some simulations are performed to verify the formula. The above calculations result in a lifetime L10 that describes a lifetime in full revolutions. In order to account for the oscillatory behaviour of the blade bearing, an oscillation factor aosc is needed. Reference [27] proposes the curve- fitted (cf) factor aosc_cf = aosc_IR_cf 1 + aosc_IR_cf aosc_OR e Bcf 1/e 1 + Bcf 1/e (3.28) using for the (curve- fitted) IR oscillation factor aosc_IR_cf for point contact aosc_IR_PC_cf = f_crit_i 90 deg 1 0.09381 0.30679 , (3.29) where θdeg is the oscillation angle of the respective bin in degrees, and with aosc_OR = f_crit_o 90 deg , (3.30) where fθ_crit_i,o = 1 is recommended for typical blade bearings. Moreover, the load zone parameter ε for each combination of inner and outer raceways can be estimated as loaded 1 cos NLB Z 2 (3.31) provided that NLB Z, where NLB is the number of loaded balls per row. The param- eter Bcf for the PC case is determined via BPC_cf = 1 1 + 5.5958 fo 2fi 1 fi 2fo 1 !1.5153 1 + 0.11491 0.36257 (3.32) The factor aosc turns the lifetime L10 , measured in revolutions, into a lifetime L10osc , measured in oscillations. Since the following formulae will use revolutions rather than oscillations, we thus introduce a corrective factor acorr = aosc aHarris = aosc 90 deg. (3.33) The above equation corrects the oscillation factor aosc by the factor aHarris = 90/deg, which is the oscillation factor that is implicitly assumed by using LRDs or ‘summing up oscillations’ and dividing them over by 360° to obtain an equivalent number of rota- tions. acorr should reduce the lifetime for rotor blade bearings by about 10% if applied correctly.
- 137. 108 Wind turbine system design The load zone parameter ε can only be determined for one row (or combination of inner and outer raceways, in a four- point bearing). This means that the factor aosc can only be determined for one row (or inner–outer raceway pair). If ISO/TS 16281 is used for the calculation of Pa , each raceway pair can indeed use a separate aosc , but if NREL1 or NREL2 are used, one may use the lowest value of aosc out of all the raceways as a conservative estimate. At this point, a lifetime L10b can be calculated for each of the bins b. These are then combined into a lifetime of the bearing over all bins L10 . This is achieved using the Palmgren–Miner rule, L10 = 1 lcoll X b lb L10,b !1 (3.34) where lb is the movement in each bin b summed up (in degrees) over the entire operating time of the turbine, and lcoll is the sum of all movement in all bins, i.e. lcoll = Σb lb . This finally results in a lifetime L10 , measured in revolutions, i.e., 360° move- ments, of the bearing. Now an average rotational speed navg (in rev/h) over the entirety of operational time must be calculated by using navg = P lb P tb (3.35) The final bearing lifetime in hours L10m,h is then finally calculated as L10m,h = L10 106 navg (3.36) The lifetimes L10b above can be modified to account for the influence of lubrica- tion and contamination by determining the modified lifetime according to ISO 281. NREL DG03 gives a summary of this procedure with a focus on blade bearings. Using the procedure as given in NREL DG03, however, results in extremely low values of aISO , typically giving the lowest possible absolute value of aISO = 0.1 [7, 20, 23]. It should thus be noted that ISO 281 specifically allows for a correction of the viscosity ratio κ for lubricants that contain effective EP- additives in applica- tions in which poor performance would otherwise be expected. This is possible if the additives have been tested under realistic lubrication conditions, e.g., in a real application or by performing an FE8- Test according to DIN 51819- 1. If this condi- tion has been fulfilled, κ can even be set to 1, which results in a significant increase of aISO . This fact highlights the importance of an appropriate grease with effective EP- additives for the best performance of a blade bearing not just with respect to wear but also raceway fatigue. If the raceways are not hardened to 58 HRC, a correction of the lifetime, e.g., as given by NREL DG03, is necessary. Likewise, the case depth of the raceways must be chosen appropriately such that at the boundary between core and hardened raceway, the allowable fatigue shear stress of the core (i.e., the fatigue strength of
- 138. Pitch system concepts and design 109 the core with respect to shear stress) is not exceeded, otherwise, an adjustment as shown in NREL DG03 must be undertaken. Ring stresses DNV GL requires the tangential hoop stresses of the bearing ring to be ‘analysed and documented’. Compressive hoop stresses can increase the raceway life of a bearing [28], but tensile hoop stresses can have the opposite effect. If the ring stresses during fatigue loading are significantly above the fatigue limit of the ring material, a fatigue lifetime calculation of the rings may be necessary to prevent ring failure. Specifically, around the bolt holes and filler plugs if present, high stresses are expected. DNV GL 2016 [5] requires a fatigue calculation of the bolt holes. Bearing rings can be considered as a structural component since they are part of the wind turbine structure that transfers significant loads. For structural components, Reference [4] refers to ISO 2394 [29]. ISO 2394 highlights the importance of accu- rate models that particularly take into account those properties of the component that have a high effect on its behaviour. For blade bearings, this particularly means that FE models with an accurate representation of surrounding structures are very use- ful; otherwise, high safety factors must be used due to high model uncertainty. Time series from aeroelastic simulations must be turned into time varying sequences of ring stresses, which, in turn, have to be transferred to bins of loads cycles using, e.g., the rain flow count approach. With these bins, S–N curves are commonly analysed using the Palmgren–Miner rule, as is also done in the following sections for bolts. S–N curves for unhardened 42CrMo4 steel taken directly from a blade bearing can be found in the literature, e.g., in Reference [30], or experimentally determined. According to ISO 2394, the calculation has to take into account various safety fac- tors for possible uncertainties in the load level and model, material model and the damage accumulation rule itself. These can be used as defined above by DNV GL or IEC. Bolt fatigue Due to the mostly axial, dynamic loading at the bearing flanges, bolts are at risk of fatigue failure. A bolt fatigue calculation is hence required by DNV GL. S–N curves of the bolts depending on the manufacturing process of the bolts and the number of load cycles are given in Reference [4] for a survival probability of over 97.7%. The S- N curves are given in Table 3.6 in terms of detail categories according to Reference [31] and negative inverse slope. The detail category of a bolt gives the reference fatigue strength in MPa, which corresponds to the point of the S- N curve that results in N = 2 million cycles. For bolts with a nominal diameter d 30 mm, the S- N curve is reduced by the factor ks = 30 d 0.25 (3.37)
- 139. 110 Wind turbine system design According to References [4] and [31]. Both sources also highlight the importance of considering bending stresses resulting from prying effects and other sources. This may be omitted if detail category 36 is used instead. Stress histories can be evaluated using a rain flow count or the reservoir method, (cf. Reference [31]). Like with other fatigue calculations, different load amplitudes can be combined using the Palmgren–Miner rule D = P b nb Nb (3.38) where nb is the number of load cycles that occurred in bin b, and Nb is the number that can occur according to the S–N curve given above at the specified stress ampli- tude. After including all bins b, the result D ≤ 1 fulfils the design requirements. The number of bins b should be high enough for sufficient accuracy, [32] recommends at least 20. Bolt loading along the circumference of the bearing flange will vary signifi- cantly, and the highest bolt loads are expected in the centre of the tension side of the bearing. Here, a sufficiently large section of bolts should be validated against fatigue failure. 3.1.7 Lubrication system The lubrication mainly ensures the function of the pitch system. In modern wind tur- bines, a system supplies the lubricant automatically and in predefined intervals. The main principals for pitch and yaw systems are similar. To avoid doubling, Chapter 7.3 gives further information regarding lubrications systems. This section explains charac- teristics for pitch systems. The gearbox itself or bearings inside the motor and gearbox must be lubricated as well. The gearbox is filled with oil, which circulates with the Table 3.6 S- N curves for different bolt types according to DNV GL [1] Detail category Negative inverse slope Bolt type For N 5·106 cycles For N ≥ 5·106 cycles 71 2 FS max F0,2 min 85 FS max= max. bolt force F0,2 min= bolt force at 0.2 % elastic strain limit 6 11 Rolled after heat treatment, no thermal coating 71 3 5 Rolled before heat treatment, no thermal coating 50 3 5 Rolled threads, thermal coating
- 140. Pitch system concepts and design 111 rotation of the hub. Some bearings inside can also be grease lubricated. The pitch drive is not considered further in this section since their lubrication is independent. Lubrication instructions are made by the supplier. First, it is important to bring up that the design of the parts is linked to each other. For example, the sealing material depends on the lubricant. Furthermore, the lubrication system characteristics are determined by the lubricant as well. In any case, the lubrication system, the lubricant and the sealings must fulfil following requirements: • • lubricate the rolling contact in the bearing and further positions like toothing • • prevent leakage • • protect against contamination from the environment (dust, particles, water, salt water) • • operate in temperature between −20 and +55°C and for a lifetime from 20 or up to 30 years • • operate safe in a spinning setting The lubrication must prevent wear of the surfaces by separating the contact partners and should reduce the coefficient of friction. Oil or grease are possibilities. In contrast to oil, grease is less prone to leak. Especially if the bearing rings deform oil could easily leave the bearing. Hence, the common lubricant for blade bearings is grease. Different commercial greases are available with different rheological and chemical properties. A grease consists of a thickener, a base oil, additives and pos- sibly other components, like solid lubricants. The composition of these can vary. Blade bearings are typically initially filled with grease by the manufacturer. During operation, pumps are regreasing the bearings continuously (cf. Chapter 8). Therefore, the pump is connected via hoses to the bearing. Grease inlets, which are basically drill holes, transmit the grease through the bearing ring to the inner side. The old grease can leave the bearing through grease outlets that are designed like the grease inlets. The old grease collects in a box or bottle. Figure 3.18 shows a drill hole of a cut segment of a four- point ball bearing. A threaded connection allows to connect hoses and collecting container to the bearing as shown in Figure 3.19. A few aspects must be considered for the design of the grease in- and outlets: • • They should be as short as possible. • • They could be axial (flange side) or radial (as shown in Figure 3.19). • • Edges and diameter changes especially on the outlet can lead to plugging. • • Accessibility for maintenance is easier if they are in the hub. Blade bearings have a rotating and a fixed ring. It is clearly easier to mount the hoses and old grease bottles to the fixed ring. It is not necessary to check pos- sible clashes and the hose routing can be simpler since the ring is not moving. Furthermore, the fixed ring does not have an interface towards the actuator, which gives more space. For maintenance it is also easier to mount these parts to the IR.
- 141. 112 Wind turbine system design Technicians can access the components without leaving the hub. If the IR is the rotating one, like for the IWT7.5- 164 blade bearing (see Section 3.1.1), a trade- off is necessary. The IWT7.5- 164 blade bearing has 20 grease inlets for each row with a size of M10 and ten outlets per row. The outlets are thicker, they are M16. The number of grease inlets and outlets is equally distributed around the bearing circumference. Both are in radial direction at the outer ring, see Figure 3.19. The relubrication intervals are experience based and can differ a lot from one bearing manufac- turer to another. One simple example is to renew the initial amount of grease in one year. The IWT7.5- 164 blade bearing has initially about 30 kg grease, which would lead to an amount of 82 g new grease every day. An additional pinion made of foam or plastic can supply the gearbox pinion and the bearing toothing. Here similar rules of thumb apply for the relubrication. 20 g every day is an exemplary value. Although the new grease blend is equally, some of the collecting boxes will be filled with more old grease than others. It is most likely that the boxes in the lower loaded arears stay empty or are filled less than the boxes in the load zone of the bearing. This requires some attention in the design. It makes sense to increase the number of grease outlets in the load zone or to increase the size of these boxes. Figure 3.18 Drill hole for grease © Fraunhofer IWES
- 142. Pitch system concepts and design 113 The sealings have two objectives. On the one side they must prevent leakage, on the other side, they should avoid any contamination. It could be particles, dust, water or salt water. Therefore, the sealing has at least one lip towards the other ring. Two or three lips increase the friction torque but perform better. The lips can be in radial or in axial direction. Large deformations can be a problem for the sealing. An FE calculation can help to understand the deformation behaviour and to identify the gap the sealing must cover, e.g., under extreme loads. 3.1.8 Coating A blade bearing must be protected against corrosion. A common way to achieve this aim is to zinc- coat the surfaces by metal spraying. According to ISO 206, this is ide- ally done with a layer thickness of 100–200 µm. For offshore or nearshore turbines, it makes sense to increase the thickness. Here, 30 µm can be added to the given values. The requirements for the non- mounting surfaces differ from requirements to surfaces towards the blade and hub, or bolt heads, or nuts. The variation of the coating height should not exceed 50 µm for these flange surfaces. It is advisable to evaluate the resistance of the coating with a salt- water test. Especially during transportation water could drip to the bore holes and cause corrosion. Thus, bore holes must have a protection against corrosion as it reduces the ring fatigue lifetime significantly. This can lead to cracks and subsequently results in ring fracture with a falling blade if not detected early enough. An oil can protect lifting points, without soiling the thread. Figure 3.19 Grease inlet with hoses and connector (left), grease outlet with collection bottles (right) © Fraunhofer IWES
- 143. 114 Wind turbine system design 3.2 Pitch actuator The actuator adapts the pitch angle by turning the bearing and thus the blade. It is connected to the bearing rotating ring (cf. Section 3.1.1). It is one of the most critical safety systems of the wind turbine. In an emergency case, it is essential to move the blade back to 90° as quickly as possible to prevent the turbine from overspeed. The kind of actuator determines the interfaces towards blade bearing and hub, as well as the power supply. The following actuator types are possible: • • electrically powered gear drive • • electrically powered belt drive • • hydraulic actuator This chapter gives further information about possible designs, operating condi- tions and aspects that must be considered for the dimensioning of an electrically powered gear drive. A hydraulic pitch actuator is presented in Chapter 7. 3.2.1 Electrical actuator A position- and speed- controlled motor turns the rotor blades to the setpoint angle, which comes from the wind turbine controller to reduce global loads and control the rotor speed and power output of the turbine. The rated torque of the electrical and mechanical drivetrain needs to be high enough to turn the rotor blade in all produc- tion load cases with a given minimum speed. The maximum torque of the drivetrain needs to be chosen to cover all emergency cases. The electrical actuator generally consists of a back- up cabinet, control cabinet, axis cabinets (including the converter), an electrical motor and a gearbox. Normally, there is an autonomous system for each blade axis and a higher- order system to control and monitor each axis. Figure 3.20 shows a schematic diagram of the electrical pitch system. Cabinets In conventional systems, electromechanical components within the control cabinet execute many functions (including safety functions) and the converter is mainly used for speed control of the motor. Besides the speed and position control, mod- ern systems provide many additional functions like condition monitoring, control of additional components inside the hub (e.g., lubrication system, hub ventilation, blade sensing systems). The connection for power and communication via the slip ring to the nacelle is made inside the control cabinet. Axis cabinets contain all necessary functions for the connected blade. The blade position or speed demand is given from the turbine controller. The pitch drive takes this as input for the overall control loop. Underlying the overall control loop is the speed feedback loop and the current loop. All those need to be adjusted for a well- tuned pitch behaviour. Depending on the overall system design the converter is used for one or more of these tasks. Depending on the electrical pitch system design the converter also charges the back- up system.
- 144. Pitch system concepts and design 115 Like the name already indicates, the back- up cabinets contain the back- up sys- tem. In an electrical pitch system, this is done typically with batteries or ultra- caps (UC). From the beginning valve- regulated lead–acid batteries had been used and are still being used. But other batteries, like lithium- ion or LiFePo4 (lithium iron phosphate) can also be found in back- up systems. Motor For electrical actuators, there are two motor types that are possible: alternating cur- rent (AC) and direct current (DC) motors. AC motors can be separated into AC asynchronous and AC synchronous motors. Classically, DC motors had been used, but meanwhile also both AC motors are common. The choice of AC asynchronous, AC synchronous or a DC motor is mainly a question of philosophy and should be made during system level design both from a cost and design/safety perspective. AC motors are significantly cheaper than DC motors but the down effect that they cannot be directly powered by the DC power storage in emergency cases has to be considered in functional safety as well as in load simulations. Table 3.7 lists advantages and disadvantages of both types. Each motor, no matter if AC or DC, is equipped with an electromagnetic brake. This brake is closed when the power supply is interrupted. This makes sure to hold the blade in safe position even during grid loss events for a long period of time. Figure 3.20 Schematic diagram of an electrical pitch system © Nidec SSB Wind Systems GmbH
- 145. 116 Wind turbine system design Gearbox A gearbox increases the torque and reduces the speed from the motor. Typically, gearboxes have multi- stage planetary gears to realise a high transmission within limited space. It is filled with an oil to lubricate the gears and bearings. The output pinion drives the blade bearing rotating ring and thus the blade. Performance level For the performance level (PL), there are two main functions that need to be fulfilled to ensure a safe stop of the turbine during an emergency operation: • • safe movement to feather position • • safe stop in feather position The function ‘safe movement to feather position’ is triggered by the safety chain, the turbine controller or by the pitch system itself. The safety chain contains two channels via the slip ring and is connected to a safety relay inside the pitch sys- tem. This safety relay initiates the emergency function of each perfect pitch axis by disconnecting 24V DC of the emergency feather command (EFC) input. The emer- gency operation will be stopped when the first limit switch is reached by the func- tion ‘safe stop in feather position’. According to the recommendation of DNV- GL Guideline [4], the PL should be ‘d’ for the function ‘Protection against excessive rotor speed’. As mentioned, this is a recommendation. The PL is categorised from ‘a’ to ‘e’ and is defined in the DIN EN ISO13849- 1 [33], where ‘a’ is showing the lowest probability of a dangerous failure per hour and ‘e’ is the highest. The real required safety level needs to be evaluated during the risk assessment of the turbine. Safety- related parts of control systems (SRP/CS) are divided into subsystems (input, logic, output) according to DIN EN ISO 13849 [34]. In this content, the SRP/ CS provides the safety function including a PL, which effectuates the necessary risk minimisation. By providing the safety function, the layout of the SRP/CS is part of the strategy of the risk minimisation. Table 3.7 Pros and cons for AC and DC motors Pros Cons AC • Low costs for motor and converter. • Motors cannot be directly powered by a battery in case of converter failure. DC • Motor can be directly powered by batteries in case of a converter failure. • High torque at low speeds. • Brushes on motor must be frequently replaced. • DC motors are costly.
- 146. Pitch system concepts and design 117 Power supply During normal operation, the pitch system is powered by main power that is trans- mitted via a slip ring unit to the turning hub. For emergency cases, e.g., during power loss, an additional power supply is needed. In many cases, batteries act as such a temporary power storage but there are alternatives like ultracapacitors as well. The size of the temporary power storage depends on the power needed to turn the blade from production position (0°) to parking position (90°). For the layout, the required torque over time and the required speed over time need to be taken into account. Especially for the layout of ultracapacitor back- up systems this needs to be checked carefully, as the layout normally is made for 1.5–2 possible emergency drives to feather position. This is mainly caused by the higher initial costs for ultra- caps (UC) back- up systems compared to battery back- up systems. Interfaces Besides the mechanical interfaces for mounting the cabinets, there are several elec- trical interfaces inside the hub of a wind turbine. Typically, there are the following lines between the hub and the nacelle via the slipring unit: • • 3–400 V AC supply • • 24 VDC signals for safety functions • • fieldbus for communication between the pitch system and the turbine controller For the fieldbus, those are the most common ones: • • CANopen • • Profibus • • Profinet • • Ethercat Sensors The following sensors are typically be found for monitoring and control of a pitch system: the pitch angle is monitored typically by an absolute value encoder and will be compared with a redundant value of different sensor. Furthermore, the pitch speed information will be derived from the pitch angle. Current and voltage for each pitch motor will be tracked within the pitch drive. Besides this, there are several temperature sensors, like: • • cabinet temperature • • ambient temperature • • motor temperature • • gearbox temperature
- 147. 118 Wind turbine system design 3.2.2 Operating conditions The pitch actuator rotates with the hub. Hence, a centrifugal force is acting at the actuator. Shocks by gusts effect the pitch actuator as well. Interfaces like bolted con- nections and electrical connectors and the components themselves must withstand this load. Furthermore, it influences liquids, like the gearbox oil. Lubrication, which includes the sealings, must fit to the rotating operation. The cut- out wind speed of the turbine and the position of the actuator in the hub allow an estimation of the centrifugal force. Under normal operation, the cut- out wind speed is the highest rotational speed. The distance between the rotational axis and the actuator is the lever arm. The IWT7.5- 164 has a cut- out wind speed of 10 rev/min. The distance from the rotational axis towards the motor and gearbox is about 2 m. Environmental conditions effect the actuator as well. It is necessary to distin- guish between the position of the blade if it is mounted to the inner or outer ring. In the second case, the actuator must be mounted to the outside of the hub and protec- tion against water or saltwater become more important. For both types, a protection against dust and salt in the air is necessary. Temperature conditions can derivate from the turbine temperature conditions, e.g., according to IEC 61400- 1 [4]. Due to active loads inside of the pitch cabinet, the experience shows a 10–15 K higher inside temperature compared to the ambient temperature. This delta can be reduced to 5–10 K if those cabinets will be equipped with a fan. Measures to keep the need protection grade of IP54 need to be done in this case. Table 3.8 lists the temperature conditions. Generally, three different temperature versions are defined: • • normal climate version • • cold climate version • • hot climate version 3.2.3 Calculation and dimensioning As mentioned in Section 3.1, loads for the dimensioning typically stem from the aero- elastic simulations (cf. Chapter 1). An aero- elastic simulation time series of IWT7.5- 164 reference turbine is available online and can be downloaded under [3]. The most important load component is the moment (Mz ), which would rotate the blade along its longitudinal axis. The pitch actuator and the brake must be able Table 3.8 Temperature conditions for the pitch actuator Normal conditions Extreme conditions T min −10°C −40°C T max +40°C +50°C
- 148. Pitch system concepts and design 119 to counteract this moment. The other loads at the blade root influence the load in the bearing and thus the friction torque. Besides the loads, the pitch speed P ' is a mandatory input for the dimensioning of the pitch actuator, as the speed and torque define the power. Power calculation The first step is the calculation of the required power of the pitch actuator. The torque (M), which is necessary to turn the blade, consists of several parts. The actuator must counteract the friction torque Mfric , the acting wind loads Mz and the inertia (J ), as 3.39 shows. M is the torque at the blade bearing ring. M = Mz + Mfric + J ' (3.39) To calculate the motor torque (MM ) the transmission from the bearing ring towards the gearbox pinion (ibearing ) and the gearbox transmission (igearbox ) needs to be con- sidered, as 3.40 shows. Information about the gearbox transmission is not available yet. The choice of a gearbox is probably an iterative process and must balance with the motor. Furthermore, losses in the toothing, the gearbox, motor and converter increase the needed power as well. They can be considered with efficiency factors. MM = M igearbox ibearing (3.40) The blade weight and its design have the highest influence on the inertia since it is the heaviest part. Nevertheless, the inertia of the rotating ring of the blade bearing and further parts like stiffener plates add to the total weight. If the design is fixed, it is possible to determine the inertia of all parts. The wind loads are dependent on the current pitch angle and the wind conditions. They can be simulated and hence esti- mated in an aeroelastic simulation. Dependent on the direction of the pitch move- ment Mz can be supportive or not. However, for an extreme load calculation, it is considered as an additional resistance. The friction in the blade bearing comes with the highest uncertainty. It consists of several parameters. The friction in the roll- ing contact is determined by the type and number of rolling elements and it is load dependant. The sealings, the cage and grease displacement have an influence as well. With ongoing degradation of the bearing the friction torque will increase. Even if the bearing is damaged, often it still can turn, just with a higher torque. A comprehen- sive overview over several test results is presented in Reference [35]. It is possible to use bearing manufacturers equations to estimate the friction torque, but often, the bearings do not exactly reflect the behaviour predicted by the models. Furthermore, the friction torque models in the literature differ as shown in Reference [36]. In addition, it is questionable which one fits best to the application. Hence it may be reasonable to add safety factors. The required torque and the required pitch speed then can be used to choose a motor type, its power- class and the gearbox. Equation 3.41 can be used to calculate the mechanical power. P = MM P ' (3.41)
- 149. 120 Wind turbine system design Besides the motor curve, a decision needs to be made for the type of motor, fitting best to the turbine needs (DC, AC asynchronous, AC synchronous). Mechanically, the motor flange and shaft need to be aligned with the gearbox flange and shaft. As pitch system suppliers not always also deliver the gearbox, this must be considered on both sides. Emergency shut down An emergency shut down of the turbine requires turning the blades to feather posi- tion. According to the load simulation, the best emergency shut down speed will be defined and is implemented into the pitch drive. For a detailed layout of the back- up system, the load requirements and the speed requirements during the emergency operation need to be known and considered. Typically, the required EFC speed is defined by the back- up voltage and the needed torque will be considered in the installed capacity. Especially for UC back- up systems this is more relevant, as the design considers, with a small safety margin, the exact needed energy. Figure 3.21 shows an example of the decreasing voltage level, during an EFC operation. Here it must be made sure that the needed voltage level for the required EFC speed is kept. Temperature The IEC 60034 [37] gives guidelines for rotating electrical machinery, like the pitch drive. Rising temperatures, because of high currents inside the drive and the motor, will be monitored by the pitch system itself. The drive temperature will be measured Figure 3.21 Decreasing voltage level of a capacitor during an EFC operation © Nidec SSB Wind Systems GmbH
- 150. Pitch system concepts and design 121 by an internal temperature sensor as well as a Ixt- function to reach the full dynamic of the drive. The Ixt- function displays the thermal load of the electronic components inside the drive, which allows to use the full dynamic without risking the drive to fail on thermal reasons. For the motor, there are typically a temperature sensor showing the real tem- perature in the winding and additionally a positive temperature coefficient close to the winding temperature limit. Brake The brake prevents unintended changes of the pitch angle. Typically, the brake is mounted at the motor and holds the motor shaft, hence acting loads must be trans- mitted to the motor. The highest Mz from the simulation data is 729 kNm. It occurs in DLC 6.2, where the turbine is in parked position and the wind speed is extreme. A safety factor of 1.1 applies for this load case. If the brake is active, speed- depended terms from 3.39 can be omitted since the blade does not pitch. Furthermore, and in contrast to an intended direction change of the pitch movement, here the pitch bearings friction helps to keep the blade in position. Gearbox The gearbox design must fit to the motor. The gearbox is probably an outsourced item. Nevertheless, a static and a fatigue strength verification according to DNV GL [5] are necessary. For the calculations of the components the individual norm applies. For example, for the bearings ISO 76 [9] and ISO 281 [22] or for the gears ISO 6336 [38]. The usage as a part of a wind turbines pitch drive comes with one additional challenge. The pitch angle adaptions are often just a few degrees. Besides a frequent direction change, it leads to an unsymmetrical load of the teeth. A small adaption of the pitch angle means that just a few teeth transmit the load. This affects especially for the stages with a low rotational speed, the pinion and the blade bearing itself. These circumstances could be considered with a correction factor. References [1] DNV GL ‘Loads and site conditions for wind turbines’. 2016. [2] Popko W. and Thomas ‘IWES Wind Turbine IWT7.5- 164. Rev 4’. 2018. [3] Popko W. ‘Aero- Elastic Simulation Time Series of IWT7.5 Reference Turbine’. 2019. [4] IEC 61400- 1 ‘C wind turbines - part 1: design requirements’. 2019. [5] DNV GL ‘Machinery for wind turbines’. 2016. [6] Stammler M., Reuter A., Poll G. ‘Cycle counting of roller bearing oscillations – case study of wind turbine individual pitching system’. Renewable Energy Focus. 2018, vol. 25, pp. 40–47. [7] Harris T., Rumbarger J.H., Butterfield C.P. ‘Wind Turbine Design Guideline DG03: Yaw and Pitch Rolling Bearing Life’. 2009.
- 151. 122 Wind turbine system design [8] Houpert L. ‘An Engineering Approach to Hertzian Contact Elasticity—Part I’. Journal of Tribology. 2001, vol. 123(3), pp. 582–588. [9] DIN ISO 76:2009- 01 ‘Rolling bearings - static load ratings (ISO 76:2006)’. 2006. [10] VDI Verein Deutscher Ingenieure ‘Systematische Berechnung Hoch- beanspruchter Schraubenverbindungen Zylindrische Einschraubenverbind- ungen’. VDI. 2015. [11] Wittel H., Jannasch D., Voßiek J., Spura C. ‘Roloff/MatekMaschinenelemente: Normung, Berechnung, Gestaltung’. in Springer Vieweg; 2019. [12] Daidié A., Chaib Z., Ghosn A. ‘3D simplified finite elements analysis of load and contact angle in a slewing ball bearing’. Journal of Mechanical Design. 2008, vol. 130(8). [13] Gao X.H., Huang X.D., Wang H., Chen J. ‘Modelling of ball- raceway contacts in a slewing bearing with non- linear springs’. Proceedings of the Institution of Mechanical Engineers, Part C. 2011, vol. 225(4), pp. 827–831. [14] Smolnicki T., Rusiński E. ‘Superelement- based modeling of load distribu- tion in large- size slewing bearings’. Journal of Mechanical Design. 2007, vol. 129(4), pp. 459–463. [15] Stammler M. ‘Endurance test strategies for pitch bearings of wind turbines’ Fakultuät für Bauingenieurwesen und Geodäsie, Gottfried Wilhelm Leibniz Universität Hannover; 2020. [16] Stammler M., Baust S., Reuter A., Poll G. ‘Load distribution in a roller- type rotor blade bearing’. Journal of Physics: Conference Series. 2018, vol. 1037. [17] Wang H., He P., Pang B., Gao X. ‘A new computational model of large three- row roller slewing bearings using nonlinear springs’. Journal Mechanical Engineering Science. 2017, vol. 231(20), pp. 3831–3839. [18] Chen G., Wen J. ‘Load performance of large- scale rolling bearings with sup- porting structure in wind turbines’. Journal of Tribology. 2012, vol. 134(4). [19] Laird D., Montoya F., Malcolm D. ‘Finite element modeling of wind turbine blades’. 43rd AIAA Aerospace Sciences Meeting and Exhibit [online]; Reno, NV, Reston, VA, 2018. Available from https://arc.aiaa.org/doi/book/10.2514/ MASM05 [20] Menck O., Stammler M., Schleich F. ‘Fatigue lifetime calculation of wind turbine blade bearings considering blade- dependent load distribution’. Wind Energy Science. 2020, vol. 5(4), pp. 1743–1754. [21] Plaza J., Abasolo M., Coria I., Aguirrebeitia J., de Bustos I.F. ‘A new finite ele- ment approach for the analysis of slewing bearings in wind turbine generators using superelement techniques’. Meccanica. 2015, vol. 50(6), pp. 1623–1633. [22] DIN ISO 281:2010- 10 ‘Rolling bearings – Dynamic load ratings and rating life (ISO 281:2007)’. 2007. [23] Schwack F., Stammler M., Poll G., Reuter A. ‘Comparison of Life calcula- tions for Oscillating Bearings Considering Individual Pitch Control in Wind Turbines’. J. Phys.: Conf. Ser.. 2016, vol. 753, p. 112013. [24] International Electrotechnical Commission ‘Wind turbines: part 4: Design requirements for wind turbine gearboxes, IEC 61400- 4’. 2012.
- 152. Pitch system concepts and design 123 [25] ‘DIN, DIN SPEC 1281- 1:2010- 05: Wälzlager - Erläuternde Anmerkungen zur ISO 281 - teil 1: Dynamische Tragzahlen und nominelle Lebensdauer (ISO/TR 1281- 1:2008 + cor. 1:2009)’. 2009. [26] ‘DIN, DIN 26281:2010- 11: Rolling bearings – Methods for calculating the modified reference rating life for universally loaded bearings (ISO/TS 16281:2008 + cor. 1:2009)’. 2009. [27] Houpert L., Menck O. ‘Bearing life calculations in rotating and oscillating applications’. Journal of Tribology. 2022, vol. 144(7), pp. 1–31. [28] Oswald F.B., Zaretsky E.V., Poplawski J.V. ‘Relation between residual and hoop stresses and rolling bearing fatigue life’. Tribology Transactions. 2014, vol. 57(4), pp. 749–765. [29] ‘ISO, ISO 2394:2015- 03: general principles on reliability for structures’. 2015. [30] Friederici V., Schumacher J., Clausen B. ‘Crack propagation modelling for service life prediction of large slewing bearings’. Procedia Structural Integrity. 2014, vol. 35, pp. 106–114. [31] ‘European committee for standardisation, EN 1993- 1- 9: eurocode 3: design of steel structures- part 1- 9: fatigue’. [32] Det norske veritas ‘DNV- RP- C203: Fatigue Design of Offshore Steel Structures’. 2021. [33] ‘DIN EN ISO 13849- 1:2016- 06: Sicherheit von Maschinen- sicherheitsbezogene teile von Steuerungen- teil 1: Allgemeine Gestaltungsleitsätze, DIN ISO’. 2016. [34] DIN ISO ‘ISO 13849- 1:2015: Safety of machinery — Safety- related parts of control systems — part 1: General principles for design’. 2015. [35] Menck O., Behnke K., Stammler M., Bartschat A., Schleich F., Graßmann M. ‘Measurements and modeling of friction torque of wind turbine blade bear- ings’. Journal of Physics. 2022, vol. 2265(2), p. 022087. [36] Stammler M., Schwack F., Bader N., Reuter A., Poll G. ‘Friction torque of wind- turbine pitch bearings – comparison of experimental results with avail- able models’. Wind Energy Science. 2018, vol. 3(1), pp. 97–105. [37] IEC ‘Rotating electrical machines: part 1: Rating and performance’. [IEC 60034- 1] 2022. [38] DIN ISO ‘ISO- 1:2006- 09: Tragfähigkeitsberechnung von Gerad- und Schrägver zahnten Stirnrädern - teil 1: Grundnorm, einführung und allge- meine Einflussfaktoren’. 2006.
- 153. This page intentionally left blank
- 154. 1 bewind GmbH, Rendsburg, Germany 2 windwise GmbH, Münster, Germany Chapter 4 Yaw system concepts and designs Christian Bulligk1 and Daniel von dem Berge2 The yaw system of a wind turbine is responsible for orientating the wind turbine rotor towards the wind. This chapter is intended to provide an insight into common yaw sys- tem concepts and designs. The focus is on active and friction- damped yaw systems with electro- mechanical drives, which are the most common concepts in multimegawatt upwind turbines onshore and offshore. First, the fundamentals are described, which are necessary for the general under- standing of yaw systems. Second, the design loads are discussed in detail to create the basis for the subsequent sections. Third, common yaw system concepts and component designs with their advantages and disadvantages are presented. Finally, a yaw system is dimensioned for the Fraunhofer IWES wind turbine IWT- 7.5- 164 and some design aspects are discussed in more detail. 4.1 Fundamentals This section is intended to briefly summarise the essential basics of yaw systems that are required for further understanding. The first subsection describes the main function, the basic structure, the functional principle and some functional requirements. Subsequently, wind direction and yaw misalignment as well as their influence on energy yield and tur- bine loads are discussed. Finally, typical key data of yaw systems are presented to get an idea of the main design parameters. 4.1.1 Introduction The main function of the yaw system is to align the wind turbine rotor with the wind direction: • • to increase the energy yield of a single wind turbine • • to reduce the fatigue load of a single wind turbine • • to maximise the energy yield of a wind farm
- 155. 126 Wind turbine system design • • to optimise the wind turbine loads in a wind farm A sub- function of the yaw system is to create a rotatable connection between the nacelle and the tower and to transfer the loads from the nacelle into the tower. Upwind turbines are usually equipped with active yaw systems consisting of the following main components (Figure 4.1): • • A yaw bearing: It rotatably connects the nacelle with the stationary tower and transfers the loads from the nacelle into the tower. The yaw bearing comprises an internal or external ring gear bolted to the tower. • • A set of yaw drives: The yaw motors provide the driving torque and rotational speed to turn the nacelle against the tower. They are usually also equipped with motor brakes that provide holding torque to hold the nacelle in position. The speed reducing yaw gearboxes mounted on the nacelle convert torque and speed. Their output pinions mesh with the teeth of the yaw bearing. • • A dedicated yaw brake system: It holds the nacelle in position when the rotor is aligned with the wind direction or when the wind turbine is maintained. In Figure 4.1 Typical arrangement of yaw system components
- 156. Yaw system concepts and designs 127 addition, it provides friction torque during yawing to reduce the alternating loads on the yaw drivetrain. The yaw brake system usually consists of a set of yaw brakes mounted on the nacelle and acting on a yaw brake disc mounted between the yaw bearing and the tower. However, there are also yaw system concepts in which the nacelle is held in position solely by the yaw motor brakes or by the yaw motors themselves (see section 4.3). • • A yaw control system: It processes the signals from the wind direction sensors and the yaw system sensors such as the nacelle position. The control system manages, commands, directs or regulates the behaviour of the yaw system. Downwind turbines have the theoretical advantage that they may be built with passive yaw systems since the rotor itself is able to yaw the nacelle into the wind. However, if a wind turbine yaws passively in the same direction for a long period of time, the cables between the nacelle and the tower are twisted impermissibly. For that reason, at least a few yaw drives are required for untwisting the cables. In addition, passive yaw systems for multimegawatt wind turbines have to be designed in a way that the nacelle does not follow every change in wind direction and the yaw speed does not become too high to avoid high gyroscopic loads. As a result, passive yaw systems are basically similar to active yaw systems as described above. When the rotor is aligned with the wind direction or when the nacelle is main- tained, the yaw system has to ensure that the nacelle holds its current position with the holding torques of the yaw system components. These torques can be: • • friction torque of the yaw bearing • • friction torque of the yaw brake system • • friction torque of the yaw motor brakes • • driving torque of the yaw motors If the external load exceeds the holding torque, the yaw system slips. Such yaw slippage events have to be carefully assessed because they can lead to critical situa- tions for the wind turbine and the yaw system components (see section 4.2.6). For maintenance and repair activities, a yaw locking device is required in case the holding torque is not sufficient, with which the yaw system is locked in place to exclude the risk of personal injury. According to the DNV standard [1], a yaw locking device is not needed if there are two independent brake systems, e.g., the yaw brake system and the yaw motor brakes, and each brake system is able to hold the nacelle in position. When the control system detects an average difference between wind direction and nacelle position that exceeds a certain value in a certain period, it initiates a yawing process. There are typically different criteria for adapting the frequency of the yawing process to the respective wind turbine operating status, such as: • • idling of the wind turbine • • power production at low wind speeds (partial load range) • • power production at high wind speeds (full load range)
- 157. 128 Wind turbine system design In addition, different criteria are monitored in parallel. For example, larger mean wind direction deviations are tracked earlier than smaller ones. If the nacelle position deviates significantly from the wind direction, the nacelle is immediately adjusted to the wind. Due to the cabling between the nacelle and the tower, the nacelle can only be rotated up to a certain angle in one direction. Depending on the installed cable loop, the cables can be twisted up to three complete revolutions in each direction. This means that the cables have to be untwisted every now and then. This is ideally done when the wind speed is below cut- in wind speed and thus the turbine is already idling to avoid unnecessary energy yield losses and loads. Some main functional requirements for yaw systems are discussed below. Further requirements can result from the respective yaw system concepts. These cannot be discussed in detail here. Although an undersizing of the yaw system is allowed to a limited extent (see sections 4.2.6 and 4.2.7), the yaw system should always provide sufficient holding torque and/or driving torque to avoid undesired and uncontrollable nacelle move- ments since these can lead to severe damages and personal injuries. This is also to be ensured in the event of faults in the yaw system or in the event of grid loss. In case of failure, the yaw system shall always come to a safe stop and a safe position. During the transition from hold mode to drive mode and vice versa, conditions can briefly occur in which the external loads can lead to undesired movements of the nacelle because the holding torque has already been reduced, but not enough driving torque is available. These should be avoided as they can lead to overload situations in the yaw drivetrain (see section 4.2.8). Unlike the pitch system, the yaw system does not necessarily have to be able to turn the nacelle against all external loads. A certain amount of undersizing is permissible since in many cases it is possible to wait briefly until the load situation is more favourable for yawing. The highest loads usually occur at the beginning of the yawing process. However, overload situations can also occur during yawing (see section 4.2.7). Overload events should be detected to protect the yaw system from these high loads as they can lead to severe damages in the yaw drivetrain. The yaw system shall at least provide the yaw control system with the informa- tion on the nacelle position and the yaw speed, so that the controller can compare the actual values with the target values and react accordingly. Depending on the chosen yaw system concept, many signals can help to get a good and reliable overview of the yaw system status and to improve the yaw system behaviour. These signals can be: • • yaw position, zero- degree position (north position), end position (cable loop) • • yaw speed, yaw motor speed • • yaw motor voltage and current • • yaw motor winding temperature, yaw motor brake temperature • • yaw motor brake status (open/closed) • • yaw brake pressure (for hydraulic yaw brakes) • • status information of the hydraulic power unit and the lubrication system
- 158. Yaw system concepts and designs 129 4.1.2 Wind direction and yaw misalignment The wind changes direction continuously. Since the wind turbine rotor does not fol- low the wind direction instantaneously as it veers, the rotor direction usually lags the wind direction by a few degrees. Therefore, the wind turbine rotor usually operates imperfectly aligned with the wind direction, which results in a lower efficiency of the rotor and higher fatigue loads. The wind direction distribution is highly dependent on the location. In addition to the global weather conditions, the local topography plays a decisive role, such as mountains and valleys. A wind rose is used to give a view of how wind speed and direction are distributed at a particular location. Some examples are shown in Figure 4.2. For the sake of clarity, the wind speeds have been omitted. There are locations where the wind blows mainly from one direction. Locations with two, three or more distinct main wind directions can also be found, as well as locations without any distinct main wind directions. It is obvious that the wind direc- tion distribution has a major influence on the operating hours of the yaw system. Locations with strong main wind directions generally have lower operating hours than others. However, the wind direction distribution says nothing directly about the frequency of wind direction changes that can lead to a yawing process. In general, it can be stated that the higher the wind speed, the steadier the wind direction. At low wind speeds, large wind direction changes take place more often than smaller ones. At high wind speeds, it is the other way round. That means that the probability of wind direction changes and thus yawing processes differ from the wind speed probability. Figure 4.2 Exemplary wind direction distributions
- 159. 130 Wind turbine system design To sum up, wind direction distribution and wind direction changes vary greatly between locations. As a result, the number of operating hours and the number of yawing processes also vary greatly. The difference between nacelle direction and wind direction is called yaw misalign- ment. During power production, the yaw misalignment corresponds approximately to a normal distribution. The standard deviation depends on the yaw control system and its criteria for initiating a yawing process. It is usually less than 10° (Figure 4.3). Yaw misalignment results in a lower efficiency of the rotor. Figure 4.4 shows the decrease in the power coefficient CP with increasing yaw misalignment. According to the momentum theory for a wind turbine rotor in steady yaw, the power coeffi- cient CP is given by the following equations, in which γ represents the yaw misalign- ment [2]: CP = 4a(cos a)2 (4.1) CPmax = 16 27 cos3 at a = cos 3 (4.2) • power coefficient [–] • max. power coefficient [–] • yaw misalignment [°] CP CPmax This cos3 γ rule is commonly used for evaluating the influence of yaw misalign- ment on the energy yield. Figure 4.3 Exemplary yaw misalignment distributions
- 160. Yaw system concepts and designs 131 If the rotor is imperfectly aligned with the wind direction, even in a steady wind, the angle of attack on each blade is continuously changing as it rotates. As a result, the loads on the rotor blades are fluctuating, causing higher fatigue loads for the blades and thus for the wind turbine. For that reason, yaw misalignment needs to be considered in the load simulation. In the partial load range of the power curve, yaw misalignment leads to yield losses and higher fatigue loads. In the full load range, however, the electrical power of the wind turbine is regulated by adjusting the pitch angles of the rotor blades. The pitching of the rotor blades only begins after the nominal power has been reached. As a result, the poorer efficiency of the rotor has no influence on the energy yield. However, the wind turbine is still subject to higher fatigue loads. 4.1.3 Typical key data In the following, some typical key data of active and friction- damped yaw systems are discussed. Since these depend heavily on the location, the yaw system concept and the yaw control strategy of the respective wind turbine, only a general overview can be given. The number of operating hours depends heavily on the wind direction distribu- tion, the wind direction changes (see section 4.1.2), the yaw speed and the control system. In its standard [1], the DNV requires at least 10% of the turbine’s service life for the number of operating hours, which corresponds to an average of 876 hours per year. This seems to be a conservative approach, because at many sites the operating hours are well below 5% of the turbine life. For the design of yaw systems, Figure 4.4 Power coefficient variation with yaw misalignment
- 161. 132 Wind turbine system design however, a design value must be specified that covers a wide range of locations. This should be done considering the factors mentioned above. The number of yaw activations is closely linked to the operating hours. Yawing processes can be carried out on average up to 100,000 times per year. This corre- sponds to a yaw activation approximately every 5.25 minutes. The yaw angles covered during a yawing process depend on the control system and the mean yaw misalignment. During power production, the angles of rotation are only a few degrees, e.g., 3°, 5° or 8°, depending on the respective criterion which initiates the yawing process. When the wind turbine is idling, larger yaw misalign- ments are usually allowed, e.g., 20°. In the event of rapid and large changes in the wind direction of, e.g., greater than 30°, the nacelle yaws immediately and covers correspondingly large yaw angles. During cable untwisting, the yaw angle can be up to three revolutions depending on the installed cable loop. The yaw speed during power production is usually less than 0.6°/s to avoid large gyroscopic forces. The larger the rotor diameter, the lower the yaw speed tends to be. When the wind turbine is idling, higher yaw speeds can be allowed if the yaw system is designed in that way. The duration of a yawing process can be derived from the yaw speed and the yaw angle to be covered. For example, at a yaw speed of 0.5°/s, 5° is completed in 10 seconds, 20° in 40 seconds and 360° in 720 seconds. If the yaw speed is halved, the durations double. As a result, the operating hours also increase as the yaw speed becomes slower. The total gear ratio of the yaw system depends on the intended yaw speed and the speed of the desired yaw motors. It usually ranges between 10,000 and 30,000. The gear ratio of the yaw gearboxes can be between 600 and 3,000. The number of yaw drives varies depending on the available installation space and the yaw system concept. The installation space is mainly driven by the nacelle and pow- ertrain layout. For example, direct drive wind turbines usually offer more installation space on the yaw bearing circumference than geared wind turbines, in which there is only space to the left and right of the powertrain. Due to a larger production volume at the supplier, many small yaw drives can also be the more economical solution compared to a few large yaw drives. Yaw systems with three to 16 yaw drives are currently common. The yaw moment of inertia of all tower head components (rotor and nacelle) of multimegawatt upwind turbines is higher than 1.0E+07 kgm². The larger the wind turbine, the higher the moment of inertia. In comparison, the moment of iner- tia of asynchronous motors of usual size in yaw drives is in the range between 4.0E−03 and 5.0E−02 kgm². Due to the high yaw ratio and the use of multiple yaw drives, however, the moment of inertia of the yaw system in relation to the tower head is usually greater than 3.0E+06 kgm² and consequently of a similar order of magnitude as the yaw moment of inertia. The ratio of the two moments of inertia influences the yaw system behaviour and the torques that can occur in the yaw drivetrain. Every gear stage requires a minimum circumferential backlash for proper opera- tion. The maximum backlash of the yaw gearbox can be assumed to be smaller than 0.5° at the output pinion. Converted to the motor side, this can result in up to four revolutions. The backlash between yaw gearbox output pinion and yaw bearing
- 162. Yaw system concepts and designs 133 teeth is added to this. The minimum normal backlash is usually 0.03–0.04 times the gear module and needs to be converted to the circumferential backlash. The efficiency of the yaw gearboxes, which usually consists of four to five planetary stages, can be assumed to be 97.0–97.5% per gear stage. The efficiency between output pinion and yaw bearing teeth lies in a similar range. The overall efficiency of the yaw system is thus roughly between 83% and 88%. 4.2 Design loads Basic knowledge of the loads acting on the yaw system is important for understand- ing the different yaw system concepts, the component designs and the dimensioning of yaw systems. For this reason, this section deals with the design loads. After a short introduction to the consideration of yawing processes and yaw misalignments in the load simulation, the loads acting on the yaw bearing and the yaw drivetrain are described. Then the focus is on the loads acting on the yaw drive- train and how the aerodynamic loads are modified to consider all loads in the design of the yaw drivetrain. Finally, some special load situations that can occur during yawing processes are discussed. The Fraunhofer IWES wind turbine IWT- 7.5- 164 serves as common thread for this book. The loads for this turbine were simulated between 2013 and 2017. The focus was not on the yaw system. As a result, the loads on the yaw system are quite high compared to today. Considering today’s possibilities for turbine load reduction and optimising the design- driving load cases, the yaw system loads would be significantly lower. For this reason, the authors of this chapter have decided to use the yaw system loads of a comparable 7 MW wind turbine with a rotor diameter of about 170 m. 4.2.1 Introduction For wind turbine certification, system and component loads are to be simulated accord- ing to IEC 61400 [3] or DNV standard [4]. The standards specify the minimum require- ments for the design load cases (DLCs) for the assessment of ultimate and fatigue strength of the wind turbine. Further information can be found in Chapters 1 and 2. Yawing processes are usually not part of the DLCs. Only a few DLCs may require yawing, depending on the definition of the DLCs by the wind turbine original equipment manufacturers (OEMs). Yawing processes are therefore often modelled in a simplified manner to simply meet the certification requirements and to reduce the turbine loads in the simulation, e.g., by yawing with constant speed. In addition, the yaw system is often modelled rudimentary in current simulation software. As a result, the yaw system loads are often given for the non- yawing wind tur- bine. For the dimensioning of the yaw system and its components, the loads must therefore be modified to consider yawing processes (see section 4.2.5). The DLCs are simulated for different mean yaw misalignments, depending on the yaw control system and possible failure load cases. In the fatigue load post- processing, the time series are weighted according to their frequencies.
- 163. 134 Wind turbine system design Appropriate assumptions are to be made for yaw misalignment. If values for the wind turbine cannot be specified, yaw misalignment of −8°, 0° and +8° evenly distributed in ±8° shall be applied in DLC 1.2 according to the DNV standard [4] (Table 4.1). Other yaw misalignments, e.g., −5°, 0° and +5°, and other percentages of time, e.g., 25%, 50% and 25%, can be found for DLC 1.2. A higher number of values for the yaw misalignment and the percentage of time would represent the yaw misalignment distribution (Figure 4.3) better. However, the number of values is generally kept small to keep the number of required load simulations and thus the simulation time low. Nevertheless, an attempt should be made to map the behaviour of the yaw control system with its different criteria for adapting the frequency of the yawing process to the wind turbine operating status. The yaw system loads are usually given for the yaw bearing coordinate system (Figure 4.5). These loads include: • • aerodynamic loads • • mass and inertia loads • • gyroscopic loads • • reaction forces, e.g., of the powertrain In Figure 4.5, the yaw bearing loads are drawn in directions according to a con- ventional coordinate system and not in the direction they usually have: XK • horizontal in direction of the rotor axis • ZK vertically upwards • YK horizontally sideways, so that XK, YK, ZK rotate clockwise • FXK radial force Fx • FYK radial force Fy • FZK axial force Fz • MXK roll moment Mx • MYK tilt moment My • MZK yaw moment Mz In the following subsections, the yaw system loads for the non- yawing wind turbine are discussed. Table 4.1 Yaw misalignment in DLC 1.2 [4] Yaw misalignment Percentage of time −8° 33.3% 0° 33.3% +8° 33.3%
- 164. Yaw system concepts and designs 135 4.2.2 Yaw bearing loads Yaw bearings are usually dimensioned by ultimate loads. Above all, failures in the pitch system that cause high aerodynamic imbalances, e.g., if a rotor blade is stuck, or large yaw misalignments at high wind speeds, can lead to very high loads on the yaw bearing. Which DLCs ultimately lead to the highest loads depends on several factors, such as the wind turbine design and the control and safety architecture. Table 4.2 shows an example of the yaw bearing ultimate loads of a wind turbine of the 7 MW class based on a statistical analysis of all DLCs for ultimate assess- ment. For each load component, the loads are listed at the point of time at which the respective load component has its maximum or minimum value. It should be noted that the values in the extreme load table are mean values of the maximum or mini- mum values from a set of time series of a certain DLC. The magnitude of the forces is up to about 5,000 kN. The moments reach values up to about 24,000 kNm. The bending moments Mx and My are particularly relevant for the design of the yaw bearing as they lead to high contact stresses in the yaw bearing. As the entire nacelle rests on the yaw bearing, it is also subject to high axial loads. The radial forces play a rather subordinate role. Figure 4.6 exemplifies the course of the resulting bending moment Mxy in DLC 1.3 (power production with extreme turbulence model) at a wind speed of 19 m/s. Figure 4.5 Yaw bearing coordinate system [4]
- 165. Table 4.2 Yaw bearing ultimate loads Mx My Mxy Mz Fx Fy Fxy Fz γF Load case [kNm] [kNm] [kNm] [kNm] [kN] [kN] [kN] [kN] – – Mx Max 16,100 10,030 19,580 9,650 890 –230 920 –4,730 1.35 dlc1p3 Min –6,890 –2,250 7,340 –3,720 580 150 600 –3,770 1.10 dlc2p2g My Max 10,730 24,110 26,510 –7,350 410 –240 490 –4,490 1.35 dlc2p1a Min 8,180 –20,440 22,030 4,070 260 100 290 –3,830 1.10 dlc2p2b Mxy Max 12,260 23,940 26,940 –3,920 730 –170 760 –4,550 1.35 dlc1p3 Min 10 –10 20 –2,280 60 –60 90 –3,060 1.10 dlc7p1b Mz Max 10,880 4,040 11,920 19,190 350 –190 400 –3,800 1.10 dlc2p2b Min 7,380 –7,200 10,450 –20,120 230 190 320 –4,360 1.35 dlc1p4 Fx Max 12,520 5,380 13,650 –990 2,050 –20 2,050 –4,850 1.35 dlc1p4 Min –1,390 –4,430 4,640 –2,650 –1,380 20 1,380 –3,640 1.10 dlc2p3 Fy Max –2,530 1,140 2,900 –6,350 –150 1,080 1,090 –3,590 1.10 dlc6p2 Min 3,390 –3,580 5,180 7,410 –160 –1,210 1,220 –3,440 1.10 dlc6p2 Fxy Max 12,520 5,380 13,650 –990 2,050 –20 2,050 –4,850 1.35 dlc1p4 Min –60 –3,000 3,000 140 10 –10 10 –3,430 1.10 dlc2p2b Fz Max 1,260 5,010 5,360 2,690 110 –410 420 –2,860 1.10 dlc6p2 Min 12,810 5,370 14,230 7,060 800 –140 810 –5,050 1.35 dlc1p3
- 166. Yaw system concepts and designs 137 The resulting bending moment fluctuates a lot, and the maximum value is only pres- ent for a very short time. If a blade axis gets stuck at a certain pitch angle and as a result an emergency stop of the wind turbine is initiated, higher amplitudes can occur due to the aerodynamic imbal- ance of the rotor (Figure 4.7). Figure 4.6 Yaw bearing resulting bending moment Mxy (DLC 1.3, 19 m/s) Figure 4.7 Yaw bearing resulting bending moment Mxy (DLC 2.2b, 20 m/s)
- 167. 138 Wind turbine system design The stochastic and turbulent wind field and the resulting highly fluctuating rotor loads result in highly fluctuating roll and tilt moments for the yaw bearing (Figures 4.8 and 4.9). This dynamic load characteristic must be considered in the fatigue strength verification of the yaw bearing. Figure 4.8 Yaw bearing roll moment Mx (DLC 1.2, 13 m/s) Figure 4.9 Yaw bearing tilt moment My (DLC 1.2, 13 m/s)
- 168. Yaw system concepts and designs 139 For that reason, the fatigue strength verification of the yaw bearing is usually carried out using the time series. However, to get a feeling for the level of the operat- ing loads and their time proportions, Figure 4.10 shows the load duration distribution (LDD) of the resulting bending moment. The resulting bending moment reaches values close to the ultimate loads (with- out partial safety factor for loads), but the durations are quite small. These are essen- tially short load peaks, which together can make up a few hundred hours in 20 years. Over 90% of the time, the load level is in the lower half of the load range. To sum up, the yaw bearing loads are highly dynamic loads whose effects on fatigue and wear are to be considered in the yaw bearing design. Failure load cases can briefly lead to very high loads that dimension the yaw bearing. 4.2.3 Yaw drivetrain aerodynamic loads The yaw moment Mz is relevant for the dimensioning of the yaw drivetrain consist- ing of yaw bearing teeth, yaw drives and yaw brake system. Sections 4.2.4 and 4.2.5 explain which other loads have to be considered when designing the yaw drivetrain. In the following, the focus is solely on the aerodynamic yaw moment Mz . The ultimate loads for the yaw moment Mz can be found in Table 4.2. High aerodynamic imbalances or large yaw misalignments at high wind speeds can lead to very high torques. The magnitude is up to about 27,000 kNm. Since the ultimate load cases are relevant for the design of the holding system of the yaw system and for the yaw slippage events (see section 4.2.6), two load cases are discussed in more detail below. Figure 4.10 Yaw bearing resulting bending moment Mxy (LDD)
- 169. 140 Wind turbine system design Figure 4.11 illustrates the course of the yaw moment Mz when a strong wind gust with a simultaneous strong change in wind direction hits the wind turbine (DLC 1.4). The wind speed suddenly increases from 12.8 to 27.8 m/s within ten seconds. In the same period, the yaw misalignment increases to almost 60°. When the cut- off wind speed of 20 m/s is reached, a turbine shut- down is initiated. As a result, the yaw moment rises sharply and fluctuates greatly. Figure 4.12 shows the yaw moment Mz when a blade axis gets stuck and as a result a wind turbine shutdown is initiated. Due to the high aerodynamic imbalance, the yaw moment fluctuates with very large amplitudes around the zero point. The stochastic and turbulent wind field, the resulting highly fluctuating rotor loads, the rotation of the rotor and the tower shadow effect result in a yaw moment Mz , which strongly fluctuates around the zero point (Figure 4.13). This means that the aerodynamic torque alternately supports or counteracts the yaw movement caus- ing high dynamic loads in the yaw drivetrain. A Fourier analysis of the yaw moment Mz usually shows two characteristic frequencies at 1P (rotor rotational frequency) and 3P (blade passing frequency) caused by the influencing factors mentioned above. These frequencies can also be seen in the yaw drivetrain and must therefore be considered in the design of the yaw system. The fatigue strength verification of the yaw drivetrain is usually done using a LDD of the yaw moment Mz (Figure 4.14). As with the resulting bending moment Mxy , the yaw moment reaches values close to the ultimate loads (without partial Figure 4.11 Yaw moment Mz (DLC 1.4, 12 m/s)
- 170. Yaw system concepts and designs 141 Figure 4.12 Yaw moment Mz (DLC 2.2b, 20 m/s) Figure 4.13 Yaw moment Mz (DLC 1.2, 13 m/s) safety factor for loads), but the durations are quite small. These are essentially short load peaks, which together can make up a few hours in 20 years. Figure 4.15 shows the same LDD with the absolute values. Over 99% of the time, the load level is in the lower half of the load range.
- 171. 142 Wind turbine system design Figure 4.14 Yaw moment Mz (LDD) Figure 4.15 Yaw moment Mz (LDD, absolute values)
- 172. Yaw system concepts and designs 143 To sum up, the yaw moment Mz is a highly dynamic load fluctuating around the zero point. In failure load cases, very large amplitudes can occur. The following points for the yaw system design can be derived from this: • • During yawing, the alternating aerodynamic torque means that the yaw motors work alternately as a motor and a generator. In addition, the gears are subject to these alternating loads. As a result, the gear backlash is run through con- tinuously, which leads to load peaks and wear. These negative effects can be reduced by generating an additional torque that always counteracts the yaw movement, such as a friction torque. • • The nacelle of an upwind turbine must be locked so that it maintains its current position during normal power production. Since very high torques can occur in the event of a failure, the nacelle should be frictionally locked. This allows the nacelle to slip, which is an overload protection. • • The nacelle of a downwind turbine should also be locked when no yawing is required. At least there should be a high level of inertia or damping in the yaw system to prevent the nacelle from following every change in wind direction and excessive yaw speeds from occurring. • • The large amplitudes in some failure load cases can lead to critical situations and severe damages in the yaw drivetrain. Yawing should be avoided in these situations. Critical situations and overloads should be detected by the yaw con- trol system to protect the yaw drivetrain. 4.2.4 Loads acting on the yaw drivetrain Several torques act on the yaw drivetrain, which can be divided into the following three categories: • • externally applied yaw torque ○ ○ aerodynamic torque (section 4.2.3) ○ ○ inertia torque • • yaw friction torque ○ ○ yaw bearing friction ○ ○ yaw brake friction • • yaw actuator torque ○ ○ yaw motor torque ○ ○ yaw drive restoring torque (in case of pretensioned yaw drives) The externally applied yaw torque contains all torques that act on the yaw sys- tem from outside. This is not only the torque generated by the aerodynamic loads acting on the wind turbine but also the inertia torque that arises when accelerating and decelerating the tower head.
- 173. 144 Wind turbine system design The yaw friction torque includes those internal torques that cannot cause the nacelle to yaw and that always counteract the yaw movement. This is mainly the friction torque of the yaw bearing and the yaw brakes. The yaw actuator torque includes the torques generated by the yaw motors, both motor and generator torque. A speciality is the restoring torque of the yaw drives. When the torsion spring of the yaw drivetrain is tensioned, it generates a corre- sponding restoring torque. This state can arise, e.g., when the nacelle rotates a little while the yaw motor brakes are already closed. The balance of these three torques is always zero. In the following, the torque distribution at the tower head during different yaw operations is discussed in more detail, using the example of an active and friction- damped yaw system with electro- mechanical drives. The induction motors are driven directly from the onboard power system. To show whether the nacelle is being accelerated or decelerated, the aerody- namic torque and the inertia torque are shown separately. Figure 4.16 shows the torque distribution at the tower head during non- yawing oper- ation. The yaw motor brakes are closed, the yaw motors are switched off and the yaw drives are not loaded (pretensioned). The following three states can be distinguished: • • Left bar: The aerodynamic torque is less than the yaw friction torque. The nacelle remains in its position and the yaw drives are still not loaded. • • Middle bar: The aerodynamic torque is greater than the yaw friction torque. Therefore, the nacelle starts to slip. Consequently, the yaw drives and the yaw motor brakes are loaded and build up a restoring torque. However, the yaw motor brakes do not slip. • • Right bar: The aerodynamic torque is greater than the yaw friction torque and the maximum restoring torque. As a result, the yaw motor brakes and the nacelle slip (see section 4.2.6). The torque distribution during clockwise yawing (nacelle top view) is shown in Figure 4.17. As a result, the yaw friction always acts anticlockwise. Depending on whether the aerodynamic torque minus the yaw friction torque supports the yaw movement or counteracts it, the yaw motors are operated as a motor or a generator. In principle, the following four operating states can occur, between which there is a continuous change due to the alternating aerodynamic torque: • • Far left bar: The aerodynamic torque counteracts the desired yaw movement. The yaw motor works as a motor. At the time shown, the yaw actuator torque is greater than the yaw friction torque and the aerodynamic torque. Therefore, the nacelle is accelerated. • • Middle left bar: The aerodynamic torque counteracts the desired yaw move- ment. The yaw motor works as a motor. At the time shown, the yaw actua- tor torque is less than the yaw friction torque and the aerodynamic torque. Therefore, the nacelle is decelerated.
- 174. Yaw system concepts and designs 145 Figure 4.16 Torque distribution at tower head during non- yawing operation Figure 4.17 Torque distribution at tower head during yawing operation
- 175. 146 Wind turbine system design • • Middle right bar: The aerodynamic torque supports the desired yaw movement. The yaw motor works as a generator. At the time shown, the aerodynamic torque is greater than the yaw friction torque and the yaw actuator torque. Therefore, the nacelle is accelerated. • • Far right bar: The aerodynamic torque supports the desired yaw movement. The yaw motor works as a generator. At the time shown, the aerodynamic torque is less than the yaw friction torque and the yaw actuator torque. Therefore, the nacelle is decelerated. The following special cases should be noted: • • If the aerodynamic torque is smaller than the yaw friction torque, the yaw motor always works as a motor since the resulting torque counteracts the yaw movement. • • If the sum of aerodynamic torque and yaw friction torque exceeds the maximum possible yaw actuator torque in motor operation, the yaw motors stall. As a result, the yaw process comes to a stop and high torques can occur in the yaw drivetrain (see section 4.2.7). • • If the nacelle drives the yaw motors, they work as a generator. The yaw motors then feed power into the onboard power system. • • If the supporting aerodynamic torque minus the yaw friction torque exceeds the maximum possible yaw actuator torque in generator operation, the yaw motors slip (see section 4.2.7). As a result, very high yaw speeds can occur. With individual pitch control (IPC) it is theoretically possible to generate yaw- ing loads in response to a measured yaw misalignment, to keep the rotor aligned with the wind direction without the use of a yaw motor. However, it is unlikely that yaw drives can be omitted entirely. They will still be needed to yaw the nacelle while the rotor is not turning and to untwist the power cables. The missing holding torque of the yaw drives would also have to be provided in a different way. In addition, the torque of the yaw motors could be required to control the yawing process. For these reasons, IPC is mainly used to minimise the external yawing loads that the yaw motors have to overcome. 4.2.5 Modification of yaw drivetrain aerodynamic loads The yaw drivetrain fatigue loads are usually given for the single non- yawing wind turbine and thus include only aerodynamic loads, which are not influenced by other wind turbines. Depending on the yaw system concept, the loads have to be modified in order to take all loads acting on the yaw drivetrain during yawing (see section 4.2.4) into account. The starting point can be either the LDD of the yaw moment Mz or the yaw bear- ing load time series, both given for the entire wind turbine service life. The results of the modifications are two LDDs, one for the yaw bearing teeth, and one for the yaw
- 176. Yaw system concepts and designs 147 gearbox. Since the modifications can be very different, the procedure is described in general below. First, the yaw friction torque has to be added to the yaw moment Mz . This should be done considering the desired direction of rotation, as it always counteracts the yaw movement. The yaw bearing friction torque can either be calculated with the time series for each point in time or for an average yaw bearing load. The yaw brake friction can be calculated with the data of the yaw brake system as a function of the braking pressure. Second, the torque has to be divided by the number of yaw drives in order to obtain the torque for the individual tooth engagement. To get the torque on the out- put pinion of the yaw gearbox, it then has to be divided by the ratio between yaw bearing teeth and output pinion. The gear efficiency should be considered. Finally, the durations per load bin have to be scaled down to the operating hours of the yaw system. This can be done simply with the percentage of the operating hours in the service life of the wind turbine. This assumes that the probability of yawing processes is almost identical to the wind speed distribution. However, sec- tion 4.1.2 has shown that this is not the case, especially when the settings of the yaw control system depend on the wind speed. The probability of yawing operations should therefore be considered accordingly when scaling down the durations. In particular during yaw start and stop events, but also during yawing, inertia torques occur that are not so easy to consider. For example, it does not make sense to add a constant inertia torque to the loads and to use these loads for the fatigue strength verification of the components, since the inertia torque is mainly present at the beginning of the yaw start. However, the inertia torque should be considered when determining the required driving torque of the yaw system. The wind turbine loads are usually given for a single wind turbine. If the wind turbine is in a wind farm, however, higher loads can occur under certain conditions. For example, if the wind turbine is in the wake of another wind turbine and the rotor is partially shaded as a result, higher yaw moments Mz can occur. These additional loads should also be considered when determining the required driving torque. If the nacelle is held in position solely by the yaw motor brakes or by the yaw motors themselves, the yaw drives are pretensioned and loaded during non- yawing operation. Consequently, fatigue loads have to be determined for this operating state. The procedure is similar to that for the loads for yawing operation. The ultimate torque should not be determined based solely on the ultimate loads from the load simulation (see Table 4.2) since yawing is not considered. Instead, possible overload situations (see following subsections) and failure cases should be investigated, considering the tolerances of the component and system parameters. How safely and quickly the yaw control system detects and reacts to overload situa- tions also influences the level of the ultimate torque. The number and arrangement of the yaw drives have a major influence on the load cycles on the yaw bearing teeth. Since the nacelle does not rotate continuously in one direction and there are distinct main wind directions, the load cycles are not evenly distributed over the circumference of the yaw bearing. By way of example,
- 177. 148 Wind turbine system design Figure 4.18 illustrates this for a yaw system with eight yaw drives and a wind direc- tion distribution assumed as normal distribution. If the yaw drives are evenly distributed over the circumference, the load cycles are distributed most evenly. However, the installation space available in the nacelle often does not allow this kind of arrangement. The more and the closer the yaw drives are arranged together, the higher the load cycle concentration. This needs to be considered in the verification of the yaw bearing teeth. To conclude, the loads from the load simulation need to be modified to reflect all loads acting on the yaw drivetrain. However, this is not fully possible. Residual uncertainties or inaccuracies remain, which should be considered in the design and verification of the components. 4.2.6 Yaw slippage events during non-yawing operation To reduce costs, the yaw system is often undersized. That means that the hold- ing capacity of the yaw system is not designed to withstand all ultimate design loads (yaw moment Mz ) and a slippage of the nacelle is tolerated up to a certain extent. In the following, the yaw slippage event during non- yawing operation is described for an active and friction- damped yaw system. • • As long as the aerodynamic torque is less than the yaw friction torque, the nacelle remains in its position. Figure 4.18 Load cycle distribution over the yaw bearing circumference
- 178. Yaw system concepts and designs 149 • • As soon as the yaw friction torque is exceeded, the nacelle starts to slip. The rotational inertia of the tower head and the yaw friction torque counteract the yaw movement. First, the torsional backlash in the yaw system is run through. Then the yaw drives are loaded. Since the motor shaft is held in position by the yaw motor brake, the yaw drivetrain is pretensioned and builds up a restoring torque that counteracts the yaw movement. • • If the torque on the yaw motor shaft exceeds the yaw motor brake torque, the yaw motor brake starts to slip. The rotational inertia of the tower head, the yaw friction torque, the braking torque of the yaw motor brakes and the rotational inertia of the yaw motors now counteract the movement. • • If the nacelle comes to a standstill, the yaw drivetrain remains pretensioned. The amount and direction of the pretensioning have a major influence on the next yaw slippage event. • • Several yaw slippage events can also take place in quick succession, e.g., in the failure load cases described in section 4.2.3. The nacelle can slip in either the same or the opposite direction as the previous yaw slippage event. The following criteria can be used to evaluate the yaw slippage event: • • friction energy of the yaw motor brake • • limit speeds of the yaw motor and the yaw motor brake • • torques on the yaw bearing teeth and the yaw gearbox The yaw motor brake is usually a holding brake. That means that the brake only engages when the nacelle has almost come to a complete standstill. If the nacelle slips a few tenths of a degree, the motor brake slips several revolutions due to the high gear ratio of the yaw system. The resulting friction energy is dissipated into heat. If the permissible friction energy of the brake is exceeded, the yaw motor brake can be damaged. Irrespective of this, friction causes wear, which could result in a replacement of the yaw motor brake. Due to the high gear ratio, high motor speeds are also reached very quickly during a yaw slippage event. The limit speeds of the components should not be exceeded, as this could lead to damages. It is often assumed that during yaw slippage the mechanical components are loaded with a constant braking moment. Even if the maximum torque during yaw slippage is carefully determined considering the tolerance of the yaw motor brake torque and additional safety factors, this approach inherits many uncertainties since information about the loads acting on the yaw drivetrain and the dynamics of the wind turbine is not really known. If the maximum permissible torque of the compo- nents is exceeded, they can break. For this reason, it is recommended to simulate yaw slippage events, also to be able to evaluate the effects on the wind turbine loads. It should always be noted that the yaw slippage event and the loads that occur depend on several factors. Therefore, appropriate safety factors should be considered when determining the maximum torque based on simulation results.
- 179. 150 Wind turbine system design 4.2.7 Overload events during yawing operation Not only the holding capacity but also the driving capacity of the yaw system is often undersized to reduce costs. That means that the yaw system is not designed to be able to turn the nacelle against all external loads. The main reason for this is that the yaw moment Mz has a very large range, but over 99% of the time the yaw moment is in the lower half of this load range (see Figure 4.15). It is therefore not economical to design the driving capacity for the highest torque that may occur. The yaw system can simply wait a few seconds for more favourable loads to yaw the nacelle. However, undersizing can lead to overload events during yawing, which are discussed in more detail below for an active and friction- damped yaw system with electro- mechanical yaw drives, which are operated directly on the onboard power system. The highest loads usually occur at the beginning of the yawing process. If the load torque on the motor shaft exceeds the starting torque of the yaw motor, the yaw system is not able to turn the nacelle (Figure 4.19). The yaw speed remains zero and the resulting high starting current causes the yaw motor to heat up quickly. If the torque on the motor shaft exceeds the breakdown torque of the yaw motor during yawing, the yaw motor stalls (Figure 4.19). As a result, the yawing process comes to a standstill. The operating state is then the same as when the yaw system starts to yaw with overload. Both overload events lead to high mechanical and thermal loads on the yaw sys- tem components. For that reason, these operating states should be detected quickly by the yaw control system, and the yawing operation should be stopped. Figure 4.19 Yaw motor torque, speed and current (motor mode)
- 180. Yaw system concepts and designs 151 The aerodynamic torque can also support the yaw movement. In failure cases with a high aerodynamic imbalance, the supporting aerodynamic torque minus the yaw friction torque can exceed the breakdown torque of the yaw motor (Figure 4.20). As a result, the motor slips. Since the motor torque then decreases significantly, very high yaw motor speeds and thus high yaw speeds can be achieved. Critical operating states in the generator mode should be avoided. It should be noted that the breakdown torque in generator mode is greater than in motor mode. In addition, the power flow is reversed, so that higher torques can occur in the yaw drivetrain. The yaw control system should detect these events and stop the yawing process. In general, it should be analysed whether a yawing process should still be started or continued during a turbine shutdown or an emergency shutdown. The high torques with large amplitudes that occur in these cases can lead to severe damages of the yaw system. If yawing is not necessary in these events, it should be avoided to protect the yaw drivetrain. The yaw motors are often operated directly on the onboard power system with- out frequency converters. According to the torque- speed curve of the induction motor (Figure 4.19), the motor speed depends on the torque. Due to the fluctuating aerodynamic torque, the yaw speed changes constantly. As a result, the nacelle is continuously accelerated and decelerated, which leads to additional loads for the mechanical and electrical components. In yaw systems with converter- fed yaw drives, the yaw speed can be controlled, and yaw speed fluctuations and inertia torques can be minimised. Figure 4.20 Yaw motor torque and speed (generator mode)
- 181. 152 Wind turbine system design To sum up, the yaw system is often intentionally undersized. The associated overload events should be detected quickly by the yaw control system to protect the yaw drivetrain from these overloads and to prevent severe damages. To be able to detect overload events, appropriate sensors are to be provided in the yaw system. 4.2.8 Yaw start and stop events This section deals with the transition from the non- yawing to the yawing opera- tion and vice versa. The focus is on active and friction- damped yaw systems with hydraulic yaw brakes and electro- mechanical drives, which are operated directly on the onboard power system. During the yaw start procedure, the holding torque is reduced, and the driving torque is increased (Figure 4.21). First, a reduction of the hydraulic pressure of the yaw brakes is initiated. Second, the yaw motor brakes are released. Finally, the yaw motors are switched on. Depending on the loads and on how the yaw start procedure is designed, it can happen that the nacelle is set in motion by the wind loads before the yaw motors are switched on. When the direction of rotation of the nacelle corresponds to the desired yawing direction, the yaw motors start in motor or generator mode depending on the yaw friction torque. However, if the nacelle is moving in the opposite direction, the yaw motors are initially used as brakes because the rotating field and the motor shaft have differ- ent directions of rotation. High torque and current peaks occur at the beginning of the counter- current braking, which are significantly higher compared to the normal Figure 4.21 Holding and driving capacity during yaw start and yaw stop
- 182. Yaw system concepts and designs 153 switch on process. After the braking process, the nacelle is then turned in the desired direction if the loads allow it. During the stop procedure, the holding torque is increased, and the driving torque is reduced (Figure 4.21). For example, first, the yaw motors are switched off. Second, the increase of the hydraulic pressure is initiated. Finally, the yaw motor brakes are closed when the nacelle has almost come to a complete standstill. This can also lead to unintentional and uncontrolled movements of the nacelle. This means that the desired nacelle position is not met and that the yaw motor brakes could also be used to brake the nacelle. In addition, the yaw drivetrain is preten- sioned after the stop procedure. The yaw drivetrain has torsional backlash that can be several revolutions on the motor side. Therefore, the yaw motor can initially start up without load and then drives into the load. This leads to load peaks, which can have a negative effect on fatigue and wear. Converter- fed yaw motors have advantages in the start/stop procedure. While motors operated directly on the onboard power system can only be switched on and off, the torque can be regulated with converter- fed drives. As a result, the start/stop proce- dure can be designed in that way that unintentional and uncontrolled yaw movements can be safely avoided. A soft start can also be implemented with frequency converters. To conclude, during yaw start and yaw stop undesired movements of the nacelle can occur when the procedures are not designed optimally. Since these movements lead to additional loads and load peaks, they should be detected and avoided. 4.3 System concepts and components This section gives an overview of the most common yaw systems available on the market. First, the differences on the system and component level are described. In addition, the auxiliary systems for the yaw system are briefly discussed. Second, a list of the usual selection and evaluation criteria is presented, with the help of which an evaluation matrix for the selection of the yaw system concept can be created. Finally, three of the most common yaw system concepts are briefly explained. 4.3.1 Differentiating features at system level At the system level, the yaw system concepts can be differentiated into active and passive yaw systems, which relates to the yaw actuator torque. A further distinguishing criterion is how the yaw friction torque required to dampen the aerodynamic loads and the yaw motion is provided. In addition, the yaw system concepts can be differentiated according to whether or not they allow limited damped motion during non- yawing operation (soft and stiff yaw system). These differentiating features are described below. The differences in the component design are described in the following subsections. The most common type of wind turbine is the three- bladed upwind horizontal- axis wind turbine with the rotor facing the wind. In the multimegawatt range, this wind turbine configuration requires an active yaw system to align the wind tur- bine rotor with the wind. Since the aerodynamic loads try to turn the rotor into the
- 183. 154 Wind turbine system design downwind position, the nacelle must also be locked so that it maintains its current position (see section 4.2.3). The active yaw system is equipped with torque producing components able to rotate the nacelle against the stationary tower. As described in section 4.1.1, it con- sists of: • • A yaw bearing that rotatably connects the nacelle with the stationary tower and transfers the loads from the nacelle into the tower. • • A set of yaw drives that provide the driving torque to turn the nacelle and the holding torque to hold the nacelle in position. • • A yaw brake system that provides holding torque to lock the nacelle and friction torque to dampen the aerodynamic loads. • • A yaw control system that manages, commands, directs or regulates the behav- iour of the yaw system. There are a few downwind horizontal- axis wind turbines in the multimegawatt range on the market. This wind turbine configuration has the theoretical advantage that it may be built with passive yaw systems since the rotor itself is able to yaw the nacelle into the wind. However, a few active yaw drives are needed to untwist the cables and to turn the nacelle when the rotor is not turning (see section 4.1.1). The passive yaw system utilises the wind force to adjust the wind turbine rotor into the wind. Small upwind turbines up to 100 kW can also be equipped with pas- sive yaw systems consisting of a yaw bearing and a tail fin (or wind vane). The tail fin is designed in such a way that it turns the rotor into the wind by applying a ‘corrective’ torque to the nacelle. However, this low- cost and reliable solution is not able to cope with the high aerodynamic loads in large wind turbines. For this reason, passive yaw systems in multimegawatt wind turbines can only be found in downwind turbines. In large wind turbines, passive yaw systems are basically similar to active yaw systems, as they have to be designed in that way that the nacelle does not follow every change in wind direction and the yaw speed does not become too high to avoid high gyroscopic loads. This is achieved by a high yaw friction torque. In addition to the friction torque of the yaw bearing and the yaw brake system, the yaw drives can provide additional damping torque to control the yaw movement. Downwind offshore turbines mounted on a floating structure may allow to omit the yaw system, as the whole turbine including tower and floating structure are aligned to the wind direction. A rotatable connection between nacelle and tower is then not required. One example is the Nezzy² floating wind turbine, developed by Aerodyn Engineering GmbH and tested by EnBW Energie Baden- Württemberg AG [5]. In the event of a grid loss, downwind turbines with a passive yaw system have the advantage that the rotor aligns itself with the wind direction, whereas turbines with an active yaw system can no longer be yawed. This can lead to very large yaw misalignments and thus high turbine loads that can be dimensioning for certain wind turbine components.
- 184. Yaw system concepts and designs 155 One way of avoiding these high loads is to install an energy storage device that enables the active yaw system to continue to operate for a certain period. Another option would be to bring the rotor into the downwind position and to switch from active to passive yawing. As shown in section 4.2.3, a yaw friction torque is needed to dampen the aero- dynamic loads. It can be provided in the following ways: • • Yaw brake system: In the case of an active yaw brake system, the pressure on the brake pads is typically reduced so that a brake torque is also generated dur- ing yawing. The pressure is increased again after the yawing process is finished. It would also be possible to vary the pressure depending on the load during yawing. In the case of a passive yaw brake system, the brake torque cannot be changed. If the passive yaw brake system is designed to hold the nacelle in posi- tion, the yaw drives have to work against a high friction torque. • • Yaw bearing: A pretensioned sliding bearing has a much higher friction torque than a roller bearing. It normally provides so much friction torque that there is no need for an additional yaw brake system. The sliding bearing can be active or passive. The active one allows to reduce the friction torque during yawing. With the passive sliding bearing, the yaw drives have to work against a high friction torque. There are yaw systems with electro- mechanical yaw drives and a low yaw fric- tion torque (section 4.3.8). The yaw motion is then damped a little by the yaw drives. To eliminate the torsional backlash, the yaw drives are then usually pretensioned. The backlash is still present but is no longer passed through at the same time. During non- yawing operation, the yaw system can be either stiff or soft. While the yaw angle remains constant in stiff yaw systems, limited damped yaw motion is possible in soft yaw systems. Studies have shown that soft yaw concepts can lead to considerable reduction in fatigue and ultimate load of the yaw moment Mz [6]. Other components, such as rotor blade and tower base, can also benefit a little from soft yaw systems. Yaw systems with a yaw brake system are usually stiff during non- yawing oper- ation since the brake system fixes the yaw angle. The yaw system may only slip in extreme load events (see section 4.2.6). In upwind turbines, yaw systems without a yaw brake system, in which the nacelle is held in position by the yaw motor brakes of the pretensioned electro- mechanical yaw drives, can also be considered to be stiff. If the yaw friction torque is exceeded, only very small yaw movements occur along the torsional stiffness curve of the yaw system. Soft yaw systems dampen wind gusts by yielding to wind loads. Depending on the external loads and the soft yaw system design, the yaw angle can change up to five degrees. Passive yaw systems in downwind turbines often include the soft yaw function. In upwind turbines, soft yaw can be achieved by hydraulic yaw systems as presented in Stubkier [6] or by electric yaw systems, in which the nacelle is held in position by the energised yaw motors.
- 185. 156 Wind turbine system design Compared to stiff yaw systems, soft yaw systems, similar to passive yaw sys- tems, lead to a significant increase in yaw movements with small amplitudes. The yaw system components, especially the yaw bearing, must be suitable for this. 4.3.2 Yaw bearing The yaw bearing is a rotatable connection between the nacelle and the tower. It transfers the loads from the rotor and the nacelle (aerodynamic loads and weight loads) into the tower. Depending on the design, it also generates a high friction torque. In principle, rolling bearings and sliding bearings can be used as yaw bear- ing. However, the bearing type has to be suitable for the high alternating bend- ing moments as well as for the axial and radial forces to which the yaw bearing is exposed (see section 4.2.2). The following rolling bearing types can mainly be found in multimegawatt wind turbines (for sliding bearings, see further down in the subsection): • • double- row four- point contact ball bearing • • three- row roller bearing The double- row four- point contact ball bearing (or eight- point contact ball bear- ing) consists of an inner and an outer bearing ring with two raceways between them (Figure 4.22). The balls are inserted into the bearing through a radial cylindrical hole in one of the rings. The hole is then closed using a removeable plug. The plugs are placed in the area of the soft spot of the raceway (see below). Theoretically, one ball has four different contact points to the rings. However, under load only two contact points in a diagonal position transfer the load. If the load changes its direction, the two contact points move to the opposite position. For this reason, this bearing type can accommodate bending moments and axial loads in both directions. Figure 4.22 Double- row four- point contact ball bearing (example) (Liebherr- Components Biberach GmbH)
- 186. Yaw system concepts and designs 157 The nominal contact angle is usually between 40° and 45°. Under load, the contact angle moves towards the edge of the raceway. This can lead to truncation of the contact ellipses. Excessive contact stresses and high stresses in the raceway edge can be the consequence. Due to the contact angle, each load component (bending moment, axial force and radial force) leads to an axial and radial load component in the bearing. This leads to greater deformation of the bearing rings and to a greater widening of the sealing gap compared to the three- row roller bearing. The two raceways can be arranged either directly one above the other or asym- metrically (Figure 4.22). The latter means that the tower- side raceway has a slightly smaller or larger raceway diameter than the upper one (depending on whether the bearing ring attached to the machine carrier is the inner or outer one). This increases the available bearing raceway in the main load direction. This reduces the risk of contact ellipse truncation and thus increases the load capacity of the bearing. The balls are guided with spacers or cages. Spacers allow higher maximum con- tact angles since they require less space in axial direction. This has a positive effect on the load capacity but leads to a higher friction torque. The raceways are equipped with lubrication grooves through which the grease can be distributed in the bearing. The grease inlets and outlets are arranged in this area. The three- row roller bearing consists of an inner and an outer bearing ring with two axial raceways and one radial raceway between them (Figure 4.23). One ring is split into a lower and an upper part. This is necessary for the assembly of the bearing so that the rollers of the axial raceways can be inserted into the bearing. The rollers of the radial raceway are inserted into the bearing through a removeable plug. The nominal contact angles are 90° for the axial raceway and 0° for the radial raceway. Therefore, the axial raceways see pure axial loads (bending moment and axial force) and the radial raceway sees pure radial load (radial forces). This leads to less deformation of the bearing rings and to a less widening of the sealing gap compared to the double- row four- point contact ball bearing. Figure 4.23 Three- row roller bearing (example) (Liebherr- Components Biberach GmbH)
- 187. 158 Wind turbine system design The rollers of the axial raceways are guided with segmented cages. In the radial raceway, segmented cages, spacers, or no guidance system can be found. The seg- mented cages are equipped with lubrication chambers in which the lubricant is col- lected and distributed on the rollers. Due to the alternating loads, the rolling bearings have a defined preload (or pretension). That means that there is no tilting, radial or axial clearance after instal- lation of the bearing. The bearing preload reduces wear and increases the lifetime of the raceways. The bearing rings have through holes or threaded holes that allow the bearing to be bolted directly to the companion structure. One of the bearing rings is geared, depending on how the yaw drives are arranged in the nacelle. The bearing rings are usually made of quenched and tempered 42CrMo4 (mate- rial number 1.7225). 42CrMo4+QT has, among other things, high strength, high toughness and good hardenability. The surface hardness of the induction hardened raceways is usually ≥58 HRC. The surface hardness depth is determined based on the loads. As a rule of thumb, 0.1 times the rolling element diameter can be assumed. The induction hardening process leads to an unhardened area between the beginning and the end of the raceway hardening zone. This unhardened zone is called hardness gap or soft spot. It is important that the soft spots are arranged in the area with the least stress on the bearing raceway. For that reason, the assembly position of the bearing rings in relation to the companion structure must be carefully determined. In addition, the hardness gaps are relief- ground to reduce the contact stresses. The rolling elements are made of rolling bearing steel, such as 100Cr6 (material number 1.3505) or 100CrMnSi6- 4 (material number 1.3520). The surface hardness of the through hardened balls is usually ≥59 HRC. The bearing teeth are also induction hardened. The surface hardness is usually ≥55 HRC. The surface hardness depth is usually determined in accordance with ISO 6336- 5. The tooth width is usually limited to ten times the gear module. In contrast to the yaw gearbox output pinion (see section 4.3.4), the bearing teeth have no pro- file or flank modifications to improve the contact pattern, for reasons of manufactur- ing and costs. Due to the large bearing diameter and the manufacturing process, the gear qual- ity according to ISO 1328 is usually ≥10 (or ≥11 according to DIN 3962). A better gear quality would lead to significant additional costs due to increased manufactur- ing effort and more scrap material. The point of the maximum eccentricity of the gearing is marked for the required adjustment of the backlash (see section 4.3.4). 42CrMo4+QT is susceptible to corrosion. Since corrosion can lead to a strong reduction in the strength properties, the bearing surfaces must be provided with a suitable corrosion protection. The outer surfaces are usually zinced by metal spray- ing according to ISO 2063, which also has the advantage of a higher coefficient of friction between the yaw bearing and the companion structure. The bore and threaded holes are either spray- galvanised or provided with temporary corrosion protection.
- 188. Yaw system concepts and designs 159 The bearing raceways and the bearing teeth are lubricated with grease. In the past, the yaw bearings were manually lubricated every 6–12 months. Nowadays, mostly automatic lubrication systems are used. These have the advantage of less maintenance effort, a longer maintenance interval and improved lubrication condi- tions. The continuous and reliable supply of fresh grease (instead of large amounts once or twice a year) results in a better mixture of fresh and used grease. In addition, condition- based lubrication is possible. The grease is applied to the bearing teeth with several lubrication pinions. The grease for the raceway lubrication is pumped into the bearing via a special distribu- tion system (see section 7.3). Therefore, each raceway has several evenly spaced lubrication holes or grease inlets. These holes are equipped with a thread (usually M10x1) so that the lubrication hoses of the automatic lubrication can be connected to the bearing with angle or swivel fittings. Sometimes the lubrication holes of both raceways are arranged one above the other and sometimes they have a circumferen- tial offset to one another. The lubrication of the yaw bearing is of particular importance. The selection of a suitable grease has a major influence on the wear of the raceways. The grease should therefore be tailored to the operating conditions. In practice, different grease lubricants with different properties are used. The base oil can be mineral, synthetic or both. The base oil viscosity at 40°C is usually between 50 and 420 mm²/s. Commonly used thickeners are metal soaps, such as lithium, calcium, lithium complex and calcium complex. The NLGI class ranges from 1 to 2. The greases contain additives and/or solid lubricants to improve their anti- wear properties. Seals are installed between the inner and outer ring on the top and bottom of the bear- ing. A distinction is made between sealing lip and dust lip. The sealing lip prevents the leakage of grease and the ingress of water and contaminants, whereas the dust lip allows the leakage of excess grease and prevents the ingress of water and contaminants. There are two different sealing concepts. The first concept consists of two seal- ing lips. This means that special radial bore holes (grease outlets) are needed to dis- charge excess grease from the bearing. The grease is collected in special containers. The second concept consists of a sealing lip at the bottom and a dust lip at the top. Excess grease is discharged via the dust lip and collected in a grease pan. In both cases, the sealing lip is usually designed as a double- lip seal. The main advantages and disadvantages of the rolling bearings are summarised in Table 4.3. These essentially result from the different rolling element contact (point contact vs. line contact). In general, rolling bearings are a proven yaw bearing technology with a large supplier base all over the world. The main disadvantage, however, is that if the yaw bearing needs to be replaced, the rotor and the nacelle have to be dismantled. This makes the design and verification of the yaw bearing all the more important. The following sliding bearing types are used in large wind turbines: • • passive sliding bearing • • active sliding bearing
- 189. 160 Wind turbine system design The passive sliding bearing consists of a massive sliding ring with internal or external gearing and with three sliding planes (upper and lower axial plane and radial circular plan) that is bolted to the tower and a set of sliding bearing units that are bolted to the main frame (Figure 4.24). The sliding planes are normally not protected against corrosion and are therefore sensitive to corrosion. Between the top of the geared sliding ring and the main frame, the upper sliding pads are located in special milled pockets on the underside of the main frame or on carrier plates for upper sliding pads, which are mounted to the main frame. These pockets ensure a defined position of the sliding pads. The sliding bearing callipers contain the radial sliding pads as well as the lower axial sliding pads that are arranged below the geared sliding ring. These axial sliding pads are pressed against the lower axial plane of the geared sliding ring by pistons with a disc spring stack. Consequently, the ring gear is constantly clamped between the main frame and the sliding bearing. The preload of the disc spring stacks can be adjusted by means of a lockable screw. The pretensioning system must be tailored to the bending moments that occur and ensure a permanent contact of the sliding pads under all operating conditions. In many wind turbines, the centre of gravity of the tower head (rotor and nacelle) is even outside the tower, which requires a corresponding design. While the lower pretensioning system near the rotor is less loaded, that on the opposite site is higher loaded and therefore may require a stronger design. This must be considered to avoid critical failures. Table 4.3 Advantages and disadvantages of rolling bearings Version Advantages Disadvantages General • Proven yaw bearing technology • Large supplier base • Low friction torque • Component replacement requires disassembly of rotor and nacelle • Additional yaw brake system or inverter- fed yaw motors required Double-row four-point contact ball bearing • Less expensive • Lower risk of wear due to point contact • Lower demand on companion structure and assembly procedure • Lower load capacity due to point contact • Higher and non- reproducible friction torque under load • Higher bearing ring deformation under load Three- row roller bearing • Higher load capacity due to line contact • Lower and reproducible friction torque under load • Lower bearing ring deformation under load • More expensive • Higher risk of wear due to line contact • Higher demands on companion structure and assembly procedure
- 190. Yaw system concepts and designs 161 Figure 4.24 Passive sliding bearing (Federal- Mogul DEVA GmbH, A Tenneco Group Company)
- 191. 162 Wind turbine system design The friction torque of the preloaded sliding bearings is usually significantly higher than that of the rolling bearings. As a result, no additional yaw brake system is needed since the friction torque is sufficient to dampen the yaw motion and hold the nacelle. On the other hand, the yaw drives have to overcome this high friction torque, which leads to a higher required driving torque. The active sliding bearing has a similar setup to the passive one. The difference, however, is that the pistons in the sliding bearing units are not spring- loaded but are hydraulically pressed against the bottom of the yaw ring gear (Figure 4.25). This allows the preload and thus the friction torque of the sliding bearing to be adjusted. As a result, the friction torque can be reduced for yawing and the yaw drives do not have to work against a high friction torque. For the geared sliding ring, the statements made above about the bearing rings of the rolling bearings apply accordingly (material, yaw bearing teeth and their lubrication). The sliding bearing can be lubricated or non- lubricated, depending on the used sliding pad types. The different sliding planes may require different material proper- ties, especially with regard to the coefficient of friction, oil and grease compatibility, shear stress and maximum surface pressure. As mentioned above, the sliding planes are normally not protected against cor- rosion, which would be an argument in favour of using sliding pads in combina- tion with grease. The grease would protect the sensitive surfaces from corrosion but would also reduce friction torque and increase service effort. Various modern non- organic pads made of polyethylene terephthalate (PET), or other resin- based materials are suitable for this kind of application. Some manufacturers use non- organic pads made of polytetrafluoroethylene (PTFE) or sintered bronze- based materials with embedded solid lubricants to provide a perma- nent lubrication layer. No additional grease is needed with these sliding pads. Figure 4.25 Active sliding bearing (Svendborg Brakes ApS)
- 192. Yaw system concepts and designs 163 The non-organic pads usually have a base layer, or a carrier plate made of special fibre material, such as glass fibre reinforced plastic (GFRP or GRP). The sintered bronze pads consist of a bronze structure with evenly distributed solid lubricants through the whole thickness of the pads. The upper axial sliding pads bear the weight of the tower head and are not only dimensioned larger for this reason. Replacing these pads requires more effort compared to the other sliding pads. They should therefore be designed for the longest possible service life. This is reflected in the material properties and in the design. The lower axial sliding pads are usually part of a pretensioning system where axial forces are applied (actively or passively) to keep the required tolerances and friction torque. This system requires some space, which reduces the size of the slid- ing pads, which also affects their requirements for the material properties. The radial sliding pads are usually made of greaseable plastic and bear the radial loads. They must ensure that the nacelle remains radially within the specified tolerances. Impermissible radial displacement of the nacelle must be avoided at all costs, as this can lead to jamming between the yaw gearboxes and the geared sliding ring. This could cause critical damage. Since it is a sliding bearing, which might be greased to protect sensitive sur- faces, the friction coefficient is more in a certain range than a stable value. The effects on the friction torque have to be considered in the design phase. The main advantages and disadvantages of the sliding bearings are summarised in Table 4.4. Compared to rolling bearings, the main advantages are that the sliding bearing units can be replaced up- tower and that no additional yaw brake system is needed, which reduces the costs. However, if the geared sliding ring needs to be replaced, a dismantling of rotor and nacelle is required. Table 4.4 Advantages and disadvantages of sliding bearings Version Advantages Disadvantages General • Sliding bearing units can be replaced up-tower • No additional yaw brake system required • Lower costs compared to rolling bearing with hydraulic brake system • Possibility of dry running (no lubrication needed except for yaw bearing teeth lubrication) • Replacement of geared sliding ring requires disassembly of rotor and nacelle • High friction torque Passive sliding bearing • No hydraulic system needed • Robust and simple design • Low maintenance effort • Preload cannot be adjusted during operation • More yaw drives needed Active sliding bearing • Preload can be adjusted during operation • Less yaw drives needed • Hydraulic system needed • Risk of leakage • High maintenance effort
- 193. 164 Wind turbine system design 4.3.3 Yaw brake system The yaw brake system can be active, passive or a mixture of both. The passive yaw brake system consists of set of disc spring loaded pistons mounted on the main frame, acting on a prepared surface of the yaw bearing and thus generating a brake torque. It can be found in some wind turbines, but active yaw brake systems are generally more common (see section 4.3.8). The active yaw brake system consists of a set of yaw brake callipers mounted on the bottom side of the main frame and acting on a yaw brake disc, which is located between the yaw bearing and the tower top flange. The brake surface area of the yaw brake disc is normally orientated towards the inside of the tower (Figure 4.1). There are two types of brake actuation: hydraulic or electro- mechanical. There are some solutions for electro- mechanical yaw brake systems on the mar- ket, but due to the greater installation space required and the higher costs, this system has not yet established itself on the market, in contrast to the hydrau- lic yaw brake system. This subsection therefore focuses on the hydraulic yaw brakes (Figure 4.26). Depending on the size of the yaw brake callipers, the brake pads are hydrauli- cally pressed against the brake disc by one or more pistons. Thereby the brake disc is clamped between the upper and lower part of the brake callipers. The brake pads are embedded in a pocket of the brake calliper and generate a brake torque through friction between the brake pads and the brake disc surface. Figure 4.26 Hydraulic yaw brake (Svendborg Brakes ApS)
- 194. Yaw system concepts and designs 165 The brake torque mainly depends on: • • the force of the pistons • • the number of pistons per yaw brake • • the coefficient of friction of the brake pad • • the effective braking diameter of the yaw brake disc • • the number of yaw brakes The nominal coefficient of friction is usually between 0.35 and 0.50. The coef- ficient of friction varies, depending on the brake pad material, the surface pressure, the contamination of the brake disc, the temperature of the friction partners and the relative speed between the friction partners. Organic or non- organic materials, in all cases asbestos- free, are often used as brake lining material, which are applied to a carrier plate made of steel or fiberglass composite. The exact components of the brake pads vary from type to type and are subject to the confidentiality of the brake pad manufacturers. Organic materials are chemically organic components, essentially carbon compounds. Conversely, non- organic brake linings con- sist of other materials that are moulded into a lining using plastic and resin compounds. As a rule, but especially in series production the brake callipers are made of a cast material, such as EN- GJS- 500- 7 or similar. For special prototypes, however, the first callipers are often made from a forged material before they go into series production. The pistons inside the brake callipers are made of stainless steel material, whereas the brake disc is often made of an unalloyed structural steel, such as S355J2G3. Standardised connecting dimensions (hole pattern, thickness of the brake disc, pressure connection, drain connection, etc.) for assembling the brake callipers have developed on the market. Although they differ depending on the size of the brake cal- liper, they are usually offered in the same way by all yaw brake suppliers. This means that brake callipers from different manufacturers can be used without major problems. The brake callipers usually have several connections for connecting the hydrau- lic lines. These are located on the side and/or on the back of the brake calliper. The connection size is usually a G1/4 thread. In addition, each half of the brake calliper has a leakage oil connection, also in G1/4. The internal leakage holes of the upper and lower brake calliper halves are sealed with an O- ring. Each piston in the brake callipers is usually covered by two or three seals – one or two high- pressure seals (usually one or two U- cups with an additional O- ring) and an additional wiper. When the high- pressure seal is worn, the oil enters the gap to the wiper and runs from there into the leakage hole. The leakage oil is then usually collected in an oil leakage bottle, which is mounted on the drain port of the lower brake calliper half. Alternatively, the pressure- less leakage can also be collected in transparent or partially transparent leakage lines. Hoses and connecting pieces from the field of pneumatics are often used for this. The hydraulic lines from the hydraulic unit to the yaw brakes and back to the oil tank, as well as from one brake to the other and between the brake calliper halves are usually high- pressure hoses or stainless steel pipes. Depending on the amount of brake callipers
- 195. 166 Wind turbine system design in the circuit, it makes sense to split the hydraulic feed and discharge line into several parallel paths. As a result, the pressure loading and unloading of all callipers is more even. Figure 4.27 shows an example of the connection of the hydraulic lines to the yaw brakes. After installing the yaw brake and connecting and commissioning the hydraulic unit, it is advisable to flush the hydraulic brake system with filtered oil to remove any dust or dirt particles that can lead to seal damage. Bleeding the brake system is also important to ensure proper function and to avoid damage to the system. The bleeding of the hydraulic circuit can be ensured by using a Minimess® adapter, which is connected to the upper pressure connection of the last yaw brake before the hydraulic line is routed to the oil tank. With a suitable adapter for this Minimess® , the air in the pressure chambers and hydraulic lines can be flushed out of the circuit. The brake pads can either be equipped with an electronic wear sensor or, as in most cases, with a wear indicator pin that is screwed into the back of the brake pad carrier plate. With increasing wear of the brake pads, the indicator pin dips deeper into a countersunk through hole of the brake calliper and shows the wear status by means of a coloured marking. Alternatively, brake callipers are also available, with an appropriately positioned inspection opening allowing a view of a color- coded carrier plate to check the state of wear. The hydraulic yaw brake system generally operates in three different modes: • • Non- yawing operation: The hydraulic yaw brakes are applied with full pressure from the hydraulic unit. Depending on the type and size of the brake, this can range from usually 160 to 190 bar – nevertheless higher pressure values can also be possible. • • Yawing operation – normal yawing: When the yaw system begins to align the rotor in the direction of the wind, the hydraulic pressure is reduced to about 10–25% of the full pressure value. With the resulting reduced braking torque, alternating loads on the yaw system can be reduced during yawing. The partial brake pressure is often a constant value. However, a number of gradations or a pressure characteristic are also possible. Figure 4.27 Yaw brake – hydraulic line connection (Trebu Technology B.V.)
- 196. Yaw system concepts and designs 167 • • Yawing operation – cable untwisting: The cable is only untwisted when the wind turbine is idling, often at low wind speeds. The external loads on the yaw system are so low that the hydraulic brakes can be completely depressurised. This also reduces wear of the brake pads. Squeaking noises, which can still be heard in the distance, may occur during yawing. The weather conditions (humidity and temperature), but also the properties of the brake disc and brake linings (e.g. roughness and cleanliness of the surfaces) often have an influence on this effect. Yaw squeaking occurs predominantly with organic brake linings. In many cases, the surface of the brake lining is glazed, which seems to promote squeaking. To reduce this problem, there are various solutions from brake system suppliers, e.g., by using other brake lining types or by using holes or grooves in the brake disc that are intended to clean the surface of the lining. Yaw squeaking should be ana- lysed in detail in each individual case and appropriate corrective measures agreed with the yaw brake supplier. 4.3.4 Yaw gearbox The yaw drive consists of a yaw motor and a yaw gearbox. The motor (see sec- tion 4.3.5) provides driving torque and rotational speed, whereas the speed reduc- ing gearbox converts torque and speed to adapt the motor curve to the yaw system, which requires high torques and low speeds. Multistage planetary gearboxes are used as yaw gearbox due to their power density and compact design (Figure 4.28). The yaw gearbox is mounted on the nacelle and its output pinion meshes with the teeth of the yaw bearing. The gear ratio depends on the speed of the intended yaw motor and the desired yaw speed. It can be between 600 and 3,000. Depending on the gear ratio, the yaw gearbox has four or five stages. A planetary stage consists of a sun gear, planet gears supported by the planet carrier and a ring gear. In the yaw gearbox, the ring gear is fixed, and the sun gear is driven. Several stages are combined one behind the other so that the sun gear of the following stage is connected to the planet carrier of the previous stage. The gearbox output shaft is connected to the planet carrier of the last gear stage. There are different design variants of yaw gearboxes (Figure 4.29). To reduce the height of the yaw drive, the first stage can be a bevel gear stage, which enables the motor to be arranged next to the gearbox. In the past, worm stages were also used. The reason for this was to use the self- locking of the worm stage to provide holding torque to hold the nacelle in position. However, the concept did not prove successful in practice since the self- locking caused problems when the yaw system slipped. In general, there are three different output versions (Figure 4.29). The output version influences the design of the machine frame, the accessibility of the gearbox, the load on the gearbox housing and the screw connection, and the possible options
- 197. 168 Wind turbine system design Figure 4.29 Yaw gearbox versions (Bonfiglioli) Figure 4.28 Yaw gearbox (Liebherr- Components Biberach GmbH)
- 198. Yaw system concepts and designs 169 for the backlash adjustment. The advantages and disadvantages of the different out- put versions are summarised in Table 4.5. • • Short version: The gearbox has little or no radial support by the main frame. As a result, the output housing and the bolted connection are exposed to the bending moments and radial forces caused by the toothing forces. On the other hand, the accessibility of the gearbox for service and maintenance is better and the main frame design in the area of the yaw gearboxes is less complex. • • Long version: The gearbox is completely supported radially by the main frame. The bending moments and radial forces are absorbed by the lowest radial sup- port. For that reason, the advantages and disadvantages are exactly the opposite of the short version. • • Pinion supported on both sides: This version is rather seldom to be found. The advantages are a more cost- effective design of the output shaft and bearing assembly and a lower deformation of the output pinion. However, the output housing is more complex. In addition, changes to the output pinion result in modifications to the housing, which makes changes more complex. Table 4.5 Advantages and disadvantages of output versions Version Advantages Disadvantages Short • Least complex main frame design • Good accessibility of gearbox • Options for backlash adjustment • Eccentric housing • Eccentric ring • Radial adjustment • Housing and screw connection exposed to bending moments • Medium to high deformation of output pinion Long • Housing and screw connection not exposed to bending moments • Low deformation of output pinion • More complex main frame design • Poor accessibility of gearbox • Options for backlash adjustment: • Eccentric housing Pinion supported on both sides • Less complex main frame design • Good accessibility of gearbox • Pinion is housed • Options for backlash adjustment: • Eccentric ring • Radial adjustment • Housing and screw connection exposed to bending moments • Medium to high deformation of output pinion • Changes to the output pinion not easy to implement
- 199. 170 Wind turbine system design A minimum normal backlash of 0.03–0.04 times the gear module is usually required between the output pinion and the yaw bearing teeth for operation (see section 4.1.3). To be able to compensate for manufacturing and assembly tolerances, it must be possible to position the gearbox radially to the yaw bearing. There are basically three ways of doing this. Their advantages and disadvantages are shown in Table 4.6. • • Eccentric housing: Output shaft and housing are not coaxial, but eccentric (1–2 mm). By turning the yaw gearbox around its longitudinal axis, the distance between the output pinion and the yaw bearing teeth can be adjusted. Since the bore pattern must be adhered to, only discrete positions can be set. As a result, the backlash requirements can sometimes not be met. In addition, the gearbox changes its orientation, which can lead to a poorer accessibility of the gearbox (e.g., oil sight glass and oil drain plugs), especially in case of gearboxes with the long output version. • • Eccentric ring: Output shaft and housing are coaxial, but there is an eccentric ring between gearbox and main frame. By turning the eccentric ring, the radial position of the gearbox can be changed. The main advantage is that the gearbox maintains its position, which can be beneficial for the accessibility of the gear- box. As with the eccentric housing, only discrete positions can be set. • • Radial adjustment: To enable a continuous adjustment, the gearbox must not be supported radially. Then the gearbox can be moved radially with the help of tools. The backlash requirements can be fulfilled, but the screw connection has to be designed for the high bending moments and radial forces to prevent the yaw gearbox from slipping. In the following, the output pinion shaft and bearing assembly for yaw gear- boxes with short and long output is described. The assembly comprises the output Table 4.6 Advantages and disadvantages of backlash adjustment options Option Advantages Disadvantages Eccentric housing • Radial support of gearbox by main frame • Discrete positioning • Positioning requires lifting of gearbox • Gearbox does not keep orientation Eccentric ring • Radial support of gearbox by main frame • Gearbox keeps its orientation • Discrete positioning • Positioning requires lifting of gearbox Radial adjustment • Continuous positioning • Gearbox keeps its orientation • Easy adjustment of backlash • No radial support of gearbox by main frame
- 200. Yaw system concepts and designs 171 shaft with the output pinion, the output housing and two preloaded tapered roller bearings in O arrangement. This overhung arrangement of the output pinion shaft is typical for slewing gearboxes. It is well suited for the high bending moments and radial forces from the tooth contact. The gearbox housing is usually made of ductile cast iron, such as EN- GJS- 400- 15 or EN- GJS- 400- 18U- LT. Non- ductile materials should be avoided to ensure sufficient material properties for cold climate applications. The forged output shaft with output pinion is usually made of a case- hardened steel, such as 17NiCrMo6- 4. The surface hardness is usually ≥58 HRC. The harden- ing depth is determined according to ISO 6336- 5. There are also solutions in which the output shaft and output pinion are two separate components. The connection is then made with a form- fit shaft- hub connection and an axial locking device. Due to the poor gear quality of the yaw bearing teeth (≥10 according to ISO 1328, ≥11 according to DIN 3962), a significantly better gear quality of the output pinion does not bring any advantage. The yaw bearing suppliers recommend a gear quality of nine or ten (according to DIN 3962). The toothing is heavily loaded. To avoid that the tip edge of the output pinion generates abrasive wear on the flanks of the yaw bearing teeth, the pinion is usually provided with a tip edge relief (Figure 4.30), based on the recommendations of the yaw bearing suppliers. The yaw gearbox is often mounted in an overhung arrangement (see Figure 4.29). Thus, deflections of the output pinion shaft are unavoidable. This leads to a poor con- tact pattern with significant excess loads for tooth root and tooth flank. To improve the contact pattern and thus increase the lifetime of the toothing, the pinion is provided with flank modifications. This is usually a combination of angle modification and crowning. The yaw gearbox is lubricated with oil and grease. For the output pinion, the statements on yaw bearing tooth lubrication apply (see section 4.3.2). Most of the Figure 4.30 Profile modifications
- 201. 172 Wind turbine system design components in the gearbox are oil lubricated. Only the bearings of the input and output shaft can be lubricated with grease. The gearbox is almost completely filled with oil. Synthetic oils with extreme pressure (EP) additives are usually used. This means that gears, bearings, shafts and shaft- hub connections are in an oil bath. Depending on the gearbox size and oil volume, an oil breather plug is used to ensure pressure compensation. Unlike the main gearbox, there is no oil injection, oil filtration or oil cooling system. Consequently, wear particles collect at the lowest point in the gearbox, which should be considered when designing the gearbox, since wear particles can damage roller bearings and seals. For changing the oil, the gearbox is equipped with oil drain plugs or valves and oil- sight glasses. The oil drain plugs should be located at the lowest point of the gearbox. The first oil change usually takes place in the first year of operation since more wear occurs in the running- in phase. The interval for all subsequent oil changes is up to five years. The tapered roller bearing close to the output pinion is usually lubricated with grease. It could also be lubricated with oil. But then an oil drain plug would not be accessible due to the attachment of the gearbox to the machine frame. The bearing can be provided with either a lifetime lubrication or a relubrication device. When it comes to seals, a distinction is made between static and dynamic seals. In a static sealing application, there is no movement between the seal surface and its mating surface. The most common static seal is the O- ring, which is used in yaw gearboxes to seal housing flanges. Dynamic seals are used between rotating shafts and the stationary housing. Yaw gearboxes are usually equipped with radial shaft seal rings at the gearbox input, between oil and grease chamber, and sometimes at the gearbox output. These can be simple or redundant. The grease- lubricated output shaft bearing is an exception. In addition to radial shaft seal rings, metal sealing plates can also be used. 4.3.5 Yaw motor and yaw motor brake The yaw motors generate the driving torque and rotational speed to turn the nacelle. Electric and hydraulic motors can be used. In today’s yaw systems, asynchronous motors (or induction motors) are the prevailing motor type. Several attempts have been made in the past with hydraulic yaw systems. Due to leakages and quality problems, these systems have so far not been able to gain acceptance in the market. However, today’s component quality could make it pos- sible to utilise the advantages hydraulic yaw systems offer (see section 4.3.1). In the following, the focus is on the asynchronous three- phase motors (Figure 4.31). These motors standardised in accordance with the IEC 60034 series are widely available, cost- effective, low- maintenance, robust and reliable. In addi- tion, the motors can be controlled with frequency converters. In multimegawatt wind turbines, yaw motors with a rated power of around 1.1 to 4 kW are used, depending on the number of yaw drives and loads. The motors provide high speeds and low torques. However, the yaw system needs high torque
- 202. Yaw system concepts and designs 173 and low speed. In addition to the gear ratio of the yaw system, the ratio of speed to torque can be influenced by the number of pole pairs of the motor (Table 4.7). The higher the number of pole pairs, the lower the speed and the higher the torque (with the same rated power). A lower speed results in a lower ratio of the yaw system. A motor with a higher number of pole pairs seems to be advantageous. However, the higher the number of pole pairs, the more expensive the motor. A motor with a small number of pole pairs and a yaw gearbox with a higher ratio would therefore be the more cost- effective solution. On the other hand, a high gear ratio leads to a high torsional backlash on the motor side. It can also be disadvantageous during yaw slippage events since the permissible values are exceeded earlier (see section 4.2.6). In addition, a very high gear ratio might require a yaw gearbox with five instead of four stages, which would reduce the efficiency of the yaw drivetrain. The choice of the number of pole pairs is therefore always a compromise. Motors with two or three pole pairs are commonly used as yaw motors. Figure 4.31 Yaw motor (Bonfiglioli) Table 4.7 Number of pole pairs, motor speed and gear ratio Number of pole pairs Synchronous speed (50 Hz) Rated speed (50 Hz) Ratio for yaw speed 0.50°/s Ratio for yaw speed 0.25°/s 1 3,000 rpm ≈2,860 rpm ≈34,320 ≈68,640 2 1,500 rpm ≈1,415 rpm ≈16,980 ≈33,960 3 1,000 rpm ≈950 rpm ≈11,400 ≈22,800 4 750 rpm ≈710 rpm ≈8,520 ≈17,040
- 203. 174 Wind turbine system design When selecting the yaw motor, its efficiency class according to IEC 60034- 30 (IE1–IE4) must be considered. In many countries, motors with standard efficiency IE1 are only permitted under certain conditions. In Europe, for example, the motors usually have to comply with at least premium efficiency IE3 (without frequency converter) or high efficiency IE2 (with frequency converter). Figure 4.19 in section 4.2.7 shows an example of motor torque and motor cur- rent as a function of the motor speed of an asynchronous motor that is operated directly on the mains (or the onboard power system). This is how the yaw motors are operated in many yaw system concepts. All motors are switched on and off at once by a reversing contactor circuit. However, the simple structure of the yaw system has some disadvantages: • • High load peaks can occur at yaw start. Due to the torsional backlash in the yaw system, the motors can initially start up without load and then drive into the load (see section 4.2.8). • • The torsional backlash is different in all yaw drives. In addition, it varies dur- ing rotation. As a result, the load is not evenly distributed among the individual yaw drives. Especially during yaw start, situations can arise in which individual motors are already loaded while the others are still running through the back- lash. As yawing continues, the loads on the individual drives become more even. • • Unintentional and uncontrolled movements of the nacelle can occur during yaw start and stop when the procedures are not designed optimally (see section 4.2.8). • • The motor speed depends on the torque at the yaw motor shaft. Due to the fluc- tuating external loads, the nacelle is continuously accelerated and decelerated, which leads to additional loads for the mechanical and electrical components (see section 4.2.7). • • If the yaw speed remains zero, the yaw motors draw high currents. This can lead to a voltage drop in the onboard power system. As a result, the available motor torque is reduced since it depends on the voltage. The voltage drop in the onboard power system can also have negative effects on the other consumers in the wind turbine. For these reasons, different strategies were developed to control the yaw motors. On the one hand, these options reduce some disadvantages. On the other hand, they increase the complexity of the yaw system and can lead to higher costs. One option is frequency converters. Due to the intermediate circuit in the con- verter, the yaw motors are decoupled from the onboard power system. By controlling the voltage and frequency, the speed and torque of the yaw motor can be controlled. Yaw start and stop processes can be programmed and the maximum motor torque can be limited. The signals of the frequency converters can be evaluated by the yaw control system.
- 204. Yaw system concepts and designs 175 These are the following configurations: • • Single frequency converter: All yaw motors are controlled by a single frequency converter. This most cost- effective configuration mainly allows soft start, soft stop and limiting the maximum motor torque. Controlling the individual drives is not possible. • • Double frequency converter: The yaw motors are divided into two groups. Each group is controlled by a single frequency converter. This configuration allows more possibilities. For example, the yaw drives can be pretensioned. • • Multiple frequency converter: Each yaw motor has its own frequency con- verter. This configuration offers the most possibilities, including even load shar- ing across the yaw drives. All motors require encoders for this, making this the most expensive solution. Another option is soft starters. The motor voltage is increased from a selected starting voltage to the nominal motor voltage within an adjustable start- up time. During yaw stop, the voltage is slowly reduced. If a certain current shall not be exceeded, a soft starter with current limitation can be selected. The yaw motors are usually also equipped with yaw motor brakes that provide holding torque to hold the nacelle in position. Depending on the yaw system con- cept, the number of yaw drives, the total gear ratio of the yaw system and the yaw motor brake friction torque, the total yaw motor brake torque is normally higher than the yaw friction torque (consisting of yaw bearing friction and yaw brake friction). The brake torque (or holding torque) is only available when the yaw drivetrain is tensioned (see section 4.2.4). Spring- applied brakes (Figure 4.32) are used in asynchronous motors. This type of brake is an electrically releasable spring- applied brake with a rotating brake disc (rotor) that is equipped with friction linings on both sides. When the brake is de- energised, the pressure springs press the armature plate against the brake disc, which can be axially displaced on the hub. As a result, the brake disc is clamped with braking force between the armature plate and the motor flange. The result- ing brake torque is transmitted between rotor and hub via gear teeth. To release the motor brake, the coil of the stator is energised with the provided direct current (DC) voltage. The resulting magnetic flux works against the spring force to pull the armature plate towards the stator. This relieves the spring force of the rotor and allows it to rotate freely. In case of a failure, such as loss of power supply, the brake is automati- cally closed. The function of the brake thus corresponds to the failsafe principle. Compared to the yaw brake system (see section 4.3.3), no further measures have to be taken to make the brake torque permanently available in the event of a fault. The yaw motor brake is usually designed as a holding brake. That means that it is normally not used to decelerate the yaw drivetrain and the nacelle. The brake only engages when the nacelle has almost come to a complete standstill. The resulting
- 205. 176 Wind turbine system design high friction energy during braking and yaw slippage events is dissipated into heat. If the permissible friction energy of the brake is exceeded, the yaw motor brake can be damaged (see section 4.2.6). When starting and stopping the yawing process, the chronological order of yaw brake system, yaw motor brake and yaw motor must be observed to avoid critical or overload situations (see section 4.2.8). The yaw motor and yaw motor brake are usually equipped with sensors to moni- tor the yaw system, depending on the yaw system concept and the yaw control sys- tem. Further information can be found in section 4.3.6. 4.3.6 Auxiliary systems Without a number of auxiliary systems, the yaw system cannot function. A detailed description of these auxiliary systems is not possible within the scope of this chapter. Some auxiliary systems, however, are described in more detail in Figure 4.32 Spring- applied brake (Kendrion INTORQ)
- 206. Yaw system concepts and designs 177 other chapters of this book. In the following, reference is made to the relevant chapters. The electrical components of the yaw system such as yaw motors, yaw motor brakes, lubrication systems and hydraulic power units require a power supply. The onboard power system usually can provide alternating and direct current at different voltage levels. However, it is advisable to limit the number of voltage levels to a minimum for economic reasons. In addition, the function of the yaw system and its components should be checked for the expected voltage and frequency tolerances as well as for low volt- age ride through (LVRT) and over voltage ride through (OVRT) events. The latter are faults in the electrical grid. During the brief undervoltage or overvoltage events, the wind turbine generator must remain connected to the grid for a limited, defined time (support of the electrical grid). A hydraulic system is required for the operation of hydraulic yaw brakes and hydraulic yaw motors. The hydraulic system is described in section 7.1. The yaw bearing (rolling bearing and sliding bearing) and its teeth as well as the output pinion of the yaw gearboxes must be lubricated. This can be done manu- ally or with automatic lubrication systems. The lubrication systems are presented in section 7.3. Depending on the yaw system concept and the yaw control system, the yaw system is equipped with various sensors (see section 4.1.1). Among other things, the sensors measure positions, speeds, temperatures, currents, volt- ages, pressures and operating states. This information is used by the yaw con- trol system to manage, command, direct or regulate the behaviour of the yaw system. Each yaw system requires a sensor that measures the nacelle position. This information is needed for the determination of the yaw misalignment. The sensor is also used for measuring the yaw speed, which can be used for the detection of criti- cal situations or for controlling the yaw speed. There is usually also a sensor mea- suring the zero- degree or north position of the nacelle. In addition, the end position for the cable loop is measured. The end positions are secured against being overrun by limit switches. The electric yaw motors can be equipped with a temperature sensor in the motor winding to monitor its temperature and to avoid overheating. More advanced control strategies may also require encoders in the yaw motors to measure the motor speed and position. If the yaw motors are operated via frequency converters, further information is available, such as the current. If not, current and voltage can be measured with separate sensors. The yaw motor brake can be equipped with a sensor to detect if the brake is released or closed. This helps to detect faulty brakes. The pressure in the hydraulic system can also be measured. This information can be used for yaw start and yaw stop events.
- 207. 178 Wind turbine system design 4.3.7 Evaluation criteria A yaw system or the constructive and planning combination of the individual parts and components can be assessed using a wide variety of criteria, e.g., in the form of an evaluation matrix. Tables 4.8 and 4.9 show a collection of these main and sub- criteria sorted into a technical and economical grouping. These criteria represent only a certain selection and therefore do not claim to be complete. The final compilation of the evaluation criteria and their weighting is the responsibility of the yaw system developer. Personal or internal company experi- ences, as well as the exact description of the objective and function can have a major influence on the selection and weighting, so that there is no numbering or weighting of these evaluation criteria here. The technical evaluation criteria cover the basic functional requirements, the turbine safety, the installation and maintenance functionality and the control options, which are described in the previous subsections. A few other criteria are discussed in the following. Table 4.8 Technical evaluation criteria Main criteria Sub-criteria Basic functional requirements • Holding nacelle in current position • Wind alignment (active or passive yawing) • Cable untwisting Turbine safety • Safe state in case of failure • Functionality in case of grid loss • Low voltage ride through (LVRT) and over voltage ride through (OVRT) capabilities (faults in the electrical grid) Installation/maintenance functionality • Locking of the yaw system during installation, maintenance and exchange of components • Availability/usability of the yaw system during installation • Exchangeability of components Control options • Soft start/stop • Nacelle slippage in case of overload • Overload protection/load limiting • Nacelle positioning accuracy • Load sharing between yaw drives • Damping of load peaks • Yaw speed Reliability/technology maturity • Proven technology/long track record • Implementation of new technology • Use of standardised components Personal-or company-internal experiences • Experience with yaw system concepts and yaw system components • Available employee skills and qualifications Environmental conditions • Temperature, humidity, etc.
- 208. Yaw system concepts and designs 179 Reliability and technology maturity are important criteria. On the one hand, proven technology with a long track record and the use of standardised components results in low developments costs and high technical reliability. On the other hand, scaling up current designs and technology is not enough as the wind turbines are then no longer competitive. Therefore, the system and component designs are con- tinuously optimised and new technologies are introduced. The associated risks are to be reduced by suitable validation measures. The personal- and company- internal experiences also play an important role. The wind turbine manufactures usually specialise in one yaw system concept and optimise it. Fundamental changes in the yaw system concept are rather rare because this means a lot of effort and a high entrepreneurial risk. Skills and abilities may need to be built up first. The economical evaluation criteria essentially cover the costs over the entire life cycle, starting with the procurement costs, through the assembly and maintenance costs, to the disposal costs. A few criteria are described below in more detail. Table 4.9 Economical evaluation criteria Main criteria Sub-criteria Costs • Material costs • Transportation costs • Maintenance costs Supplier base • Supplier diversity • Local sourcing possibilities • Experiences in wind industry Assembly/installation effort • Effort for indoor assembly and indoor commissioning • Effort for outdoor assembly and outdoor commissioning • Effort for special handling or assembly tools • Mounting space requirements and restrictions Storage effort • Indoor and outdoor storage requirements • Long- term storage requirements and effort Service effort • Effort/complexity for maintenance • Maintenance time Transport requirements/restrictions • Transportability on land or sea • Restrictions based on maximum transport dimensions and weight Modularisation/standardisation potential • Use of standardised components or components from other turbine platforms • Adaptation effort to new turbine variants or new loads Disposal effort • Recyclability of component material Energy consumption • Energy consumption
- 209. 180 Wind turbine system design The supplier base is a very important criterion. Components should be available from multiple suppliers to reduce risk. Large wind turbine manufacturers produce worldwide. Local content requirements must also be met in several markets. For that reason, it should be possible to source the components locally. The operating and environmental conditions place special demands on the com- ponents used in wind turbines. The suppliers should therefore have experiences in the wind industry and have trained service staff available for service work on wind turbines. The wind turbines are growing ever larger. For the transport of onshore wind tur- bines,certaintransportdimensionsshouldnotbeexceeded(seesection4.4.1).Otherwise, the transport costs will increase sharply. As a result, the power density of the yaw system increases continuously. The yaw system and its components must fulfil this. The potential for modularisation and standardisation is becoming more and more important. In case of yaw gearboxes, yaw motors and yaw brakes, e.g., attempts are made to use standardised components that are manufactured by the suppliers in very large numbers and to use these in different wind turbine plat- forms and variants. By changing the number of yaw drives and yaw brakes, the yaw system can be adapted to different loads (due to different rotor diameter, rated power, wind conditions, etc.). 4.3.8 Common system concepts Table 4.10 summarises the three most common yaw system concepts used in multimega- watt upwind turbines from major wind turbine manufacturers. The data were compiled from freely available manufacturer documents and certification reports. However, the availability of information varies greatly between manufacturers. In addition, detailed information is often very difficult to find. The overview therefore does not claim to be complete. It is only intended to show how the concepts are spread. It can be seen from Table 4.10 that all three concepts are active, stiff and friction- damped yaw systems, but in which the friction is generated in different ways. In Concept 1, the sliding bearing provides the friction torque, whereas in Concept 3, the friction torque is generated by the yaw brake system. Concept 2 has a very low yaw friction torque and thus the yaw drives are pretensioned to eliminate the backlash. In all three concepts, electro- mechanical yaw drives are used. They consist of planetary gearboxes and asynchronous motors, which are operated directly on the mains or with a frequency converter. Vestas, Siemens Gamesa and Enercon only use one yaw system concept in their current turbines, whereas GE and Nordex use different concepts. Nordex changed the concept from its Delta platform to its Delta4000 platform. GE uses a different yaw sys- tem concept for its onshore and offshore wind turbines. GEs offshore technology comes from the former wind turbine OEM Alstom, which was acquired by GE. GE uses active and passive yaw brakes in its current onshore wind turbines. Nordex relies on hydraulic brakes in Concept 3.
- 210. Yaw system concepts and designs 181 4.4 System dimensioning and design aspects This section is intended to give an exemplary insight in the dimensioning of a yaw system for a multimegawatt upwind turbine. A few design and calculation aspects are also discussed in more detail here. First, the basic process of dimensioning and the boundary conditions are explained. Second, further requirements resulting from the selected yaw system concept are presented. Finally, yaw bearing, yaw brake system and yaw drive sys- tem are dimensioned. The aim is not to completely dimension and design the best possible and most competitive yaw system. On the one hand, this would go beyond the scope of this chapter. On the other hand, the design and calculation details differ between the wind turbine manufacturers and component suppliers. Therefore, only a general insight can be given here. Table 4.10 Common yaw system concepts Concept 1 Concept 2 Concept 3 System level Active/ passive Active Active Active Soft/stiff Stiff Stiff Stiff Damping Friction-damped by yaw bearing Friction-damped by yaw drives Friction- damped by yaw brake system Component level Yaw bearing Sliding bearing Rolling bearing Rolling bearing Yaw brake – – Active, hydraulic and/or passive Yaw drive Electro-mechanical Electro-mechanical Electro-mechanical Yaw gearbox Planetary gearbox Planetary gearbox Planetary gearbox Yaw motor Asynchronous motor Asynchronous motor Asynchronous motor Examples Vestas EnVentusTM platform 4 MW platform 2 MW platform Offshore platform Siemens Gamesa 5.X platform 4.X platform 3.X platform 2.X platform Offshore turbines GE Haliade-X platform Nordex Delta4000 platform Enercon EP5 platform EP3 platform EP2 platform Nordex Delta platform GE Cypress platform 3 MW platform 2 MW platform
- 211. 182 Wind turbine system design 4.4.1 Introduction and general requirements A yaw system is never dimensioned and designed independently from the rest of the wind turbine. In this respect, the following is an artificial design scenario in which many design assumptions are made. The Fraunhofer IWES wind turbine IWT- 7.5- 164 serves as a common thread. Missing or deviating data are explicitly stated and explained accordingly. The dimensioning of a yaw system is an iterative process. The various development methodsarenotdiscussedhere.Thefocusisratherontheinterdependenciesandinterfaces within the yaw system and to the companion structure. Figure 4.33 shows an example of a possible development process for a yaw system. In the first step, based on the general requirements for the wind turbine and the yaw system, the system concept is selected using the evaluation criteria from section 4.3.7. This results in additional requirements for the yaw system and the individual components. In the next step, yaw bearing, yaw brake system and yaw drive system are pre- dimensioned. This usually takes place in parallel due to the interdependencies. In Figure 4.33 Yaw system – exemplary development process
- 212. Yaw system concepts and designs 183 addition, several variants are examined to come to the best solution. The system and component designs are further detailed in several loops. The number of variants is reduced in the process. The auxiliary systems are dimensioned at a later stage. Detailed information about the other components is often required here, which are not available at the beginning of the development process. Of course, the auxiliary systems must be considered when deciding on a variant. Changes in component designs can affect other yaw system components and the companion structure. Changes from the outside, such as new loads, changed installation space or new requirements, can lead to major changes. For these reasons, there is often an iterative jump back and forth to one or more work steps. Table 4.11 summarises the key data of the Fraunhofer IWES wind turbine IWT- 7.5-164, Table 4.12 the requirements for the yaw system. These are explained in more detail below. Table 4.11 Wind turbine data IWT- 7.5- 164 [7] Parameter Value Rated power 7.5 MW Rotor diameter 164 m Hub height 120 m Rated rotor speed 10 rpm Drivetrain concept Direct drive Design life 20 years Operating temperature −20 to +40°C Survival temperature −30 to +50°C Yaw moment of inertia 5.531E+07 kgm² Table 4.12 Boundary conditions for the yaw system Parameter Value Active/passive Active Soft/stiff Stiff Damping Friction-damped Type of yaw drives Electro- mechanical drives with asynchronous motors Yaw system loads Section 4.2, maintenance loads see section 4.4.2 Operating hours 10% of turbine design life Yaw speed 0.25–0.50°/s Yaw slippage Allowed to a limited extent in case of extreme load cases Power supply 400 V, 50 Hz Max. machine frame width 4.2 m Available installation space for yaw drives 225° of 360° (left, right and rear)
- 213. 184 Wind turbine system design A turbine design life of 20 years is assumed based on the given loads (see section 4.2). The temperature range is not specified for the IWT- 7.5- 164 [7]. Since the temperature range has only a minor influence on the yaw system design and the loads were simulated for normal temperatures, a normal temperature application is assumed. Cold climate or hot climate applications must be considered in the component designs accordingly. For example, materials, seals and lubricants must be suitable for the extreme temperatures. The yaw system for the IWT- 7.5- 164 has not yet been detailed. Since it shall be based on common yaw system concepts (see section 4.3.8), the yaw system shall be an active, stiff and friction- damped one using electro- mechanical drives with asynchronous motors (Table 4.12). As explained and presented in section 4.2, loads of a comparable 7 MW wind turbine with a rotor diameter of about 170 m are used instead of the IWT- 7.5- 164 loads. These loads correspond better to the current state of the art. The operating hours of the yaw system are assumed to be 10% of the turbine design life. This corresponds to the conservative specification from the DNV standard [1]. The yaw speed of the IWT- 7.5- 164 is given as 1°/s [7]. This is comparatively high for such a large wind turbine (see section 4.1.3). For this reason, the target cor- ridor is set to 0.25–0.50°/s. Yaw slippage (see section 4.2.6) is only allowed to a limited extent in extreme load cases and should be avoided if possible. A simulation of yaw slippage events is not carried out as part of the example dimensioning. The outer diameter of the tower top flange is specified as 3 m [7], which is com- paratively small for a 7 MW wind turbine with the given loads. With a larger diam- eter, more yaw drives and yaw brakes would fit in the wind turbine. In addition, one yaw drive and one yaw brake generate a higher driving and holding torque due to the larger diameter, which reduces the number of drives and brakes required. A larger diameter is also beneficial for the yaw bearing. Smaller balls can be arranged on a larger raceway diameter. This reduces the bearing height and improves the lubrication conditions in the bearing. For the same adjustment angle, the balls cover a greater distance in relation to their diameter compared to a smaller bearing with larger balls. As a result, the grease is better distributed in the yaw bearing and is thus better mixed. The lubricating film is also renewed more quickly, thereby reducing wear. For these reasons, the specified diameter of the tower top flange is not used here. It should rather result from the yaw system design. The IWT- 7.5- 164 is an onshore wind turbine. Since this is a reference wind turbine, no target markets are defined from which requirements for the permissible nacelle dimensions could be derived. However, to be able to transport the nacelle as cost- effectively and flexibly as possible, a certain width should not be exceeded. Larger widths are possible, but limit means and routes of transportation. This makes transportation more expensive and reduces the number of possible wind turbine sites. For the example dimensioning, the maximum permissible width of the machine frame is set at 4.2 m. This allows road transport in many markets as well as rail transport in the United States of America. It is assumed that the nacelle cover and secondary structure can be transported separately if necessary.
- 214. Yaw system concepts and designs 185 Machine frame design and drivetrain concept have a major impact on the avail- able installation space for the yaw drives. Wind turbines with a geared drivetrain usually only allow yaw drives to be arranged to the left and the right of the pow- ertrain since rotor shaft and bearing assembly, main gearbox and generator require a lot of space in the nacelle above the yaw system. In comparison, direct drive wind turbines, where the generator hangs in front of the tower, often offer more space for yaw drives. For the direct driven IWT- 7.5- 164, an available installation space of 225° is assumed for simplification. The yaw drives are located on the left, right and rear. This would be specified in more detail as part of an overall wind turbine development. The standards IEC 61400- 1 [3] and DNVGL- ST- 0361 [1] specify the minimum requirements for the design of wind turbines and yaw systems intended to ensure an acceptable safety level. Any of the requirements of these standards may be altered if it can be suitably demonstrated that the safety of the system is not compromised. The example sizing of the yaw system, however, follows the minimum requirements of IEC 61400-1. 4.4.2 Step 1: yaw system, holding torque and driving torque As described in section 4.3.7 and shown in section 4.3.8, wind turbine manufac- turers often specialise in one yaw system concept. A detailed selection of the yaw system concept in this artificial example dimensioning process makes no sense and is beyond the scope of this chapter. Therefore, it is determined that the yaw system shall correspond to Concept 3 consisting of a double- row four- point contact ball bearing, a hydraulic yaw brake system and electro- mechanical yaw drives operated with frequency converters. In the following, the required minimum driving and holding torques are deter- mined based on the given loads, which form the basis for the dimensioning of the yaw brake system and the yaw drive system. According to the DNV standard [1], the two brake systems (hydraulic yaw brakes and yaw drives) shall be designed in such a way that each brake system is able to hold the nacelle in position during wind turbine installation, service and maintenance. Therefore, a yaw locking device is not required. The relevant yaw moment Mz values are summarised below, with the first value given in Table 4.2: • Max. yaw moment Mz with partial load factor: 20,120 kNm • Max. yaw moment Mz without partial load factor: 17,440 kNm • Max. yaw moment Mz in individual time series: 19,220 kNm • Max. yaw moment Mz during installation/maintenance: 8,720 kNm • Max. yaw moment Mz in LDD: 15,600 kNm Since yaw slippage should be avoided in this example, the total holding torque of the yaw system should be greater than or equal to 20,120 kNm. If yaw slippage is
- 215. 186 Wind turbine system design allowed, the minimum total holding torque should be carefully determined, and the yaw slippage events should be simulated and analysed in detail (see section 4.2.6). It should be noted that the values in the extreme load tables are mean values of the maximum values from a set of time series for a specific DLC. This means that in the individual time series values can occur that are higher than the maximum yaw moment Mz without partial load factor (see above). For maintenance purposes, the yaw brake system and the yaw drives must gener- ate a minimum holding torque of 8,720 kNm. If the hydraulic brake system has leak- age or is maintained and thus pressure- less, the yaw drives hold the nacelle safely in position. It is very unlikely that all yaw drives fail at the same time. However, should this be the case, the hydraulic yaw brake system keeps the nacelle in its position. During turbine installation, it needs to be checked whether and when the hydrau- lic yaw brake system can provide its holding torque. In some cases, it is possible to generate the required hydraulic pressure with a manual pump. Permissible wind speeds for wind turbine installation as well as service and maintenance activities are usually specified considering the requirements from IEC [3] and DNV [1]. In principle there are two possibilities. Either the allowable wind speeds are specified, and the maintenance loads are simulated for these wind speeds, or the permissible wind speeds are determined on the basis of the available holding torques of the yaw system. Another criterion could be how often the holding torque of the hydraulic yaw brake system is exceeded in the fatigue DLCs. In this example, this is less than 20 hours in 20 years (Figures 4.14 and 4.15), which is considered acceptable. Therequiredminimumholdingtorquesaresummarisedbelow.Itisobviousthatthetwo holdingsystemsneedtogeneratehigherholdingtorquestoachievethetotalholdingtorque. • Total holding torque: ≥ 20,120 kNm • Holding torque of yaw brake system: ≥ 8,720 • Holding torque of yaw drives: ≥ 8,720 The determination of the partial brake pressure during yawing is always a com- promise. The higher the partial brake pressure, the higher the damping of the exter- nal yaw moment Mz , but also the higher the load on yaw drive and yaw bearing teeth. In this example, a constant partial brake pressure is assumed. It is often in the range between 10% and 25% of the full brake pressure during non-yawing operation. This means that in this example the yaw brake friction torque would be at least 872–2,180 kNm. Figure 4.15 shows that the yaw brake friction torque would then be higher than the external yaw moment Mz between about 50% and 85% of the time. In this example, 10% is set as the starting point. The actual yaw brake friction depends on the yaw brake used and is determined in section 4.4.2. • Partial yaw brake friction: ≥ 872 kNm
- 216. Yaw system concepts and designs 187 Section 4.2.7 explains that the driving capacity of the yaw system is often under- sized to reduce costs. The LDD of the yaw moment Mz in Figure 4.15 provides the reason for this. Over 99% of the time, the load level is in the lower half of the load range. In most cases, the yaw system can simply wait until more favourable loads are available to yaw the nacelle. When determining the required driving torque, yaw bearing friction, yaw brake friction and inertia torques need to be considered (see sections 4.2.4 and 4.2.5). However, these can only be estimated at the beginning. Based on experience and on equations from the bearing suppliers, a conservative yaw bearing friction torque of 150 kNm is assumed. The total estimated yaw drivetrain torque is shown in Figure 4.34. Compared to Figure 4.15, the torque values are increased by 1,022 kNm (yaw brake friction 872 kNm, yaw bearing friction 150 kNm). There are no fixed criteria for the required driving torque. In this example, based on experience from similar applications, the minimum driving torque is set at 7,262 kNm. This means that only about 0.25% of the time the driving torque would not be sufficient. However, the maximum yaw drivetrain torque is 16,622 kNm (Figure 4.34). The required maximum driving torque must be checked again when the actual friction torque is known. • Max. available driving torque: ≥ 7,262 kNm Figure 4.34 Yaw drivetrain torque (LDD, absolute values)
- 217. 188 Wind turbine system design With the specification of the holding and driving torques to be achieved, the basic requirements for the yaw system are defined. The subsystems and components can now be dimensioned (see following subsections). In section 4.4.8, it is finally checked whether the yaw system meets the requirements (Table 4.13). 4.4.3 Step 2a: yaw bearing, yaw brake and yaw drive Due to the interdependencies, yaw bearing, yaw brake system and yaw drive system are usually dimensioned in parallel. Depending on the boundary conditions, a differ- ent starting point can be useful. In this dimensioning example, the yaw brake system seems to dimension the size of the yaw system. Hence, the design is started with this system. Before that, however, the components used are presented. As described in section 4.3.3, yaw brakes from different suppliers have stan- dardised connecting dimensions. In multimegawatt wind turbines, two different brake sizes are common: • • a small brake with two pistons with a diameter of 90 mm • • a large yaw brake with three pistons with a diameter of 120 mm The larger one shall be used for the IWT- 7.5- 164. With the large yaw brake, a higher brake torque can be generated compared to the small one on the same diam- eter. This allows to keep the yaw system diameter small. The main data of the yaw brake used in this dimensioning example are summarised in Table 4.14. There are no standardised yaw gearboxes. However, certain gearboxes are used more often than others by several wind turbine OEMs. To benefit from the cost advantages, it is worth checking whether existing gearboxes that are produced in large quantities can be used. In this dimensioning example, two different yaw gear- box sizes are considered to show the influence on the yaw system design. The main data are shown in Table 4.15. The asynchronous motors are standardised according to the IEC 60034 series. Suitable yaw motors are selected so that the static load capacity of the yaw gearbox and yaw bearing teeth are not exceeded. The maximum motor torque is limited to a maximum of 1.8 times the nominal torque by a frequency converter. The yaw bearing teeth usually limit the maximum allowable motor torque, which is already considered here. The motor data are summarised in Table 4.16. Table 4.13 Yaw system – fatigue and static strength analysis (Sub-)component Standards, guidelines, methods and criteria General IEC 61400- 1 [3], DNVGL- ST- 0361 [1] Yaw system Holding torques (yaw slippage events), driving torque (overload events, availability of yawing), yaw start and stop events Components See sections 4.4.4–4.4.6
- 218. Yaw system concepts and designs 189 Table 4.17 shows the holding and driving torque per yaw drive at the output pinion as well as the extreme torques that can occur during holding and yawing. There are still some reserves, e.g., for the additional loads that can occur during yaw slippage events or overload events during yawing. The extreme torques on the yaw bearing teeth still have to be compared with their load capacity. Table 4.14 Yaw brake data Parameter Value General Piston diameter DPiston 120 mm Number of pistons aPiston 3 Number of friction pairs 2 Overall dimensions Approx. 500 × 307 × 274 mm Weight Approx. 180 kg Brake pressure p Full pressure – holding 180 ± 2 bar Partial pressure – yawing 18 ± 5 bar Partial pressure – untwisting 0 bar Coefficient of friction μ Coefficient of friction – holding 0.45 Coefficient of friction – yawing 0.50 Coefficient of friction – untwisting n/a Table 4.15 Yaw gearbox data Parameter Value small gearbox Value large gearbox General Gearbox type Multistage coaxial planetary gearbox Multistage coaxial planetary gearbox Number of stages 4 4 Gear ratio 945 1,254 Dynamic efficiency 0.885 0.885 Static efficiency 1.000 1.000 Static load capacity at pinion Approx. 50 kNm Approx. 137 kNm Output pinion Gear module 16 mm 20 mm Number of teeth 11 14 Tooth width 165 mm 205 mm Profile shift coefficient 0.5 0.5 Dimensions and weight Motor frame size IEC100 IEC132 Output flange diameter 425 mm 530 mm Gearbox height Approx. 705 mm Approx. 1,200 mm Weight Approx. 275 kg Approx. 825 kg
- 219. 190 Wind turbine system design In case of the yaw bearing, it can also make sense to check whether the suppli- ers already have a suitable bearing on offer that can be used to meet the yaw system requirements. In this dimensioning example, however, the yaw bearing is dimen- sioned roughly (see section 4.4.5). 4.4.4 Step 2b: dimensioning of the yaw brake system With the yaw brake data given in Table 4.14, the yaw brake system can be easily sized using the equations below. Fc = DPiston 2 2 10 pholding or yawing 1, 000 aPiston (4.3) Table 4.16 Yaw motor data Parameter Value IEC100 Value IEC132 Motor Motor type Asynchronous motor Asynchronous motor Rated voltage 400 V, 3~, AC 400 V, 3~, AC Frequency 50 Hz 50 Hz Rated power 1.85 kW 4.00 kW Rated speed 930 rpm 950 rpm Rated torque 19 Nm 40.21 Nm Maximum torque 34 Nm 72 Nm Motor brake Motor brake type Spring-applied brake Spring-applied brake Motor brake torque 40 Nm 80 Nm Brake torque tolerance ±20% ±20% Dimensions and weight Motor flange diameter 250 mm 300 mm Motor diameter 195 mm 258 mm Motor height 398 mm 523 mm Weight Approx. 30 kg Approx. 58 kg Motor inertia 0.0099 kgm² 0.0305 kgm² Table 4.17 Yaw drive data Parameter Value small drive Value large drive Load capacity Gearbox static load capacity Approx. 50 kNm Approx. 137 kNm Driving torque Nominal driving torque at pinion 15.89 kNm 44.64 kNm Max. driving torque at pinion 28.44 kNm 79.93 kNm Extreme torque at pinion (+10%) 31.29 kNm 87.92 kNm Holding torque Nominal holding torque at pinion 37.80 kNm 100.32 kNm Extreme torque at pinion (+20%) 45.36 kNm 120.38 kNm
- 220. Yaw system concepts and designs 191 FB = FC 2 static or dynamic (4.4) MB = a FB Deff 2 (4.5) • number of yaw brakes [–] • number of pistons per calliper half [–] [–] • effective braking diameter [m] • piston diameter [mm] • braking force [kN] • clamping force [kN] • braking torque [kNm] • hydraulic pressure [bar] • coefficient of friction DPiston Deff aPiston FB FC a MB pholding or yawing static or dynamic First, the clamping force FC is calculated from the total piston area of a calliper half and the hydraulic pressure. Second, the braking force FB results from the clamp- ing force, the number of friction pairs and the coefficient of friction. Multiplying the braking force by the number of yaw brakes and the effective braking diameter finally gives the braking torque MB . The manufacturer’s documents usually contain a specification for determining the braking diameter Deff depending on the brake cal- liper design and inner diameter of the brake disc. When arranging the yaw brakes care must be taken to ensure that there is sufficient space between the yaw brakes. The space is needed for tools and equip- ment during assembly and disassembly of the yaw brakes. Ideally, the brake cal- liper halves can be pivoted via one of the outer fastening bolts which remains in place as pivot. This enables an easy brake pad replacement. In addition, sufficient space must be provided between the yaw brake and the yaw bearing, e.g., for lubrication lines. It often makes sense that both brake systems each provide about 50% of the holding torque. For this example, this means that the yaw brakes must generate around 10,000 kNm of holding torque. This requires 14 yaw brakes and a brake disc inner diameter of 2,625 mm. The results are summarised in Table 4.18. The hydraulic pressure is subject to tolerances. This is considered in Tables 4.14 and 4.18. For yaw slippage events, the minimum braking torque during non- yawing operation is of interest. For yawing, the maximum braking torque defines the load on yaw gearbox and yaw bearing teeth. The yaw brakes are standardised components that have to fulfil the requirements of the IEC or DNV standard (Table 4.19). The strength of the brake callipers is verified by finite element method (FEM) calculations and bench tests. The strength verification of the bolted connections is also done using FEM. The friction coefficients as a function of pressure, temperature and sliding speed as well as the wear and yaw squeaking behaviour are examined in bench tests. The tightness of the seals is also validated in bench tests. To prove the suitability for cold climate applications, the tests are also carried out in climate chambers.
- 221. 192 Wind turbine system design 4.4.5 Step 2c: dimensioning of the yaw bearing When designing a yaw bearing, experience from other wind turbines can be used. Based on the loads and the available installation space, the yaw bearing can be pre- dimensioned with little effort. In the further course, the yaw bearing design is detailed, and extensive calculations are carried out. In the previous subsection, the yaw brake system was dimensioned. The size of the yaw brake system defines the inner diameter of the yaw bearing in this Table 4.18 Yaw brake system Parameter Value General Number of yaw brakes 14 Brake disc inner diameter 2,625 mm Effective braking diameter 2,727.5 mm Clamping force FC Nominal clamping force – holding 610.73 kN Minimum clamping force – holding 603.94 kN Nominal clamping force – yawing 61.07 kN Maximum clamping force – yawing 78.04 kN Clamping force – untwisting 0 kN Braking force FB Nominal braking force – holding 549.65 kN Minimum braking force – holding 543.55 kN Nominal braking force – yawing 61.07 kN Maximum braking force – yawing 78.04 kN Braking force – untwisting 0 kN Braking torque MB Nominal braking torque – holding 10,494.25 kNm Minimum braking torque – holding 10,377.65 kNm Nominal braking torque – yawing 1,166.03 kNm Maximum braking torque – yawing 1,489.92 kNm Braking torque – untwisting 0 kNm Table 4.19 Yaw brake – fatigue and static strength analysis (Sub-)component Standards, guidelines, methods and criteria General IEC 61400- 1 [3], DNVGL- ST- 0361 [1] Brake callipers Finite element method, bench tests (extreme load test, fatigue load test) Brake pads Bench tests (coefficient of friction, wear behaviour and squeaking behaviour), heating calculation Seals Bench tests (tightness) Bolted connections Finite element method, VDI 2230
- 222. Yaw system concepts and designs 193 dimensioning example, which is set to 2,995 mm here. Based on similar yaw bear- ings with a ball diameter of 60 mm, the bearing dimensions are defined (Table 4.20). With yaw bearings of this size, M36 screws are usually used for the attachment to the companion structure. The maximum possible number of screws results from the bolt circle diameter and the tightening tools used. Manufacturing tolerances must be consid- ered with a safety margin in the required minimum distance between two screws. The verification of the bolted connections will show how many screws are actually required. As described in section 4.3.2, removable plugs in the inner bearing ring are needed to insert the balls. For this reason, one bore hole is used to fix the filling plugs and cannot be used for bolting the inner ring to the main frame. When determining the number of screws on the inner ring, the number of grease inlets required should also be considered. For an even distribution, the number of screws should be divisible by the number of grease inlets per raceway. It might also make sense to position the grease inlets between the 14 yaw brakes to ensure good accessibility. Table 4.20 Yaw bearing Parameter Value small drive Value large drive General Bearing type Double-row four-point contact ball bearing with external gearing Double-row four-point contact ball bearing with external gearing Guiding system Spacer Spacer Ball diameter 60 mm 60 mm Number of balls per row 152 152 Amount of grease in the bearing approx. 14 kg approx. 14 kg Dimensions Inner diameter 2,995 mm 2,995 mm Inner bolt circle diameter 3,085 mm 3,085 mm Raceway diameter 3,212.5 mm 3,212.5 mm Outer bolt circle diameter 3,340 mm 3,340 mm Outer diameter 3,500.8 mm 3,516 mm Ring heights 217 mm 217 mm Bearing height 230 mm 230 mm Weight approx. 3,500 kg approx. 3,650 kg Bolted connections Screw size M36 M36 Bore type – inner ring Through hole Through hole Bore type – outer ring Threaded hole (blind hole) Threaded hole (blind hole) Number of screws – inner ring 125 (pitch 126) 125 (pitch 126) Number of screws – outer ring 136 136 Gearing Gear module 16 mm 20 mm Number of teeth 216 173 Tooth width 160 mm 200 mm Profile shift coefficient 0.5 0.5 Tip reduction −1.6 mm −2.0 mm
- 223. 194 Wind turbine system design Yaw bearing weight and amount of grease are usually determined by the bearing manufacturer. Here, an estimate is made based on the CAD model and experience. In the past, yaw bearings were designed purely analytically for extreme loads. This approach is no longer up to date. The deformation of the companion structure and in particular changes in stiffness have a major influence on the rolling element forces and contact angles as well as on bearing ring and bolt stresses (see below). An analytical calculation should only be used as part of a pre- dimensioning of the yaw bearing. The following shows how the yaw bearing raceway can be verified analytically. The extreme load case with the highest yaw bearing Mxy is used (Table 4.2), as this leads to the highest rolling element force. First, the maximum ball force is calculated using (equation 4.6). The maximum contact angle is usually assumed to be 60°. The influence of the companion structure is considered by the excess load factor by which the maximum ball force is conser- vatively increased. The results are shown in Table 4.21. Q = 4 Mxy Dpw 2 Z + ˇ ˇFZ ˇ ˇ 2 Z sin ˛ + 4 ˇ ˇFxy ˇ ˇ 2 Z cos ˛ ! K (4.6) • Dpw raceway diameter [m] • Fxy yaw bearing Fxy [kN] • Fz yaw bearing Fz [kN] • K excess load factor • Mxy yaw bearing Mxy [kNm] • Q maximum ball force [kN] • Z number of balls per raceway • maximum contact angle [°] [–] [–] Table 4.21 Yaw bearing raceway verification Parameter Value Loads Yaw bearing Mxy 26,950 kNm Yaw bearing Fxy 760 kNm Yaw bearing Fz 4,560 kNm Bearing and calculation data Raceway diameter Dpw 3,212.5 mm Ball diameter Dw 60 mm Number of balls per row Z 152 Max. contact angle α 60° Constant cp 934.45 (N/mm²)2/3 Excess load factor K 1.30 Calculation results Maximum ball force Q 214.10 kN Permissible surface pressure pperm 4,200 N/mm² Maximum surface pressure p 3,648.14 N/mm² Safety factor 1.15
- 224. Yaw system concepts and designs 195 Then, the contact stress according to Hertz can be calculated using (equation 4.7). The required constant cp results from the radii and the material properties of the roll- ing element and raceway. A calculation is not discussed in detail here. p = cp 3 s Q D2 w (4.7) • cp constant [(N/mm²)2/3 ] • Dw ball diameter [mm] • p contact stress [N/mm²] • Q maximum ball force [N] The maximum permissible contact stress is usually defined by the bearing manufacturer considering material, surface hardness and hardness depth. Often, the permissible surface pressure of 4,200 N/mm² from ISO 76 is used. The static safety factor is the ratio between the permissible and actual contact stress. According to the DNV standard [1], it shall be at least 1.1. The IEC standard [3] allows a safety factor of 1.0. The results are summarised in Table 4.21. The static safety factor is 1.15, which should be sufficient. However, the finite element calculations considering the com- panion structure need to confirm this. The fatigue and static strength analysis according to the IEC or DNV stan- dard is carried out using finite element calculations, considering the influence of the companion structure (Table 4.22). The focus is on the raceways and the bolted connections. The structural integrity of the bearing rings should be verified as well. For a sufficiently high number of unit load cases, the ring stresses are determined using Table 4.22 Yaw bearing – fatigue and static strength analysis (Sub-)component Standards, guidelines, methods and criteria General IEC 61400- 1 [3], DNVGL- ST- 0361 [1] Bearing rings Finite element method (ring stresses and seal gap widening) Raceways Finite element method (rolling element forces, contact angles, Hertzian pressure, pressure ellipse truncation, plastic deformation, stresses in raceway edge, core crushing and lifetime) Gears ISO 6336 series (surface durability and tooth bending strength) Bolted connections Finite element method, VDI 2230, IEC 61400- 1 [3], DNVGL-ST-0361 [1]
- 225. 196 Wind turbine system design FEM calculations. These ring stresses form the basis for the fatigue strength analy- sis based on IEC, DNV or other standards using the time series. Special software programs are usually used here. The static strength analysis is carried out with FEM calculations for the yaw bearing ultimate loads. Rolling element forces and contact angles are calculated. In submodels, the pressure ellipse, plastic deformation and stresses in the raceway edge are then calcu- lated for the most stressed balls. The toothing is usually verified according to the ISO 6336 series. Due to load- depending factors in the gear calculation, it is recommended to carry out the cal- culations with the load collective according to ISO 6336- 6 for this highly stressed gearing. The calculation with equivalent torques usually leads to less precise safety factors. A finite element calculation of the yaw bearing of the IWT- 7.5- 164 wind turbine cannot be carried out here. However, the general influence of the companion struc- ture is shown below as an example. With an ideally stiff companion structure and a pure bending moment load on the yaw bearing, the curve of the rolling element forces over the circumference of the bearing ideally corresponds to a sine or cosine curve. In reality, however, the companion structure and the bearing itself are flexible. In addition, the stiffness of the companion structure is not constant over the circumference. Stiffness jumps can occur, particularly with the machine carrier. As a result, there is an uneven load distribution between the two bearing race- ways and larger ball forces and contact angles can occur compared to the rigid cal- culation (Figure 4.35). This leads to higher Hertzian pressures. Figure 4.35 Ball forces and contact angles (example) (Liebherr- Components Biberach GmbH)
- 226. Yaw system concepts and designs 197 A truncation of the pressure ellipses at the raceway edge can also occur. This means that the pressure ellipse cannot form completely (Figure 4.36). As a result, there can be higher surface pressures and high stresses in the edge of the raceway. This can result in plastic deformation and chipping of the raceway edge. 4.4.6 Step 2d: dimensioning of the yaw drive system After the yaw brake system and yaw bearing have been designed, the required num- ber of yaw drives can be determined. In principle, there are three criteria that must be met: • • Holding torque: The yaw drives must provide sufficient holding torque so that the requirements from section 4.4.2 are met. • • Driving torque: The yaw drives must provide sufficient driving torque so that the requirement from section 4.4.2 is met. The required driving torque should be redefined based on the actual yaw brake friction during yawing. • • Yaw bearing gear strength verification: The yaw bearing is usually the weakest part. The minimum safety factors according to the IEC standard [3] must be fulfilled. The required yaw system parameters are summarised in Table 4.23. It also shows that the nominal yaw speed is in the desired range stated in Table 4.12. The difference between the required total holding torque and the holding torque of the yaw brakes results in the required holding torque of the yaw drives (Table 4.24). The holding torque per yaw drive is given by the multiplication of motor brake torque, total ratio and static efficiency. With these two values, the required number of drives can be easily determined. The yaw brake friction torque during yawing is given in Table 4.18. The yaw bearing friction depends on the loads acting on the yaw bearing. It can be estimated using the equations from the bearing manufacturers. For reasons of simplification, a Figure 4.36 Pressure ellipse truncation
- 227. 198 Wind turbine system design constant and conservative value of about 150 kNm is assumed here for the yaw bear- ing friction torque. The total friction torque is then about 1,320 kNm (Table 4.25). It is thus higher than the yaw moment Mz more than 65% of the time. In this example, the 99.75% criterion is used to determine the required maxi- mum driving torque. That means that the motor torque should be at least 99.75% of the time higher than the yaw drivetrain torque. This leads to a required motor torque of at least 7,560 kNm (Figure 4.37 and Table 4.25). The maximum torque that can occur in the yaw drivetrain, however, is 16,920 kNm. The maximum driving torque per yaw drive is given by the multiplication of maximum motor torque, total ratio and dynamic efficiency. The required number of Table 4.23 Yaw system – system parameters Parameter Value small drive Value large drive Gear ratio Gearbox ratio 945 1,254 Pinion – no. of teeth 11 14 Yaw bearing – no. of teeth 216 173 Total ratio 18,556.36 15,495.86 Efficiency Number of gearbox stages 4 4 Efficiency per stage 0.97 0.97 Efficiency pinion/yaw bearing 0.97 0.97 Dynamic efficiency 0.859 0.859 Static efficiency 1.000 1.000 Speed Nominal motor speed 930 rpm 950 rpm Nominal yaw speed 0.050 rpm 0.301°/s 0.061 rpm 0.368°/s Table 4.24 Yaw drive system – holding torque criterion Parameter Value small drive Value large drive Required holding torque Required total holding torque ≥21,120 kNm ≥21,120 kNm Holding torque of yaw brakes 10,494.25 kNm 10,494.25 kNm Rqd. holding torque of yaw drives ≥9,625.75 kNm ≥9,625.75 kNm Available holding torque Motor brake torque 40 Nm 80 Nm Total ratio 18,556.36 15,495.86 Static efficiency 1.000 1.000 Holding torque per yaw drive 742.25 kNm 1,239.67 kNm Number of yaw drives Required number of yaw drives ≥12.97 ≥7.76 Chosen number of yaw drives 13 8
- 228. Yaw system concepts and designs 199 drives results from dividing the required maximum driving torque by the maximum driving torque per yaw drive (Table 4.26). The previously determined number of yaw drives serves as the starting point for the fatigue strength analysis of the yaw bearing teeth. The calculation is carried out with the load spectrum (Figure 4.37), as this leads to more precise safety fac- tors than the calculation with equivalent torques. The equivalent torques are given in Table 4.27 for information only. The load spectrum for the individual gear mesh needs to be truncated. Otherwise, the required safety factors (Table 4.27) cannot be achieved. In this example, the truncation torque is determined using a static gearing calculation, so that the mini- mum safety factor for pitting is higher than 1.10. Table 4.25 Required driving torque Parameter Value Friction torque during yawing Yaw brake friction torque 1,166.03 kNm Yaw bearing friction torque ≈150 kNm Total friction torque ≈1,320.00 kNm Required driving torque Criterion 99.75% Required max. driving torque ≥7,560 kNm Max. yaw drivetrain torque 16,920 kNm Figure 4.37 Yaw drivetrain torque (LDD, absolute values)
- 229. 200 Wind turbine system design Table 4.26 Yaw drive system – driving torque criterion Parameter Value small drive Value large drive Required driving torque Required max. driving torque ≥7,560 kNm ≥7,560 kNm Available driving torque Max. motor torque 34 Nm 72 Nm Total ratio 18,556.36 15,495.86 Dynamic efficiency 0.859 0.859 Max. driving torque per yaw drive 541.79 kNm 958.09 kNm Number of yaw drives Required number of yaw drives ≥13.95 ≥7.89 Chosen number of yaw drives 14 8 Table 4.27 Yaw drive system – yaw bearing gear strength verification Parameter Value small drive Value large drive Number of yaw drives 14 8 Fatigue assessment Truncation torque 625 kNm 1,225 kNm Maximum LDD torque 1,208.57 kNm 2,115.00 kNm Torque exceedance factor 1.93 1.73 No. of hours above truncation limit 8.52 hours 0.05% 2.10 hours 0.01% Angle between the yaw drives 14° 17° Load cycle factor 59.451 44.768 Number of load cycles 3.1331E+06 2.8871E+06 Equivalent torque – pitting 290.133 kNm @ 3.1331E+06 cycles 511.555 kNm @ 2.8871E+06 cycles Equivalent torque – bending 327.179 kNm @ 3.0000E+06 cycles 578.339 kNm @ 2.8871E+06 cycles Type of calculation Collective Collective Required safety factor – pitting ≥1.10 ≥1.10 Safety factor – pitting 1.105 1.106 Required safety factor – bending ≥1.25 ≥1.25 Safety factor – bending 1.524 1.343 Ultimate assessment Holding torque per yaw drive 742.25 kNm 1,239.67 kNm Ultimate torque – holding (+20%) 890.71 kNm 1,487.60 kNm Max. driving torque per yaw drive 541.79 kNm 958.09 kNm Ultimate torque – driving (+10%) 595.97 kNm 1,053.90 kNm Ultimate torque at yaw bearing 890.71 kNm 1,487.60 kNm Required safety factor – pitting n/a n/a Safety factor – pitting 0.926 1.004 Required safety factor – bending ≥1.20 ≥1.20 Safety factor – bending 1.679 1.596
- 230. Yaw system concepts and designs 201 If a load level of the load spectrum were to exceed the truncation torque, the minimum safety factor for the fatigue strength analysis would be below 1.10. In 20 years, the truncation torque is only exceeded for 8.52 hours, which is 0.05% of the time. It must be ensured that the maximum driving torque per yaw drive (Table 4.26) is smaller than the truncation torque (Table 4.27). A sufficient safety distance between the torques depending on the yaw system concept and the yaw control system should be considered. Overload events need to be detected and overloads in the yaw drive- train need to be avoided. In this example, this is ensured via the frequency converters and other sensors that measure the motor speeds, for example. The arrangement of the yaw drives has an influence on the number of load cycles of the most heavily stressed yaw bearing tooth. In the first calculation, all drives are arranged close together. This is the most conservative variant. In the further course, the arrangement is detailed considering the available installation space. The individ- ual optimisation steps are not shown here. Only the results of the final arrangement are shown in Table 4.27. The fatigue strength analysis confirms the number of yaw drives. Fourteen small or eight large yaw drives are needed to fulfil the requirements. There are higher safety factors for the smaller drives. The yaw system configurations seem to be well balanced. All criteria lead to almost the same number of yaw drives. A higher partial brake pressure would lead to higher loads and thus to a higher number of yaw drives in this criterion and in the driving torque criterion. The static strength analysis is done with the maximum holding torque per yaw drive. However, the upper tolerance of the motor brake torque of 20% is consid- ered. Since the motor brake was chosen to match the yaw gearbox and its pinion, the required safety factors are achieved. The fatigue and static strength analysis of the yaw gearbox is done according to the IEC or DNV standard considering the calculation standards for the different machine elements (Table 4.28). Structural components such as housing and planet carrier are verified via finite element calculations. Table 4.28 Yaw gearbox – fatigue and static strength analysis (Sub-)Component Standards, guidelines, methods and criteria General IEC 61400- 1 [3], DNVGL- ST- 0361 [1], bench tests (extreme load test, fatigue load test, measurement of backlash and torsional stiffness) Gears ISO 6336 series (surface durability and tooth bending strength) Bearings ISO 76, ISO 281 Shafts DIN 743 Shaft-hub connections DIN 5466, DIN 6892, DIN 7190 Bolted connections VDI 2230, finite element method Planet carriers, housing Finite element method, IEC 61400- 1 [3], DNVGL- ST- 0361 [1]
- 231. 202 Wind turbine system design In addition, bench tests are often carried out to validate the fatigue and static load capacity of the gearbox. Backlash and torsional stiffness are also measured. Asynchronous motors and their motor shafts are standardised. So, there is no need for a separate fatigue and static strength analysis (Table 4.29). Depending on the load characteristic and the required operating hours, however, it can make sense to have closer look at the bearing and the shaft- hub connection. Torque- speed curve, current- speed curve and the other motor data according to the IEC 60034 series are determined in bench tests. 4.4.7 Step 3: auxiliary systems The design and dimensioning of the auxiliary systems are not discussed in detail here. The power supply is assumed to be given. The hydraulic system and the lubri- cation system for the example yaw system are described in Chapter 7 of this book. In the following, the sensors are briefly discussed in general. In section 4.3.6, possible sensors are presented. The following signals shall be measured in the example yaw system: • • yaw position and yaw speed • • zero- degree position (north position) and end position of the cable loop • • current and voltage via the frequency converter • • motor winding temperature • • hydraulic pressure at the yaw brakes • • status information of the lubrication and hydraulic system These signals ensure the proper function of the yaw system. In addition, over- load events and failures can be detected. 4.4.8 Summary The yaw system of the IWT- 7.5- 164 wind turbine is dimensioned in the previous subsections. Some design and calculation aspects are discussed in more detail. However, a complete and detailed design is not possible within the scope of this book. The example dimensioning is therefore to be understood as a first insight Table 4.29 Yaw motor – fatigue and static strength analysis (Sub-)component Standards, guidelines, methods and criteria General IEC 61400- 1 [3], DNVGL- ST- 0361 [1] Motor IEC 60034 series, bench tests Bearings ISO 281, ISO 76 Bolted connections VDI 2230 Shaft-hub connections DIN 6892
- 232. Yaw system concepts and designs 203 into a possible design process. In the following, the results are briefly summarised. Figures 4.38–4.41 show the developed yaw system with the small yaw drives in various representations. Figure 4.38 Yaw system – top view Figure 4.39 Yaw system – diagonal view 1
- 233. 204 Wind turbine system design Table 4.30 summarises the yaw system parameters at the system level. It shows that the requirements for yaw speed, holding torque and driving torque are fulfilled. Yaw slippage events are very unlikely. The maximum driving torque is very rarely exceeded. However, it must be noted that the torque required for accelerating the nacelle is not considered. The small yaw drives can be arranged in any way. The maximum allowable width of the machine frame is not exceeded. The large yaw drives, however, can- not be arranged on the sides of the nacelle. Otherwise, the machine carrier width is exceeded. Therefore, they can only be arranged in the rear area. For the sake of completeness, the yaw moments of inertia are also shown in Table 4.30. The magnitude of the moments of inertia of the tower head and yaw drives is similar. In the yaw system configuration with the small yaw drives, the inertia of the yaw motors is smaller than the tower head inertia, with the large yaw drives it is the other way round. Figure 4.40 Yaw system – diagonal view 2 Figure 4.41 Yaw system – sectional view
- 234. Yaw system concepts and designs 205 References [1] Machinery for wind turbines. DNV GL standard DNVGL- ST- 0361, DNV GL AS; 2016 Sep. [2] Jenkins N., Burton T., Bossanyi E., Sharpe D., Graham M. Wind Energy Handbook. Third edition. Hoboken, NJ: Wiley; 2021. Available from https:// onlinelibrary.wiley.com/doi/book/10.1002/9781119451143 [3] Wind energy generation systems – part 1: design requirements. Geneva, Switzerland: IEC standard 61400- 1:2019- 02, International Electrotechnical Commission; 2018. [4] Load and site conditions for wind turbines. DNV GL standard DNVGL- ST- 0437, DNV GL AS; 2016 Nov. Table 4.30 Yaw system – summary Parameter Value small drive Value large drive General Number of yaw drives 14 8 Number of yaw brakes 14 14 Total ratio 18,556.36 15,495.86 Required yaw speed 0.25–0.50°/s 0.25–0.50°/s Nominal yaw speed 0.050 rpm 0.301°/s 0.061 rpm 0.368°/s Dynamic efficiency 0.859 0.859 Static efficiency 1.000 1.000 Centre distance 1,831.5143 mm 1,889.2759 mm Max. machine frame width 4.2 m 4.2 m Actual machine frame width Approx. 4.15 m Approx. 4.15 m Holding torque Required total holding torque ≥20,120 kNm ≥20,120 kNm Total holding torque 20,885.81 kNm 20,411.60 kNm Required partial holding torque ≥8,720 kNm ≥8,720 kNm Holding torque of yaw brakes 10,494.25 kNm 10,494.25 kNm Holding torque of yaw drives 10,391.56 kNm 9,917.35 kNm Driving torque Required max. driving torque ≥7,560 kNm ≥7,560 kNm Maximum driving torque 7,585.05 kNm 7,664.73 kNm Exceedance of maximum torque 0.17% 0.17% Rated driving torque 4,237.79 kNm 4,280.28 kNm Exceedance of rated torque 9.97% 9.97% Inertia Inertia of tower head (rotor + nacelle) 5.531E+07 kgm² 5.531E+07 kgm² Inertia of yaw motors 4.773E+07 kgm² 5.859E+07 kgm² Ratio yaw motors/tower head 0.86 1.06
- 235. 206 Wind turbine system design [5] EnBW Energie Baden- Württemberg A.G. lernt schwimmen in der ostsee [online]. 2020. Available from https://www.enbw.com/unternehmen/presse/ windkraftan- lage-nezzy-lernt-schwimmen-in-der-ostsee.html [Accessed 2 Apr 2022]. [6] Stubkier S. ‘Hydraulic Soft Yaw System for Multi MW Wind Turbines’. [Ph.D. dissertation]. Aalborg University, Denmark, Institute of Energy Technology. [7] Fraunhofer Institute for Wind Energy Systems. IWES wind turbine IWT- 7.5- 164 rev. 2.5. Bremerhaven: Fraunhofer Institute for Wind Energy Systems; 2017.
- 236. 1 Fraunhofer- IWES, Fraunhofer- Institute for Wind Energy Systems, Bremerhaven, Germany 2 Institute for Electrical Drives, Power Electronics and Devices, University of Bremen, Bremen, Germany Chapter 5 Drivetrain concepts and developments Jan Wenske1,2 Traditionally, the drivetrain (DT) of a wind turbine (WT) is defined as the rotating, mechanical linkage, transmitting torque between the wind rotor as an entire subsystem, which includes the blades, the hub with blade bearings, and the pitch system, toward the generator. The generator converts the mechanical into electrical energy by the use of electromagnetic fields, forces, and induction between the rotor and the stator. Besides the torque and the dead- weight forces, the DT is furthermore exposed to parasitic loads due to aerodynamic and mechanical loading, mainly from the WT rotor. These can be thrust forces, imbalances, gravity, centrifugal and gyroscopic loads, as well as mostly unwanted axial and radial generator air- gap forces, reaction, or constraining forces from the respective DT suspension system and supports. Those are designed to transfer DT reaction forces toward the fixed structure of the nacelle, the machine bed, also referred to as the main frame. The focus of this chapter is on general DT concepts for horizontal axle WTs, which are common today. After a few more general explanations on the subject, this chapter provides an overview of the DT concepts already implemented and their variants. This is fol- lowed by comments on basic design rules, technical characterization, discussion of platform concepts, and scalability. The various developments for onshore and offshore applications of leading WT manufacturers are described in more detail, with reference to the presented basic DT concepts and in some cases compared with each other regarding performance indicators. The chapter concludes with a brief outlook. 5.1 Fundamentals A modern and state- of- the- art definition of the entire DT of a WT includes its mechanical as well as its electrical part from the hub interface to grid connection.
- 237. 208 Wind turbine system design Whereby the main electrical parts such as generator, main converter, electrical fil- ters, and transformer forms a subsystem, the so- called electrical DT. This approach mainly results from the strong dynamic interaction between the mechanical and the electrical parts with their inherent small time constants. The relevant frequency range of electromagnetic oscillations and thus the cross coupling to the mechanics is from to 2.5 kHz. The cross- coupling characteristics are defined by the air- gap forces, current, and magnetic induction within the generator. This and the following sections summarize some general definitions and essential basics of DT concepts such as typical components, briefly functional descriptions, and an introduction to the simplified classification system for DTs, based on specific technical characteristics as well as some crucial points of the DT design. Indeed, the complete design of a new DT system for a modern multi- megawatt WT is a very complex process, and many technical requirements have to be considered during conception, design, and optimization, not only for each component or subsystem (e.g., main shaft support, gearbox) separately but also for the entire system. In this chapter, the focus is more on the entire system then on subsystems or specific components; for more information on these, please refer to the corresponding chapters of this book. The entire DT of a WT consists of the following main functional modules; depending on the DT concept, some of them are optional: • • rotor main shaft suspension/bearings (refer to Chapter 10) • • main shaft, sometimes referred to as rotor shaft (refer to Chapter 9, 10) • • rotor lock (for details, refer to Chapters 6 and 7) • • low- speed shaft coupling, flexible or rigid (e.g., shrink disk and flange) • • gearbox with support (refer to Chapter 6) • • rotor brake (service brake) • • safety clutch (referred to as torque limiter) • • high- or medium- speed shaft coupling (flexible or rigid) • • generator with support and the electrical subsystem of the DT (refer to Vol. 2) Also, the rotor with the rotor hub, usually made from nodular cast iron (incor- porate blade bearings and pitch system, refer to Chapter 3), more precisely the mass and stiffness distribution as well as damping properties of this rotor system, which induce by far most of the loads for the WT and DT, play an essential role on DT dynamics but formally are not parts of it. Therefore, for load and dynamic analysis of the DT, the eigenfrequencies and torque- relevant modes of the entire rotor and its coupling to the DT must be considered. The main functions of the already named, single modules within the DT will be briefly explained as follows. a. The main shaft (typically a hollow shaft, casted or forged), sometimes referred to as low- speed shaft, transfers rotor torque to either the gearbox or the gen- erator. To transfer radial, lateral, and axial forces as well as bending moments toward the nacelle structure, the main shaft is supported by a suspension system (main bearing arrangement).
- 238. Drivetrain concepts and developments 209 b. The rotor, respectively, main shaft suspension system is an arrangement of roller bearings for interfacing with the nacelle structure and the low- speed (with rotor speed) rotating parts of the DT. The bearing configuration depends on whether the DT concept has a dedicated main shaft, or the rotor hub is linked directly, respectively, with a kind of spacer/adapter within the DT. c. The rotor lock (refer to Chapters 6 and 7), due to the form- locked joint (e.g., by electrically or hydraulically actuated bolts) of the nacelle structure and rotor, serves to safely carry out maintenance and repair work (e.g., gear replacement and service work in the rotor hub). There are various concepts for positioning the rotor before the rotor lock is triggered (targeted braking from idle or using an auxiliary drive on the gearbox). d. The low- speed coupling as a flexible (for non- torque loads refer to company publications of Co. CENTA and Co. Geislinger) but torsional rigid linkage on the rotor side of the DT depends on the concept, and thus the usage is optional. However, at least a rigid coupling is usually required for the linkage between the main shaft and the generator or gearbox, in any case. e. The generator high- or medium- speed coupling is a flexible (for non- torque loads) but torsional rigid coupling on the gearbox output side, sometimes referred to as high- speed shaft. The usage depends on the concept and is optional. Usually, this coupling also provides electrical isolation, due to a glass fiber spacer or ceramics plates, to avoid the occurrence of bearing currents. f. The usage of a safety clutch (torque limiter) depends on the concept and is therefore optional. Normally, it is only applicable in multi- stage gearbox con- cepts placed on the gearbox output side to protect the gearbox from extreme torque peaks that can be induced by electric faults or grid events. Figure 5.1 Schematic of a WT nacelle with “classic” geared DT, from the hub with pitch system to generator, © Fraunhofer- IWES
- 239. 210 Wind turbine system design g. Depending on the concept, the electrically or hydraulically actuated rotor brake can be installed on the low- speed or high- speed side of the DT. It is used to brake the rotor from low speed or idle operation, e.g., in the event of a grid fault, until standstill. Once stationary, it also acts as a parking brake—keep atten- tion—this is not equivalent to a rotor- lock system described above. h. The gearbox normally steps up the rotational speed from the wind rotor and transmits torque to one or splits between more generators (refer to Chapter 6), and therefore is usually applied one to four gear stages with a reaction force support at the housing (Figure 5.1). Depending on the concept, the gearbox has a flange connection or torque arms to handle reaction forces. The gearbox in a DT is optional (Figure 5.2). Usually, the whole DT of WT is tilted upward in the rotor direction under an angle of 4–6°, to serve enough space between blades and tower under operation. All DT schematics shown in section 5.2 have a DT tilt of 5°. 5.2 Drivetrain concepts 5.2.1 Drivetrain diversification and classification The simplest form of characterization or differentiation of WT DTs is that in gearless and geared DT concepts, regardless of how many gear stages are used. Concepts with continuously variable- ratio transmission were developed and built as prototypes or in smaller series but have not been established on the market until now. These concepts combine a mechanical gear transmission solution with a hydraulic (the DeWind D8.2 WT from 2006 was equipped with such a hydrodynamic WinDriveTM [1] transmis- sion) or electrical actuator for setting the continuously variable ratio (Co. SET, refer to Chapter 6). Since such solutions are no longer offered on the turbine market today, Figure 5.2 Schematic of a WT nacelle with gearless, direct drive from the hub with pitch system to generator, © Fraunhofer- IWES
- 240. Drivetrain concepts and developments 211 detailed explanations are not given here. The interested reader is referred to specific literature and company publications. Another special form of variable transmission by means of the purely hydraulic solution using a closed- circuit hydraulic system with a variable displacement pump/motor combination is also not further considered, too. However, these designs did not get beyond a demonstrator status. Early WTs had rather “classic” DT concepts with three- stage gearboxes to step up the slow rotational speeds of the wind rotor to rated output speeds in the range of 1000–1800 rpm, corresponding to the main grid frequency of 50 or 60 Hz, for direct grid coupled standard generators, either induction generators (IGs) or synchronous generators (SGs) with fixed four, six, or eight poles or pole- changing designs. Those turbines were very limited regarding rotor speed variations and belong to the class of stall- controlled, so- called fixed- speed WTs. In these early years of commercial wind energy utilization (1980s) with less turbine capacity, this type of DT had advantages; the main reason for combining multi- stage gearboxes and so- called high- speed gen- erators was the availability of off- the- shelf industrial products for such components. Since the 1990s, more and more variable- speed WTs using full or partial power converters for grid coupling and active pitch control became a standard. With the followed, fast- growing size (rotor diameter and rated power) of turbines more and more frequently failures especially within suspension system and the gearboxes cropped up and revealed the necessity of more detailed DT analysis due to the complexity of site conditions, loads, and entire system dynamics [2]. In contrast to the beginnings of the DT designs in the 1980s, for over 30 years, more and more high- fidelity models (FE Finite Element, Multi Body Simulation incl. flex- body parts, and multi- domain/physics models) have been used, which now include nearly the entire mechanical DT properties down to the rolling contact of the bearings, tooth contact of the main gears, and non- linear, distributed stiffness and damping details of individual modules. However, this does not mean that the usage of such tools automatically leads to reliable and more efficient designs. Broad experience and comprehensive technical understanding remain indispens- able and key elements for this. But these sophisticated models are nowadays standardly applied in the design process for WT DTs, for load assessment and strength verification under stationary, transient, and long- term dynamic operating states [so- called DLC, design load cases, refer to Chapters 1 and 2, (IEC 61400- 1, GL Rules and Guidelines Part 1 Wind turbines; 2010 and Part 2 Offshore Wind Turbines; 2012, DNV GL loads and site conditions 2016 technical guidelines)], as well as for controller synthesis, structural- borne noise assessment, and optimiza- tion in tailored variants. In the first time of extensive multi- megawatt turbine development, since the late 1990s, the Original Equipment Manufacturers (OEMs) as well as the suppliers react with specialized component designs for single types of turbines. So, in general, the failure rates of the mechanical DT system decreased continuously with outliers upward due to the extensive growth of the new WT sizes, introduction of new DT concepts, and still lacking knowledge or underestimated parasitic effects. Increasing rotor diameter causes inevitably exponentially growth of mechanical loading by forces, bending moments, and torques applied on the DT in general, as well as the
- 241. 212 Wind turbine system design resulting deflections and displacements. The even higher loading on the mechanical DT components required even more innovative and reliable concepts and compo- nents within even shorter design periods. The development of gearless DTs for WTs started in the early 1990s. The German manufacturer ENERCON designs Direct- Drive (DD) WTs, using multipole electrical excited synchronous generators. So, the first E30/E40 turbines marked a historical point from that point of view. Although these early gearless DTs were typically heavier than those with a classical geared DT, they seemed to have superior reliability at that time. The last stage of DD technology nowadays uses permanent magnet excited generators [e.g., OEMs Siemens Gamesa Renewable Energy (SGRE), Goldwind, Lagerwey, and Vensys], originally introduced by the inventor and wind pioneer Prof. Klinger in early 1997 with the Genesys 600- kW turbine and developed further on to a first- megawatt series turbine by the company Vensys Energy. Later on, this IP serves the Chinese OEM Goldwind as a technical basis. This general differentiation leads to the simple DT classification mentioned above, which, however, is less suitable for directly comparing the various designs. Here a smarter differentiation seems helpful in order to correctly compare the applied DT concepts of the OEMs and the technical details. Of course, finally, regardless of the respective DT design, it is the entire turbine design, which counts as a whole, e.g., in terms of tower head mass, serviceability, or technical reliability. In order to describe different types of DTs and to compare their properties, the following clas- sification, according to four main characteristics, can be useful. 1. gear transmission (corresponds to gear stages/ratio in general) • No gear stage (i=1:1) → gearless DT/referred to as DD • one to two gear stages (i=1:9–50) → medium- speed DT/referred to as Hybrid- Drive Remark: Latest OWT DT designs with three gear stages (all planetary, i=50–100) sometime still referred to as Hybrid- Drive • three to four gear stages (i=1:60–150) → high- speed DT / “classic” Geared- Drive (GD) • Variable- ratio gear stage → variable- ratio DT (eg, “WinDriveTM ” from Voith) 2. Generator type (corresponds to type of electrical DT): • IG + full rated converter • doubly- fed IG (DFIG)+ partially rated converter • synchronous generator build as permanent magnet excited (PMSG) or as electrical excited (EESG) type + full rated converter 3. Rotor/-Main shaft suspension • three- point suspension → one main bearing at the front (upwind side), second support bearing in the gearbox • four- point/dual suspension → classic dual bearing support, two separated bearings • (dedicated) bearing unit → dual bearing support, two bearings with one shell • single bearing → only one bearing support, sometimes referred to as “Moment bearing”
- 242. Drivetrain concepts and developments 213 4. Level of integration: • non- integrated → all components separated, no functional dual use of parts • low- integrated → at least one functional dual use of DT parts • semi- integrated → multiple functional use of DT parts → high compactness of the DT • fully integrated → highest integration, hardly no single functional use of DT parts As already mentioned, roughly since the early 2000s, theWTmanufacturers have entered the multi- megawatt rated power class, but that does not mean that a favored DT concept has emerged so far. But on the contrary, until mid of the 2010s, the num- ber of DT concepts and variants further increased dramatically, which resulted in a high diversification (refer to Figures 5.3–5.5). During that time, manufacturers have mainly defined themselves and their innovative capacity by very special concepts. The lot sizes of non- variable parts were correspondingly low, sometimes with a negative impact on the turbine costs and reliability. Not much later, since the mid of the 2010s, there was a clear trend appearing, especially on the onshore market that the WT became more of a commodity product with a rather low entry level, in terms of risk, price, and technology. In consequence, even more players were entering the market (e.g., Chinese OEMs were catching up). Various, quite PR- driven terms (e.g., “HybridDriveTM ,” “Compact Drive,” “PureTorqueR ) made the rounds, and for the last time, some OEMs wanted to set themselves apart from one another in this phase with explicit unique designs. Nowadays, standardization, modularization, and platform strategies are common wordings also in the wind industry, whereas these were frowned at OEMs in the former times and were sometimes ridiculed, when it was still being discussed in the early 2010s. Thus, a design classification based on at least some design features promise to bring more order to this diversity and can help to compare various DT designs in a better way. Figure 5.3 Diversification of WT DT concepts [3]
- 243. 214 Wind turbine system design So, the main reason to shift from more technology to a consistently economical- optimized- based view was simple costs and the competition for mar- ket shares. Within a MAKE Consulting report of 2011 (Figure 5.3), they called it, “The need to push wind power towards grid parity while coping with the size and cost increases associated with larger turbines is the core impetus for the diver- sity of drivetrain choices available today.” On the other side, a strong technologi- cal driver was the upcoming offshore wind applications. As an example, some of the OEMs and system suppliers sought to combine the predicted advantages of DD (higher overall reliability and high DT compactness) and “classic” geared high- speed DTs (track record, availability and experience of existing transmission suppliers, and lower component weight) by developing medium- speed gearbox DTs, respectively, sometimes referred to as Hybrid- Drives. These Hybrid- Drive designs should end up to reliable, lightweight, and compact DTs without using the third (high- speed) gear stage, typically a spur gear stage, of classical WT gear- boxes, which is presumed to as the most error- prone gear stage. At the electrical part of these DTs, mid- size, medium- speed, reliable synchronous generators are applied. In case of utilizing PM technology, this means the same performance Figure 5.4 Combinations of main DT classes and different electrical subsystems [5]
- 244. Drivetrain concepts and developments 215 Figure 5.5 Technical combinations of different key elements for DT concepts [4]
- 245. 216 Wind turbine system design with a significant lower need for critical permanent magnet material, in terms of costs and supply chain, compared to DDs. Transmission suppliers Moventas and Winergy introduced semi- integrated, hybrid (medium- speed) DT solutions nearly at the same time in the years 2012 and 2013. It should be recognized that the engineering company Aerodyn Engineering introduced the basic principle for the Multibrid (later Areva) M5000, a very early multi- megawatt OWT with 116 m rotor diameter, years before in 2004, and thus marked a historical milestone for these kinds of geared DT concepts with a very high level of integration and medium- speed generator. The so- called SCD for the “Nezzy” OWT marks their latest design. Although the medium- speed DT concepts never really surpassed the current DD concepts regarding lower tower head mass weight, these types of DTs are, from a pure market point of view and due to their favorable suitability for platform concepts, still one of the favorite concepts for current and next genera- tions of high- capacity WTs. Currently and in the near future, all three main DT concepts (classic GD, Hybrid- Drive, and DD) seem to be remaining in use. Thus, two development processes can be observed: (1) pure economically driven process due to consolidation of OEMs in highly competitive markets (WT as a commod- ity product) and (2) the standardization and modularization process for high- end WTs, mainly driven by manufacturers with a multi- branch industrial background and corresponding experiences. Furthermore, in the last 15 years, somewhat gen- eralized, the onshore and offshore DT developments clearly leave a common line of design and technology. 5.2.2 Drivetrain concepts and design principles In the early days of modern wind energy utilization, the requirements for the DT and its design for WTs were initially very functional. The focus in the areas of research and development was on the rotor blades and aero- elastic modeling. Challenges regarding dynamics, parasitic loads, grid events, environmental influences, and suf- ficient service life of the DT were commonly considered by a supposedly robust and oversized component design. However, the retroactive effects of the entire DT concept in the sum of its parts and more or less standard industrial components on the overall turbine behavior tended to be underestimated. Particularly, this becomes obvious in the periods of strong turbine growth (in terms of rotor size and rated power). With the spread of modern simulation and design tools, the possibilities for analysis and optimization had expanded signifi- cantly and thus supported specific component designs for dedicated use in WT design. This subsequently led to higher power and torque densities of components as well as more highly integrated solutions with components that needed match- ing characteristics in detail with each other. The optimized design of components for WTs is quite demanding; on the one hand, it does not seem to differ that much from other special applications (railway, marine, aviation, etc.), but on the other hand, it is not really comparable to any of these. The technical and economic bound- ary conditions for WT applications (refer to Figure 5.6) are partly completely
- 246. Drivetrain concepts and developments 217 different (e.g., acceptable material costs, fatigue strength, service intervals, service life requirements, dynamics). Of course, basic rules of technical mechanics and mechanical engineering should be considered for the basic DT conception and design. Especially, influ- ences of parasitic stationary and dynamic loads as well as possible load- dependent deformations and displacements of machine parts and structural components should always be analyzed and taken into account. 1. The suspension system of the DT components and the entire support and fitting system should be designed statically determined. 2. If (1) is not applicable or practicable, additional flexible elements or assemblies should be integrated, so that statically overdetermined DT parts can be structur- ally separated into statically determined, coupled ones. 3. In case of unavoidable static over- determination, displacements, deformation, tolerances, and bearing clearness should be precisely evaluated by means of MBS and FEM analyses and then taken into account accordingly. 4. A WT DT should be neither absolutely rigid nor very flexible designed in all possible six degrees of freedom (DOF). As a more general design recommen- dation, a functional distinction and assessment between the different parasitic loads, force, and torque flows through the DT is advisable in order to optimize dead weight and to minimize constraining forces, structure- borne noise propa- gation, and the tendency to oscillate (resonances, torque, and longitudinal oscil- lations) as well as suppressing whirl effects, which can be excited, e.g., due to gyroscopic rotor forces or unbalances. Figure 5.6 Broad range of general requirements and sometimes contradictory aspects for DTs of WTs
- 247. 218 Wind turbine system design For the introduction and discussion of the possible, especially the most common mechanical DT concepts, we first orientate ourselves on the “geared” or “gearless” design characteristic. The possible combinations with generator–converter concepts will not be discussed in every detail here (for more details, refer to Vol. 2), since these have mostly less effects on the inherent mechanical system properties. 5.2.2.1 “Classic” geared drivetrain concepts (GD) This section starts with the “classic” geared DT concepts for WT, sometimes referred to as high- speed DT, based on a three- to four- stage gearbox with step- up transmis- sion ratios i normally between 60 and 150 toward the generator (refer to Figure 5.7). The gearbox output shaft is called the high- speed shaft, which is also the eponymous property of this basic concept. As already mentioned, generators with four- to eight- pole designs (two- to four- pole pairs, respectively) are common in industry, and thus all different types of generators (IG, DFIG, and PM/EESG) can be used. With some geared drive concepts for multi- megawatt WTs, attempts are made to work with sig- nificantly higher overall gear ratios in combination with power split between several generators, which then rotate at rated speeds of 5 000–7 000 rpm (refer to Chapter 6, “Rapid Wind”—RWTH Aachen University). Figure 5.7 Non- integrated, “Classic” geared high- speed concept with four- point suspension (e.g., GE 2.x/3.x platform, V80, SWT- 3.6, REpower 5M, Senvion 6.xM, Siemens Gamesa SG 2.x,3.x platform)
- 248. Drivetrain concepts and developments 219 There are two more design characteristics of this DT class that remain to be discussed, which are the main shaft, respectively, rotor suspension system, and the level of integration. The non- integrated GD DT concept utilizes a dual- bearing sup- port, sometimes referred to as a four- point suspension system (Figure 5.7), dedi- cated to the rotor main shaft support. Typically, the design philosophy is to realize a common and robust suspension system with a high level of stiffness for all non- torque loads within a statically determined support configuration. Mostly, the first bearing (upwind bearing next to the rotor) is a fixed one and the rear one (down- wind side) is a floating bearing, which allows axial clearance for thermal expansion of components [6]. To handle the torque reaction moments, the gearbox is usually equipped with two torque arms (only one torque arm can cause additional bending under torque loading) and top mounts realized, solely to support itself with respect to rotation about the longitudinal gearbox axis and not to carry the dead- weight forces. Additionally, the trunnion supports are designed slightly elastic (elastomer supports or bushings, respectively) for vibration damping and to provide structural- borne noise decoupling. The lowest level (absence) of integration of this concept is quite obvious, each component has a specific functionality, and dual functionalities are strictly avoided. An advantage of this concept is the good serviceability due to the sepa- rated structures (e.g., for a gearbox change). Disadvantages are the comparatively large mechanically DT dimensions and thus passive and active weights due to large, solidly built machine beds, main shaft, critical tolerance chains, etc. The generator and the gearbox output shaft are connected by a special high- speed coupling that, seen from a pure mechanical aspect, shall compensate for a certain range of static and dynamic misalignments without the formation of constraining forces. Since the generator is mounted separately on flexible and vibration- damping supports on a machine support (can be designed in one or more parts), mechanical tolerances, misalignments during DT assembly, and operating vibrations can be neglected to a certain extent. Thus, constraining forces between components, due to misalign- ments, should be mitigated. In order to clarify the difference, a related “classic” DT concept is presented next, which uses a different type of main shaft suspension, according to Figure 5.8. In order to save dead- weight and installation space, a separated second main shaft bearing (on the downwind side) is not used here. The bearing of the first gear stage (e.g., usually the bearing of the planet- wheel carrier) is used for this instead. Thus, it has a dual functionality in this configuration. The support of the first planetary stage and the function of the second main shaft support point on the downwind side (refer to four- point suspension) lead to a characterization as a low- integrated concept. The main bearing on the upwind side is designed as a fixed bearing, and the gearbox has two torque arms with trunnion bearings [6]. These perform a comparable func- tion as with the four- point suspension, besides carrying dead- weight force here, too. In consequence, this DT concept is more compact, which saves weight (partially compensated by the necessarily reinforced gearbox housing and support) but has a more complex dynamic, due to parasitic loads in the gearbox and possibly induced deformations within the first gear stage. The type of generator connection toward
- 249. 220 Wind turbine system design the gearbox does not change. The typical linkage between the main shaft and the gearbox is usually realized by means of a shrink disk coupling (rigid coupling, refer to Figures 5.7 and 5.8). For a couple of years, so- called bearing units (refer to Figure 5.9 and to Chapter 10) are used to avoid some disadvantages (e.g., tolerance sensitivity, bearing play, and overall length) of the classic separated design of the four- point suspension and at the same time improve the robustness and stiffness properties. These bearing units typically combine two tapered roller bearings, which are usually preloaded against each other, on a compact main shaft with smaller axial dimensions, in a compact, tubular housing. These parts can then be mounted as one unit on the machine bed. The bearing unit has two mechanical interfaces, one on the rotor and one on the generator side. Due to the preloaded installation of the bearings, special care must be taken to consider thermal expansion in order to avoid thermal run- away of the bearings. These bearing units have greatly improved axial rigidity compared to the “classic” four- point suspension and minimize dynamic axial shaft movement, which can affect the gearbox input stage. Furthermore, the total bending stiffness improves compared to separate supports, too. In a first step, the installation principle of the generator just can be taken over from the classic four- point suspension. Figure 5.8 Low- integrated, “classic” geared high- speed concept with three- point suspension (e.g., DeWind D8.x, REpower MD77/MM82, V112- 3.0 MW, Vestas newer 2–3 MW platform, Nordex Delta- and 4000- Series, tailored three- point suspension GE 5.x MW “Cypress” and Vestas 4 MW platform)
- 250. Drivetrain concepts and developments 221 However, this simple implementation of bearing units offers hardly a lot of real advantages. Although it results in a level of low integration (per definition) through the joint use of an outer bearing shell for the two main bearings, this dual use is marginal, DT dimension becomes hardly smaller, and dead weight remains rather high. A more compact DT design can only be achieved, if the gearbox somehow is integrated into the concept, e.g., with a connection to the bearing unit by flange interface; thus, classic torque arms with trunnions bearings can be omitted (refer to Figure 5.10). Experience has shown that, despite a more compact design, there are barely any weight advantages compared to the classic concept (four- point suspension) due to the stiff but comparatively heavy bearing shell and the slightly more complex machine bed structure interfacing. However, the advantages of high rigidity regarding non- torque loads, the total reduced bear- ing clearance and tolerances, as well as inherent good alignment are given and make this DT solution comparatively very robust, especially for high- end, highest capacity DT designs. The high stiffness in all non- torque DOF harbors the risk of generating con- straining forces due to static over determination, if the gearbox is also linked rigidly throughout, so at least elastomer bushing for flange connection should be utilized (refer to Figure 5.10). As an alternative, a flexible coupling is some- times used in realized designs for connecting the main shaft to the gearbox input shaft. Due to the very high torques, this type of torsional rigid but deflection- and Figure 5.9 Classic” geared high- speed concept with bearing unit (similar to four- point, dual bearing suspension), per definition “low- integrated” due to the common bearing shell, the bearing unit is a dedicated unit fixed to a support structure at the main frame or directly on the main frame
- 251. 222 Wind turbine system design Figure 5.10 Low- integrated, “compact” geared high- speed concept with bearing unit (similar to 4- point suspension) and flange- connected gearbox Figure 5.11 Low- integrated, “pure torque” geared high- speed concept with a bearing unit, slow- speed shaft flex- coupling, and flange- connected gearbox
- 252. Drivetrain concepts and developments 223 misalignment- compensating slow- speed shaft coupling (pure torque transfer as the main objective) requires a great effort due to design and manufacturing com- plexity. Furthermore, additional axial space is needed for integration, according to Figure 5.11, which means less compactness of the entire DT, but at least creates a mechanical determinated system. Some DTs for WTs still apply a very special form of double- bearing suspen- sion. A double- bearing support on a so- called “king- pin” structure, which is hollow- type like rotor main shafts. The function of the main shaft is partly taken over by the hub structure and, as shown in Figure 5.12, by a slim, flexible main shaft, which serves to transmit torque and, due to its flexible characteristics, to decouple parasitic loads from the gearbox input side. The gearbox is attached separately to a support structure within the nacelle. This special main shaft takes over the additional task of a flexible coupling on the low- speed shaft, due to the dual use of low- integration level results per definition [7, 8]. The next logical step toward higher integration is to replace the bearing unit or the dual bearing king- pin solution with a single- bearing solution, often referred to as moment bearing. The idea of a moment bearing was introduced by bearing suppli- ers. With an additional larger diameter and reduced overall length, two tapered roller bearings that are set against one another in an O- configuration (also referred to as back- to- back, B2B) are integrated into one bearing within the SKF solution, the so- called “NautilusTM” bearing. Other bearing manufacturers (e.g., Co. ThyssenKrupp Rothe Erde) used an alternative design and also created a single- bearing solution Figure 5.12 Classic” geared, high- speed, low- integrated DT concept with four- point suspension realized by a dual bearing rotor hub as a common shell on a hollow king- pin structure and an inner, angular flexible, “PureTorqueR” shaft (GE Alstom- Ecotècnia ECO 3.0 MW)
- 253. 224 Wind turbine system design Figure 5.13 Low- integrated, geared high- speed concept with single- bearing suspension and minimized main shaft (Fuhrländer FL 2500, Bard 5.0 using torque arms with partial flexible trunnion supports instead of flange connection with bushings). The flexible support of the gearbox is optional, and a rigid coupling is possible, if tolerances are taken into account. Figure 5.14 Semi- integrated, geared high- speed concept with gearbox- integrated single- bearing suspension (Vestas V90- 3.0 MW)— Remark: special attention has to pay to unwanted gearbox effects (e.g., dynamic housing deformation and own- weight loads)
- 254. Drivetrain concepts and developments 225 based on a three- row cylindrical roller bearing arrangement with comparable prop- erties in terms of high radial, axial, and bending stiffness. The use of moment bear- ings takes place at all conceivable DT concepts, which primarily aim for a high level of integration and thus a shortened, compact DT design. It remains to be noted that the WTs with the lowest specific nacelle weights (related to rated power and torque) currently have DTs with single- bearing solutions. This applies to geared (refer to Figures 5.13.–5.14) and gearless concepts (refer to Figure 5.23). A potentially already very compact DT, but only with a low level of integration, results from this use of the moment bearing as a separate unit, as a replacement for the aforementioned bearing unit, respectively, dual- bearing arrangement. The clas- sic rotor main shaft is omitted here and is replaced by a kind of short connection adapter from the inner ring of the moment bearing to the gearbox input shaft; the hub is mounted at the opposite side to this bearing ring, too. As it is shown above (refer to Figure 5.14), the level of integration can also be further increased for this concept by integrating the single bearing or compact dual bearing (refer to Chapter 6) in the gearbox. In case integration is unwanted, by applying direct flange connections between the gearbox, bearing, and struc- ture, the parts are still separated, but without any additional adapters. With a smart, deformation- considered design, it might be also possible to omit the flexible high- speed shaft coupling by using a flange connection with flexible bushings between the gearbox and high- speed generator, but then the gearbox has to carry the genera- tor dead weight, too. 5.2.2.2 Geared, medium-speed concepts (referred to as Hybrid-Drive) As already explained, medium- speed concepts belong to the main class of geared WT- DTs. Most commonly, the last gear stage, which is usually designed as a spur gear stage, is omitted here; in some cases, the second planetary gear stage too. In contrast, we see a new trend applying up to three planetary stages also for Hybrid- Drive DT for highest- capacity WTs (10 MW and very large rotor sizes). For the medium speed range, the generator operates at rotational speeds between 150 and 800 rpm at step- up gearbox ratios i between 9 and 50 (with three gear stages i up to 100). Due to the physical principle, only synchronous generators are sensible options for these applications. Induction machines or DFIG are not suitable for this nominal speed range, due to weight and efficiency reasons. As a result of the lower genera- tor rotational speeds, the generator torque requirement increases inversely propor- tional. As a rule of thumb, the so- called air- gap volume (volume of the cylinder, which fits into the air- gap diameter and has the length of the laminated iron core of the generator) is proportional to the torque- generating capability, so in consequence with a reduced gear ratio, the installation space for the generator increases inversely proportional and thus its weight too. The necessary air- gap volume will be advan- tageously implemented by a larger generator diameter with an almost unchanged or slightly reduced laminated stack length.
- 255. 226 Wind turbine system design These DT concepts, sometimes referred to as Hybrid- Drives, were originally introduced from gearbox system suppliers (Winergy HybridDriveTM and Moventas FusionDriveTM ) and OEMs to increase the level of integration (higher compactness) of WT DTs, i.e., to decrease tower head mass, increase reliability, and thus lower “Levelized Cost of Energy” (LCoE). Commonly gearbox and medium- speed gen- erators are rigidly linked at their housings by flange connection, so the gearbox housing carrying the dead weight of the generator too (Figure 5.15). An addition- ally higher integration level of gearbox and generator is rather rare but technically possible e.g. by using the rigid bearing unit concept for the support of the planet carrier of the first gear stage. Even a generator without own bearing support would be technical possible (“fly wheel concept”), but mostly, we see still self- supported concepts for each component. In general, this higher the risk of thermal run away of bearings and unwanted constraining forces due to critical tolerance with the double fit of shaft and housing flange connections. These issues can be solved by the use of flexible shaft coupling, flexible bushings for flange connection, or at least a more flexible bearing support for the generator shaft, which allows some angular mis- alignments (“Spherical Roller Bearings,” SRBs). Other solutions could be designed with an overall lower bending stiffness of the shaft [e.g., face- to- face (F2F) arrange- ment and single- bearing solution] or with enough bearing clearance to ensure a stable (from a thermal point of view), static determinated suspension in connection with the gearbox. Almost the same issues occur at the linkage between the bearing unit, respectively, single- bearing support and the gearbox. So, in the past, OEMs Figure 5.15 Low- integrated, geared, medium- speed concept with double- bearing suspension (bearing unit), rigid slow- speed shaft/gearbox coupling, and flange- connected gearbox and generator
- 256. Drivetrain concepts and developments 227 still use complex, flexible coupling (e.g., Adwen AD8 and Vestas V164/174) to achieve a pure torque loading or equivalent measures even on the low- speed shaft, which is quite costly and makes the DT slightly less compact (Figure 5.16) Within most of these concepts, the connection between medium- speed gearbox and main shaft is realized via a flange connection or shrink disk, too. If the shaft connec- tion is designed to be rigid, then classic flexibly mounted torque arms (Moventas “FusionDriveTM” ) should be provided instead of a flange connection between the gearbox and the nacelle structure, to avoid double fits and thus constraining forces under static and dynamic operation. Such designed- in flexibilities usually result in a reduction of non- torque load influences from the rotor on the gearbox. In case of additional flange connection, inevitable dead- weight loading is applied to the bear- ing unit, e.g., the AD8 10 MNm transmission unit (gearbox, coupling, and housing) has a dead weight of 86 tons, plus the weight of the generator (fluid cooled generator ~20 Nm/kg for medium- speed DTs). By definition, these concepts achieve a low to medium level of integration, since the individual components are directly connected with each other and thus housings sometimes have a dual function (support of connected components) but are still real- ized as separate parts (e.g., generator and main bearing). These concepts therefore offer a compromise between compact design and service or repair options without complete dismantling of the DT or exchange of the entire nacelle. Figure 5.16 Low-integrated, geared, medium- speed concept with double- bearing suspension with bearing unit and slow- speed shaft coupling (Adwen AD8, Vestas V164/174- 7.10 MW), a bearing unit for an 8- MW DT weighs ~80 tons
- 257. 228 Wind turbine system design The “HybridDriveTM” from Winergy, on the other hand, uses a flange connec- tion, which is designed using flexible bush bearings on the shaft and housing side to ensure some decoupling and for structural- born noise reduction (Figure 5.18). Figure 5.17 Top: 5.x MW HybridDriveTM with oil -cooled generator and main bearing unit after pre- assembly (semi- integrated, all parts rigid flange connected); the low- speed shaft flex- coupling is visible between bearing unit (blue) and (gray) first planetary gear stage, © Flender GmbH, 2022; Bottom: generic installation situation of a HybridDriveTM within a WT nacelle, © Flender GmbH, 2022.
- 258. Drivetrain concepts and developments 229 Nowadays, especially for the next- generation WT platforms, several of the OEMs seem to prefer these compact medium- speed DT solutions (e.g., Vestas EnVentusTM platform, SG onshore platform, MingYang offshore platform up to 16 MW, Vestas V236 new offshore platform, and Goldwind GW 242/12000) in combination with bearing units. Figure 5.17 shows such a Hybrid- Drive (here is an example from Co. Flender) of 5+ MW class. The main difference between the solutions for different OEMs, applications (onshore, offshore), and rated power classes, besides internal design features, is the presence or absence of the flex coupling at the low- speed shaft. As far as the author’s known experience with these kinds of couplings is good up to now, but as already mentioned, installation space, DT dead weight, and costs are slightly higher. Only the complex low- speed shaft coupling for high power can be easily in the range of 100+ k€. On the other side and that should not be Figure 5.18 Top: Winergy HybridDriveTM from 2012; Bottom: basic principle of this low- integrated, geared, medium- speed concept with single -bearing suspension, minimized rotor shaft, and noise decoupling gearbox flange support with elastomer bushings (planned for Fuhrländer FL 3000).
- 259. 230 Wind turbine system design underestimated, this coupling delivers proper DT damping and proven decoupling from parasitic loads from the gearbox module and therefore serves the overall reli- ability of the WT. Currently, what we rarely see in the market are solutions, which apply single- bearing concepts integrated into a gearbox–generator assembly, like was introduced in the Multibrid (later Areva) M5000 or the SCD (Figure 5.19). Such a concept would correspond to the definition of a fully integrated geared, here especially Hybrid-Drive, concept. 5.2.2.3 Gearless concepts (referred to as Direct-Drives) With gearless DD concepts, the generator must convert the low rotational wind rotor speed and the high torque into electrical energy without step- up geared transmission. For that, already realized DD WT generators have diameters between 3 and 12 m with nominal power ratings between 500 kW and 14 MW. The nominal radial air- gap length of these generators is between 4 and 10 mm, and the iron stack is between 0.5 and 2.5 m. Deformations (eccentricity and tilting) of a few millimeters in these very large structures are therefore already critical. In general, all DD concepts can be realized with inner or outer rotor design depending on the overall concept of suspension and gen- erator technology (permanent magnet or electrical excited generators). Nowadays, the outer rotor design in combination with PM technology is more common. Outer rotor designs for electrical excited synchronous generators are quite unfavorable regarding Figure 5.19 Fully integrated, geared, medium- speed concept with integrated single- bearing suspension, with having to handle at least partially the rotor, gearbox, and generator dead weight (Multibrid/Areva M5000, SCD)
- 260. Drivetrain concepts and developments 231 their mechanical construction, weight, and thus cost aspects. On the other hand, from a mechanical point of view, for inner rotor concepts, there is no big difference between these generator types. However, differences pop up in the technical details and con- straints such as slip ring systems, restrictions of pole pitch, controllability, and power density. Common arguments for the use of EESG are that it has one more degree of free- dom for control, due to variable electrical excitation and lower material costs (no rare earth material needed). PMSG utilize rather expensive rare earth material (price shock for neodymium and dysprosium started 2011 [5]), as a rule of thumb 600–900 kg mag- net material (NdFeB, with ~30–40 % share of rare earth materials) per installed MW for typical DD with nominal rotational speeds from 7 to 11 rpm in the multi- megawatt power range. If we discuss the possible characteristic features of DTs utilizing DDs, we are starting again with the classic four- point suspension (double- bearing) system that can be applied here, too. A 4 MW OWT (Figure 5.20) from GE in 2011 designed by ScanWind with a back- pack fly- wheel inner rotor PM generator concept uses this in combination with a long main shaft, two separated main bearings, fixed on the up- wind, floating on the down- wind side, torque arms, and trunnions supports at the generator housing to handle torque reaction forces. Thus, the dead- weight forces of the wind rotor and the generator affect opposite sides of the shaft and machine bed. Though some advantages, this concept with separated components could not survive on the market and was also comparatively heavy with a top head mass of 280 tons (including genera- tor own weight of 84 tons). Just to compare it, this weight is fairly equal to the two- stage medium- speed 10 MNm gearbox of the Adwen AD8- 180 OWT from 2015. The medium- speed (medium voltage) PM generator (8.6 MW@~350 rpm, rated) has a dead weight of ~32 tons. Figure 5.20 (bottom) presents the equivalent DD drivetrain concept to PureTorqueR design for geared DT, with a slim and therefore flexible main shaft to decouple rotor load- induced deformations from the generator and thus its air gap. Also, the corresponding geared DT variant, according to Figure 5.12, achieves that by means of a non- torque flexible main shaft and separated support structure of the gearbox. When it comes to WT DDs, ENERCON was undoubtedly the pioneer in intro- ducing this concept in the 1990s. These first DD concepts were designed with an electrically excited synchronous generator. Sometimes, it was referred to as ring generators because of its outward appearance, with an internal rotor salient pole design, a common double- bearing solution for the wind rotor hub on a “long” king- pin structure, and the hub structure rigidly coupled to the rotor of the generator (refer to Figure 5.22). The type of bearing support and the inner rotor concept make it inherently difficult to access the rotor hub via the nacelle. Later concepts that apply other main suspension concepts offer clear advantages in direct comparison here. The design of this rotor suspension system can best be compared with a kind of hub integrated bearing unit, already presented for the gearbox concepts, but less compact, and the outer common bearing shell is part of the hub structure. With a suitable choice of bearings, it offers high axial rigidity and low bearing clearance. The weight forces from the wind rotor can be advantageously distributed and dis- sipated over the multiple- part king- pin structure using a comparable small bear- ing diameter. The Achilles heel and therefore disadvantage of many DD concepts compared to geared transmission concepts is the large influence of loads and thus structural deformations on the air- gap dimensions and geometry of the generator
- 261. 232 Wind turbine system design Figure 5.20 Top: Low- integrated, DD (inner rotor) concept with four- point suspension by separated main shaft bearings and generator (“fly wheel” concept) torque arms (GE 4.1 MW, designed by ScanWind from the predecessor, a 3.5- MW model). Bottom: Non- integrated, DD (inner rotor) concept with double- bearing suspension for both the main shaft and the generator (Envison E128- 3.6 MW PP 2B).
- 262. Drivetrain concepts and developments 233 (air- gap load sensibility). Thus, this bearing concept previously used by ENERCON in combination with the mechanical generator arrangement and air- gap diameter requires a very rigid overall design of the king- pin structure in order to reduce load induced air- gap sensitivity. Figure 5.22 shows the realized construction principle in more detail. What is presented there is the effect of titling of the generator rotor in the stator under rotor loads (dead- weight force dominated), caused by different lever arms and bending stiffness of the entire structure, greatly exaggerated as illustrated in an Fini Element (FE) analysis. An asymmetrical air- gap distribution has a nega- tive effect on the efficiency and thus on heating, noise, and magnetic tensile force distribution within the generator and should be minimized. Due to the long period of ENERCON’s market leadership for DD turbines, the DD concept in general was under the prejudice, that it was much heavier in principle than comparable geared transmission DT concepts for a long time, since ENERCON relied on the inherent heavier, electrically excited synchronous generator technology and the very solid casted king- pin structures. Figure 5.21 shows the same concept in principle, but in an external rotor version (realized by means of PM generator technology from Co. Vensys, based on the idea of Prof. Klinger from the HTW Saarland, Germany). In the turbine design with outer rotor PM- generator technology, a slightly more compact and therefore overall lighter structure is already achieved. The air- gap sen- sitivity regarding the rotor- induced loads is reduced already by that. To reach the next level of integration and as a logical consequence in parallel to the developments on the geared DTs, a very compact single- bearing solution or a dedicated compact bearing unit can be used. In order to achieve the highest possible integration level, Figure 5.21 Semi -integrated, Direct- DriveDD (outer rotor design) with double -bearing rotor hub suspension, direct, rigid generator coupling on a “‘king- pin”’support structure (e.g., Vensys and thus Goldwinds first designs up to 1.5 -MW WT class)
- 263. 234 Wind turbine system design the external rotor generator is fitted with a moment bearing, to which the rotor hub is then flanged at the inner ring by a bolted connection. This single bearing is then usually connected to the outer ring with the stator directly by bolts in front of a shortened king- pin support structure. Thus, the generator with integrated bearing forms the center component carrying itself and the rotor hub. This design currently represents the highest degree of integration across all DT variants with the lowest number of components at the same time and can therefore undoubtedly be described as fully integrated (refer to Figure 5.23). Figure 5.23 illustrates the much lower rotor dead weight/air- gap sensitivity of the fully integrated DD DT concept. A similar integrated design can also be achieved by using a very compact bearing unit instead of the moment bearing for the Figure 5.22 Semi -integrated, Direct- DriveDD (inner rotor design) with double -bearing rotor hub suspension, direct, rigid generator coupling on a “‘king- pin”’support structure (older Enercon WT E40- E126, EP1, 2, 4, MTorres)
- 264. Drivetrain concepts and developments 235 generator bearing. With only a slightly larger overall length, this alternative concept offers the advantage of a broader supplier base for more standardized large bearings. In the past, there have certainly been supplied bottlenecks for highly specialized moment bearings [5]. The new Enercon EP5 and the Lagerwey LP4 use that concept with EESG and PMSG respectively, but both with inner rotor design. To avoid load- induced deformation of the generator air gap, OEM Alstom (now GE) extended their PureTorqueR philosophy from geared DTs (refer to Figure 5.12) toward DDs within the “Haliade” OWT. This DT concept (refer to Figure 5.24) combines the classic dual- bearing rotor in- hub suspension on a solid king- pin struc- ture with a specific flexible coupling adapter (angular flexible, torque rigid but with damping elements) between the rotor hub and the generator rotor. The concept shall decouple the non- torque loads from the separately supported (with additional bear- ings) PM generator. Thus, the air- gap sensitivity issue was solved here in the classic way (mechanical decoupling). But consequently, this leads to a higher number of Figure 5.23 Fully integrated DD (outer PM rotor) with integrated single- bearing suspension, king- pin support structure (SG- 3.0–14.0 MW DD, Co’s Goldwind, EWT, Lagerwey, Leitwind, XEMC/Darwind, and STX) with an inherent very low wind rotor load induced air- gap sensibility
- 265. 236 Wind turbine system design specialized parts, more axial installation space, and thus significantly higher tower head weight. The equivalent design for the geared DT is shown in Figure 5.16. Another concept with the highest possible level of integration should not be unmentioned here. The so- called in- hub- generator concept (e.g., IMPSA IWP 100) Figure 5.24 Low - integrated, DD (outer rotor) with double- bearing rotor hub suspension, flexible hub/rotor coupling, and separated generator bearing support (GE Halliade 6 MW and X) Figure 5.25 Fully integrated, DD (outer rotor) with double, in- hub bearing suspension (Project NaGeT—FiTTg in cooperation with Fraunhofer- IWES 2011–2013—funded from BMWI), prototype NaGeT 3.0, scale 1:5
- 266. Drivetrain concepts and developments 237 uses an outer rotor PM generator as the central component in the hub on a king- pin support structure (Figure 5.25). The common rotor/generator suspension is realized by the classic form of a robust double- bearing solution, which can be realized with comparable small bearing sizes (diameter). Finally, the full integration level in this case is achieved by the fact that the generator rotor and the rotor hub merge into one functional unit. The disadvantage of the concept, which due to the principle also has a very low air- gap sensitivity, is the larger rotor hub diameter required to integrate the pitch systems in the radial direction above to the generator and to enable service activities at them. Despite some advantages, this concept has not yet been imple- mented with one exception, IMPSA IWP 100. 5.3 General design rules and procedures The design of modern WTs and thus the design of DTs are based on the fundamental technical rules and standardized procedures. Since the 1990s at the latest,WTs and their structural, mechanical, and electrical subsystems and components have been designed in accordance with national and international standards (ISO, IEC, DIN, DKE, and ANSI/AGMA) and technical guidelines issued by certification companies [e.g., DNV (formerly DNV- GL), TÜV, Bureau Veritas, DEWI- OCC, Wind- FGW]. The standardi- zation by means of binding, harmonized, and generally accepted technical regulations must be clearly distinguished from the process of certification (refer to Figure 5.26). In principle, the idea of certification is an independent review by an accredited author- ity with regard not only to compliance with the binding regulations but also to the best state- of- the- art experience (technical guidelines, e.g., DNV GL 2010, VDI, and Wind- FGW) in the design, development, production, and grid integration. It serves to establish product and operational safety as well as an assessment of quality and technical risks, to ensure transparency and trust among all stakeholders, besides other Figure 5.26 Key elements of WT certification process (different tasks/ assessments)
- 267. 238 Wind turbine system design things by carrying out accredited procedures, e.g., for tests and measurements (refer to Chapter 9). Fundamental requirements for predefined service lifetime design of a WT/DT are as follows: • • mitigate or avoid loads due to proper basic design, e.g., by lightweight construc- tion or through compliance with general design rules, • • withstanding loads (external, internal, and environmental conditions), which means sufficient reliability as well as robustness with a predefined level of safety, • • control or manage loads (mechanically, electrically, by controls) with passive (e.g., twist- bend coupling, flex- pin gearbox, and torque limiters) or active (e.g., individual pitch control IPC, peak shaving, and active damping) approaches, • • mitigation of environmental impact (e.g., noise emissions, shadow effects, ice drop, environmental pollution by leakage of operating fluids, and esthetics), • • acceptable serviceability (monitoring and diagnostics, maintenance, and repair) due to the fact that technical systems need service and can fail in general, all in respect of the boundary condition of general manufacturability and operation at the lowest life cycle costs, grid compliance, and the highest possible, site- specific energy yield. 5.3.1 Safety, protection, reliability and control In discussions, the term safety sometimes remains unclear due to its ambiguity and its use in the context of protection. The simple rule in this context is protection serves the operational safety. Safety in terms of design characteristics (design safety margins) serves the reliability. In general, all required strength verifications within the design process need extensive calculations, analyses, and simulations as well as some validations through field measurements and tests (material, component, system, and field tests) (refer to Figure 5.27, [9]). This is standard to ensure safety and reliability. For even higher reliability, an overall product validation (refer to Chapter 9) with a test and valida- tion concept that accompanies the development, already starting with the first devel- opment phase, is strongly recommended. Within the last 10–15 years, a rethinking and higher acceptance of development- accompanying and -embedded test strate- gies can be noticed in the wind industry in general. Perhaps, this was triggered by the increasing number of technical problems in the intensive growing phase (e.g., damage of gearboxes, bearings, and generators) of modern multi- megawatt WTs from the late 1990s onward. The causes were diverse and often complex, but from a retrospective, e.g., by proper environmental testing, test bench campaigns, and test- based model validations; these problems most probably could have been avoided. In general, experts from OEMs, suppliers, and science are still discussing whether the series defects are mainly arisen by the reason of the overall system development and integration, thus at the OEMs side or at the supplier’s side, due to quality issues of the mostly very special and large components (large bearings, gears) that are pro- duced in relatively small quantities.
- 268. Drivetrain concepts and developments 239 From a more independent perspective, however, the reason can be partially pinned down to the lack of technical exchange between the OEMS and suppli- ers and sometimes poor technical component specifications and less experimental validation, too. Another aspect that should not be underestimated is the still ongo- ing market pressure regarding costs, which we already know from other industries (“López- Effekt” 1993) and that can cause massive product quality issues. From an operational but also from a design (down to component level) related point of view, the control and protection system of a WT is determinant for load management (minimizing material stresses, [10]) on the one side and on the other side crucial for the turbine protection against hazardous situations in case of system failures or extreme external conditions. Here a clear distinction between both of these has to be made. The control system shall operate the turbine in a controlled manner within the permissible operating range, e.g., during all normal operating maneuvers, during start- up, ramp- up, partial, full, and transient overload operation as well as during controlled shutdown or power reduction mode. This includes clas- sic continuous and discrete controls (closed loop control) and condition- based oper- ation mode transitions (state machine with state transitions); therefore, the turbine Figure 5.27 General design process with strength verification and model validation starting from entire turbine servo- aero- elastic concept simulation for interface load assessment to component load analysis and optimization
- 269. 240 Wind turbine system design control system has interfaces to the primary controls of internal component and system controllers (pitch controller, main converter, pre- heaters, etc.). Due to that, the task of the classic turbine control system covers no dedicated protection and therefore safety aspects; however, in terms of load reduction, it increases WT safety in terms of reliability. Furthermore, it is important to understand that the load analysis and also a vibra- tion analysis [11] for the DT should only be carried out with the turbine controls taken into consideration, because the influences on most of the loads are essential (e.g., [10, 12–15]). The inherent vibration characteristics of single components are important but should only be used for the preliminary design phase in that detached way. Cross- couplings between the mechanical DOF and force feedback effects of connected components/interfaces should never be underestimated in WT design. The control system of a turbine usually has feedback of the following internal sig- nals, especially but not only from the DT: • • rotational rotor speed respectively generator speed • • wind speed and direction from the nacelle anemometer • • nacelle/tower vibration level • • temperatures (external, internal, oil, generator, converter, coolant, etc.) • • voltage, current, and frequency and connection status at PCC, output power • • cable twist status • • yaw position, respectively, yaw error • • further binary status information, e.g., about brake wear, from hydraulic auxil- iaries, lubrication and cooling systems (refer to Chapters 7 and 8) The control system uses the following main actuators to operate the turbine in stationary and transient operation: • • yaw drive system (drives and hydraulic brakes, refer to Chapter 4) • • pitch system (single- blade angle adjustment, refer to Chapter 3) • • activation of the mechanical brake system (on low- or high- speed shaft) • • electrical grid connection (controlled on and off of the auxiliary and main switches, refer to Vol. 2) ○ ○ switched on—after synchronization ○ ○ switched off—at loss of grid connection or permanent grid voltage or fre- quency, outside the prescribed limits • • Active, reactive power and generator torque adjustment (refer to Vol. 2) by means of the main converter (B2B configuration of the generator side and grid side converter internally connected by a DC- voltage link). Typical time constants of the actuators or dynamic response times, respectively, for the entire turbine and thus the DT are shown in Table 5.1. On the other hand, the dedicated turbine protection systems take over, if the control system, critical sen- sors, subsystems, or components fail or as a result of an event, so that the turbine and thus the DT are no longer operating in a normal range. Once activated, the protection
- 270. Drivetrain concepts and developments 241 system can be based on aerodynamic, mechanical, or electrical principles to ter- minate an abnormal operation or hazardous situation, transfer the turbine to a safe condition, and maintain the system in this condition. Therefore, a protection system consists of a detection, an activation, and an actuating unit to fulfill this task. A fur- ther key requirement to the protection system is its fail- safe characteristic (e.g., in case of power supply/grid failures). Furthermore, according to the IEC61400 stan- dard [16], the requirement is to bring the WT rotor to a full stop from any hazardous operation. The control and the protection system shall be independent systems as far as this is technically possible. The protection system shall run on a dedicated safety PLC and use dedicated redundant sensors and/or signal lines; furthermore different physical principles should be used to avoid systematic failures. As a standard, a redundant braking system must be implemented (at least one system on the low- speed shaft); in modern multi- megawatt WT, this is usually real- ized (and generally accepted) by the three independent blade pitch actuator systems with energy storage units. The aerodynamic break effect of at least one rotor blade in complete feather position must be sufficient to bring a turbine in uncritical idle speed. The optional requirement to bring the turbine and therefore the DT in full stop condition can be managed by a mechanical service braking system on the main or generator shaft (usually from max. 10–15% of rated speed). After full stop, the mechanical service break continues to act as a parking brake. An absolute minimal requirement is to bring the turbine into a safe and controlled idle mode in a worst- case multi- failure scenario. In general, the protection system must be designed to override the control system in any case; its design shall according to the necessary safety class (safety integrity level) result from a combined process of failure mode assessment and failure criticality assessment. The detected level for safety system activation has to be defined; that design limits are not exceeded, which means such a situation has to be simulated and ana- lyzed intensively during the design process, especially if they can cause ultimate loads at the turbine and its DT components (e.g., torque reversal or overload within gear stages during emergency stop or grid loss [15, 17–20]). Such scenarios will be identified (if not already specified by standards, e.g., IEC) and checked within safety analysis process [failure mode and effect analysis (FMEA), failure mode, effects, and criticality analysis (FMECA), and fault tree analysis (FTA)] and optional within turbine type certification again. Table 5.1 Main actuator systems of a WT and the ranges of dynamic response Actuator Response time/dynamic Remarks Yaw drive max. 1°/s (dynamic) Discontinuous operation (dead band) Pitch drive max. 8–10°/s (dynamic) Discontinuous, collective pitch control Continuous individual pitch control Generator torque 10–50 ms (response) Air- gap generator torque by generator side main converter Grid side converter 2–20 ms (response) Active and reactive power by grid side main converter
- 271. 242 Wind turbine system design Some of the following WT safety- relevant situations must be detected, which also affect the entire DT in different ways: • • critical over- speed (rotor, generator) • • critical overload (generator, gearbox indirect) • • grid fault—permanent loss of grid connection • • abnormal cable twist between the nacelle and tower • • abnormal nacelle/tower vibration level • • fire or spark and smoke detection within the nacelle/cabinets (to avoid uncon- trolled fire spread) • • critical temperature of DT- components or operating fluids • • permanent loss or disturbance of safety- related measurement signals • • fault of WT control or communication system (loss of a “life- signal”) Product safety is grabbing increasing attention not only in the consumer sector but also in the industrial sector and especially in safety- critical applications with a high criticality in terms of property damage and personal injury in the transport and energy supply sectors. Therefore, established standards already exist and proce- dures had been developed to support development processes for complex products already at the beginning of the product life cycle in terms of reliability, availability, maintainability, and safety. The corresponding method referred to as RAMS process sums up different tools and merges them into a generic procedure to avoid essential failures already in the product planning and development phase. The EN- Standard 50126- 1 describes the generic RAMS process. The methods FMEA, FMECA, FTA, and LCC (Life Cycle Costing) are primar- ily used as tools for RAMS verification. The results show analytically under which conditions the determined RAMS parameters are met. The RAMS calculations are based on input values that, if possible, are specified and confirmed by the suppliers of the components used. If no supplier information is available, suitable reference values from comparative products or expert estimates are used as an approximate solution. In addition to pure technical safety requirement fulfillment, development using the RAMS process should achieve compliance with general relevant or generic but not dedicated WT safety standards, such as: • • EN 50126 • • IEC 61025 • • IEC 61508 • • IEC 61511 • • EN ISO 12100 Risk assessment and risk reduction of machines In the following, a brief explanation of the main safety- related tools of the RAMS process is given in general. For more detailed information, please refer to the dedicated EN and ISO standards and related literature.
- 272. Drivetrain concepts and developments 243 FTA describes a probabilistic safety analysis, which is based on Boolean alge- bra and is used to determine the probability of a plant or overall system failure. The method is described by the International Electrotechnical Commission as the inter- national standard IEC 61025 (EN 61025) under the term “fault state tree analyses.” FMECA is a methodology, developed originally to change from an approach of “find failure and fix it” (reactive approach) to “anticipate failure and prevent it” (preventive approach). The methods developed focused on qualitative and quantita- tive risk identification for preventing failures. Therefore, FMECA involves quantita- tive failure analysis. Within FMECA, a series of linkages between potential failures (failure modes), the impact on the operation (effects), and the causes of the failure (causes and mechanisms) is generated. The methods and techniques associated with the FMECA were published in a series of Military Standards. MIL- STD- 1629A is the most prominent of these standards and is still in use today. It is inductive or data- driven and linking elements of a failure chain as follows: “Effect of Failure— Failure Mode—Causes/Mechanisms.” The FMECA shall be performed prior to any failure actually occurring. It analyzes risks, which are measured by criticality (as a combination of severity and probability), to take preventive action and thus provide an opportunity to reduce the possibility of failure and its criticality. FMEA and FMECA are closely related tools. Each tool resolves to identify failure modes that may potentially cause product or process failure. The FMEA is qualitative, just exploring “what- if scenarios,” where FMECA includes a degree of quantitative input taken from a source of known failure rates. Reliability as a measure of failure probability can be explained as follows. During the lifetime of a structure or mechanical component, these are subjected to loads and environmental effects. In consequence, this leads to a change of health state condition, which means lifetime consumption up to the point where deteriora- tion, microstructural damage, corrosion, or wear out causes a failure. The reliability of a part can be defined by the probability of the part reaching a limit state and thus, real or per definition, entering a state of failure [21]. There are several different types of limit states, but two types are very common, the ultimate limit state, which directly corresponds to the limit of the load- carrying capacity of the structure or the part and is characterized by material failure and fracture or related failure modes (e.g., extensive plastic yield, brittle fracture, instability, and buckling). On the other hand, it is the fatigue limit state. Here the failure probability increases as a function of time. A prediction of the failure probability as a function of time and method of equivalent loads can be used to predict the time when the failure probability will exceed (consumption of fatigue budget) a critical threshold (by definition, e.g., a maximum acceptable failure probability). Additional limit states can be defined, e.g., the serviceability limit state, which can imply that there are deformations out of tolerance without material failure (e.g., cracks, wear, corrosion, permanent defec- tion, or vibration), so the further operation becomes unsafe. This corresponds to a failure mode without critical consequences if detected, and the turbine will be pre- ventively and permanently shut down. The structure, component, or system enter the state “unreliable,” which may result in a defined action, e.g., shut down or inspection with subsequently revaluation if applicable.
- 273. 244 Wind turbine system design Typically, the design safety of mechanical components and thus DT components will be determined as described above, based on the material properties. However, as these components are parts of complex mechanical systems, there are additional aspects of safety to be evaluated. Due to the complexity, a further range of possible influencing factors for aging, wear, and component failure (temperature, humidity, salt mist, sand, etc.) have to be taken into account. The resulting, more complex (combined) failure modes and a nonlinear temporal progression of damage make it even harder, sometimes not feasible, to find a probabilistic approach to assess mechanical safety. Experience- based, empirical methods or condition monitoring based on measurements must be used here, in order to record the current status and the progress of the damage to assess whether the defined limit states have been reached. 5.3.2 Loads and load cases As part of the design process for subsystems and components, load cases, which the turbine and its DT are exposed during the anticipated service life and thus are relevant for its strength and characteristically for the planned site class (e.g., IEC wind class), have to be defined as well as their probable proportion of the designed service life. As already explained, all components have to withstand these loads with a sufficient safety margin, defined by partial safety factors. Design load cases, as already explained in Chapter 1, are predefined design situations under various expected operations, typical WT maneuvers, and external conditions, mainly wind events. The IEC 61400- 1 standard [16] gives a guideline and lists the design relevant load cases with a description of condition, turbine maneuver, and other parameters. The previously explained FMECA is a useful tool to decide which load cases are design relevant for the turbine, to determine sectional strains for individual systems and components, e.g., main shaft hub flange, main shaft suspension, gearbox, gen- erator, of the entire DT. It should be pointed out here that this identification of sectional strains for the definition of load input functions especially for the DT should only be a first step in the design and analysis. In fact, the state- of- the- art tools for load simulation and thus load calculation for the entire WTs are already quite powerful. Whereby for WT sub- systems like the DT, the level of detail for the models is often limited (e.g., DT simplified as a two- mass oscillator). These limitations affect the number of mechani- cal DOF as well as the modeling depth (rigid, flex- body, friction, and damping) and thus the consideration of linear and nonlinear deformation (nonlinear stiffness, back- lash, bearing clearance, point- line contacts, position- dependent parameter changes, cross- couplings, and micro movements). Conducting of detailed sub- system simu- lations (e.g., MBS of the DT, refer to Chapter 2) just using pre- calculated (by the use of simplified sub- system models within a WT simulation [22]) sectional forces and moments at system interfaces, e.g., on the main shaft flange, lead to systematic errors. But it is still quite common for the previously calculated section loads to be used for the detailed design, for example, for highly detailed FE- calculations of components.
- 274. Drivetrain concepts and developments 245 However, it is essential to consider the dynamic feedback of reaction forces and thus also changes in displacements and deformations in relation to adjacent systems (Figure 5.28). Influences and errors in “worst- case scenarios” can then be estimated by means of a sensitivity and tolerance analysis with regard to these effects in the overall model. The general solution of simulating everything as detailed as possible seems obvious, but it reaches its limits in practical implementation beside others due to the huge number of different load cases and the simulation of comparatively long- time series (per bin 600 s). In addition, the DT system as described above is not only embedded mechanically in the overall system but also dynamically coupled with the overall turbine system via thermal, electromagnetic, tribological, and hydraulic links, since even very small displacements or not expected, parasitic loads due to side effects can lead to insufficient gear load balancing or not optimal tooth contact, causing extensive gear or bearing wear with early failure pattern. Thus, all load cases or transient operations should be assessed carefully regarding their relevance to the specific design (Figure 5.28). Loads can occur in different life cycle and operating situations in combination with varying external conditions. Relevant situations during WT lifetime are as follows: • • testing • • storage and transportation and construction/installation • • normal operation (production mode at partial and full power) • • stand-still, idling, cut-in/out • • abnormal and faulty conditions (e.g., high wind with extreme yaw error) • • extreme external condition (e.g., typhoons, freak waves, earthquakes) • • maintenance and repair Figure 5.28 WT design process starts with structure modeling and simulation of the entire turbine, the site condition (3D- wind fields), and WT control
- 275. 246 Wind turbine system design Wind conditions can be roughly divided into normal and extreme wind con- ditions and thus inflow condition for the WT. The condition and the assumption if it is a normal (mean frequently used) operation mode or a temporary (may be very rare or unique) situation lead to a specific kind of analysis, either fatigue or ultimate strength related (refer to Chapter 1). As a rule of thumb, if the specific turbine (means hardware and software of the turbine) operation situation can be either characterized by a fault, respectively, malfunction, or the external conditions (mostly wind) are extreme, then the situation belongs to the class of ultimate load cases; otherwise, it is a fatigue load case. In general, the resulting DT load are as follows: • • aerodynamic- induced input loads, means 6- DOF rotor, respectively, rotor hub flange loading (torque, thrust, radial forces, plane bending moment); for more details, refer to Chapter 2 • • grid fault- induced input loads, due to grid events dynamically coupled to the generator air- gap torque (torque peaks, reversals, oscillations, [17, 18 and 15]) Loads, here referred to as aerodynamic induced loads, are caused by aerody- namic forces from the dynamic 3D- wind field, structural properties causing aero- elastic interaction, as well as by the blade- pitch control concept for power control, load reduction, and balancing. The aerodynamic forces and aero- elastic modeling with a detailed explanation of these loads are not intended to be the subject of this book, but for some more details, refer to Chapter 2 [23–27]. Load case simulation with the entire turbine model using commercialist tools, e.g., Bladed, Flex5, FAST, as already described, will provide coupled aerodynamic load input data for the rotor flange as time series, distributions sometimes referred to as load level distribution, and extreme values. Besides that, dead- weight forces, icing, wave (site specific), internal excitation, or reaction loads are always present, and DT components are exposed to them indirectly or directly. Briefly other static and dynamic loads result from: • • gravity loads (rotor weight, in general components dead weight) • • centrifugal forces and Coriolis forces, forces of inertia (due to rotation and unbalance of mass) • • gyroscopic forces (due to yawing) • • external excitation reaction forces (structural deformation or oscillation of the bedplate and support structure) • • internal excitation forces (can cause oscillations due to generator cogging, uni- lateral magnetic pull forces due to eccentricity or axial not centered generator rotor, gear meshing, misalignments, imbalances, cross couplings, e.g., due to kinematics of flexible shaft couplings, gear reaction forces) One example of additional dynamic loading is strongly increasing gyroscopic force effects for flexible main shaft or rotor shaft supports. In particular, gyroscopic forces induced from the rotor mostly occur during the nacelle yawing operation
- 276. Drivetrain concepts and developments 247 (refer to Chapter 4). This happens in any case regardless of the level of flexibility of rotor/main shaft suspension and leads to an additional yawing moment (only non- three bladed rotors) about the vertical axis of the rotor plane and to a tilting moment about its horizontal axis. Besides the rotor speed and inertia, the angular yaw velocity affects this proportion. Therefore, yaw speed is typically limited, and strong acceler- ation and deceleration (e.g., due to yaw drive jerks and friction slip- stick effects) are avoided by yaw drive design and control for modern multi- megawatt WTs. Adding then flexibility to the main shaft support can cause significant forces and oscillations [28] (whirl effect) with dynamic feedback and possibly self- reinforced loading on bearings, bedplates, the main shaft, yaw bearing, and dependent from concept at the gearbox, generator, oil lubrication system, if strong tower oscillations are induced. 5.3.2.1 Drivetrain with three-point main shaft suspension To calculate, usually design relevant, the bending moments along the main shaft as well as the resulting loads at the main bearings and the gearbox trunnions (gear- box supports in general), the equations for force and moment balance along the axes have to be established. The number of equations and their complexity can be reduced, due to some boundary conditions and pre- assumptions, which are neces- sary to come to an explicit result. For a three- point main shaft suspension (refer to Figure 5.29), according to Figure 5.8, the following assumptions are made. The main bearing does not support any moments (also bearing friction is neglected): MBy = MBz = MBx = 0 (5.1) The generator coupling (flex coupling) does not carry any loads (forces and moments), besides torque: MCz = MCz = 0 (5.2) Figure 5.29 DT forces and moments for classic three- point rotor shaft suspension as a basis to calculate balance of forces and moments, with FcogXz part of dead- weight forces along the z- axis FcogXz = WXcos , whereas WX the dead- weight force of the component X (refer to Index).
- 277. 248 Wind turbine system design FCy = FCz = Fcx = 0 (5.3) The main shaft bending moments are supported by the main bearing and the gearbox bearing of the first gear stage; thus, no trunnion moments around the lateral DT axes derive (trunnion are rather flexible for such moments) according to Figure 5.30. MGy = MGz = 0 (5.4) The main bearing (fixed bearing) carries all rotor axial forces because the gearbox integrated bearing support in combination with the trunnion is assumed as ideal floating in that direction. 5.3.2.2 Drivetrain with four-point main shaft suspension To analyze the conditions for four- point suspension DT, according to Figures 5.7 and 5.30, the boundary conditions have to be slightly modified to avoid over deter- mination of the mechanical.Again, for the gearbox supports, the assumption is made that they are flexible in titling and axial direction; thus, no additional constraining forces are generated here, and a tolerancing study (for real designs) should be con- ducted in any case. Their first purpose, in a realized construction, is to provide the gearbox with constraints of mechanical macro displacements, and the second pur- pose is damping noise and vibrations. Figure 5.30 shows the orientation of the used nacelle coordinates (non- rotating, main shaft oriented). With FcogXz = WX cos , whereas WX is the dead- weight force of the component X (R, S, Sh, G, BD). Figure 5.30 DT forces and moments for four- point (double- bearing) main shaft suspension as a basis to calculate the balance of forces and moments. Within a common four- point suspension design, the rear (downwind) bearing is the fixed one (different from three- point suspension).FcogXz is part of dead- weight forces along the z- axis FcogXz = WXcos , whereas WX is the dead- weight force of the component X (refer to Index).
- 278. Drivetrain concepts and developments 249 a) Gearbox, shrink disc, brake disc dead- weight carried by the trunnions: FGz + WG cos + WSh cos + WBD cos = 0 (5.5) b) Trunnion force in the y- direction is negligible compared to the bearing forces: FGx + WG sin = 0 (5.6) FGy FBfy, FBry (5.7) 5.3.3 Loads analysis and strength verification Usually, the rotating parts of a DT are more prone to reliability issues than structural parts. For the design and strength verification of DT components, one must distin- guish between different analysis methods in general. There is the fatigue analysis, well known for structural parts, which is usually conducted with aggregated damage equivalent loads. Similarly, in the process but with another focus, the analysis of extremes is supplemented by a quasi- static (“worst- case”) analysis, which covers unique or rare ultimate loading cases. And somehow special for specific mechani- cal parts (gears, bearings, sealings, etc.), the analysis regarding temporal propor- tions of load levels or operating states utilizes load duration distributions (LDD) for the calculation of aggregated operational loading or specific conditions, which not exclusively address fatigue issues [29]. Mechanical parts can fail due to many different failure modes resulting from different physical effects, often somehow related to classic fatigue, but not only [30]. Typical failure modes are, e.g., bearing fatigue, bearing axial cracking, gear- tooth cracking or fracture, fretting corrosion, micro/macro pitting, scuff- ing, true or false brinelling, cage fracture, element or ring fracture, skidding, sliding and electrical discharge, surface wear- initiated fatigue, subsurface/ surface fatigue, and fatigue spalling from Hertzian stress (max. shear stress) under the contact surface. To deal with that, specific influencing factors, partial safety, and safety factors are used for the individual failure modes or loads, respectively. However, the use of general partial safety factors often leads to rather conservative designs, especially where several factors are multiplicatively linked. The advantage of smart validation strategies (refer to Chapter 9) should be pointed out here again. The current partial safety factors were mostly deter- mined on scaled components or specific material samples for each type of load. Superposition of loads or influencing factors is then mostly considered conser- vatively. Tests on real or smart scaled components (refer to Chapter 9) can help to adjust the safety factors for the respective application and parts without losing reliability but saving costs. The required partial safety factor is defined as the ratio between allowable stress and permissible stress, whereby the calculated actual stress (including influencing factors, e.g., stress concentration, scaling, surface treatment, etc.) based on the nom- inal stress should never exceed the permissible stress.
- 279. 250 Wind turbine system design 5.3.3.1 Cycle analysis (fatigue analysis) Cyclic loads referred to as fatigue loading cause cumulative damage to the mate- rial that can lead to component failure after a certain number of cycles. Fatigue loads are usually loads well below the static material strength load level, and many load cycles are required before a fatigue failure will occur. This is commonly called high- cycle fatigue (range of cycles ~105 to 107 , refer to Figure 5.31, material tests according to ISO 1099, ASTM E466- 15, and DIN 50100). However, for some mate- rials with particular high S/N- curve slopes (range of cycles ~103 to ~104 , refer to Figure 5.31, material tests according to ISO 12106, ASTM E606, and BS 7270), the loads of importance for fatigue are close to that range that will cause static fail- ure. For such materials, fatigue issues become more of an extreme value problem, regarding loads, with only a few load cycles required to cause failure. This is com- monly referred to as low- cycle fatigue, whereby there is no exact definition of the limits. The stress in the low cycle fatigue test consists of an elastic and a plastic strain component. While there is a linear relationship between stress and strain in the elastic range (Hooke’s law), it is not linear in the plastic range. As a result, a hysteresis loop results in cycling loads. For standard fatigue assessment, a cumulative damage for the specific applica- tion and load scenario is commonly determined using the Palmgren–Miner rule. This method requires knowledge of the number of cycle distribution over the stress ranges. For fatigue analysis, typically cycle counting methods are used to generate the numbers for the cycle distribution of loads or stress ranges, respectively, from a time response of loads gained by measurement or simulation either. The rain- flow counting method is one of the most common and by far the most developed one. It is applicable for low- and high- cycle fatigue assessments, especially if stress ranges and mean stress values change during time. Furthermore, it is able to identify both slowly varying stress cycles and more rapid stress changes superposing those. As the only procedure, the rain- flow method identifies the mean value range and the number of cycles for a defined amplitude range, which means discretization and characterization (somehow classification respectively) of the loading or stress. The result of rain- flow counting is a distribution matrix with rows representing the mean stress level and column for the stress amplitude ranges. So, each element nij of this matrix (refer to Figure 5.32) contains the number of stress cycle associated with a particular stress amplitude range and a particular mean stress level. Each row represents the stress range distribution for a specific level of mean stress. Thus, if S/N curves are available for various ratios R between the compressive stress ampli- tude and the tensile stress amplitude, this representation will allow prediction of par- tial damage for each element of the matrix by applying the appropriate S/N curve. The total fatigue damage D can subsequently be determined by summing up the partial damage over all elements of the matrix. Reference is made to Figure 5.32 schematically describing the rain- flow matrix and how it is determinated. However, the choice of a reasonable order of magnitude of the discretization regarding the number of classes is left to the user.
- 280. Drivetrain concepts and developments 251 Figure 5.31 From material probe testing (here without mean stress, R = −1 [31]) to the principle S/N- curve shape for steal parts with main characteristics
- 281. 252 Wind turbine system design D = P i P j nij Sj Nij Sj (5.8) where Sjis the jth stress range (amplitude range), Sj , and the ith mean stress, nijis the number of stress cycles (matrix element row i and column j), and Nijis the number of stress cycles to failure according to S/N curve. This approach is applicable for damage predictions for components, which are exposed to non- zero mean stress. For some DT components, e.g., the main shaft, which are exposed to load cycles with a mean stress of nearly zero, the Miner rule in its simple form can be used. Thus, only the row representing the “no mean stress” level has to be applied for analysis. In reality and in most cases, the mean stress is not zero or negligible. Due to the mean stress influence, the fatigue strength of the material is reduced in general. A so- called amplitude transformation can be carried out in order to use the particular Figure 5.32 Principle of time signal analysis by means of rain- flow counting (e.g., main shaft bending with notch influence at the main bearing within a high- speed- geared DT with 3- point suspension), © Fraunhofer- IWES
- 282. Drivetrain concepts and developments 253 mean- stress- free S/N curve in the damage calculation. Here, the mean- stress afflicted cycle amplitudes are converted to the respective weighed mean- stress- free ampli- tudes either using a linear function (“Goodman line”) based on the tensile strength or the yield point (“Soderberg line”) or using the “Gerber parabola,” depending on the material characteristics. Furthermore, notches and surface roughness have an essential influence on the fatigue characteristics as well as the material itself. So, in consequence, we come from the basic (synthetic) material S/N curve to modified S/N curves for components, which includes additional factors (e.g., surface rough- ness, scaling factors, treatments, stress concentration, material support rate). For a more realistic assessment of specific stress cycle ranges near and below endurance strength, e.g., the FKM Guideline [19, 31, 32] suggests three variants of the Palmgren–Miner formula. The “miner elementar” uses all amplitudes also those below endurance strength of the component S/N curve, the “miner original” neglects those amplitudes, and the “miner consequent,” applies a flatter slope in the endur- ance strength area than in the fatigue strength area, due to the previous damage. In recent decades, continuous progress has been made in the reliable assessment of the damaging effects of dynamical loads in components (Figure 5.33). We see a development from the “classic” strength verification over state- of- the- art FE simula- tion with fully elastic (linear, nonlinear) material properties to even more modern methods like general fracture mechanics with the local and the probabilistic crack propagation concept. Fracture mechanics describes the process of crack growth, crack propagation, crack arrest ability, and fracture in a component or material under operating condi- tions (load cycles) [33, 34]. By estimating the lifetime or remaining useful life of Figure 5.33 Comparison of classic and advanced methods and approaches performing the strength analysis of components: (a) load profile over time with constant load limits vs. (b) with variable load limits (real load time history) (c) synchronous phasing of the load shares vs. (d) non- synchronous, arbitrary phasing (e) ideal linear- elastic behavior vs. (f) elastic- plastic and nonlinear material behavior (i.e., hysteresis) (g) material free of cracks vs. (h) consideration of defects (i.e., cracks) → fracture mechanics methods: LEFM and YFM, probabilistic approaches
- 283. 254 Wind turbine system design crack- affected components (or materials), inspection intervals, the reached service- ability, or fatigue limit state can be defined in a targeted manner. A distinction is made between two general concepts: linear- elastic fracture mechanics (LEFM) and yielding fracture mechanics (YFM). In LEFM (for brittle materials), the material behavior is linear elastic until deformation- free fracture (unstable crack propaga- tion) occurs. The characteristic value of LEFM to distinguish between stable and unstable is KIc , which describes the critical (C ) stress intensity (K) factor during crack opening mode “I” (means region “I,” within the crack propagation diagram). The standard crack propagation diagram is divided into three regions of propagation behavior. • • Region I: Threshold value ΔKth • • Region II: fatigue crack growth da/dN • • Region III: critical stress intensity factor KIC (fracture toughness) In case of cracks within ductile material with distinctive yielding properties, i.e., with plastic deformation around the crack tip, the concept of flow fracture mechan- ics is used. There are two definitions usual: the determination of the characteristic values via the energy stored near the crack tip (J integral concept) and via the crack tip opening (CTOD “crack tip opening displacement”). The stress intensity factor describes the stress distribution around the crack tip. It can be calculated for common geometries (test specimens and components) using approximation formulas or FE simulations. Then the type of crack stress is determined by the global stress, i.e., from the nominal stress of the cross sec- tion affected by the crack, the crack length, and a correction factor (a kind of “notch” factor) depending on the component geometry or specimen geometry. This stress intensity factor in turn is used within the Irwing–Williams series and can be compared then with material properties such as critical stress intensity. This predicts whether the crack will come to a standstill or whether the crack will grow in a stable manner or become unstable ending in a sudden fracture. These Irwing–Williams series (sometimes referred to as the Sneddon equations) describe stress distribution in the immediate vicinity of the crack tip. For this, it is assumed that the mechanical crack dimensions (crack length a) are small compared to the component dimensions. The relevant, general standards for material testing, to determine the fraction mechanic properties, are ASTM E399- 09, ASTM E647, ASTM E1820- 11, and ISO 12135. To define the representative stress range distribution for a mechanical part at a specific point, it is essential to consider all relevant, various conditions and opera- tional states of the turbine. Some of those situations are transient conditions, e.g., start and stop for which, in contrast to continuous operation conditions (production mode at a specific mean wind speed with a specified turbulence), the stress distribu- tions are not determined by stochastic processes. However, during the stationary production condition, constant stochastic parameters can be assumed to be valid in short term, e.g., during the common agreed 10- min measuring and simulation
- 284. Drivetrain concepts and developments 255 times for different DLCs. These 10- min periods have their origin in meteorological modeling and bin classification of wind measurements. For some of the mechanical DT component such description and clustering does not fall into place, as mentioned above, even when most of the loading results from wind, but nevertheless it is com- mon in use. So as mentioned within a 10- min period, stochastic wind characteristics such as the mean wind speed and the turbulence intensity at hub height are assumed to be constant. During such short periods of stationary conditions, the load response processes that finally cause the stress cycle in the considered components then can be taken as stationary stochastic processes, too. Therefore, the wind field- induced stress range distributions under stationary condition are often seemed to have similar distribution characteristics that are close to a Weibull distribution [21, 31, 32]. Also, the principle of S/N- curve modeling is based on the assumption that most fatigue failure characteristics of materials for specific load conditions (stress cycle ampli- tude and mean value) can be described by Weibull distribution, due to its inherent “memory” properties: FS s = 1 e s/SA ˇ (5.9) For the representation of stress range distribution, it will often be sufficient to apply specific Weibull distributions. The simplest distribution model among this family of distorted distribution is the three- parameter Weibull distribution [21, 31]: FS s = 1 e sa /SA ˇ (5.10) The coefficients a, β, and SA are distribution parameters, and they can often be expressed as a function of wind characteristics (average wind speed and turbulence intensity) [21]. When the short- term (means here periods of 10- min) load cycle dis- tribution, conditions by the assumed wind regime have been established and the long- term distributions of average wind speed and turbulence intensity are known for the site, and furthermore, the anticipated service life of the turbine has been defined, then the cumulated distribution of all load cycle ranges during the lifetime can be established. As already mentioned, this cumulated distribution of different production modes times of a turbine can itself often be represented by a Weibull distribution, which forms a significant contribution to the loads that cause fatigue damage. To obtain the distribution of all relevant fatigue loads in the design life, the cumulated distribution has to be supplemented by the load cycle distributions from transient conditions such as start–stop, standstill, idling, and abnormal yaw mis- alignment. For this purpose, information, e.g., about the duty cycle or grid events for the WT is necessary. The duty cycle is a repetitive period of operation, which is characterized by a typical sequence of different modes of different durations for consecutive 10- min periods. Information about the duty cycle should, as a mini- mum, contains the number of starts and stops during a representative period of time, together with the corresponding wind characteristics [21]. Starting and stopping the turbine are potentially critical with respect to fatigue. One reason for this is that connection (after synchronization) of the electric WT
- 285. 256 Wind turbine system design system to the grid can cause high transient loads in the entire DT. Specific loads dur- ing standstill and idling below cut- in wind speed with respective low aerodynamic loading (especially thrust) can be significant in comparison to nominal operating condition, due to rotor gravity forces, low speed, poor lubrication, and harmful kine- matic conditions for gear and bearings. Yaw misalignment as well as yaw movement can be critical in any case with respect to DT loading and fatigue. It should be noted that cumulated stresses at particular locations considered critical with respect to fatigue or static strength may be due to a combination of single stresses arising for different reasons or load processes (wind and yaw movement). For a detailed stress assessment on a component level, one approach is to assume load peaks (torque, axial forces, bending, etc.) of two load processes appear simultaneously. Thus, high- amplitude cycles can be combined with other high- amplitude cycles, even if they have slightly different frequencies. The mean values of two superposing load pro- cesses should be combined in a kind of “worst- case” scenario to create the largest possible mean stress value. Once the load spectrum has been established over the design lifetime, it is common to define so- called damage- equivalent load ranges S0 to be used with an equivalent number of cycles neq . A damage- equivalent load range is defined as that constant load range S0 , which in neq cycles will lead to the same accumulated damage as the distributed load spectrum that consists of many different load ranges. When the equivalent number of cycles and the specific material fatigue characteristic (S/N curve) are chosen or specified, the equivalent load range can be found as [21, 31, 32] follows: S0 = 0 B @ P i niSk i neq 1 C A 1/k (5.11) in which k denotes the S/N- curve slope. This definition of a damage- equivalent load range is well known in fatigue analysis and is often used in WT strength verification. It should be noted here that this method is only applicable for materials whose S/N curves can be described by one slope k (e.g., not valid for composite materials). As discussed, load spectra are usually not known with certainty but are somehow pre- dicted, e.g., from a limited number of time series of load responses, obtained from aero- elastic and servo- mechanic simulations or field measurements. The already introduced time series of 10- min duration are used for this purpose and one estimate of the equivalent load range can be calculated from each simulated time series of a DLC. Then, time series of load response are generally simulated for various wind characteristics, operation modes, and duty cycles. The already introduced cumulated load spectra are established on this basis by appropriate weighting according to the expected long- term distribution over the design lifetime. 5.3.3.2 Statistical and deterministic extremes (ultimate loads) When extreme load responses are of interest, such as for design against failure in ultimate loading regimes, extreme value distributions are needed. So, let us assume
- 286. Drivetrain concepts and developments 257 that a total of n 10- min time series of the load response have been generated by servo- aero- elastic simulations. Let us continue to assume the maximum value of the load response within the 10- min is of interest (for ultimate load analysis). Due to the stochastic process nature of wind regime within 10- min intervals, there will be no determinated time response but will have a variability, which again can be represented by a probabilistic distribution. The natural variability results in differ- ent values for the particular extreme value in the n simulated time series. Then the characteristic load response is usually taken as some quantile of the distribution of the maximum load responses from the simulated time intervals. The simple but statistically correct way is to predict the maximum load responses and their particu- lar quantiles of distribution from a statistical model, which utilizes the information about the maximum load responses obtained from the n simulated maximum val- ues. That means, in consequence, a high number of simulations n are necessary to achieve a sufficiently accurate estimate (high confidence coefficient) of the extreme values or other stochastic key parameters. Thus, it is recommended always to predict extreme loads based on the statistics of simulated load response processes, rather than using the observed maximum values of one exemplary simulation run only, because two or more arbitrarily selected simulated (e.g., 10- min) time series may serve considerably different extreme values. This implies that the practice of per- forming one or only a few simulations and selecting the average extreme load or the absolute largest extreme load as the ultimate load without proper consideration of the stochastic nature of the extremes will not give reliable results. Furthermore, in these cases, the results cannot be extrapolated to a characteristic value defined by a quantile or to a different duration of the load case than the original one (e.g., 10-min). There may be almost wind- independent deterministic operating cases, for which one simulation of the specific scenario is sufficient. But even here, prior knowledge is required, e.g., when calculating a transient short circuit torque peak (grid fault with a DFIG), so the result is strongly influenced by the phase angle or operat- ing state (magnetization, torque, and speed) at the time the fault occurs. A “worst- case” analysis is appropriate here, whereby stochastic input variables (wind speed progression during the fault) are neglected, or their most unfavorable progression is assumed, respectively. If dependencies are too complex, the possibility to con- duct Monte Carlo simulations remains, assuming that the influencing variables have a normal distribution, an extreme value can be adequately estimated again, if the sample size is sufficient. 5.3.3.3 Load duration distributions The loads or operational conditions are divided into several ranges, and an evalua- tion is made of how long the respective load stage occurs during a time series (load/ condition duration time) of an operation scenario. Multiplied by the frequency of occurrence of the respective time series during the entire anticipated lifetime of the component or system, this results in a certain number of hours for the respective load level or condition, on the basis of which the damage to a component can be
- 287. 258 Wind turbine system design determined. The frequency of occurrence results from the statistical distribution of the individual load cases that are assigned to specific mean wind speeds. Especially for the design of some mechanical components like bearings, gears, or hydrau- lic parts, it is common to use these aggregated operational loads or conditions (to assess, also but not exclusively fatigue issues) based on LDDs. The single load dura- tion can be used separately or weighted individually, summarized to an equivalent load or operating condition, respectively. Thus, certain environmental conditions (temperatures and contamination states) can also be linked to different load points. In contrast to the classic fatigue analysis, here it is not directly about the recording of load cycles, but the relative duration of specific, complex loadings and conditions. It should be noted that the mere consideration of LDDs does not describe the dynamic, transient transitions between the individual operating points (neither the dynamics nor the frequency of the changes), so the use of LDDs for certain mechanical parts (e.g., bearings and gearboxes) is common, but if characteristics of dynamic opera- tion are essential (e.g., resonances, transient excitations), additional information and analysis will be needed. 5.3.4 Modularization, standardization, and platform concepts As already mentioned, the wind industry is very cost- sensitive, among other things due to strong international competition of OEMs and specific energy market regu- lations. This is also evidenced by the consistently low EBIT margins of Western OEMs. The cost pressure on the market for WTs has grown significantly over the years and seems to remain high. The expectations of operators regarding perfor- mance, availability, and reliability of WT increase, too. The currently comparatively high raw material costs and critical supply chains further exacerbate the situation. Especially in the last years, OEMs introduced new market- specific products at always shorter periods of time to gain market shares. But in the medium- and long- term perspective, turbine manufacturers are faced with the question of how to adapt product and development portfolios to stabilize profit margins in nearly unchanged conditions. The steadily growing world market will continue to demand new, improved turbines for still lower LCoE and currently with a strong focus on the part of project CAPEX. Even if the key factors that influence the energy production costs of a WT are the total lifecycle costs, which include costs for the overall WT development, construction, maintenance, service, and costs for decommissioning as well as the lost profits due to reliability issues. This fact alone should strictly exclude a simple and development cost- saving upscaling of existing turbine types of a manu- facturer’s portfolio, even if that is technologically possible within some limits and seems to gain short- term results. OEMs who used such a strategy in the past have nearly disappeared from the international market. All well- known manufacturers have been relying on so- called platform strategies for some time, as we know them from other industry sectors. Basically, a platform strategy always requires a certain degree of modularization of the product as a whole or at least in its essential parts. Of course, this is also reflected in the corresponding WT DT developments, as the “heart and the motor” of a WT in combination with the
- 288. Drivetrain concepts and developments 259 rotor, for the current and future generation of WTs. It should be noted at this point that the general designation “platform” or “platform strategy” for WTs and their DTs is not a universal sign of product quality. The manufacturers sometimes pursue differ- ent goals with it, and the degree of supplier involvement varies greatly. The industry sector with the highest level of experience and know- how in this area is probably the automotive industry (e.g., MQB, referred to as “modularer Querbaukasten,” MSL, MSQ, and MEB from VW). Their product lines are initially based on consistent, deep modularization for a technical platform that can be used as broad as possible (in terms of product range), with standardized module interfaces if applicable, the highest pos- sible proportion of identical parts, as well as with the highest possible flexibility for small individually adaptations, future developments, and products. Of course, “a WT is not an automobile” is the most common response to comparisons with the automo- tive sector with its consumer product focus, much higher production numbers, etc., but anyway, some WT OEMs are still rather away from the fundamental idea of a real platform concept, except the wording. From a generic perspective, a platform strategy for WTs and their DTs should take into account the following technical product requirements, at which this list does not claim to be complete: • • different WT applications, onshore, near- shore, and offshore sites, etc. • • different wind regimes for specific sites or markets (mean wind speed, turbu- lence intensity, extreme condition, e.g., typhoon) • • additional project (site)- specific requirements (complex terrain and high alti- tude), cold and hot climate (dry and/or humid), other customizations • • specific energy market conditions or requirements • • grid requirements (e.g., grid codes, ancillary grid system services) • • specific requirements for noise emissions (limits and flexible regimes) and other environmental impacts • • market limitations regarding transportation, construction, and service • • market and competition requirements for flexible power upgrades • • cost- efficient and short- period future developments and optimizations So, let us have a look at the more experienced automotive sector again. Here, the supplier industry reacts accordingly and offers suitable modules for the different OEM platform approaches. Suitable modules in this context mean the OEM defines the interfaces of the modules, “standardized” them (just for the single OEM or in agreement with other OEMs) if applicable. Then the suppliers can bring all their innovation potential and specific know- how into module design, but modules and thus suppliers remain exchangeable. For nearly 10 years, the wind industry sector starting increasingly using their own ideas of platform strategies [35, 36] of course, focused to reduce manufacturing costs, optimize production and construction pro- cesses, faster upgrades to stay competitive, and to achieve fast development cycles for the markets in general. In this process, fundamentally different approaches are conceivable with regard to the role of the supplier industry. Especially for DTs, currently, an increasing change
- 289. 260 Wind turbine system design in the role of the supplier companies can be observed. For example, the large gearbox manufacturers more often slip into the role of system providers, offering the entire DT and no longer individual components. The essential parts of a WT platform idea are thus transferred to the system supplier, who applies then an internal platform strategy that suits him, according to the specific turbine platform design of a single OEM. You could call it platform concept within another platform concept or “flat” modularized platform concept. Since the specific DT system module forms an essential part of the respective turbine platform, the OEM and system supplier are tightly linked and must work closely together technically. But this type of platform strategy shows only a lower degree of modularization for the OEM, and it seems to have more benefits for the supplier side. It is noted that the module, in this case, the entire DT system module nor supplier’s DT sub- modules neither, is easily interchangeable for the OEM, and the supply chain remains somehow critical with such a single source. From the OEM’s point of view, there is a lot of competence from the system supplier in the module “entire, custom- made drivetrain,” which cannot be transferred 1:1, if once necessary. Indeed, the OEM’s development costs can be reduced by the partial shifting of costs and risks toward the system supplier, but sustainable cost reductions, including all other benefits of that flat modularization of the entire turbine within this “flat” modu- larized platform concept remain at least questionable. From the OEM’s point of view, the fundamental idea should be a much deeper modularization of a complex product WT that in fact leads to a specialization and concentration of know- how among the suppliers, but here in a much more focused and specialized area of single components or systems. The idea of the outsourcing process is that individual value- added activities should be distributed at the best among the core competencies of the supplier companies. For component suppliers, the chances of development partnerships lie in expanding their own, more specific areas of com- petence and being able to offer comprehensive product- specific services, so at least in theory. For the WT OEMs, the focus should be on their core competencies, i.e., the WT as an entire complex system itself. This means that the OEM core know- how is to take into account and to merge all requirements (list above) into a platform strategy, the individual operational management, advanced turbine controls, system integra- tion, and market know- how, logistics for on- site turbine construction, services, and costumer relationship. In contrast, the suppliers develop into module suppliers with high- specialized technological competencies. Because of an industrial product like a WT and not an emotionally charged consumer product like a car, the OEM must create sufficient individually added value to be able to generate profit margin in a competitive industrial market. With a product development by means of the “flat” modularized platform concept, mentioned at the beginning, the OEM dependency on the specific supplier in terms of know- how, pricing, and supply chain is very high and is often influenced by the relative economically strength of the companies (OEM vs. suppliers) and the competitive situation (OEM vs. OEM and suppliers vs. suppliers). So let us briefly summarize the idea of a “deep” modularization for a platform strategy. With modularization, complex overall structures are divided into individual, sep- arately coordinatable modules in order to then reassemble them into a complexity- reduced overall structure. Further advantages of the modularization result, for
- 290. Drivetrain concepts and developments 261 example, in an increasing strategic flexibility due to the decrease in complexity. In general, the possibility of outsourcing “individual” modules to a corresponding specialized supplier market increases the OEM’s ability to innovate and enables cost reductions and competitive advantages by concentrating on factors critical to suc- cess for the entire turbine. The market should benefit from individualized product solutions and a greater variety of variants that fits best to specific requirements with falling prices at the same time. The entire product architecture should be divided into smaller modules that are connected to each other via clearly defined interfaces with requirements and thus define their expected interaction, too. Ideally, this ends up with a standardization of modules. In the automotive industry, standardized basic modules in the form of a uniform body and base plate are used for this purpose in order to adapt them to customer- specific requirements using specific equipment variants of engine, auxiliaries, gearbox, coupling, brakes, suspension systems, con- troller, battery, cockpit, and seating groups. To transfer this approach to the wind industry, following the example of the automotive industry, means more or less a class of WTs (of one or more OEMs) are bundled under the umbrella of a modularized platform. Each system (e.g., DT) con- sists of a number of modules that are used in different system variants within one or more OEM platforms. The question arises now, for which modules definition, we will have the greatest synergies and the fewest restrictions in terms of platform design. As expected, the answer is not such simple. However, just simply defining modules in a so- called turbine platform such as rotor blades, hub, pitch system, DT, nacelle with bedplate and tower, and some “other parts” like pitch bearings, yaw bearing, converter, transformer, etc., seems not covering the full idea presented from automobile OEMs and is not innovative. A deeper modularization with the above- mentioned objectives is very demanding and typically conducted in several steps, possibly also loops. For this reason, even the possible basic principle can be presented here in a greatly simplified, excerpted, and exemplary manner only (Figure 5.34). a. In the first step, the existing product structure is completely, functionally ana- lyzed and decomposed. Because an existing product structure with its (sub- ) assemblies and components has often been created over a longer period of time in the past. To gain deeper (“smart”) modularization for a platform concept, this existing product structure must be completely broken up and analyzed. The first step must therefore be devoted to the scope of the structure to be considered and its breakdown into the smallest units to be considered. b. Subsequently, these units are analyzed with regard to their suitability for group- ing into modules. The parts especially their interfaces to other units and as part of the entire system (product) are examined (mechanical, structural, electrical, thermal, etc.). It must be determined which reasons exist (technical, supplier- related, complexity, etc.) for possible combining different units into different modules. One goal can be to minimize the criticality of interaction (deflection, parasitic loading, excitation of vibrations, thermal loading, etc.) between the modules to be defined.
- 291. 262 Wind turbine system design c. In the next step, based on the results, the units should finally be grouped into defined modules. As a rule, modules should finally have well- defined interfaces with low complexity and low risk of uncertainties of external influences or inherent re- action (feedback). The description of module interfaces and criti- cal module characteristics is essential for success within the whole process. If possible, modules’ interface descriptions and general characteristics should be standardized also for cross- platform benefits. Figure 5.34 Simplified example of “smart” modularization process for gaining a more generic WT platform concept and DT modules; source: IWES
- 292. Drivetrain concepts and developments 263 d. In the final step, after the definition and standardization of the modules, regard- ing their technical characteristics and interfaces and first developments by sup- pliers (prototype and/or model), a re- check, respectively, validation process according to V- Model (according to VDI/VDE 2206) has to take place. This is possible by means of calculation, simulation, module, and system testing as well as supplier assessment, by the OEM, who must be still the process owner. The grade of technical cooperation between OEM and suppliers can vary in order of the criticality of the respective module. e. In case of technical findings, identification of weak points, or optimization potentials as well as for new requirements, the OEM or a standardization group of different OEMs has to start the process partly or in total again. The number of original variants of DTs over all OEMs, which were previ- ously developed separately, could be greatly reduced through a concept of deeper (“smart”) modularization. On the component side (suppliers), the smart modular strategy should make it possible to push innovations on the component/module level, e.g., for individual generators or gearboxes, couplings (thus the higher the number of variants of modules with defined “standardized” interfaces) and at the same time achieving for the OEMs a high flexibility for specific product ranges or markets, in the best case. Thus, the smart modularization of product’s platforms enables strategies for the WT manufacturers, to offer a wide variety of products with a highly flexible variance of the individual modules worldwide at the lowest costs and highest quality. Also, this idea can be transferred further from the product to the production and product service. 5.3.5 Scalability of designs and performance indicators In many technical areas, the application of the similitude theory plays a major role in the planning and implementation of scaled model tests (wind tunnels, flow and towing basins, subsoil test pits, and all kinds of electromechanical laboratory test stands). Furthermore, similarity considerations generally allow statements about the properties and behavior of scaled components and systems. A prerequisite for this is the corresponding physical and/or system- theoretical descriptions of the system under consideration, not mandatory but often supplemented by a verification/test series on an example system based on measurements. In control engineering, the normalization of the transfer function, e.g., of the con- trolled system, is a related procedure, which enables us to simplify the controller syn- thesis for a specific class of controlled systems, e.g., electrical machines, mechanical DTs, and to limit the parameter spaces for the controller settings. The calculations and interpretations are conducted here with so- called per unit variables, i.e., dimension- less parameters as well as control and measured variables. The similitude theory in its original form also uses largely dimensionless parameters. The best known for WTs are the dimensionless coefficients for the rotor power cp (power coefficient), the rotor thrust cT (thrust coefficient), and the rotor torque cM (torque coefficient), as well as the
- 293. 264 Wind turbine system design aerodynamic blade profile coefficients cL (lift coefficient) and cD (drag coefficient). In mechanics, dimensionless parameters such as damping coefficients, e.g., Lehr’s damping factor, as well as the description of vibrations using amplitude and ampli- fication factors, as well as normalized mode shapes, are known generally. Electrical engineering also uses dimensionless parameters, e.g., in the form of stray numbers, winding factors, pole overlaps, relative short- circuit voltages. With the help of the similitude theory, scaling rules, also known as modeling rules, can now be derived for the WT as a whole but, of course, also for its subsys- tems and components [37]. In concrete terms, this can mean developing a model family with different sizes and corresponding performance characteristics from an existing (complete) system by scaling it up. In the simple but not trivial case, e.g., the (conceptual) upscaling of an existing 1 MW WT has already been tested in the field to a multi- megawatt turbine. The scaling as a rule has to be done with a scal- ing factor φ according to strict geometrical rules. All mechanical dimensions of the original plant design are then scaled with the same scaling factor, i.e., increased or decreased accordingly, and only then result in a strict geometric similarity. For certain operating parameters (e.g., nominal speed) of the original system, there are necessary changes for the scaled systems if certain technical boundary condi- tions (e.g., an unchanged top speed number, nominal wind speed, and wind class) of the system for the sake of compliance with design features (e.g., rotor coefficients) are physically required, due to the similarity theory. This scaling, considering bound- ary conditions (laws of similarity and environmental conditions), then leads to corre- sponding model laws (laws of scaling). These apply then to the model family under the specific boundary conditions and can be regarded them as basic model laws. Starting from this consideration of the overall system, the derivations (specific model laws) for the subsystems and components within this application result, i.e., specifically for this WT under these conditions. This understanding is important because, for example, a direct comparison of systems with different sized rotors in the current development does not necessarily aim at a higher turbine capacity (power), according to the proce- dure described above regarding the basic model laws. Increasingly, turbines are being designed with a similar design, but larger rotors for better energy yield, respectively, a better capacity factor, at low- wind sites, i.e., in the similarity analysis and thus the model laws, there are deviations from the basic rules, due to a changed nominal rotational rotor speed, wind speed, as well as rated generator power. The same applies, as an example, when reducing the design blade tip speed for the noise- optimized system design. Of course, a distinction must be made as to whether this is a special operating mode of the basic design or whether the system design as a whole has been optimized for this requirement. The theoretical scaling in conjunction with the similarity theory makes it possible to compare development lines from OEMs and sometimes also between OEMs with regard to performance indicators [specific weight- to- power ratio (tons/MW), specific power rating (W/m2 ), specific torque- per- weight ratio (Nm/kg), and weight- per- swept area ratio (kg/m2 )] without knowing all the design details. Due to a large number of additional influencing variables and uncertainties, these comparisons usually do not provide exact values, but at least usable qualitative statements (Table 5.2).
- 294. Drivetrain concepts and developments 265 Table 5.2 Modeling laws for basic scaling with φ (boundary conditions such as design tip speed, nominal wind speed, and wind class design remain constant) Physical size Sign Proportionality according to scaling factor Remarks Scaling factor φ ~φ1 Scaling of each mechanical dimension Area A ~φ2 e.g., cross section and surfaces Volume V ~φ3 Also for complex- shaped bodies Weight M ~φ3 Also for costs, as a first approximation Power P ~φ2 Capacity of a WT Torque M ~φ3 Torque from the wind rotor Rotational speed N ~φ −1 Torque of the wind rotor Lift forces FL ~φ2 For constant Reynolds number Drag forces FD ~φ2 For constant Reynolds number Rotor thrust T ~φ2 Thrust from the wind rotor Centrifugal forces FCe ~φ2 Only for constant lambda Gravity forces FG ~φ3 Almost critical point attention! Bending moments Μb ~φ4 Those related to gravity forces Natural frequencies ωn ~φ −1 longitudinal p c/m , torsional p c'/J Frequency ratios n/ωn ~φ0 Only for constant lambda Damping factor D ~φ0 Only for constant lambda Stiffness (torsional stiff.) c (cφ ) ~φ1 (φ3 ) Refer to natural frequencies Moment of inertia J ~φ5 Refer to natural frequencies Second moment of area IP ~φ4 Refer to torsional stiffness Section modulus WP ~φ3 e.g., important for strength Performance indicators Proportionality according to scaling factor Remarks Mass to swept area (kg/m2 ) ~φ1 Increasing (without innovation) Specific weight (tons/MW) ~φ1 Increasing (without innovation) Specific power (W/m2 ) ~φ0 Indicates the wind class design Specific torque (Nm/m2 ) ~φ1 Indicates the wind class design Torque density (Nm/kg) ~φ0 Constant (without innovation)
- 295. 266 Wind turbine system design 5.3.5.1 Wind turbine performance indicators Despite the clearly described scaling laws and similarity rules, it is not trivial to compare turbines of different sizes, stages of development, and DT topologies with one another in terms of their performance and level of innovation. As already men- tioned, the reasons for this often lie in the subjectively small but nevertheless rel- evant deviations of the individual main design variables (e.g., nominal wind speed, tip speed at design speed factor, wind classes) but also in the partly inconsistent, publicly accessible technical data sheet information regarding, e.g., nacelle and tower head masses, nominal speeds, power curves, from the turbine manufactur- ers or database providers. In the meantime, especially for the latest generations of turbines in the onshore and offshore sectors, there are hardly any clear technical, publicly available details, apart from the rotor diameter and nominal power of the turbines. The details of nacelle weights and tower head weights are for the most part based on indirect information from third parties, e.g., crane manufacturers and other sources that are difficult to verify and are therefore always subject to a certain degree of uncertainty. Clear reasons for these information restrictions on the part of the OEMs are not clear; one assumption is that transparent and detailed informa- tion would allow conclusions to be drawn about a possible future capacity of the platforms and a general comparability between different OEM products. Probably, these are not desirable from an OEM competitive point of view. Occasionally, how- ever, there is technical information directly from the manufacturers, which provides exemplary insights into the applied technologies and current physical limitations. The articles written by Eize de Vries in the Windpower Monthly magazine should be emphasized here, which always shine with a high degree of technical details and represent an important and trustworthy source of information about WTs and DTs. As an example, a short excerpt from an interview at “WindEurope 2019” in Bilbao about details on the then brand new V164/174 OWT from MHI Vestas shall be presented here. MHI Vestas CTO senior product manager Anders Bach Andersen talks to Eize de Vries about technical innovations: “Because 10.0 MNm gearbox input torque is a constant both turbines operate with 90 m/s rated tip speed, the resulting rotor speeds are 9.9 rotations per minute for the V174 and 10.5 rpm for the V164. Since power is a function of torque and rotational speed, …,” “the bigger V174 rotor turns slower, and the maximum power output must be reduced. To manage the generator temperature even during sustained operation with many full- load hours and/or in high- temperature regions, the V174 gearbox step- up ratio is increased from 1:38.1 to 1:40.8. This offers higher generator speed with reduced internal heat production and thus extra thermal reserves due to lower generator cur- rents for a given output” [38]. This example of slightly adjustable gear ratios shows quite impressively a major advantage of the geared DT concept compared to DDs. Furthermore, this example also illustrates the essential influence, means benefits, and disadvantages of the potential to increase the rated rotor- blade tip speed, especially for offshore applica- tions. The debate about proper blade tip- speed limits has a longer history, predicting that offshore- turbine tip speeds would increase within years from 80–90 m/s of the
- 296. Drivetrain concepts and developments 267 second and third generation of OWT to 100–110 m/s levels, to improve aerody- namic efficiency and to limit the input torque for big rotor diameter. MHI Vestas also stated in that interview, that due to reliability issues (rain erosion), they decided therefore to retain the 90 m/s rated tip speed of the V164 series for the V174. For the latest OWT generation, however, Vestas has dropped this limitation. The reason today as before lies more in the limitations of the current transmission technologies, whereby an important design criterion here is of course always the maximum input torque, which once exceeded would usually require a new development of the trans- mission for the higher requirements. For an analysis or comparison of WT developments using performance indica- tors, some basic explanations regarding the technical/physical assumptions to be made are necessary for the reasons mentioned above. The theoretically achievable turbine rotor power of a WT at a constructively defined nominal wind speed vN is calculated according to the following well- known equation (5.11), whereby cpmax here is not Betz’s optimum of 16/27 (refer to Chapter 2) but an estimated physically achievable value of roughly 0.5, at its best point, due to general limitations and losses (wake, tip/hub losses, friction losses, etc.) for conventional, modern three- blade WT: PR th. = 1 2 air DR 2 2 cpmax v3 N (5.12) air1.2041 (kg/m3 is the air density (standard value at 20 °C and 1 013 mbar) and DR(m) is the rotor diameter. This theoretical rotor power PR th. must now be evaluated with the usual efficien- cies in the DT and additional losses through auxiliary units (hydraulics, control, cooling, etc.). For this purpose, an aggregated maximum overall efficiency for the nominal point at a nominal wind speed of ηtotal = 88% is assumed. This was chosen rather arbitrarily but corresponds to the author’s average empirical values for vari- ous generic turbine design calculations and test bench tests. This results in a theoret- ical nominal turbine power PWT th (corresponds to electrical output power of the WT). PWTth. = total PR th. (5.13) This theoretically achievable turbine power also marks the theoretically achiev- able upper specific rotor area power under given conditions (best physical points, nominal wind speed, and normal atmosphere) and could theoretically serve as a best- in- class comparison value. Real turbine designs could be compared with their corresponding design data (assuming a known performance curve), and a theoreti- cal utilization factor of the respective design at a specific wind speed could be cal- culated (ratio of values). Typically, the utilization factors for nominal wind speed would be lower for the low- wind WT (IEC III, (IV)old , S) that have been developed more frequently in recent years, due to the design combination of large rotors and moderate turbine- rated power. Here the primary turbine values at the mean wind speed of the corresponding IEC wind class should be taken for comparison. With
- 297. 268 Wind turbine system design the current state of technical development of the WT, the characteristic values of the specific rotor power are suitable for a characterization, if calculated once about the WT nominal wind speeds and once again with regard to the mean wind speed of the respective IEC wind class of the WTs. If the ratio to the theoretical maximum value is calculated for both values according to (5.13), differences for a specific tur- bine show the degree of the design as a real low WT and, in the comparison between two turbines, the possible performance difference between those two WTs. Remark: Be aware the nominal specific rotor power of WT today (datasheet value) is just given as the ration of rated power and rotor swept area, independent of the WT IEC wind class design and its nominal wind speed, where rated power is reached [24]. So, this value gives only a general design, no WT performance indica- tion. Values around 350 W/m2 for offshore WTs and well below, as well as values around and below 250 W/m2 for onshore WTs clearly speak for a today “low- wind design,” although there are no official low WTs for offshore application by definition (no specific IEC wind class): = PWT_th AR = 1 2 air total cpmax v3 N = 0, 265 v3 N [W/m3 ] pAth. (5.14) tAth. min., max. = pAth. 1 !min.:::.!max. [Nm/m3 ] (5.15) This means an assessment of realized turbines with their data sheet- specific infor- mation for the specific rotor area output should be at least conducted with regard to the calculated theoretical value pAth. (5.14) at the nominal wind speed of the turbine. According to the model laws for WT scaling, however, the specific rotor power density has no dependencies on scaling. Therefore, it cannot be used for a charac- terization of innovations in turbine design. However, in literature or studies [35], sometimes, the classification “best in class” is used for the lowest values according to the IEC design wind class and the nominal power of the turbine, which is quite irritating and not reasonable. The specific rotor torque rating, which is unusual in the literature, seems more suitable for this purpose as a technological assessment. Due to the speed depen- dency at a constant design blade tip speed, a proportional increase is to be expected here when scaling WTs applying the same concept (no innovations), causing the special torque- dependent component weights to increase, e.g., gearbox, generator (especially DD), and also the passive structures have also to be reinforced [39]. These component dead weights (masses) applying a pure scaling of existing WT technologies, as already known, show a cubic increase. However, due to the lack of mass reference (and thus implicit costs), the specific rotor torque rating does not also have any pronounced evaluation properties but rather serves to ensure the general comparability of designs, e.g., if the design nominal speed is reduced for noise- optimized operation with the same rotor size and nominal wind speed. Such turbine designs would be recognizable not only by their specific rotor power but also by their comparatively high specific rotor torque tAth. (5.15). In simplified terms, it can be stated that the specific rotor power and the specific rotor torque serve more as a general characterization of the turbine design.
- 298. Drivetrain concepts and developments 269 Mass- related parameters, on the other hand, are better suited as performance indicators. The reason for this is that we generally expect a cubic increase of nacelle and tower head mass. This, in turn, leads (for pure scaling) to a lin- ear increase of tower head mass (rotor+hub+nacelle) to rotor swept area ratio, which, viewed on its own, refutes the economic argument of larger WTs solely through scaling without innovations (refer to [40]). Previous investigations (refer to [24, 27 and 41]) have shown that when looking at different generations of turbines, this ratio moves within certain, relatively narrow limits, which can only be achieved through the consistent introduction of innovations in always bigger WTs. Typical values here are in the range slightly below 20 kg/m2 (typi- cally 18 kg/m2 ); previous experiences show that it has not yet been possible to fall below 15 kg/m2 for multi- megawatt WT, at least when using the current technologies. For that reason and because no really disruptive developments can be identified on the market in the short and medium term, this should cur- rently mark our lower limit (best point). It is even more astonishing that turbines with ratios of 22 kg/m2 and more (~26 kg/m2 ) are currently able to assert them- selves on the market. The very simplified cost estimate, which increases linearly with the weight, does not seem to play such a decisive role here or is superim- posed and compensated, respectively, by other effects at the OEMs (production, assembly, supply chain, logistics, etc.). Nevertheless, the ratio provides a good first estimate of the resource consumption (level of utilization) of the respective turbine technology (especially in the rotor and nacelle area) with regard to the required specific use of material resources. Another interesting parameter from the author’s point of view is the specific torque density (Nm/kg), which is also used to assess the force density and thus the specific performance of gears (range of high- end WT transmission at 200 Nm/kg and slightly higher). Just to emphasize the technological improvements and inno- vation, let us state here on a component level, 10 years ago, 100 Nm/kg marked the best point for WT gearboxes. For the entire turbine, this proposed performance parameter (Nm/kg) is related to the tower head mass or nacelle weight and there- fore includes the aerodynamic and structural design of the rotor as well as the DT/ nacelle design. Therefore, it means this key parameter is a strongly aggregated vari- able. Within modern WTs, it is in the range of 11–25 Nm/kg (rotor torque to tower head mass ratio), whereby the current values for DD concepts are currently roughly estimated 10–15 % higher than those for the best- geared WT and thus currently mark the best- in- class point, in terms of general (not evaluated) resource efficiency of the material. According to the laws of pure scaling, this performance parameter should be quite constant. If we compare older with newer realized designs, we see a clear trend in implementing innovations, a higher utilization of material, based on the same and the introduction of new materials with higher strength properties, documented in rising values over time (5–10 Nm/kg in 1996, 8–15 Nm/kg in 1999, refer to [24]). For the specific weight- per- power ratios (tower head mass/rated power) due to the currently very similar blade tip- speed interpretations, we would expect a linear
- 299. 270 Wind turbine system design increase in terms of pure scaling. However, caution is advised if these differ sig- nificantly, e.g., Vestas V90 nominal blade tip speed 76 m/s (37 tons/MW), the SG- 8.0- 167 DD (~47 tons/MW), and the V236- 15 (~55 tons/MW) both with ~104 m/s. Then a direct comparison with the help of this performance indicator is less reason- able or just shows the existence of innovation. As an example, considering scaling in its simple form without corrections, we would expect roughly 1.8–2.5 times higher values, so 68 tons/MW for the SG- 167 and 93 tons/MW for the V236. Indeed, the SG- 167 and V236 stay well below these values, which means innovations must be present. Here, the use of the previously mentioned ratio of rotor torque related to tower head mass is the better choice for a direct comparison. In concrete num- bers, this means for the turbine types over different generations: V90 (15 Nm/kg), SG- 8.0- 167 DD (17 Nm/kg), and the brand new V236- 15.0 (21 Nm/kg), which draws a quite more realistic and direct picture of lines of innovations for WTs over the time. 5.4 Onshore wind turbines and drivetrain developments At this point, a short outline of the development history of onshore WT and espe- cially their DTs for the most well- known western OEMs should be given as an example. This should in no way belittle the development achievements of other manufacturers in other regions, e.g., Asia, Brazil, or those that are no longer on the market; the restriction is simply necessary to limit the size of this section. As in the introduction of this chapter already explained, looking back since the early 1980s, there has been an immense variety, especially with regard to DT developments and variants in the onshore sector. This applies to specific, technical solutions not only for main components (e.g., transmissions, refer to Chapter 6) but also for the DT concepts as a whole (combinations of different main components). Today, in the 2020s, this diversity is still considerable. Due to various technical developments and boundary conditions (e.g., growth in size and specific markets) as well as ongoing market consolidation among OEMs, suppliers, and operators, a certain saturation and focus (e.g., compact, medium- fast DT for very large onshore/offshore platforms, DD in particular for large offshore systems, classically designed DT for mid- size commodity WTs at emerging markets and regions) is clearly visible in the concept variants for DTs. This does not mean that new developments or re- engineering of concepts that are already known but not yet established on the market (see Chapter 6, multi- generator concepts) are not taking place. The specific variety of vari- ants (NPIs) in the platform concepts of the manufacturers Nordex, ENERCON, Vestas, SGRE, and GE, which are also examined in more detail here, is still high but not extreme. This is primarily about rotor- specific and rated power variants (specific rotor area power density) for certain wind and site classes and regional markets, but each is based on the same DT concept or WT platform of the respective OEM.
- 300. Drivetrain concepts and developments 271 5.4.1 ENERCON ENERCON started producing WTs in the mid- 1980s. The first turbines E- 15/16/17 in the power range of x0kW and a little later the E- 32 with at least 300 kW already equipped with a synchronous generator and full converter, until that time ENERCON turbines were still equipped with two- stage gear- boxes. With the first gearless series WT E- 40 (500 kW) from 1992, the wind pioneer ENERCON (the founder Aloys Wobben, respectively) achieved the technical and economically breakthrough already in 1993, again 2 years later, ENERCON played with the E- 66 in what was the former WT top class of 1.5–2 MW (the number behind the Enercon “‘E-” always defines the rotor diameter of the WT type, respectively). Later on, ENERCON introduced the designation EPx as an additional type designation, which indicated the belonging to a spe- cific xMW platform (with the corresponding rated power range) of the turbine. The E- 66 thus formally belongs to the EP1 class, the successor E- 70 2004 with 2–2.3 MW already to the EP2 class. Until 2017, ENERCON always relied on the tried- and- tested basic nacelle and DT concept of the E66 for all new WT developments, with the characteristic egg- shaped nacelle, first made of GRP and later on made of aluminum, since the introduction of the E- 82 series. This consistency is also a reason why the practical verification of scaling laws can be shown and verified particularly well on this WT up to 2017. During that time, there were no fundamental changes in concept, but evolutionary, partial improvements only (e.g., cooling conversion to partial water cooling within the E- 82, 2006- 2014, EP2 series); Figure 5.22 (bottom) shows the basic representa- tion of the long- established ENECON “old” DT concept. This concept applied a combined bearing of the rotor hub and the generator rotor on a “long” king- pin structure with a two- bearing bearing solution [small bearings in diameter with large bearing spacing, see FEM Figure 5.22 (top)] in the hub. The laws of scaling and the adherence to the generator technology as an electrically excited internal rotor synchronous machine and the rotor suspension principle resulted comparatively high nacelle or tower head masses for the ENERCON WTs until the recent past. The top marked, until the introduction of the V164- 8.0 MW Offshore WT from Vestas, the uprated E- 126 (7.58MW) EP8 from 2010 was the most powerful onshore WT ever. This type had a nacelle weight includ- ing the hub of 364 tons; with blades, the tower head mass was ~650 tons. Its “grandfather” model the E- 66 had 67 tons for the nacelle and ~100 tons for the total head mass. Besides other ENERCON thus underlined the prejudice, which is still rampant in some cases, that DD WTs generally have a very high nacelle mass, especially for high- rated power WT. The generator of the E126 alone already weighed ~220 tons, but the high mass in total to large extent resulted from the special, necessary rigid construction of this Enercon turbine series. Until the EP4 series from 2014 with the type E- 126 and E- 141 remained true to the “old” ENERCON design principles, so the E- 141 still had a tower head mass of ~490 tons. The 2018 less powerful but very innovative, optimized
- 301. 272 Wind turbine system design platform EP3 with a separated generator (just in front of the nacelle and without a long king- pin structure) with integrated bearing unit already showed a sig- nificant reduction in the specific nacelle weight with 256 tons (more compact structure, GRP introduced for the nacelle cover, Al winding system, etc.). Here for the first time, the first technological influence of the turbine manufacturer Lagerwey, alongside Vensys and Goldwind, a specialized OEM for WT with DD PM synchronous generators, which ENERCON took over at the end of 2017, is quite obvious. In 2019, ENERCON then developed the new EP5 series with types E- 136, E- 147, and E- 160 partially based on the Lagerwey WT plat- form L136. Already the ENERCON EP3 and Lagerwey LP4 platforms show a remarkable similarity in DT and structural technology choices, apart from the generator technologies (EESG vs. PMSG, see below), but both apply pre- formed, casted Al winding coils. Both compact DD designs feature an inner- rotor generator at the 690 V level (which is also new for ENERCON, applying 400 V before) and incorporate a main bearing unit with a hollow shaft and two pre- stressed tapered- roller bearings instead of a single bearing, according to Figure 5.23. Both concepts allow easy service access to the rotor hub, by the hollow bearing unit structure, in contrast to older ENERCON designs. With similar functionality, the Lagerwey PMG has a diameter of “only” 5.5 m and is mainly passive air- cooled. The main bearing solution has, by comparison, a longer hollow shaft, much smaller bearing interspacing, and uses oil lubrica- tion. ENERCON’s EP3 uses a conventional externally excited generator with slightly above 6 m outer diameter, smaller- size bearing spaced further apart, and grease lubrication. The E- 160, the first ENERCON WT with PMSG (inner rotor design) with 5.56 MW, currently marks the peak performance type of the series (prototype 2020/21). Already in early 2022, ENERCON built up the next evolution of the EP5 with a new E- nacelle concept, which includes, for the first time for ENERCON, all electrical equipment (converters and transformers) within the nacelle and not split between the nacelle and the tower base. This basic modularity promises an advantage for the production and commissioning of turbines; also tower base constraints (dimensions) are omitted. Thus, for the new EP3 and also for the EP5, ENERCON changed its nacelle appearance from egg shape to the “vintage” E40 design of course in a more modern look. 5.4.2 Nordex In 1987, this German OEM with Danish roots produced the N- 27/250 with 250 kW nominal power, which was then the most powerful series WT in the world for a short time. Just like ENERCON, the Nordex Company has remained true to its original DT concept. Nordex relied for their WTs on classic three- stage gearbox in combi- nation with high- speed asynchronous generators up to the year 2000, which were designed for still quasi- fixed speed operation and passive stall control. However, the systems had pole- changing generators so that at least two speeds for low wind (partial load) and high wind (full load) speed ensured an acceptable energy yield.
- 302. Drivetrain concepts and developments 273 The early systems N43/600 and N54/1000 from 1994 to 1996 used a three- point bearing suspension and thus a low- integrated DT concept according to Figure 5.8. Until 1998 with the N60/62, Nordex continued to apply on stall control and directly coupled asynchronous machines without converters (only soft starter AC connection by means of Thyristors) with system outputs up to 1300 kW. The variable- speed operation in conjunction with active pitch control for rotor- side power control (refer to Chapters 3 and 7) and speed control by the generator was first introduced with the licensed builds S70 to S82 (formerly Südwind) 1.5 MW from 2001 and later with the AW70 toAW82 (formerlyAcciona). On the generator side of these WTs, partial con- verters were applied in conjunction with double- fed asynchronous machines. The basic mechanical DT topology (classic high- speed gear concept with three- point rotor main shaft suspension, with low- integration level, refer to section 5.2.2) has been retained to this day. Nordex is using the three- point suspension concept and the principle- related introduction of a high proportion of parasitic wind rotor forces into the gearbox structure, currently for rotor diameter up to 163 m and WT rated power between 6 and 7 MW. This is remarkable, 5 years ago, the 3- MW WT class was considered the magic limit for the technically reliable use of this DT topology. The year 2000 marked a turning point in the development of turbines for Nordex, with the N80/2500, a completely newly developed turbine, however, based on the well- known DT topology, being put into operation. At that time, again the most powerful series turbine was available on the market. Following a redesign of the tur- bine family, which Nordex calls “generation Beta,” different rotor diameters are now available for the platform (N80, N90, and N100) with the same rated power class. The third stage of evolutionary development, the so- called “Gamma class,” took over the technical DT features from the beta version again. The technical innova- tions and evolutionary adjustments were related, among other things, to improving the ease of maintenance. The Gamma class includes the N90/2500, N100/2500, and N117/2400. The Nordex Company experimented with the N90/2500 in a nearshore environment, but Nordex withdraw all offshore plans in very early concept phases (e.g., N150/6000, which was planned with a DD PMSG concept, quite remarkable for an OEM that solely used high- speed geared DTs with three- point suspension and DFIG system, until that time). The last stage of development, the development of the Delta class, began in 2013 and can also be described again as evolutionary. The DT concept remained unchanged, but the system output was increased to the 3+ MW (3.3–3.9 MW) range. A variety of site- specific rotors from 100 to 131 m are offered, which requires a certain flexibility on gearbox input torque/loads and generator torque and speed characteristics. In 2017, the even more powerful Delta 4000 platform for rated outputs from and over 4 MW made its debut, one key differ- ence in comparison to the normal “Delta” Platform was the installation of the entire E- System (converter and transformer) within the nacelle. As already mentioned above, currently, the N163/6.x marks the top (rotor diameter and rated power) of that WT platform with a reinforced gearbox and an electrical system with improved cooling. The series production is scheduled to start in 2023. With regard to system performance variants, Nordex relies on site- specific adjustment options and is able to take optimal account of local conditions.
- 303. 274 Wind turbine system design 5.4.3 General Electric wind energy (GE) The history of WTs from GE started with the companies Tacke Windtechnik (Germany) and Zond (United States) which were bought by Enron in 1997. After the Enron bankruptcy in 2002, GE took over their wind division. GE Renewables was and is particularly well represented on the onshore WT world market, the rather changeable offshore division only gain traction with the takeover of Alstom’s energy division in 2015. However, in terms of market share, GE occupies third place behind Vestas and Siemens Gamesa among the western manufacturers. Especially in the onshore sector, GE was influential with its turbines on the way that the standard WT developed into a commodity with the largest possible quantities, standardized production, and low prices. GE’s 1.5- MW WT already reached the limit of 10,000 manufactured and installed systems in 2008. A total of ~16,500 GE 1.5sle turbines were installed. Since the takeover of Enron’s wind division and from a historical perspective, GE has rather relied on reputed robust DT concepts, which means non- integrated designs with separated DT parts (no dual functionalities of parts) as far as possible (refer to Figure 5.7). The 3 MW platform is still being produced using a four- point main shaft suspension in principle; however, in the case of the 3 MW platform, it is not designed with two completely separate bearing housings but integrated into a cast structure at the nacelle frame. The gearbox then hangs on the comparatively short main shaft, supported by two torque arms. The design is relatively compact compared to the standard four- point bearing design, but not a dedicated lightweight construction due to the massive nacelle support structure. In contrast, the older 2 MW GE platform, still a high seller in specific markets, applies a three- point suspension for the rotor shaft. A three- stage gearbox with torque arms and a high- speed DFIG system is the standard configuration for the other DT parts of this smaller platform, too (refer to Figure 5.8). Since the introduction of variable- speed turbine control, GE has relied almost exclusively on double- fed asynchronous machines with partial converters as a particularly cost- effective solution for the elec- tric DT. The WTs of the former 1.x MW range (especially the 1.5 MW types) also used the classic three- point suspension for the rotor shaft. After gearbox damage increasingly occurred in the field at that time, comprehensive test and validation pro- grams started at the NREL with regard to DT dynamics and reliable main bearings configurations [20, 42], which continue till now. The development of the very suc- cessful 2 MW platform has its roots in the design of the 1.5 MW (1.5i) in 1996. The 1.5sle with its large numbers was presented in 2004, followed by variants up to 1.85 MW and with rotor diameters up to 100 m until 2013, in comparison to the Western OEM competition, comparatively small systems (rotor sizes) with lower capacity (rated power). During the development of the 2 MW platform and the introduction of the 2.5 MW DT, high- speed PM generators with a full power converter were used for the first time. But already with the introduction of the GE 2.5- 120 variant, GE returned to the internally valued, above all cost- effective, DFIG technology. The 2 MW and 3 MW platforms are primarily aimed at emerging markets and mass markets, where robust systems with uncritical specific wind farm site area to power density appear to be proving their worth. GE now develops on the basis of the 2 MW
- 304. Drivetrain concepts and developments 275 platform the so- called “Serria” platform, especially for the onshore US market with high- capacity factors and rated power in the 3 MW range. The high- end WT in the onshore area at GE mark the 5–6 MW class of the so- called CypressTM platform presented in 2017, with the currently largest turbine, a GE6.x- 164. Surprisingly, GE again uses a “kind of” three- point suspension system for this largest of all platforms, which in the past seemed to be more suitable for smaller and medium- sized (more cost- sensitive) turbines. Like Nordex, GE is now also using the concept of three- point suspension in the 5–6 MW class. In the case of GE, however, in a kind of modified form, the second main bearing (downwind posi- tion) is in the gearbox housing, but this is not only additionally reinforced, but the classic, rather flexible torque arm supports are also missing. Here, the gearbox con- nection to the supporting nacelle structure is more massive, rather like the second bearing support of a four- point main shaft suspension, which means a kind of hybrid design out of both concepts, but formally, it belongs to three- point suspension class. Also, for its biggest onshore WT, GE uses DFIG technology with a partial power converter for the grid connection. 5.4.4 Vestas The Danish company Vestas, originally a manufacturer of agricultural machinery, is one of the oldest OEMs (production of WT started in 1979) and has been the world’s largest manufacturer of WTs for a number of years. It started in the early 1980s with a few 55- kWWTs and quickly developed, thanks to large orders from the US market and culminated in the founding of Vestas Wind System A/S in 1987, which focused exclu- sively on wind energy. Vestas has been active in the offshore wind energy sector since 1995. In 2004, Vestas merged with the Danish manufacturer NEG MiconA/S, which in turn was the result of a merger between Micon A/S and Nordtank Energy Group A/S. The first Vestas WTs from a few 10 kW up to the power range of 1–2 MW use asyn- chronous generators for their electrical DTs as standard and were connected to the grid in the larger systems via thyristors (AC soft starters). Thus, these stall control turbines ran more or less at a fixed speed (only slip frequency changes within a few percentages from nominal speed were possible with the concept) on the grid. These first Vestas turbines in the lower power range usually had a three- stage gearbox, in which the main rotor bearing was integrated as a double bearing. This was of course manageable with rotor diameters of 20–30 m. One of the most successful turbines from the earlier Vestas portfolio was the V52- 850 kW (the so- called 1 MW platform), the production and license were sold to Gamesa (a Spanish OEM) in 2001. The design of the V52 was quite remarkable in many respects and marks the transition of former high- end WTs to the MW or multi- megawatt WTs of today’s design like no other. The rotor shaft had a robust double bearing, which means a very early bearing unit (with a common bear- ing housing), and the gearbox input shaft was connected to the main shaft by means of a shrink disk. The gearbox had two torque arms, and the weight of the transmission rests largely on the output side of the main shaft. The already applied DFIG generator here had the OptiSlipTM control patented by Vestas, which can change the rotor resist- ance and thus the slope (slip characteristic) of the generator characteristic near the
- 305. 276 Wind turbine system design synchronous point by means of thyristor controllers. This allowed the DT to be oper- ated ±15 % of the respective synchronous speed of the generator, i.e., with a quasi- variable speed. The V52 had a pitch system, but compared to today’s systems, it only had a central pitch cylinder for the mechanically synchronized blade angle adjustment of the three blades, actuating through the hollow main shaft and partially the gearbox (the first gear stages), too. Today, for safety reasons, a central pitch for large WTs is no longer permitted because the service brake can no longer be dimensioned to be suffi- ciently strong, and redundancy would therefore be lacking in the event of a simple fault. With the acquisition of NEG Micon, there was also a merging of technology strands. Micon already used the less costly three- point suspension for the rotor shaft support of larger WT. Thus, the mechanical design of the Vestas V72- 1.5 MW and V82- 1.5 MW was derived from the corresponding NEG Micon WTs. For the very successful V80- 2 MW, which was also used in early offshore projects, Vestas relied on the robust double bearing, but here in a separated design (no bearing unit), with two separated main bearings, and again with high- speed OptiSlipTM generator system, which means a DFIG with controlled rotor inner resistance linked to a three- stage step- up gearbox. The following series at Vestas were then also designed with DFIG but in combina- tion with a partial converter for better efficiency and controllability. To that date, there have been comparatively few innovations for the DTs of the Vestas WTs; the designs were almost entirely no- risk designs based on proven predecessor concepts. With the introduction of the V90- 3 MW 2003, however, Vestas presented a real boom of inno- vations. The details of the V90 are explained in more detail in the following section about offshore turbine DT development. Also, the following backward role of Vestas to “safe” designs with the type V112- 3.0 MW which, however, was equipped with a three- point bearing in comparison to the 2 MW platform, as was already common at Micon years before. For the onshore sector, Vestas developed its 2 MW platform start- ing with the V80- 2 MW continuously from the V90- 2.0 MW to the V120- 2.2 MW and experimented around 2011 with the use of high- speed PM synchronous generators for the first time. Just like the 3 MW, the current 2 MW platform from Vestas uses a three- point suspension and a high- speed generator concept with a three- stage gearbox. At the end, a DFIG system is applied for the electrical DT of the 2 MW platform, and the 3 MW platform first used a PM synchronous generator with full converter, after the world price of rare earth materials increased around the year 2013 Vestas switched to the asynchronous generator with full converter. The 3 MW platform starts with the V105- 3.45 and currently ends with the V155- 3.6 MW, which was set up as a prototype in 2021. From 2017 on, the development of the 4 MW platform began, which is based on the 3 MW platform and covers outputs of up to 4.5 MW. The 4 MW platform also has a three- point bearing, but in a special, hardened form. Indeed, very similar to GE’s CypressTM - platform concept, for the 4 MW Vestas platform, the bearing supports are reinforced against each other. Vestas uses an additional structure here in the form of a kind of half- shell, which positively connects the front bearing point to the gearbox housing above the main shaft. This type of suspension system can be seen as a kind of hybrid between three- point bearing and four- point bearing. Due to the integration of the second main bearing (downwind side) in the gearbox housing, however, it is formally considered a three- point suspension.
- 306. Drivetrain concepts and developments 277 The 7- MW onshore platform EnVentusTM , on the other hand, was a fundamen- tally new development, which currently marks the peak performance for onshore WT with the V162- 6.8 MW and the V172- 7.2 MW (presented in April 2022). The EnVentusTM (from 2019) DT (mechanical and electrical) jointly developed by Vestas and the transmission specialist ZF is trimmed for maximum modularity. Therefore, the design of the machine house follows a kind of container design in which the mechanical DT and the electrical subsystems are housed separately from each other. The mechanical DT relies on a low- integrated solution with a bearing unit (includ- ing a prestressed tapered roller bearing arrangement), which is directly bolted to the nacelle main frame. Here, in turn, the experiences from the offshore turbine develop- ment V164/174 at Vestas had an influence. The flexible coupling that was still there on the rotor shaft/main shaft was omitted here. The gearbox is bolted to the bearing unit via a flange connection; the same applies to the medium- speed PM synchronous generator on the gearbox output side. The bearing unit thus bears the dead weight of the gearbox (with two planetary stages) and generator (refer to Figure 5.15). Overall, the structure appears compact but very solid due to the separated, massive bearing unit. Vestas seems to have done without introducing dedicated flexibilities (cou- plings) for damping or decoupling the components from one another and to mitigate the parasitic rotor or constraining forces as far as possible, which makes very pre- cise handling and assembly of the entire EnVentusTM unit necessary. This platform, together with the experience from the V164/174, forms the basis for the new super- class offshore turbines up to 15 MW (refer to section 5.5). 5.4.5 Siemens Gamesa Renewable Energy The roots of this traditional OEM based in Spain lie with the former Danish WT manufacturer Bonus Energy A/S (founded 1980) and its German cooperation part- ner and licensee AN Windenergie (since 1989). The basis for the early systems in the 1980s and 1990s was classic high- speed geared DTs with directly grid cou- pled asynchronous generators. For the first larger WTs still below 1 MW, the OEM quickly relied on three- point bearing suspension for these stall- controlled, quasi- fixed speed WTs (refer to Figure 5.8). Just like the former ENERCON turbines with their characteristic egg shape since the E- 66, the AN Bonus turbines from 450 kW upward are easily recognizable by their torpedo- like nacelle shape with a relatively pointed hub spinner end. The DT concept was followed at Bonus Energy up to the AN Bonus 2300/82 type. With the takeover of Bonus Energy by Siemens (2004), only the naming changed to SWT- 2.3- 82, and the DT topology remained unchanged. The next stage of evolution was the introduction of the “VS” variant, which now enabled variable- speed operation by applying a full converter between the asynchronous generator and the grid. Siemens (or Bonus Energy) took this technological step comparatively late compared to its competitors. These older 2.x MW WT were among the last and largest stall- controlled series WTs. With the SWT- 3.6- 107/120, which is also described in the following OWT section, Siemens launched one of the first dedicated offshore turbine families on the market, which differed from the early 2.x MW platform essentially in the four- point bearing
- 307. 278 Wind turbine system design suspension of the high- speed geared DT. In 2008, Siemens started developing its first DD WT platform with inner rotor PMSG prototypes. But later on the result- ing series of different 2–4- MW onshore turbines started with the SWT- 3.0- 101 DD looked completely different. The comparatively low nacelle weight of 78 tons and total tower head mass of (78 + 60) tons (the IEC III wind class type SWT- 2.3- 113 DD had 73 tons nacelle and 67 tons rotor weight) was already evident here not only in comparison to the corporate group own gearbox WTs (SWT- 2.3- 101 with rotor mass 62 tons and nacelle mass 82 tons) but also in direct comparison to the Vestas best in class V90- 3.0 MW with 70 tons nacelle +41 tons rotor but a 10% smaller rotor and at a higher rated wind speed. What also become evident here is that the assessment of performance (specific weight to power, etc.) of WT/DT designs by scaling becomes quite tricky, due to different rotor power densities and rated wind speeds of the WT types. In this respect, the DD proved to be extremely competitive for the first time in comparison to the classic, high- speed gearbox DTs that had dom- inated to date. All DD series from Siemens are still equipped with a Moment bearing integrated in the generator, which means they all have very compact and fully inte- grated DTs (refer to Figure 5.23). On that early DD onshore basis, Siemens devel- oped the D7 offshore platform, the precursor of the latest 14 MW OWT (refer to section 5.5). With the merger of Siemens and the Spanish OEM Gamesa, an onshore specialist for low- cost markets, the DD platforms were no longer built for onshore applications but exclusively developed and offered for OWT. The onshore market, on the other hand, was completely taken over by Gamesa within the group in 2016. Their 8- MW OWT platform with Hybrid- Drive, very similar to the Vestas V164, from the joint venture Adwen from AREVAWind (formerly Multibrid) and Gamesa, was quickly abandoned in favor of Siemens DD Platform in the offshore sector. The AD8- 180 thus remained the only dedicated OWT ever built by Gamesa. Gamesa was formerly associated with Vestas through a holding and used some of Vestas technologies under license (2001 small onshore WTs G52- G80). Gamesa made a major technological step in 2009 with the G128- 4.5 MW, which had the world’s largest rotor at that time. Gamesa applied the DFIG with a partial power converter for the electric DT in the previous power range of 2–3 MW. However, the G128- 4.5 MW was designed with a Hybrid- Drive (two- stage planetary gear, i = 1:38) and a PM synchronous generator with a full converter. The rotor was supported by a robust bearing unit (double bearing) with a flanged gearbox and a medium- speed generator (refer to Figure 5.15). After the merger with Siemens and the lack of success of the Hybrid- Drive platform G1xx- 4.5 on the market, Siemens Gamesa today concen- trates on the classic high- speed DT with a three- stage gearbox and DFIG system again. The DT of the largest platforms 4.x and 5.x (from SG 5.0- 132 to SG 6.6- 170) is mounted via a compact bearing unit (double- bearing suspension). The three- stage gearbox is mounted on two torque supports (according to four- point suspension) and is connected via flange by a shaft adapter (refer to Figure 5.9). It remains similar to the G128- 4.5 MW in some aspects. While the smaller 2.x platforms (up to G114) apply “classical” separated four- point suspension, as does the 3.X platform (G132), too (refer to Figure 5.7).
- 308. Drivetrain concepts and developments 279 5.5 Offshore wind turbines and drivetrain developments Due to the subsequent chronological development of offshore WTs, the diversity of their DT concepts and variants used over time is somewhat less manifold than those for onshore turbines, but nevertheless impressive. Even if the current market shares of the two main DT concept speak rather clearly in favor of the DD concept, the high- power Hybrid- Drive for offshore applications with two to three planetary stages, which was mainly pushed and built by Vestas, has won a respectable posi- tion. For OWT, the competition for the best DT concept remains open, too, whereby a disappearance of the DD technology in this market segment is unimaginable from today’s perspective. As in the onshore sector, some OEMs, including pioneers in the offshore sector (e.g., BARD,Areva, Senvion/former REpower, Sinovel) disappeared for different reasons from the market after some initial success. Other well- known turbine OEMs decided at a very early stage not to get involved in the “adventure” offshore wind energy, such as ENERCON or Nordex; some appeared undecided, such as GE, where the time phases of commitment and abandonment alternated. This draws a quite realistic picture of the offshore wind industry as a modern adven- ture for manufacturers, suppliers, and operators. Little or poorly usable experience from the offshore and maritime sector paired with major technical challenges (sea- bed foundations, very large WTs, complex electrical shore connection, complex logistics for construction and maintenance, and limited accessibility) and the most adverse environmental conditions (sun, salt, ice, ocean currents, and maritime veg- etation cover) standing for a few challenges, which can be somehow compared to other great technical “adventures” like the moon landing. Therefore, all pioneers in this field deserve a high degree of respect for the achievements, which are already taken for granted today in some cases. Strictly speaking, the era of offshore wind energy started in 1991 with the con- struction of 11 turbines of the 450 kW class (hub height 35 m, rotor diameter 35 m) in the first offshore wind farm Vindeby in Denmark; 25 years of experience from operating the systems in shallow water (water depth below 4 m, distance from shore 2 km) under rough offshore conditions is the reward for this Danish pioneering work when the farm was decommissioned in 2017, due to economic reasons. Important findings were missing in other countries, e.g., during the construction of the first German offshore wind farm, indeed with a completely different turbine size, water depth, and distance from the coast, when they started. The contracted OEM for the wind farm Vindeby at that time was Bonus Energy. They supplied a modified ver- sion of their standard Bonus 450 kW WT for the project. The turbines were modi- fied for offshore use by sealing the towers, controlling the humidity inside with air conditioning to extend the life of the machinery, and hardening the gearboxes using heavy- duty gearbox concepts with an input planetary stage. The first industrial phase of offshore wind energy started in the year 2000 with the first noteworthy, comparatively xx MW near- shore farm installations in the North Sea (DK/Middelgrunden, UK, Sweden) consisting of smaller multi- megawatt turbines ([43], Table 5.3). Viewed from the history of development, the
- 309. 280 Wind turbine system design early “standard” offshore WT DT (Vestas V66- 2 MW/V80- 2 MW, Bonus B76- 2 MW), based on onshore turbine DTs with non- or low- integration level, had either two separated main shaft bearings (classic four- point suspension) or a three- point suspension concept (Enron EW70- 1.5MW, NEG Micon NM72, and Siemens SWT- 2.3- 82), a three-/four- stage gearbox, and mainly high- speed asynchronous genera- tors (DFIG or squirrel cage IG). The DD technology was not up for discussion at this early stage, probably also due to the rejection of the DD pioneer ENERCON regard- ing a clear offshore commitment. From a technical point of view, the ENERCON turbines, which at that time were generally equipped with open air- cooling systems, would probably have had been assessed extremely critically for pure offshore envi- ronmental use. None of that early offshore turbine was equipped with a full con- verter system. Various power control concepts such as thyristor- based AC- converter solution with full pitch control or pole- change concepts with fixed speed and active stall control started this first age of commercial use of offshore wind energy. This has changed relatively fast to full pitch control (blade adjustment in direction of blade feather position) and the use of self- commutated converter (e.g., Siemens with their “VS” upgrades for the SWT- 2.3 or DFIG with partial power converter like later V80 version) for variable- speed operation. It is essential to understand that all of these first OWTs were not real dedi- cated offshore turbine developments, but slightly modified onshore technology and designs, which could be one explanation for some of the technical reliability prob- lems that crop up not exclusively but more frequently at the multi- megawatt WT DTs during these early phases of commercial used offshore wind farms. Maybe in some cases, the missing design bases for OWT at that time left no other practi- cal alternatives to gain further experience for the upcoming multi- megawatt turbine classes and thus lead sometimes to underestimation of requirements (loads, dynam- ics, and environmental conditions). From the beginning, in the wind industry and especially in the offshore area, a rapid turbine growth was repeated mantra- like as the key solution for falling level- ized cost of energy. This led some newcomers in the offshore WT manufacturer’s segment to a technologically daring entry into 5+ MW turbine development from the scratch without any track record or field experience with the applied technology, especially in the area of the DT (e.g., use of highly integrated concepts, Hybrid- Drives, and single- bearing suspension). Table 5.3 Typical OWTs of the first phase of commercial offshore wind farms Turbine type Rated power (MW) Rotor diameter (m) Rated wind speed (m/s) Rated rotor speed (rpm) Tower head mass (tons) (nacelle + rotor) Tip speed (m/s) SWT-2.3-82 SWT-2.3-93 2.3 2.3 82.4 93 14 17 (16) 82 + 54 (82+60) 73 (78) V80-2.0MW 2.0 80 15 19 61 + 34 80
- 310. Drivetrain concepts and developments 281 Nevertheless, the need for dedicated offshore turbine development was quite obvious, and the era of such turbines, for the second phase of offshore wind farms (Table 5.4), started with Enrons 3.6 MW (later GE 3.6s) turbine that was one of the first offshore- dedicated machines. Originally designed with a 100 m rotor, this was replaced by a 104 m rotor after GE bought Enron’s assets in April 2002. The tur- bine incorporated a conventional non- integrated high- speed geared DT with DFIG and 4- point main shaft suspension (refer to Figure 5.7). As an OWT, this machine already had innovative equipment features that are customary and necessary to this day in order to take into account typical offshore requirements (rough environmental conditions, on- site logistics). Some of them we see nowadays within some of the latest OWT developments again, e.g., a container solution beneath the nacelle and behind the tower where the electrical/electronic equipment gets a protected environ- ment, for main component exchange a 40- ton foldable portal crane in and a helicop- ter platform installed above the nacelle. Seven units were produced for operation at the Arklow Bank project of Ireland in 2003. Plans for a mass- and cost- optimized GE 3.6sl successor with an enlarged 111- m rotor were dropped, and GE rather stayed away from the offshore sector, except for a short intermezzo with the upcom- ing DD technology also for offshore application in the early 2010s. At that time, GE canceled its development of a 4.1 MW offshore WT with 113 m rotor and a non- integrated DD (PMSG) concept with double suspension main shaft support [refer to Figure 5.20 (top)]. The only prototype built, owned by Goteborg Energi, was erected in Goteborg in 2011. Then again GE disappeared from the offshore wind segment for years until its acquisition of Alstom in 2015. From then on, GE began to catch up offshore technology, but more details on that later. First, we want to follow some of the historical timelines for offshore turbine and especially DT development. The second innovative turbine development with a really ground- breaking DT concept was the V90- 3.0 MW from Vestas. It was not originally designed for off- shore purpose only but used for it and with some design features that enabled the advance into the 5+MW offshore turbine class and at the same time pioneering for the later development lines of high and fully integrated DTs. This early and cou- rageous design owns the Krone for the best (lowest) weight- to- power ratio (tons/ MW) of a nacelle with DT at that time and even considered scaling rules even some Table 5.4 Typical OWTs of the second phase of commercial offshore wind farms Turbine type Rated power (MW) Rotor diameter (m) Rated wind speed (m/s) Rated rotor speed (rpm) Tower head mass (tons) (nacelle + rotor) Tip speed (m/s) GE 3.6s (sl) - version II 3.6 104 (111) 14 (15.5) 13,6 185 + 83 73 (79) V90-3 MW 3.0 90 15 16,1 70 + 41 76 SWT 3.6-107 SWT 3.6-120 3.6 107 (120) 13.5 (12.5) 13 125 + 95 (125 + 100) 73 (81)
- 311. 282 Wind turbine system design years later. The V90- 3.0 MW nacelle (24 tons/MW, tower head mass also remark- able 37 tons/MW) was the attraction at the 2003 Husum Wind technology trade show with a new main shaft suspension system concept using only a single main bearing (Moment bearing), fully integrated within the gearbox housing (refer to Figure 5.14). Despite having 50% more capacity and a 27% increase in rotor- swept area over the predecessor the V80- 2.0 MW, the nacelle mass and dimensions remain nearly unchanged, also for offshore applications. Inside the high- speed DT, the gear- box with the integrated single main bearing for the rotor was directly flanged to a cast main carrier (main frame), by that, eliminating the “traditional” main shaft of “classic” geared multi- megawatt DTs, in series WTs the first time. The gearbox unit consisted as usual of a combination of a two- stage planetary and one helical output gear stage. The gearbox lubrication system was a forced feed system (dry sump lubrication, refer to Chapters 6 and 7) without the use of an integrated oil sump. Thus, the V90- 3.0 gave an example of the technical potential of a high- integrated DT design. Scaling along proven conventional Vestas V80 technology lines with a classic distributed DT construction (with a basis of ~31 tons/MW for the 2 MW nacelle and a tower head mass of 48 tons/MW) would have increased head mass and thus material costs substantially. Nevertheless, there is always an increased failure potential when introducing several major innovations in a new development at the same time. Unfortunately, and despite a very extensive field test program at 15 prototype sites, the V90 series soon developed major gearbox issues and establishing the root cause proved time- consuming. In early 2007, Vestas was forced to withdraw the offshore version before being reissued for offshore use in May 2008. A redesign (gearbox, main bearing, and carrier structure) pushed its total head mass up from 104 to 112 tons, which still counted as lightweight against equivalent- sized competitors, but some reputational damage was already done. One consequence for Vestas at that time, the successor the V112- 3.x MW marking a return to a very conventional (no- risk) DT with a three- point suspension, this WT type was foreseen for offshore and onshore appli- cations. Another consequence at Vestas was ongoing huge investments in testing infrastructure to mitigate technical risks of innovation or quality issues of suppliers in early phases and not to carry them into series. To summarize 122 new developed up, the V90- 3.0 MW can be seen as a real milestone of WT DT development and has pushed innovation in general in that phase of rather conservative (“no- risk”) turbine designs, despite its technical problems. The third player in that early offshore market segment was Siemens (former Bonus Energy) with its SWT- 3.6- 107 (later -120), which was a turbine type with an absolute no- risk, conventional distributed, geared high- speed DT with separated, double- bearing main shaft (four- point) suspension system, which, according to Siemens, was hardened (means reinforced) for dedicated rough offshore use. The story of offshore wind energy in Germany started in that phase with the approach of some near- shore prototype OWT installations. So the first offshore pilot projects were started up as near- shore plants in 2004 (Enova Offshore Ems- Emden), 2006 (Rostock offshore plant), and 2008 (Hooksiel), in the year 2010 followed the Alpha- Ventus offshore test field, Germany’s first offshore wind farm with 12 of the
- 312. Drivetrain concepts and developments 283 most powerful offshore turbine (REpower/later Senvion 5M- 126 and Multibrid/later Areva M500- 116) available on the market at that time, which heralded the era of the first 5- MW OWTs (Table 5.5). With the Areva M5000, medium- voltage technology with IGCTs for a full power converter system was used in WTs for the first time. In terms of converter reliability, this innovation has so far shown noticeable advan- tages over common, conventional low- voltage installations based on low- voltage IGBT (1 200 or 1 700 V types) modules. The big OEMs Vestas, Siemens, and GE remain a power class below with their 2–3.6 MW turbines during this period. The BARD Offshore 1 Park was Germany’s first commercial offshore wind farm when it went into operation in 2013. This 400 MW wind farm is located 100 km off the northwest German coast in the North Sea and consists of 80 BARD 5.0- 122 newly developed WTs. The REpower/Senvion 5M- 126/6.xM- 152 utilized a robust, conventional (non- integrated) four- point suspension system with high- speed three- stage gearbox and DFIG system. Roughly speaking, REpower was supplying a well- engineered but not highly innovative offshore turbine. Accordingly, specific performance data such as the specific weight- to- power ratio (tons/MW) and the spe- cific rotor- swept area performance (W/m2 ) are relatively modest in direct compari- son to more innovative concepts from other OEMs. Also, the BARD OWT followed a rather conservative turbine concept, partially documented by the high tower head masses, too, but showed some innovations within the DT. The conventional high- speed DT was combined with a separated single- bearing (OTRB and moment bear- ing, respectively) solution with a greatly shortened main shaft (refer to Figure 5.13) and thus a comparatively small overall length. A special feature here is the separate, comparatively soft support of the gearbox via torque arms and elastomeric hydraulic supports, similar to “classic” three- point suspension solutions, which usually apply double raw SRB for the upwind bearing without capabilities to carry moments in the rotor plane. Measurements on the turbines in the field [44] documented still rather high dynamically displacements of the entire gearbox during operation, taking into account the actually bending resistant main bearing. The generator, a DFIG system, Table 5.5 Era of first 5+MW OWT class introducing the third phase of OWT development Turbine type Rated power (MW) Rotor diameter (m) Rated wind speed (m/s) Rated rotor speed (rpm) Tower head mass (tons) (nacelle) + rotor/hub + blades Tip speed (m/s) REpower 5M 6.2M-126 6.2M-152 5.08 6.15 6.15 126 126 152 13 14.5 11.5 12.1 12.1 11.1 (290) + 120 (325) + 134.5 (325) + 156.5 80 80 88 Areva M5000-116 M5000-135 5.0 5.0 116 135 12.5 11.4 14.8 13.4 (200) + 111.5 (235) + 140 90 95 BARD 5.0 5.0 122 12.5 15.7 (280) + 70 + 3*28.5 100
- 313. 284 Wind turbine system design is also supported separately. The Multibrid (later Areva) M5000 had the highest degree of innovation in this turbine class and showed for the first time in this 5+ MW OWT capacity range a fully integrated DT design with a medium- speed PM generator. An additional special feature here was the very low speed level with a gear ratio (one- stage gearbox) of less than 1:10 compared to today’s Hybrid- Drive concepts. In combination with a single- bearing solution for the rotor, the M5000 marked a compactness of the DT in the offshore turbine sector that has never been realized again, besides the later SCDTM (aerodyne engineering). Innovative journal bearing solution was introduced in the Renk gearbox within the planetary gear stage. So far, at least no gearbox problems have been reported in the systems, which are still in operation. However, in the area of the main bearing, there was extensive rework necessary, probably due to quality issues, which made the replacement of entire nacelles on the offshore park site necessary. With regard to the specific weight- to- power ratio of the nacelle/(tower head mass), for a time the M5000 sets the reference value of approx. 40/(62) tons/MW for the first 5+ MW OWT class. However, this does not apply for the rotor swept area- mass ratio with 26–29 kg/m2 and the entire turbine torque density of 11 Nm/kg of the M5000, due to the comparatively small rotor sizes for an OWT. On the other side, even a simply scaled V90 (rated power, rotor diameter, applying basic model rules) would be slightly below the values of the weight- to- power ratio, which exemplary documented not only the performance for high- integrated geared DTs in general but also the absence of a real technology leap compared to the Vestas V90. Some years later, the fully integrated DDs from SGRE took the lead in terms of lightweight OWT nacelle/(tower head) mass with values of ~30/(47) tons/MW and up to ~17 kg/m2 entire turbine torque density. The latest and still ongoing phase (fourth phase) of OWT development (super class OWT with potential capacities of 10 MW and above, Tables 5.6–5.8) started in the mid of the 2010s with a bang and a technical innovation step similar to the launch of the Vestas V90. Siemens Wind Energy (later SGRE) introduced a new offshore turbine DD (PMSG) product line with the SWT- DD 6 MW and a 154 m rotor design. Siemens offshore DD technology roadmap started about 2011. This DD concept applies a PM synchronous generator with an outer rotor design and an integrated single main bearing as a central suspension system (refer to Figure 5.23). The turbine was upgraded first to 7 MW, later on to 8 MW capacity, and was avail- able from 2019 on with a new 167- m rotor as SG 8.0- 167 DD. According to Siemens Gamesa, the SG 7.0- 154 DD turbine was a slight evolution of the 6- MW model, with some minor optimizations and release of design margins. Both versions incor- porate the already proven in- house generator family (~6 m diameter and 2–2.5 m length) design, which offers easy internal service access to the hub via a hollow gen- erator pin- king shaft and the integrated single rotor/generator moment bearing. Two separated power- electronic converters are mounted inside the nacelle, and a 33 or 66 kV medium- voltage transformer is located in a separate compartment underneath and behind the tower. The latest incremental evolutionary upgrade with a nominal power of 8 MW provides a power mode option up to 9 MW under certain condi- tions and for a certain period of time. It is equipped with an enlarged 167 m rotor.
- 314. Drivetrain concepts and developments 285 The generator with improved closed- loop internal air cooling operates at a slightly higher rated speed, because the tip speed increases roundabout by 10% compared to its predecessors with a 154 m rotor. At Siemens Gamesa, offshore turbines with gearboxes [the so- called internally (older) Siemens G4 platform] do not seem to matter anymore in the future. The top of development there was marked by the SWT- 4.0- 130 (former 3.6 MW design). The usage of gear technology at SGRE nowadays is focused on onshore WT on the basis of Gamesa 3.x, 4.x, and 5.x platforms. On the other hand, gearless onshore sys- tems, originally based on the Siemens D3 platform (4.2 m outer rotor PMSG) with capacities from 3.3 to 4.7 MW and rotor diameters of 120–130 m, were only built in small numbers until around 2017, seem to be no further under ongoing development. Following the purchase from Ecotècnia by Alstom and later on the acquisition of Alstom by GE, the production for the 6MW Haliade OWT started in 2016 at the St. Nazaire factory. For its DT, the Haliade featured an inner rotor DD PMSG con- cept with a very high mechanical robustness (based on Alstoms PureTorqueR design principles, refer to Figure 5.24). The prototype started an extended test period in spring 2016 at Østerild Wind Turbine Test Field. At that time, the most powerful turbine under test (Prototype from 2014 to 2016 at Østerild Wind Turbine Test Field) was MHI Vestas V164- 7.0 MW, with a power Table 5.6 Era of the latest (fourth phase) super class OWT—development line of SGRE Turbine type Rated power (MW) Rotor diameter (m) Rated wind speed (m/s) Rated rotor speed (rpm) Tower head mass (tons) (nacelle + hub) + blades Tip speed (m/s) SWT-6.0-154 DD 6.0 154 13.0 11.0 (275) + 3*25 89 SWT-7.0-154 DD 7.0 154 13.5 12.0 (285) + 3*25 96 SG-8.0-167 DD 8.0 167 12.0 12.0 ~(285) + 3*30 104 AD8-180 8.0 180 15.0 8.5 ~(356+ 107) + 3*34 80 SG-10.0-193 DD 10.0 193 ? ? ~(400) + 3*40 ? SG-11.0-200 DD 11.0 200 ? ? ~550* ? SG-14.0-222 DD 14.0 222 ? ? ~600* ? SG-14.0-236 DD 14.0–15.0 236 ? ? ~650+ ? “?” means, no reliable public information available, “~”means, slightly more, “*” assessment by the author, no public data available.
- 315. 286 Wind turbine system design mode option up to 8.4 MW. The main shaft of the V164 was supported by the main bearing unit with two main shaft bearings. Flange connections between main DT components eliminate misalignment risks, while a flexible shaft coupling on the low- speed side should supply “pure” rotor torque transfer to the medium- speed two- stage gearbox and flange linked SG, to avoid, respectively, minimize parasitic gear- box loads. This ongoing fourth phase of OWT development (Tables 5.6–5.8) was first characterized by the race for the first WT with a double- digit capacity, which was already in full swing, even if the key figures of the first turbines (6–8 MW) suggested a certain distance from this target. The speed of development and introduction of new uprated WT types were impressive high at that time. Siemens erected its first Table 5.7 Era of the latest (fourth phase) super class OWT – development line of Vestas Turbine type Rated power (MW) Rotor diameter (m) Rated wind speed (m/s) Rated rotor speed (rpm) Tower head mass (tons) (nacelle + hub) + blades Tip speed (m/s) V112-3.3 3.3 112 13.0 12.8 (17.7) (157.0) + 3*11.9 104 V164-10.0 10.0 164 13.0 10.5 (375) + 3*35.0 90 V174-9.5 MW 9.5 174 13 9.9 (390) + 3*35.0 90 V236-15.0 MW 15.0 236 13* 8.4* ~800* 104* “?” means no reliable public information available, “+” means slightly more, and “*” assessment by the author, no public data available. Table 5.8 Era of the latest (fourth phase) super class OWT—summarized development line of GE Turbine type Rated power (MW) Rotor diameter (m) Rated wind speed (m/s) Rated rotor speed (rpm) Tower head mass (tons) (nacelle + hub) + blades Tip speed (m/s) Haliade 150-6 MW 6.0 151 12–13* 11.5 ~(400) + 3*26 91 Haliade X12–14MW 12–14 220 12–13* 7.81 ~(600+85) + 3*55 89 “?” means no reliable public information available, “~” means slightly more, and “*” assessment by the author, no public data available.
- 316. Drivetrain concepts and developments 287 wind park equipped with its brand new offshore DD turbine SWT- 6- 154 DD in 2015 in UK, shortly afterward, in 2016, GE followed with the newly developed Haliade 6 MW in the USA. Since 2017, MHI Vestas with the new V164 flagship first with 7 MW, later 8 MW rated power (Burbo Bank wind farm, UK), could boast of building the most powerful offshore series WT in the world for some years (refer to Figure 5.16). Senvion (former REpower) and Siemens have carried out further increases in performance (upgrades) and, in some cases, adjustments to the rotor diameter on their existing platforms (refer to Figure 5.7), so that capacities of 6.3 MW- 152 m (Senvion) and 7–9 MW- 167 m (SGRE) could be achieved without fundamentally new or really up- scaled designs. It was then again MHI Vestas that brought the original V164 concept in its final stage up to 10 MW and was the first manufacturer ringing in the double- digit megawatt class era for OWTs. So MHI- Vestas made that remarkable single step from V90- 3.0 MW/V112- 3.3 MW OWT class to a dedicated 8–10 MW super class OWT. ZF Wind Power supplied the differential- type medium- speed planetary gearboxes for both the initial 8 MW V164 and the latest 9.5/10 MW upgrades. Their differential- type gearboxes differ from conventional planetary designs through the incorporation of torque- splitting technology for the low- speed gear stage, also patented by ZF. Now, in terms of time, we are entering the real area of super class OWTs, which are not disruptive new developments, but clearly based on the technological concepts of their predecessors (proof of technology concepts- the former 6–8 MW class). Vestas again launches the world’s first 15 MW OWT, setting new standards with a 236- m rotor size. The Danish company is aiming to install a V236- 15.0 MW prototype in summer 2022, followed by a series of ramp- ups in 2024. The announcement follows the full acquisition of the former offshore joint venture with Mitsubishi Heavy Industries (MHI). However, even during the MHI Vestas era (2014–2020), the joint venture relied on Vestas know- how for their offshore tur- bine developments. The V236- 15.0 MW has been developed for high- wind IEC I/S/T condi- tions, including typhoon- prone markets (IEC T, operating in typhoon- prone conditions means a turbine must be able to withstand extreme wind speeds). As mentioned, this latest- generation offshore platform (refer to Figure 5.15) is based on proven technical solutions from the proven V164/174- 9 MW offshore as well as the EnVentusTM onshore platform, clearly obvious for the DT topol- ogy and the modular nacelle concept, which features main power converters housed in container- like side compartments. Not new but different from V236 direct predecessors is the increased rated tip speed from ~90 m/s standard value for the V164 and V174, and also somehow common for current offshore wind application—to above 100 m/s. An important benefit of a faster- spinning rotor is a slightly better aerodynamic performance (efficiency) and much more relevant from a design perspective, since the reduction in gearbox input torque allows for a more compact design. But of course, higher tip speeds also harbor the risk of an accelerated blade leading edge erosion and thus inherent higher OM (Operational Maintenance) costs.
- 317. 288 Wind turbine system design Vestas itself stated that the entire DT V236 is functionally an upscaling along proven EnVentusTM design principles. The V236 gearbox utilizes three planetary step- up stages versus two within EnVentusTM design, to compensate for the much higher rating and larger rotor with an inherently reduced rotor speed with an overall higher gear ratio. According to Vestas, the specific gearbox layout further enables smaller dimensions, especially for the first low- speed gear stage. Similar to the smaller EnVentusTM unit, the V236 gearbox has journal bearings in all reliability- prone positions. Its torque density exceeds somewhat 200 Nm/kg. Furthermore, the low- speed coupling between the main bearing unit and the gearbox, deployed with the V164/V174 series, originally integrated to prevent rotor- induced bending moments from entering the gearbox input side, was eliminated, which increases inherently risks for DT main components. On the other hand, Vestas has gained some experience because this expensive DT element had already been omitted from the EnVentusTM DT. In consequence, a reinforced interface between the main bearing unit and gearbox seemed to be necessary to handle the non- torque reaction loads and the overall much higher turbine capacity. Furthermore, accord- ing to Vestas, the in- house- developed permanent magnet synchronous generator, also based on experiences with the EnVentusTM platform, has a bigger diameter and is slightly longer because the rating is 2.5 times higher. They stated again the advantage of slightly higher generator speed in order to optimize using the rather scarce and expensive rare earth Dysprosium, characterized by higher price volatility and more supply- chain risk compared with neodymium. However, adding dyspro- sium allows a generator to run at much higher magnet operating temperatures with less risk of demagnetization, so in consequence, just leaving it away, means higher effort for cooling and thermal control at all. Vestas supported this by connecting the fully sealed air- cooled PMG with an air- water heat exchanger to the passive rooftop cooler, as it is already common for DD (e.g., Siemens Gamesa) and comparable high- performance electrical machines for years. Already in April 2018, GE announced that they will start testing the world’s larg- est WT prototype at that time—the Haliade- X— at facilities in Blyth, England. GE’s renewable energy department signed a 5- year contract with the British government- funded Offshore Renewable Energy Catapult to begin trials of the 12 MW turbine. Plans in September 2020 already called for a new upgraded version of 13 MW GE Haliade- X turbine to be installed at Dogger Bank Wind Farm by 2023. A prototype was installed at Port of Rotterdam, and a test run with an already uprated 14 MW pro- totype started there in October 2021. In the same year, this GE flagship in the super class OWT area, the upgraded Haliade- X, was certified for typhoons. The new super class OWTs show around similar capacity ratings 14–15 MW (maybe with not stated yet, but inherent design margins for up to ~17–18 MW) more differences with respect to their specific power rating relating to the swept rotor area, most probably due to the ongoing, very international development of the offshore mar- ket for different regions. Something is already established in the onshore market with its specific and broad requirements for single- market segments or regions. So, the new V236- 15.0 MW comes with a 343 W/m2 specific power rating, set against 311 W/m2 for the V236- 13.6 MW, and 400 W/m2 for the current V174- 9.5 MW flagship. GE’s
- 318. Drivetrain concepts and developments 289 uprated 14- MW rated Haliade- X offers 405 W/m2 , and Siemens Gamesa’s upcoming SG 14- 222 DD in 14- MW “standard” mode 362 W/m2, respectively, 320 W/m2 for the latest announced SG 14- 236 DD. In line with wind industry practice for onshore and offshore, newly introduced platforms always have future scalability in their DNA. The initial V164- 7.0MW (331 W/m2 ) platform was followed by a V164- 8.0 MW prototype and commercial model (379 W/m2 ), then again uprated to the V164- 9.5 MW (450 W/ m2 ), and finally to a V164- 10.0 MW (473 W/m2 ). Perhaps further scaling of the cur- rent platforms and OWT flagships along the same evolutionary development pathway could therefore in future result in uprated outcomes toward 20 MW or site- specific extremely low specific power ratings of 250–300 W/m2 (refer to MingYang latest MySE16.0- 242 OWT, 16 MW power rating using a low- integrated Hybrid- Drive with bearing unit and a three- planetary gear stage gearbox, and to Goldwind GW12- 242 with 261 W/m2 , also applying a low- integrated Hybrid- Drive PMSG DT). So, this Top OEM (No. 4 worldwide) takes the step and seems to leave the proven former path as a DD specialist, exactly the other way around Siemens did it in the past. This shows once again very clearly, nothing is fixed, and the race for the best DT concept in the wind industry remains open. 5.6 Outlook and potential development trends A look back at the technical developments and the global growth of WT installations shows, on the one hand, clear, consistent long- term trends with regard to various aspects [3, 5, 45]. This applies to the growth of onshore and offshore WTs, which has already been mentioned several times in the chapter, in terms of the rotor swept area and the output capacity (rated power). In 2022, the current prototypes that are planned for the upcoming wind farms will have 14–15 MW for offshore and almost 6–7 MW for onshore applications. In retrospect, this growth of turbine sizes did not always run at the same speed, but generally always with considerable dynamics and with clearly positive gradients. However, the OEM’s commitment to take technical risks for the development or adaptation of the turbine key systems (rotor, DT, and tower) varied greatly during this period from the 1980s to the 2020s (i.e., within 40 years of development history). Roughly speaking, cycling periods of high and low risk developments are clearly visible, sometimes temporal shifted between markets, regions, and OEMs. Especially in recent years, the gradient of growth for turbine rotor swept area and capacity has again increased significantly. A clear end even a leveling off is not yet observable. But the last few years seem to have been shaped more by evolutionary developments than by groundbreaking developments, even if the turbine sizes are really remarkable. Further optimizations were carried out, espe- cially for the DT, and existing design margins were exploited. This approach is also documented by the performance indicators, which do not indicate any technological leaps for the latest developments of OEMs. On the other hand, the lack of such disruptive innovations can be clearly rec- ognized. But what would be such a disruptive technological leap? For example, the introduction of HTS [46] technology is suitable for series production, especially
- 319. 290 Wind turbine system design in the area of DD, in order to significantly reduce the specific installation space for a higher capacity and the dead weight of the generators compared to the cur- rent state. In recent years, the company Envision in cooperation with partners has developed a prototype- ready, scaled 3 MW generator for field testing in a WT. The first test runs at Fraunhofer- IWES on the 10 MW generator/nacelle test bench DyNaLab in Bremerhaven (Germany) were quite promising. HTS technology for WTs [46–50], which currently remains limited to DC sub- applications within the generators (means DC excitation of the SG) and offers the technological poten- tial to increase the value of the air- gap flux density (currently with state- of- the- art PM technology limited to ~1 T) to significantly higher values clearly above 2 T. In order to achieve this, the entire generator must be designed accordingly (e.g., materials and magnetic flux paths within the generator). The constant power require- ment for the cooling capacity of considerably above 100 kW for very large systems 15 MW remains disadvantageous. To date, the construction and weight advantage has been decompensated by the additional systems required and the expensive spe- cial materials, which are additionally not easy to handle within generator produc- tion process. The cooling system for the HTS has to run for the stand- by function of the WT, which does not negligibly reduce the overall efficiency, flexibility, and availability of operation. Finally, nothing or very little is known about the techni- cal reliability of the systems (cryogenic systems in offshore use, rotary transmitter for cooling medium, HTS materials in dynamic operation, etc.). So, the hurdles to introduce new technology are always quite high. A related question is whether the current growth will continue in this form. The main argument of the proponents for even bigger turbines is the lower number of required foundations and associated internal inner- array cabling. As well as lower service costs due to the smaller number of turbines (fewer personnel resources, etc.) should enable decreasing LCoE, as far as the theory goes. The increasing complex- ity and border lining utilization of material and physical principles within the largest turbines speak against this. So, components become sometimes specifically more expensive; thus, a 5 MW gearbox has lower specific costs than a high- end 20 MW gearbox; the same applies to generators, pitch systems, etc. Summing up some key factors from the view of the author that could enable further substantial growth of turbine size and power rating: • • new materials and designs for generator air- gap flux densities above 2 T • • internal operating temperature ranges of 200°C and above for electrical compo- nents and power electronics (permanently and with high reliability) • • new HTS materials for higher temperatures and at lower prices, combined with a good processibility characteristics (e.g., less brittle) • • efficient cooling for higher current density in generators and converters/cooling in general in combination with sealed systems (protection against environmen- tal influences) • • new material surfaces (processes and coatings) supporting significant higher wear resistance and surface pressure (e.g., for more than 25- year bearing life- time design and clearly exceeding dynamic stress limits of 1 650 MPa as well
- 320. Drivetrain concepts and developments 291 as extreme stress limits of 4 000 MPa [6, 51]). Metal- composite materials with higher high- cycle fatigue capabilities for dynamic loading (e.g., main shaft, support structures, hub, and gearbox housing) • • fatigue strength of connection elements/systems (gluing, screwing, shrinking, etc.), more reliably quality (process), and monitoring technics • • new materials/coatings with high strength and surface wear- out resistance for mixed friction operation • • further introduction of journal bearing (hydro- static, -dynamic, combined) for further critical application in the WT (e.g., for main shaft suspension, yaw-, pitch-bearings) • • wireless energy supply and DD technology for blade- pitch systems • • continuous, in- situ residual lifetime assessment of individual DT/WT compo- nents—call it “real” condition monitoring of individual components Furthermore, there are reasonable assumptions for even broader application of wind energy utilization in the future. This means some more specific applications, e.g., in the offshore sector with dedicated hydrogen production in off- grid operation directly in the wind farm or the requirements for new operational regimes for WT, e.g., to compensate for the lack of dynamic operation capability (in terms of degra- dation) of electrolyzers or for the direct use as mechanical/hydraulically drives for heat pumps applications. So, at the end, the share of WTs with similar designs to current turbines dedicated to low wind speed applications could increase, including special operational strategies, just to achieve the highest capacity factors for specific applications. Currently, there seem to be no materials (which have already left the laboratory stage) available to enable further middle- term uninterrupted growth of the turbines. In the very personal opinion of the author, simple upscaling the currently existing and now rather outbid technologies and platforms of the 15 MW offshore class and 7–8 MW onshore class makes less sense if the LoCE shall be further reduced according to the general growth and scaling rules (refer to section 5.3.4). Maybe if also in future even less sites are available, and the overall installed capacity has to be maximized under this boundary condition. The not new but interesting idea of Multi- Rotor- WT [52] should not be underestimated for application with special requirements for highly concentrated power per footprint. The same is for the idea of splitting the rotor power internal into smaller portions, which are easier to handle with more standardized components (e.g., multi- generator concepts, refer to Chapter 6). However, the scaling idea could also push a reverse approach, which has so far been little discussed until now. This is about downscaling, i.e., scaling down the already optimized technology of the super class turbines in order to use their tech- nological potential, e.g., in terms of costs not only CAPEX (lightweight design) but also OPEX (reliability) for special applications, e.g., in the onshore area and markets or the already mentioned multi- rotor concepts. This could end up with much smaller high- end turbines combining maximum efficiency, reliability, and minimized envi- ronmental emissions (noise, shadow, and hazards potential). Just as a compromise in order to take into account the general problem of acceptance through a special
- 321. 292 Wind turbine system design design and layout and still generating at an attractive LCoE, especially in densely populated areas of industrialized nations, i.e., a certain departure from commodity (low- cost) WTs toward a technological even more optimized, smarter, renewable power plant, at least for specific global regions. Of course, further growth of standard turbines will certainly remain attractive for other markets (mass markets or markets with other boundary conditions). Future Mega WT with outputs 20 MW, on the other hand, could remain the clearly favored direc- tion of developments for floating [53, 54] and autonomously swimming WT (“free floater”). Here an optimized availability and reliability as well as the ratio between the mass of the supporting and floating structure and the tower head mass may be more relevant. References [1] Voith Turbo Wind GmbH Co. KG Windstrom einzigartig erzeugen. Die Windrive-Technologie. technical report; 2018. [2] Schulze T. Ganzheitliche Dynamische Antriebsstrangbetrachtung von Wind- energieanlagen, dissertation. Göttingen: Sierke Verlag; 2007. [3] Rohrig K., Berkhout V., Callies D, et al. ‘Powering the 21st century by wind energy—options, facts, figures’. Applied Physics Reviews. 2019, vol. 6(3), p. 031303. [4] Karimi H.R. Structural control and fault detection of wind turbine systems[online]. IET Energy Engineering, Band 117. ISBN- 10:1785613944; 2018 Aug 31. Available from https://digital-library.theiet.org/content/books/ po/pbpo117e [5] Wenske J. Direct Drives and Drive- train Development Trends [Technical Report]. Fraunhofer; 2012 Apr. [6] Chovan C., Fierro A.J. Improving Bearing Life in Wind Turbine Main Shafts and Gearboxes [technical artical, The Timken Company; 2021]. technical report, 2012. [7] Bergua R., Jove J., Echarte J. ‘Pure torque drivetrain design: a proven solution for increasing the wind turbine reliability’. Brazil Windpower Conference; technical report, 2014. [8] Guo Y., Bergua R., van Dam J., Jove J., Campbell J. ‘Improving wind turbine drivetrain designs to minimize the impacts of non- torque loads’ [online]’. Wind Energy. 2015, vol. 18(12), pp. 2199–2222. Available from http://doi. wiley.com/10.1002/we.v18.12 [9] Zhang H., Ortiz de Luna R., Pilas M., Wenske J. ‘A study of mechanical torque measurement on the wind turbine drivetrain - ways and feasibilities’. Wind Energy. 2018, vol. 21(12), pp. 1406–1422. [10] Fanzhong M., Wenske J., Gambier A. ‘Wind turbine loads reduction using feedforward feedback collective pitch control based on the estimated effec- tive wind speed’ [online]’. American Control Conference (ACC); Boston,
- 322. Drivetrain concepts and developments 293 MA, 2018. Available from https://ieeexplore.ieee.org/xpl/mostRecentIssue. jsp?punumber=7518121 [11] Srikanth P., Sekhar A.S. ‘Wind turbine drive train dynamic characterization using vibration and torque signals’. Mechanism and Machine Theory. 2020, vol. 98, pp. 2–20. [12] Bartschat A., Morisse M., Mertens A., Wenske J. ‘Analysis of dynamic inter- actions between different drivetrain components with a detailed wind turbine model’. Journal of Physics. 2016, vol. 753, p. 082022. [13] Wenske J., Beckert U. ‘Sensorless control for active damping of torsional vibrations in wind turbine drivetrains with doubly- fed induction generator’. Renewable Energy and Power Quality Journal. 2016, pp. 242–248. [14] Chen L., Xu H., Wenske J. ‘Active damping of torsional vibrations in the drive train of a DFIG wind turbine’. Renewable Energy and Power Quality Journal. 2014, pp. 270–275. [15] Xu H., Chen L., Wenske J. ‘Active damping control of DFIG wind tur- bines during fault ride through [online]’. 3rd International Conference on Electric Power and Energy Conversion Systems (EPECS); Istanbul, Turkey, 2013. Available from https://ieeexplore.ieee.org/xpl/mostRecentIssue. jsp?punumber=6704878 [16] IEC standard 61400-1:2019-02. Wind energy generation systems – part 1: design requirements. Geneva, Switzerland: International Electrotechnical Commission; 2018. [17] Röder J., Jacobs G., Duda T., Bosse D., Herzog F. ‘Investigation of dynamic loads in wind turbine drive trains due to grid and power converter faults’. Energies. 2016, vol. 14(24), p. 8542. [18] Helsen J., GuoY., Keller J. ‘Gearbox high‐speed‐stage bearing slip induced by electric excitation in a test facility [online]’. Wind Energy. 2018, vol. 21(11), pp. 1191–201. Available from https://onlinelibrary.wiley.com/toc/10991824/ 21/11 [19] Gallego- Calderon J., Natarajan A. ‘Assessment of wind turbine drive- train fatigue loads under torsional excitation’. Engineering Structures. 2015, vol. 103, pp. 189–202. [20] Guo Y., Keller J. ‘Combined effects of gravity, bending moment, bearing clearance, and input torque on wind turbine planetary gear load sharing’. Conference Paper NREL/CP- 5000- 55968; Dearborn, MI, 2012. [21] Veritas D.N. Guidelines for design of wind turbines. 2nd Edition. Denmark: Wind Energy Department, Risø National Laboratory. ISBN 87- 550- 2870- 5; 2002. [22] Helsen J., Peeters P., Vanslambrouck K., Vanhollebeke F., Desmet W. ‘The dynamic behavior induced by different wind turbine gearbox suspension methods assessed by means of the flexible multibody technique’. Renewable Energy. 2014, vol. 69, pp. 336–346. [23] Jenkins N., Burton T., Bossanyi E., Sharpe D., Graham M. Wind Energy Handbook. 3rd ed. Hoboken, NJ: Wiley; 2014 Sep. Available from https:// onlinelibrary.wiley.com/doi/book/10.1002/9781119451143
- 323. 294 Wind turbine system design [24] Gasch R., Twele J. (eds.) Windkraftanlagen. Grundlagen, Entwurf, Planung und Betrieb. 7. Aufl. Wiesbaden: Vieweg + Teubner Verlag; 2011. [25] Manwell J.F., McGowan J.G., Rogers A.L. Wind Energy Explained. 2nd Edition. West Sussex: Wiley; 2011. Available from https://onlinelibrary.wiley. com/doi/book/10.1002/9781119994367 [26] Hau E. ‘Windkraftanlagen, Grundlagen, Technik, Einsatz, Wirtschaftlichkeit.’ 4 Auflage. Heidelberg: Springer; 2008. [27] Hau E., von Renouard H. Wind Turbines, Fundamentals, Application, Economics. 2nd edition. Berlin, Heidelberg: Springer. ISBN- 10 3- 540- 24240- 6; 2006. Available from http://link.springer.com/10.1007/ 3-540-29284-5 [28] Tan J., Zhu C., Song C., Han H., Li Y. ‘Effects of flexibility and suspension configuration of main shaft on dynamic characteristics of wind turbine driv- etrain’. Chinese Journal of Mechanical Engineering. 2019, vol. 32(1), p. 36. [29] Li Y., Zhu C., Chen X., Tan J. ‘Fatigue reliability analysis of wind turbine drivetrain considering strength degradation and load sharing using survival signature and FTA’. Energies. 2019, vol. 13(8), p. 2108. [30] Kirsch J., Kyling H. ‘Optimized cast components in the drive train of wind turbines and inner ring creep in the main bearing seat’. Forschung Im Ingenieurwesen. 2021, vol. 85(2), pp. 199–210. [31] Bork C.-P. ‘Leitfaden für eine betriebsfestigkeitsrechnung’ in Gudehus H., Zenner H. (eds.). Materials and corrosion. Vol. 47. 3. Auflage. Düsseldorf: S., Verlag Stahleisen mbH; 1995. p. 112. [32] ForschungskuratoriumM. (FKM), Rennert R., KulligE., VormwaldM. (eds.) “Analytical strength assessment of components in mechanical engineering: made of steel, cast iron and aluminium materials” [VDMAVerlag]. Frankfurt; 2021. Available from https://d-nb.info/1238617476 [33] Pyttel B., Varfolomeyev I., Luke M., Berger C., Siegele D. ‘Welding in the world’ in “Fracture mechanics proof of strength for engineering compo- nents” — overview and extension topics. Vol. 51; 2007. pp. 5. [34] McEvily A., Suresh S. Fatigue of Materials. Cambridge: Cambridge University Press. 1992. Available from http://doi.wiley.com/10.1002/adma. v5:4 [35] Barla S. ‘Global wind turbine technology trends 2021 – market report’ [on- line]’. Wood Mackenzie. 2021. Available from woodmackenzie.com [36] Gille D. ‘Schneller schrumpfen. Einige Turbinenbauer erschließen mit weni- gen Handgriffen andere Leistungsklassen. Eine Wende im Verständnis der Plattformstrategie’. ERNEUERBARE ENERGIEN. 2011. [37] Nam J.S., Park Y.J., Kim J.K., Han J.W., Nam Y.Y., Lee G.H. ‘Application of similarity theory to load capacity of gearboxes’. Journal of Mechanical Science and Technology. 2010, vol. 28(8), pp. 3033–3040. [38] Vries E. de WindTech: MHI Vestas goes global with 9.5MW typhoon- class V174. Denmark: WindPower Monthly; 2019. [39] Guo Y., Parsons T., King R., Dykes K., Veers P. An Analytical Formulation for Sizing and Estimation the Dimensions and Weight of Wind Turbine Hub
- 324. Drivetrain concepts and developments 295 and Drivetrain Components [Technical Report]. NREL/TP- 5000- 63008; 2015 Jun. [40] Klinger F. UPSCAL Beschreibung von Gesetzmäßigkeiten des Größenwachs- tums von Windenergieanlagen und Lösungsansätze für die sich daraus erge- benden Problemstellungen, Forschungsbericht. Hochschule für Technik und Wirtschaft des Saarlandes, University of Applied Sciences; 2002. [41] Molly J.P. ‘Technical trends in wind turbine development’ [online]’. DEWI Magazine. 2002. Available from www.dewi.de [42] Guo Y., Keller J. ‘Investigation of high- speed shaft bearing loads in wind tur- bine gearboxes through dynamometer testing’ [online]’. Wind Energy. 2018, vol. 21(2), pp. 139–50. Available from http://doi.wiley.com/10.1002/we.v21.2 [43] Zhang Z., Matveev A., Ovrebo S., Nilssen R., Nysveen A. ‘State of the art in generator technology for offshore wind energy conversion systems, 2011’. IEEE International Electric Machines Drives Conference (IEMDC), Conference Paper; Niagara Falls, Ontario, Canada, 2011. [44] Gellermann T. Investigation into the dynamic behaviour of drive trains in wind energy converters. Bremen: VDI Tagung, Schwingungen von Windenergieanlagen; 2011. [45] Vries E. de ‘Top turbines of 2020- spotlight on new hardware in our annual awards for the best technology’. WindPower Monthly. 2021, vol. 37(1). [46] Tsukamoto O. ‘Present status and future trends of RD for HTS rotational machines in Japan’. Physica C. 2014, vol. 504, pp. 106–110. [47] Abrahamsen B., Mijatovic N., Seiler E, et al. ‘Superconducting wind turbine generators, IOP PUBLISHING, supercond’. SciTechnol. 2010, vol. 23(8pp), 034019. [48] Ohsaki H., Queval L., Terao Y. ‘Design and characteristic analysis of 10 MW class superconducting wind turbine generators with different types of stator and rotor configurations’. 4th International Conference on Clean Electrical Power (ICCEP 2013); Alghero, Italy, 2013. [49] Liang Y., Rotaru M.D., Sykulski J.K. ‘Electromagnetic simulations of a fully superconducting 10- MW- class wind turbine generator’. IEEE Transactions on Applied Superconductivity. 2019, vol. 23(6), pp. 46–50. [50] Le T.D., Kim J.H., Kim D.J., Boo C.J., Kim H.M. ‘Status of the technology development of large scale HTS generators for wind turbine’. Progress in Superconductivity and Cryogenics. 2015, vol. 17(2), pp. 18–24. [51] SKF Group Bearing damage and failure analysis. technical report: PUB BU/ I3 14219/2 EN; 2017 Jun. [52] Giger U., Kleinhansl S., Schulte H. ‘Design study of multi- rotor and multi- generator wind turbine with lattice tower—a mechatronic approach [online]’. Applied Sciences. 2021, vol. 11(22), p. 11043. Available from https://doi.org/ 10.3390/app112211043 [53] Moghadam F.K., Nejad A.R. ‘Evaluation of PMSG‐based drivetrain technol- ogies for 10‐MW floating offshore wind turbines: pros and cons in a life cycle perspective’ [online]’. Wind Energy. 2020, vol. 23(7), pp. 1542–63. Available from https://onlinelibrary.wiley.com/toc/10991824/23/7
- 325. 296 Wind turbine system design [54] Nejad A.R., Torsvik J. ‘Drivetrains on floating offshore wind turbines: les- sons learned over the last 10 years’ [online]’. Forschung Im Ingenieurwesen. 2021, vol. 85(2), pp. 335–343. Available from https://doi.org/10.1007/ s10010-021-00469-8
- 326. 1 GGS GmbH 6490, Andermatt, Switzerland Chapter 6 Gearbox concepts and design Urs Giger1 6.1 Introduction In 1997, the Swiss gear company MAAG Getriebe AG was taken over by the Danish company F.L. Smidth- Fuller Engineering A/S in Valby, Copenhagen. As a young engineer at MAAG, the author was not aware at the time of how these two compa- nies would have a long- term influence on his career path. The combination of gears and wind turbines became a fascinating challenge. Both companies come from quite different backgrounds. F.L. Smidth’s three- bladed wind turbine on the island of Bogø (built in 1942) already looked very similar to the classic ‘Danish’ wind turbine. It was part of a combined wind- diesel system that provided power to the island, with a rotor diameter of 24 m and rated power of about 60–70 kW. Max MAAG, the founder of the company of the same name in Zurich, dis- covered the profile addendum modification in 1908 and thus made very strong teeth in the root circle possible for the gearing world. The company ZF in Friedrichshafen, co- founded by Max MAAG, was one of his first customers of these strong teeth, for modern drive units in airships. Today, ZF is one of the most important manufacturers of high- performance wind power gearboxes up to 15 MW. The fascination for wind power increased for the writer in 2001 at the first meeting of the wind gearbox industry at NEC Micon in Randers DK. Load gear- boxes in wind turbines suddenly got a bad reputation around the turn of the millen- nium. The massive increase in the number of units overtaxed the manufacturers. An increase in production was attempted by dangerously increasing the grinding speed of profile grinding machines. Massive grinding burns in the tooth root are the worshipping consequences. The faster profile grinding process compared to the gear generating process became the big nightmare of several gear manufacturers overnight.
- 327. 298 Wind turbine system design This is how the author experienced his entry into the wind world as a young gearbox designer. And where does the industry stand today? Some 20 years later, the author of these words was asked to write a chapter in a German wind reference book about the concepts and design of wind power gearboxes. Without any academic degree worth mentioning, but with a large backpack of collected experience, I would like to write this chapter for those young people who have also succumbed to the fascination with gears and wind power. Manufacturing a gear wheel is handcraft! Every gear must be perfectly designed, produced, measured and installed so that it does not become the weakest link in the chain. This handcraft can only be learned in a gear manufacturing company. The knowledge of the interrelationships in the design, manufacture and operation of gear drives cannot yet be fully represented in modern computer programs. Less critical engineers quickly become enthusiastic about these tools. There is a great danger that they will unreservedly believe the numbers and figures in the solution template. But especially the expensive damage cases of the last 20 years in wind power gear- boxes show how close the borders between success and failure are. Was it external circumstances, such as higher loads than assumed, or was it the company’s own failures in the manufacture of the gearbox that led to the premature failure of the turbine component? This question must be approached very openly and neutrally in the case of any damage. Every gearbox failure in a wind turbine is one too much. A betrayal of the sustainability of wind energy and nerve- wracking for the operator of the turbine. The following lines are intended to further encourage curiosity and enjoyment of wind power gearboxes and their reliable construction and to help wind power remain a success story in the renewable energy family. 6.2 Challenge for load gearboxes in wind turbines A wind turbine gearbox is a speed- increasing gearbox that increases the relatively slow speed of the wind rotor to the speed of the generator. It is a demanding applica- tion that requires careful consideration of the load spectrum to ensure that the gear- box has adequate load capacity and is within size and weight limitations. Gearboxes must be designed to maximize efficiency and minimize noise levels. Due to a large number of wind turbines and limited accessibility for maintenance in the nacelle, reliability and serviceability are important factors. The operating environment requires gearboxes that are resistant to extreme temperatures, contamination and corrosion. The wind energy industry is constantly evolving, and industry professionals have reduced gearbox design to a single configuration. This configuration and its design iterations have been around for many years; consequently, design and man- ufacturing errors have been progressively minimized. However, regardless of the sophistication of the gearbox design and design process, most wind turbine down- time is due to gearbox- related problems. In addition, gearbox replacement and lubri- cation account for one- third of the spare parts costs for the entire turbine.
- 328. Gearbox concepts and design 299 A wind turbine converts the kinetic energy of the wind into electrical energy. The transformation machine (generator) is subject to an important design law. The faster this machine is made to turn, the smaller it can be built. Or the other way around, the faster an electric machine turns, the less expensive materials have to be built in. Therefore, an increase gearbox will still be of great importance. According to estimated figures from the wind energy industry, about 53% of main gearboxes are used in the drivetrain of wind turbines worldwide. Accordingly, wind turbines without step- up gearboxes have a share of 47%. Through a study on the competi- tion between direct drive and gear drive in wind turbines, empirical evidence was found that it is possible for different concepts to coexist. In the wind turbine market, this does indeed appear to be the case. Because of constant frequency electrical grid transmission systems and low- cost generators manufactured in large quantities, direct drive generator prices are necessarily much higher because they are manufac- tured in much smaller quantities as special purpose machines. Assuming materials of similar density, a direct- drive generator has 100 times the torque and weight of the geared version, with a gear ratio of 100:1. If the genera- tor rotor were approximately the same length, it would have the same polar moment of inertia as the generator, which turns a hundred times faster. Relative to the tur- bine rotor, the inertia is multiplied by the gear ratio squared. The power- to- weight ratio of the direct- drive generator will decrease disproportionately, as will that of the turbine, while the power- to- weight ratio of the gearbox will remain constant. Gearboxes that are designed to be flexible in torsion allow the rotor to accelerate while reducing the load on the gears. A rotor is optimized for the range below the rated power of the generator. An impor- tant parameter for the design of any turbomachine is the tip speed ratio λ. It indicates the ratio of the circumferential speed of the rotor to the (in this case) wind speed. At the same high- speed number, large rotors appear to turn at a leisurely pace compared to smaller ones. Common three- blade rotors today have tip speed ratios of 7–8. On a wind turbine, the fluid energy source, the wind, does not impact in a con- trolled manner. This factor is the major challenge in the design of the gearbox and other drivetrain components. The instantaneous wind speed produces an instanta- neous rotor torque that is proportional to the square of the wind speed. This instanta- neous torque is applied through the gearbox to accelerate the inertia of the generator, which appears large when related to the turbine shaft via the gear ratio squared. Since the power is proportional to the wind speed to the power of the cubes, a tem- porary increase of 50% in the speed results in a doubling of the torque and a tripling of the power. For example, if the power is increased from 750 to 3 000 kW, i.e., by a factor of 4, the rotor diameter doubles and the speed halves. It follows that the weight and torque of the rotor increase by a factor of 8. By the way, the moment of inertia (related to the fifth power of the diameter) increases by a factor of 32. The trend towards large turbines has led to very expensive gearboxes. These gearboxes are more expensive per kilowatt than smaller turbines because the rotor shaft torque increases sharply. Rotor torque increases with the cube of the rotor diameter, while power increases with the square of the diameter (Figure 6.1). This is
- 329. 300 Wind turbine system design due to the fact that the rotational speed of large wind turbines is limited because of aeroacoustics noise. An important goal of the wind industry remains to design wind turbines in a way that reduces energy costs and extends the service life of system components. 6.3 Historical drivetrains in wind turbines From the analysis of the existing, evaluation of all experiences, verification of the theoretical knowledge, looking for new solutions, testing of new approaches, arise with an inventive spirit the product of the future. Let us begin with the analysis of the existing. One important remark should be allowed before we start. Different languages use different terms to denote all this class of gear trains: some call them planetary gear trains (PTGs), while others prefer epicyclic gear trains. The aim is to provide the reader with the most commonly used types of simple PGTs. It is not always pos- sible to match the naming of the technical practical terms with the standards. The first documented large wind power turbines in the 1930s with a capacity of up to 20 MW foresaw a great future. And the state of the art today is again in this order of magnitude, which makes wind energy a leader in renewable energy production in terms of low initial costs of the energy hour. Of course, in the scope of this chapter, a narrow selection of examples must be made. However, every great prototype deserves to be mentioned. The exciting history of gearboxes and their suc- cess or failure has always fascinated. This nearly 100- year review of the wind world begins with a graphic overview of turbine sizes (Figure 6.2). Figure 6.1 Weight follows cubic model law
- 330. Gearbox concepts and design 301 Very early Kleinhenz (1937) introduced plans for wind power systems, which should be able to handle power up to 20 MW [1–3]. In cooperation with Maschinenfabrik Augsburg- Nürnberg (MAN) in Gustavsburg, Kleinhenz was able to further develop the concept of his large- scale wind power plant in 1938, result- ing in the MAN- Kleinhenz project, the technical features of which still appear very modern today. Namely, a three- or four- blade rotor with a diameter of up to 130 m was to sit on a braced tubular steel tower with a hub height of 250 m. The approach of the gearbox was very interesting. The power distribution was followed by means of helical gears and four generators. However, the concept was never built. The very high torque of almost 20 000 kNm at the input shaft of the gigantic wind turbines would have required a PGT of almost a 4 m large ring gear with a module of 35 mm (with eight planets), but this was not manufacturable at that time. Epicyclic gearing was freely used during the course of the 19th century in a wide variety of industrial applications but did not keep place with the demand for large powers, because improvements in the technique of cutting external gears outstripped developments in the production of internal gears. The ingenious gear designer W.G. Stoeckicht was working on large planetary gearboxes at the same time in 1937 and awarded a Diploma of Honour (Diplome D’Honneur) at the Paris World Exhibition for a planetary gear in a 1 400 hp Diesel Locomotive, exhibited by the German Railways. Shortly afterward followed a speed reducer for an underwater application of 5 000 hp power. We are talking about a weight of only 1 000 kg and a ring gear of 0.9 m diameter at relatively high speeds from 3 370 to 580 rpm [4]. However, such fast speeds at the input shaft for a wind generator do not seem to be appli- cable. Toward the end of this chapter, let us recall these marginal conditions. The Stoeckicht design appeared in the wind world later nevertheless in the 1980s. With very compact gears and little use of materials, he pursed precisely the requirements that are necessary for a nacelle high above the ground. An early example with parallel shaft gears can be found in the two wind tur- bines WE 10/G6 and StGW- 34 from Allgaier Werke, a manufacturer from Germany and designed by Prof. Hütter. The StGW- 34 rated power was 100 kW. The turbine started working at a wind speed of 3.7 m/s. The wind turbine was equipped with a Figure 6.2 Development of turbine size
- 331. 302 Wind turbine system design total of two rotor blades, and the diameter was 34 m at a maximum speed of 36 rpm. The smaller rated power with 6 kW was supplied by the Allgaier WE 10/G6, with a 78 m² rotor area. The early gearing concepts, based on parallel shaft gears, are too large and too heavy as the power increases. Light and compact also apply to wind turbines, as the weights in a nacelle must be kept to a minimum. It is therefore significant that the gearboxes that were redesigned in the aircraft industry in the 1950s were often used as a model for wind energy. The Ivchenko AI- 20 is a Soviet turboprop engine devel- oped by the Ivchenko design bureau in the 1950s. The compactness of the engine is achieved by splitting the torque between two gear stages, with the two- stage plan- etary gear having a differential stage, as shown in Figure 6.3. In the 50s of the previous century, a split epicyclic gearbox approach is known from France, in which two individual stages were spatially separated and connected with a long coupling. The nominal apparent power of 800 kVA or 650 kW. The pro- peller was connected to the alternator by a coaxial mechanical linkage involving two sets of epicyclical spur gears. In order to make the gearboxes favourable as a sin- gle stage, they were designed as epicyclical gearboxes (Figure 6.4). The turbine was installed in the field of Institute Aérotechnique from Saint- Cyr L'Ecole Nogent- le- Roi [5]. The rotation speed of the rotor was 47 rpm. The alternator rotation speed was at 1 000 rpm (invariable as regulated by the 50 Hz frequency of the EDF grid). An interesting example was built in Holland in the 1980s at STORK- FDO (rated power: 300 kW, NEWECS 25) Holland. The planetary gear, flanged to the double- bearing main shaft, shows a very modern and up- to- date drive technology (Figure 6.5). The engagement in a ring gear of the high- speed shaft is chosen quite favourably in terms of epicyclic technology, which employed an annular gear, with the intention of taking advantage of that design’s ability to engage more teeth than Figure 6.3 Planetary coupling gears from aviation 1950s
- 332. Gearbox concepts and design 303 the normal spur gear, and therefore reduce the load and wear factor on the individual teeth. There is a concave contact pressure. And the ring gear serves as a very good bucket wheel for the lubricating and cooling oil. The most exciting drivetrain development story in the wind industry in the last 40 years is firmly associated with the 3 MW wind turbine LS- 1 in Burgar Hill, England, in 1987. 3 MW British Aerospace/GEC/Taylor Woodrow turbine LS- 1 is a production of Wind Energy Group (WEG) Ltd., a manufacturer from the UK (Figure 6.6). This manufacturer had been in business since 1978. However, WEG Ltd. has not been in business since 1998. The manufacturer was also acquired by NEG Micon A/S. The irony of history, the much smiled at Danes with their small wind turbines 100 kW, which were quickly repaired over the weekend and thus achieved a fabulous availability, took over the highly convinced megawatt pioneer wind turbine manufacturer from England. Frequently cited and reported, this experimental system had been ground- breaking in the powertrain. Ray Hicks from Wales (GB) was the designer of the Figure 6.4 L'Eolienne de Nogent Le Roi (France) 1955–1966 Figure 6.5 AII- PGT gearbox [6]; A external meshing, I internal meshing
- 333. 304 Wind turbine system design gearbox system with the superimposed stage, and GEC Energy Systems Ltd. was the gearbox manufacturer. The main gearbox had 10 planets in the first two stages, all ground and preloaded in tapered roller bearings directly with raceways on the planet. The subsequent 2:1 ratio bevel gear stage fed torque vertically into the tower to the generator below. This was common practice in the 90s. But here comes another novelty. In order to be able to use a simple robust synchronous generator with a fixed speed, a superimposing stage was placed in front of the generator as an out- put stage after the bevel gear stage. This interesting combination worked perfectly. Unfortunately, the turbine was blown away in 2001. As we will see, the step- up ratio of 18:1 input stage compound two- carrier PGT serves as a template for the gearbox example 7.5 MW at the end of this chapter. Interestingly, this compact gearbox system by Ray Hicks proved to be excellent in the LS1 wind turbine. However, this concept was not later adopted by the Danish turbine manufacturers, despite the acquisition of WEC. Here I suspect the lack of understanding of the design, calculation and construction of this type of gearbox. It is easy to understand single tooth contact in parallel shaft gear units. As soon as a planet carrier turns relatively around a sun gear, the first difficulties in understanding and calculation arise. And the large number of planets in one stage was little known on the mainland. This was a great deterrent at first. Nevertheless, the German Federal Ministry for Research and Technology had also investigated the use of variable- speed, electrically controlled superimposed gearboxes in wind turbines under grant number 0329121 A and published the final report in April 1993. What was not properly considered, the English machine con- trolled the sun shaft (small torques) and the German ring gear (large torques) and thus missed to draw the right conclusions in the final report. But also, the English machine was not copied further, too complicated seemed the whole development to unite in a robust simple series machine. A short time later, a 300 kW two- blade machine with a 33 m rotor diameter was built by the same company, WEG Ltd. Again, Ray Hicks was the designer and new, company Compact Orbital Gear (COG) the manufacturer of almost 500 units. The Figure 6.6 The 3 MW Wind Turbine Project on Orkney 1981–1995 (ETSU- R- 95)
- 334. Gearbox concepts and design 305 gearbox was designed with an application factor KA of 2.0 and 340 kW nominal power at 48 rpm input speed. Unfortunately, the story ended badly for COG. The pitch adjustment routed through the gearbox was faulty. In addition, the main shaft occasionally sheared off during emergency stops. Both failures caused major repairs and ultimately court cases that did not end well for COG. On December 26, his patent WO91/19916 was published (Figure 6.7). Ingeniously solved was the bearing of the rotor in the main gear. The planetary stages mounted on the main shaft were efficiently arranged and the installation space was optimally utilized—daring design, but lightweight construction in its most extreme form. Of course, the mainland gearbox builders had never seen such a design before and also judged it skeptically, actually completely wrong from today’s point of view. Nevertheless, 30 years later, we have arrived at multi- planetary solutions. From 1988, the Danish electricity company ELSAM operated a 2 MW research turbine at Tjæreborg, which was in operation until 2001. The development was based on the second Danish wind energy program that started in 1982, which focused on the construction of two 750 kW turbines and one 2 MW turbine. The large turbine had some problems with the control system and the gearbox. The gearbox came from the famous TGW gearbox manufacturing company. Remarkable were the integrated planetary bearings (Figure 6.8). Very modern at the time, these bearings have higher load ratings and are freed from rotating outer rings and their problems of wear. The major gearbox problems started with a break in the cross section of the first stage ring gear. The ring gear was later modified and bolted to the side of the housing wall. This proved to be successful. Bearing modifications to the fast stage followed, until a tooth break at the fast stage meant the end after 45000 h. This tower was also blown away. Figure 6.7 MS- 3 from patent sketch WO91/19916 (300 kW)
- 335. 306 Wind turbine system design Somewhat later, an interesting 3 MW design was added by the same gearbox manufacturer. The version of TGW shown here in the WTS 3 MW wind turbine was built in a collaboration between Hamilton Standard from the USA and Wind Turbine Systems Corp. (Swedyards) of Sweden. The overall gear ratio was 1:60. The large input moments were worked through by a special trick. It was decided to split the first stage. A common dual- bearing sun shaft summed the half- stages and fed the total torque via the coupled planet carrier into the second planet stage. As with the 2 MW Tjeaborg machine, the housing was welded. Remarkable was the planet bearing, this time with the outer rings inserted in the planet. The MAAG gearbox design in 2003 for the power class looked different from the state of the art at that time (Figure 6.9). A compound two- carrier PGT was selected for the 1.3 MW power rating. In a joint partnership, the bearing company Timken [7] stood aside and used the newly created Integrated Flexible Pin in both epicyclic stages. A pseudo- PGT so- called ‘Star epicyclic stage’ combines with a differential Figure 6.8 TGW Thyssen Getriebe- und Kupplungswerke GmbH, Herne, Germany Figure 6.9 MAAG DPPV- 7- 79, turbine type: Nordex N- 60
- 336. Gearbox concepts and design 307 stage, by a so- called ‘floating member’, which is a total support (bearing) free cou- pling shaft. The second stage is formed as a reaction plate, a planet carrier equipped with up to eight planets, presenting a innovative concept. Two gearboxes were built. In Orkney (Kirkwall district), the first gearbox is still running in an N60 Nordex turbine, after almost 20 years. MAAG has chosen the concept that was already successfully used in 1987 in the LS1. Of course, it was proposed, designed, and implemented with the help of Ray Hicks with a very high- quality standard as in a Swiss watch. The inventor of the elastic equalizing pin (flexible pin) already had consulting mandates at MAAG in the 1970s. For a first time, in 1983, MAAG and Hicks worked together in wind project for a Dutch general contractor FDO for turbines rated at 1 000–1 500 kW power (NEWECS 45). Presented at the same time as the MAAG solution in 2003 and also comparable, Bosch Rexroth also uses a stage as a differential stage in the third stage (1c) in Figure 6.10. In the same figure, only three planets per stage were deliberately drawn. For a long time, good load balancing in a rotary stage was considered acceptable only with three planets. The manufacturer Bosch Rexroth, later taken over by ZF, built these gearboxes very successfully. The concept is explained in more detail in a disclosure document DE19963597A1 from 2007. The colour illustration from the company brochure and an abstract from the disclosure document DE19963597A1 are listed one after the other. Wolf [9] intro- duced symbols in 1958 that represent an epicyclic gear unit by a circle with three lines going outward (the three connecting shafts). These symbols, types and uses are defined in more detail, e.g., in the VDI 672 guideline. It is not difficult to see how the layman is easily overwhelmed by Wolf’s scheme. Because also the calculation of Willis does not contribute necessarily to the dismantling of the question marks, at that time, Prof. Kiril Arnaudov from Bulgaria [10] developed a simple and practical teaching method for his students. A history worth mentioning was the PSC 1002 gearbox from the Jahnel- Kestermann company in Bochum. The power was 750 kW, and about 570 units Figure 6.10 GPV gearbox principal Bosch Rexroth [8]
- 337. 308 Wind turbine system design were built, mainly for the NEC Micon turbine NM48. The basic design of the PSC1002 wind turbine gearbox is characterized by the use of a planetary stage on the input side and two helical gear stages on the output side. The entire gear unit is connected to the rotor and the machine carrier via a so- called three- point support. This design is used in numerous wind turbines up to approximately 2.5 MW and has proven itself in compliance with certain narrow bearing clearances in the planet carrier. The technical innovations in the PSC1002 wind turbine gearbox are primarily to be found in the planetary stage bearing concept of the sun gear shaft and were pat- ented by Jahnel- Kestermann. On the generator side, the bearing is accommodated in the housing as standard. The sun shaft was newly supported in the planet carrier on the rotor side. This mounting was made in a comparatively thin- walled sleeve on the planet carrier, which protruded towards the sun wheel. The required adjustment flex- ibility is achieved exclusively by the clearance of the radial roller bearing, which is matched to this, and by the deformability of the planet carrier support, which specifi- cally encompasses the radial roller bearing. As a further advantage, this bearing con- cept offers the possibility of integrating the wheel of the first spur gear stage on the sun gear shaft. This eliminates the need for an additional coupling between the planetary and helical gear stages, thus enabling a shorter gear unit design. The shorter gear unit design also results in a smaller gear unit housing and thus in weight and cost savings. The American laboratory NREL has consulted this gearbox PSC 1002 for extensive investigations (GRC) and modified it several times. Probably, the best- documented wind turbine gearbox is publicly available through a great number of measurements and reports [11]. No organization has contributed as much to the bet- ter understanding of the kinematic relationships in a wind turbine gearbox as NREL located at the foot of the Rocky Mountains near Boulder, Colorado. Hidden here is great will for wind power to become better and more reliable. Due to gearbox prob- lems in the US turbines, even these machines did not reach the required lifetime. At first, general turbine problems were suspected, not manufacturer- dependent ones. The basic quality deficiency was not suspected as the main cause. Even compliance with the state of the art was not sufficient according to the initial findings, because problems nevertheless occurred. Furthermore, the bearings were mainly suspected as the big evil, less the gear teeth. The first culprit was the bearing failures and only secondarily the teeth, so seen as a secondary cause. And the failures in the 500–700 kW wind turbines with gears continued in the megawatt class. For these reasons, a large investigation program with its own large test rig and sophisticated measure- ment technology was started. In the year 2009, still another built and also patented solution of the company GGS is presented (Figure 6.11). To avoid indefinitely loaded roller bearings, the sun of the first stage was mounted on a plain bearing mandrel. On the one hand, the coupling shaft is guided in this way, and on the other hand, the second stage is supplied with the necessary lubricating and cooling oil. The solution was exhibited at the Hanover Fair in the same year. The concept was installed in various licensed machines in the 2.5 MW power class.
- 338. Gearbox concepts and design 309 6.3.1 Hybrid systems Hybrid systems are a middle path between the conventional solution with three gear stages in the megawatt range and direct drive solutions, which usually require a generator with a relatively large diameter. The goal is a simpler and more reliable gearbox with a generator of comparable size, resulting in a dimensionally balanced and compact powertrain. Around the turn of the millennium, the medium- speed concept gained a lot of attention. Based on the many gearbox failures at the fast output stage, it was learned that the omission of this HSS stage should result in a considerable increase in avail- ability. At the same time, permanently excited generators with 150–300 rpm were developed and directly coupled to the two first gear stages. The first known produc- tion turbine in the megawatt class was the WinWind WWD- 1 wind turbine from Finland. The hybrid drivetrain of the WWD- 1 wind turbine consists of a single- stage planetary gear and a low- speed synchronous generator. This Multibrid® con- cept combines the reliability of a direct drive and the compactness of a gear system. This period also witnessed the installation of the Falcon 1.25 MW wind turbine in Nordenham (Germany), which was commissioned in December 2009. The com- pany GGS from Andermatt [12] developed and built the new patent drivetrain for it (Figure 6.12). A compound two- carrier PGT unit was inserted into the main shaft and sur- rounded by two main taper roller bearings. The driveline consisted of two stages, each with seven and five planets. The tooth forces were transmitted on flexible pins, which ensured perfect load balancing. The gear solution incorporated in the main shaft is called an integrated tubular gear system (ITGS). Integrated systems eliminated misalignment between the rotor main shaft and the gearbox/generator. The same gearbox lubricant was used to lubricate the enclosed rotor shaft bearings, eliminating the need for grease- lubricated main bear- ings. In an integrated system, the seals must be carefully designed because of the possibility of gearbox oil entering the generator or vice versa and wear debris or other contaminants entering the gearbox. Figure 6.11 Compound two- carrier AI- PGT; source: GGS
- 339. 310 Wind turbine system design Today’s modern drivetrains with gearboxes of the class up to 15 MW mostly belong to this category and are specially developed combinations of the main shaft, main bearing, coupling, gearbox and medium generator. 6.3.2 Exceptional developments in the drivetrain Essentially, these are ways to implement a variable speed in the gearbox that allows a synchronous generator with sinusoidal power generation to be connected directly to the output, eliminating the need for an electric converter. We remember that the LS1 in Orkney demonstrated this in 1987 successfully. WinDrive from Voith Craislheim was essentially a mechanical solution for variable- speed operations, based on a torque converter combined with a planetary gear based on the Stoeckicht principle. As a fluid machine, the torque converter is well matched to the wind turbine rotor, and via the fluid in the converter, the system decouples the input and output shafts by absorbing the peaks in input torque and providing vibration damping. With the WinDrive solution, the added mechanical complexity and cost of the transmission system were offset by the elimination of the cost, mass, and losses of an electric converter. The damping and compliance inher- ent in the hydrodynamic coupling ensure that a synchronous generator can be used. Voith technology had long been established in industrial drives, but the wind energy application presented new challenges, particularly in terms of service life and effi- ciency, which Voith addressed but ultimately failed to bring to market. The counterpart to the hydrodynamic solution is the electromechanical solution of the LS1 drivetrain at first and after also SET company [13]. Again, preferably externally excited medium- voltage synchronous generators directly connected to the mains were to be used. To compensate for the disadvantage of the fixed speed of synchronous generators directly connected to the mains, a differential PGT was placed in front of the generator. This system from SET [13] was also difficult to establish itself on the market. The complexity of controlling the superimposed gear- box was a major challenge for the turbine manufacturer. Figure 6.12 1.25- MW Falcon wind turbine with PMG system (2010); source: GGS
- 340. Gearbox concepts and design 311 6.3.3 A Swiss geared wind turbine For years, the Swiss landscape has prevented the development of an independent wind industry, because the claim to protect the landscape has priority over the issue of domestic renewable energy production. Wind project durations of 20 years are not uncommon in Switzerland. In this difficult environment, the company GGS from Andermatt is trying to develop and build its own wind turbine for the demanding Alpine region. The distributed generation drivetrain (DGD) drive opens up a very broad solu- tion (Figure 6.13). The DGD drive reduces the gear load by consistently utilizing the formula, power P equals speed times torque, and applying it several times at the gearbox output. At very high speeds, over 5 000 rpm at the generator shafts, up to 12 generators work together in a distribution gear to deliver the incoming power package from the wind rotor. The two input stages correspond to the template from the LS1 and the ITGS patent of the company GGS from 2011. It is probably the most compact power drive in the wind world. Enormous advantages result from maintenance. A generator–inverter unit weighs only 140 kg. In case of failure or malfunction, these lightweight units can be replaced within a very short time. The basic idea was supported in Switzerland, and a report on it was published [14]. Due to the ultra- short overall length, the entire drivetrain can be tilted downward by 90° — just right for installation and maintenance in the Swiss mountains. 6.3.4 State of the art Exemplary, for the current state of the art in wind power gearboxes to see, shows a whole gearbox series (Figure 6.14) on the website of the manufacturer Flender (former Figure 6.13 Tilt turbine Altanus and right, general view of the ground test
- 341. 312 Wind turbine system design Winergy). Of course, one learns little about the internal structure of the gearboxes. But, to achieve more torque per kilogram of gear weight (up to 200 Nm/kg), as mentioned on the website, between four and seven planets per stage will have to be installed. This is not new knowledge, let us remember again the LS1 machine. The input torque of 913 kNm was successfully shared among 10 planets. The weight of the gear parts was 4.6 tons, resulting in 198 Nm/kg of torque per kilogram of gear weight, and this was already in 1987. The ‘modern’ gearbox concepts actually all go back to existing and proven gearboxes from earlier wind turbines. For the large power range, these planetary gearboxes have become established. The transmission of high specific torques/gear- weight numbers of up to 200 Nm/kg is now state of the art. Thus, we have arrived in the middle of the subject area. We remember hav- ing read at the beginning of the subchapter that from the analysis of the existing, the evaluation of all experiences, the verification of the theoretical knowledge, the search for new solutions, the testing of new ideas, the product of the future results with an inventive spirit. To understand the actual gear design, a little more we need to look at some theories. 6.4 Basic gear tooth design The calculation and dimensioning of gearings are complex and fill entire books. It cannot be described in detail in this short publication. For further information, please refer to the relevant standards and technical literature [15, 16]. The gear wheel is not a spontaneous invention; it is rather a phenomenon (simi- lar to the wheel) and has evolved over two millennia. It is still evolving today and is certainly one of the products that can be called ‘high tech’. The famous Basel mathematician Leonard Euler already recognized the advantages of involute gearing in 1762. For a uniform and shock- free transmission of the tooth forces in an engage- ment of two gears, he referred to the rolling curve ‘circular involute’ as a suitable shape for the tooth flanks. The gear mechanism has historically been the most effective and efficient mech- anism for coupling machines with different optimum speeds. In its simplest form, a Figure 6.14 Gearboxes for modern wind turbines; © Flender GmbH, 2022
- 342. Gearbox concepts and design 313 fixed ratio gearbox consists of a pinion with a smaller number of teeth meshing with a wheel with a larger number of teeth, whose respective axes are parallel. The ratio of the number of teeth, diameters, torques of the driven to the driving gears (pinion and wheel) to each other is equal to the gear ratio. For example, a wheel with 80 teeth drives a pinion with 20 teeth at four times the speed. Thus, with the help of the number of teeth, the rotation can be reduced or increased accordingly. This rather simple principle is used in almost all mechanical drives and plays a particularly important role in our lives. Mechanical clocks may be mentioned as examples. The total gear ratio of several parallel gears is obtained by multiplying the indi- vidual gear ratios. Intermediate gears (R gear) only change the direction of rotation. During tooth meshing, the point of contact of the tooth flank moves along a straight line, and a sliding and rolling movement takes place between the tooth flanks them- selves (refer to Figures 6.15 and 6.16). The transmission of torque in tooth meshing Figure 6.15 Simple wheel and pinion Figure 6.16 Base tangent contact path [17]
- 343. 314 Wind turbine system design is not directly metallic, but always through a few thousandths of a millimetre- thick oil film interposed. To ensure a constant speed ratio, the respective teeth must have exactly the same circular pitch and a geometric shape that allows the torque to be transmitted from one tooth to the next by a sliding/rolling mechanism that ensures a constant peripheral speed. Figure 6.16 shows the pitch circles of a pinion and a gear which touch at the pitch point on the line connecting their respective centres. The circumference of the respective pitch circles is equal to their number of teeth multiplied by the common pitch. As shown, the line of meshing between the contacting gears is a straight- line common tangent to their respective base circles, from which the involutes, the tooth form is generated. It passes through the pitch point at an angle to the line of contact called the pressure angle (usually 20°). Its length is limited by the respective tooth tip diameters that intersect the common meshing line. To ensure continuity of trans- mission, the normal distance between successive tooth flanks (the basic pitch) must be smaller than this length by a factor called the overlap ratio εα. In most standard gears, this is between 1.4 and 1.7, so there are two pairs of teeth in mesh (double mesh) at the beginning and end of the mesh path but only one in the middle. The commonly chosen tooth form is an involute, the characteristics of which are clearly described in any textbook on gearing. Although gearing is very simple in principle, it is very difficult to realize in practice. As mentioned at the beginning of this article, Max MAAG succeeded in 1908 in decisively improving this gear meshing for good. His accidentally discovered modification of the geometrical meshing ratios was aimed at selecting the part of the involute curvature in such a way that it influenced the shape of the teeth. By mov- ing the tool during rolling (manufacturing method), the tooth was given a different shape, i.e., a different part of the involute was used for meshing. This can be seen very clearly in Figure 6.17. Figure 6.17 Influence profile shifting tooth profile (root)
- 344. Gearbox concepts and design 315 Without modification, the part of the involute starting directly at the base circle is used, with modification, a part of the involute further away is selected. This is the great secret that MAAG accidentally discovered by a mistake in the manufacture of a gear. Through long trial and error, he subsequently obtained a new strong gear system that made the teeth at the root stronger. Even today, these addendum modi- fications are important in choosing the right roll and glide ratios in each tooth mesh. The calculation of the pressure distribution is performed for each support point using the formulas according to Hertz [18]. Based on the local line loads, a pressure distribution is calculated for each load level on the contact lines of the considered engagement position. Since each discrete individual force Fi from the force vector Fi acts on a small section Δl of the contact line, it can be assumed for simplicity that the force is distributed uniformly along the partial section Δl of the contact line. Along the meshing section, the existing local equivalent radii of curvature ρi of the tooth flanks are determined at the local support points of the contact lines in the mesh. The relative radius is the product of the respective tangent lengths at the points of contact divided by their sum, i.e., the constant length of the common tangent. For a given common tangent length, the product of the respective pinion and wheel tangent lengths would be maximum if they were equal. Of course, this would only be the case if the pinion and wheel are of the same size. It follows that the relative radius of curvature is minimum at the lowest point of contact between the gear tip diameter and the pinion root. However, this lies in the area of the double- tooth con- tact, so that the selected load point for calculating the highest surface tension is at the lowest point of the single- tooth contact on the pinion flank. In order to illustrate the essential design criteria for gear teeth, the gears and their possible damage patterns should be studied first. Classical tooth damage can be roughly divided into three groups, tooth fracture, pitting, and scoring. Tooth fracture usually occurs near the base, caused by excessive bending stress. Pitting appears as more or less extensive material chipping on the load- bearing tooth flank. The cause here is excessive pressure on the contact line of the two tooth flanks. This is associated with excessive stresses immediately below the surface, which causes the material on the tooth flank to literally flake off. The third cause is called scuffing and occurs when there is insufficient oil film between the tooth flanks. The oil film breaks down, and direct metallic tooth contact occurs. This in turn leads to micro- welding of the material, which is immediately torn out, marking the typical seizure at the head and root. As a designer, I have a choice of materials and heat treatments that determine the allowable loads in tooth contact (Figure 6.18). The following is an excerpt from the old DIN 3990 [4], the great role model of ISO 6336 [19]. The design of a gear must now be optimized in such a way that the desired safety results with regard to the three damage phenomena. The effects caused by the three damaging factors often behave in opposite directions when the essential parameters of a gear are changed, so that an optimum compromise must always be aimed for (Figure 6.19). The surface pressure is the criterion that determines the volume of the pitch cylinder of a gear pair, i.e., the square of the respective diameter multiplied by the width of the tooth surface. The compressive stress generated by the normal force
- 345. 316 Wind turbine system design between the teeth is determined by dividing this force by the width of the meshing surface and the relative radius of curvature at the point of contact, which varies from the beginning to the end of the contact path. The toothing can afterwards be optimized, the variation of the following param- eters being the most usual: • • addendum modification (influences specific slippage and strength) • • tooth depth (reaching an optimum transverse contact ratio) • • helix angle (reaching an optimum overlap ratio) • • fillet optimization (e.g., bigger radius by tool module different than gear module) • • profile and lead modification (improving contact pattern, reducing meshing shock) Figure 6.18 Different materials and their application limits Figure 6.19 Optimization triangle in the design of the gearing
- 346. Gearbox concepts and design 317 A consideration, the KMAAG value by hand formulas is an understandable way for the gear design to roughly determine the necessary gear volume and therefore the size and weight of the gearbox. This method from the MAAG Gear book 1985 (page 104 ff.) deliberately avoids great calculation effort and promotes simple handcraft- ing. The following KMAAG values are generally recommended for gearing: EH case- hardened gearing: KMAAG 7.5 N/mm2 , V quenched and tempered gearing: 4 N/ mm2 at 1 200–1 300 N/mm2 and 3.5 N/mm2 at 1 050–1 200 N/mm2 . When the limits are exceeded, the gear width is simply increased. With this method, here as for a planetary stage with a fixed ring gear, we can start with the gear stage design. Subsequently, we evaluate after by using ISO 6336 method B. The formulas distinguish convex and concave pressing, and material differences and heat treatment methods must correspond to the same design. Then corresponding limit values for KMAAG from known designs are used for comparison (Figure 6.20). This simple method allows the gear design, i.e., the individual tooth contact with always the same easily determined geometric boundary conditions to be reliably evaluated. In a planetary stage, the sun- planet mesh is usually in the critical posi- tion. The above historical wind turbine gear stages are recalculated with the simple method and compared in Table 6.1. Two values are notable, the high KMAAG values for the type 1250 with 15 N/ mm² and the very low L10h values of the LS1 planetary bearings. In the first case, the entire gearing was nitride, including the ring gear. In the second case, the bearings were preloaded directly in the planets without outer rings and showed no signs of wear even after disassembly, whereby the entire operating time of the LS1 turbine came to just under 15 000 full load hours. A typical value for wind power gearboxes 10–15 years ago was 11–13 N/mm². Many standards already know the K- factor as a specified reference value in various industries. Why the wind world did not bother to extend it for wind power gearboxes is to be found in the same corner, which rejected Figure 6.20 Calculation sheet for KMAAG
- 347. 318 Wind turbine system design planetary gearboxes with more than five planets in wind power gearboxes and still punished them with penalty factors for uneven load sharing. Thus, the weight advan- tage of many planets was always immediately destroyed. 6.4.1 PGT planetary stage in detail The planetary gearboxes (PGT) play an important role in wind turbine gearboxes. However, thanks to their numerous possibilities, these gearboxes cause difficulties in theory, calculation and design [9, 20–22]. Characteristics are three central shafts, which together distinguish this compact gear design to the ideal speed or torque con- verter (Figure 6.21). Thus, the power to be transmitted is distributed over several tooth meshes. This advantage comes into its own where very high torques have to be trans- mitted at medium and low speeds. Ideally, the largest possible number of planets is used in one stage. This makes full use of the compactness. Large overall gear ratios are then preferably realized in successively compound multi- carrier PGTs. In each case, two of the three central shafts are coupled together. There are three possibilities, full input power, power sharing, or circulating power. In favourable arrangements, the cou- pling shafts can carry only a partial power, in unfavourable cases, however, a multiple of the input power as internal power. The theoretical relationships for power flow and speed ratios have been compiled by Prof. Dr.-Ing. Kirill Arnaudov in a very practical system [10]. But that is only the theoretical side of the story. In practice, the transmission sys- tem must be designed to withstand the many loads, including the chosen bearings. Table 6.1 Post calculation with KMAAG of historical gear stages Year 1986 1987 1998 Type/name 1250 DK LS1 GB PEAC 4440 Country D Stage # and type 2 PU* 1 PU 2 PF† 1 PF 1 PU P, nominal power kW 2205 2205 2252 996 1660 n, wind rotor RPM 21.9 110.8 34.0 607.8 19.0 Mt , input torque kNm 960 190 912 51 834 Stage ratio - 5.053 5.727 2.767 4.481 5.647 kγ , mesh load factor 1.15 1.00 1.03 1.03 1.00 b, tooth width mm 250 130 250 240 330 kappa b/d' - 0.81 0.48 0.44 0.88 1.17 m, module mm 16 12 13 10 16 KMAAG N/mm² 15.02 7.27 4.19 4.61 11.7 σFMAAG N/mm² 112 68 55 51 83 Gear rim—outside-Ø mm 1500 1440 1760 1380 1520 n, sun RPM 58.2 281.1 109.3 337 48.4 L10h , planet h 50 367 101 348 2 186 2 435 31 762 Total weight stage kg 2 372 905 3 032 1 572 2 430 *PU annulus fix † PF carrier fix
- 348. Gearbox concepts and design 319 6.4.2 PGTs have a number of advantages and applications They are distinguished for being very compact; i.e., they have small dimensions and low weight—two to three times lower than the common non- PGTs thanks to the adoption of the multi- flow principle; i.e., several planets are used to split the power flow. The diminished dimensions have a number of beneficial consequences, such as reduced material consumption and a light construction, respectively. A small mass moment of inertia is important for fast- paced drives. The diminished dimensions of the gears allow for both heat treatment and achieving higher accuracy in their production, which combined with the lower pitch line velocity leads to lower inter- nal dynamic loads and to a quieter operation of the gear train, which is particularly important nowadays. Due to their compactness, the required smaller gear train bearer is important in some cases. Especially for lifting equipment, such as bridge cranes, this accounts for a substantial lightening, not only for the trolley but also for the entire construction, and hence reduction in the price. Another substantial advantage is the very high efficiency of some PGTs and vice versa—the possibility of self- locking when the efficiency is low. The coaxially of the input and output shaft also has advantages in some cases (e.g., vehicles, wind turbines, aircraft engines). The PGTs offer new layout possibilities that do not exist with the other types of non- PGTs. They are used as follows: • • gear trains, both with F = 1 and with F = 2 degrees of freedom • • reducers or multipliers • • differentials, i.e., power division or power summing gear trains • • part of systems to make a stepless change of the angular velocity • • change gears (gearboxes) in vehicles: cars, buses, tugboats, tractors, tanks, etc. • • reversing gears in ships, locomotives, etc. Figure 6.21 AI PGT simplest form
- 349. 320 Wind turbine system design 6.4.3 Difficulties in using PGTs The application of the PGTs despite the advantages has, on the other hand, some shortcomings and difficulties. Their theory is more complex than one of the non- PGTs. These are the processes that run inside the train, and hence problems such as differentiating the types of internal power—absolute, coupling, and relative (rolling) power, internal division and internal circulation of power, and load sharing between planets. These processes, especially in the complex compound planetary gears, are not so clear and easy to understand and have contributed to the reputation of PGTs as something complicated and difficult to understand. Some unsuccessful technical solutions and failures, apropos, have also contributed to this reputation. This leads to the difficulties with the accurate determination of the loads as a prerequisite for the proper calculation of the gear train elements, and the difficulties with the correct determination of efficiency, which is crucial for some cases. All in all, the theory and practice of planetary gears have quite a lot of ‘pitfalls.’ The great compactness of PGTs, which is itself a considerable advantage, otherwise may mean a reduced cooling surface, which in some cases leads to difficulties in heat removal and com- plicated and costly arrangement due to forced lubrication and cooling. The price per kilogram of PGTs is higher than that of non- planetary ones. However, due to their lower weight, with a successful design, the cost of PGTs may get lower eventually. It should also be noted that planetary gears require a higher precision of manu- facturing. There is also the danger of complete destruction of planetary gears when a single tooth is broken, which unlike the non- planetary gears cannot be discarded into a safe place. All this means that the design and production of PGTs must be executed with extreme responsibility. It should be taken into consideration that the number of the different types of PGTs is relatively large and this fact alone makes it hard for the designer to select a suitable gear train type. In addition, tooth geometry causes some difficulties as well. 6.4.4 Increasing the power sharing Planetary gears offer the advantage of an extremely low power- to- weight ratio due to the property of power sharing among a larger number of teeth meshes. The power increase depends essentially on the number of planets installed per stage. In this regard, there are important points to consider. Increasing the number of planets from three to eight or ten causes significant changes. First, the transmittable torque can be increased by 166% and 233%, respectively, compared to the three- planet solution, and second, the centre distance is increased as a result of the sun becoming larger. This means a reduction of the bearing force on the individual planetary axes. Another feature worth mentioning is the reduction of the gear width. Often, the gearing limits can now no longer be utilized. The planetary bearing reaches the load limit before the gear teeth due to the limitation of the rolling bearing load carrying capacity (refer to Table 6.2). But in the case of the Q- 10 stage, ten interventions compared to the Q- 3 stage result in six times the bearing life L10h according to ISO 281 of the individual planetary bearing. By increasing the number of meshing points
- 350. Gearbox concepts and design 321 in an orbital stage, the sun grows and the possible ratio becomes smaller. This disad- vantage, which appears at first glance, can be optimally compensated for by adding a new PGT stage. With this nesting of orbital stages, single- walled planetary carriers are advantageously used in the individual stages. With this arrangement, the space available for the maximum number of planets is not restricted by unnecessary con- necting webs. The tip circles of the planets can be moved together to within a few millimetres. In principle, an identical planet can be installed in several stages. This simplifies the production lot and limits the variety of components. Four possible compound two- carrier three- shaft PTGs show in Figure 6.22 how two orbital stages can be combined in a meaningful way. All four combinations exist in wind turbine step- up gearboxes. The four solutions can be summarized into two groups. The first two connections A and B (Figure 6.22), sequentially coupled PTGs, each stage transmit 100% power. The total reaction torque is dissipated once via both ring gears and in the other case via ring gear and planet carrier. The next two types C and D according Figure 6.22, with internal division of power, split the incoming power in a ratio influenced by the number of teeth or type of coupling of two central shafts. A very clear distinction between the power division between the shafts and the power sharing between the planets has to be made. This results in the great advantage that the auxiliary stage (two- shaft gear) only has to be designed for the differential power. This is enforced because the main stage (three- shaft gear) acts as a differential stage. All three central shafts of this main stage have a different speed. The common reaction torque of the coupling stages C and D (Figure 6.22) is Table 6.2 Results when increasing the number of planets from Q3 to Q10 Symbol Unit Q-3 Q-8 Q-10 i, transmission ratio - 5.57 2.87 2.70 a, centre distance mm −415 −517 −535 b, face width mm 380 160 130 Ft kN 442 133 103 KMAAG % 100 63 63 σF , ref MAAG % 100 73 75 ISO 281, L10h h 26 000 71 000 148 000 Figure 6.22 Compound two- carrier three- shaft PTGs
- 351. 322 Wind turbine system design derived in each case via the retained planet carrier of the auxiliary stage. This leads to generally thicker planetary pins in this auxiliary stage, because it has to act for the whole compound multi- carrier PGT. The example in Figure 6.23 will briefly show exactly how much material can be saved with the right combination of coupling gears and optimum number of planets in a stage. Three possibilities of a gearbox design with a ratio of i = 40 were com- pared at the end with the resulting weight. The variation of the number of planets and the number of stages allows for an increasingly smaller weight. The interesting thing is that the gears are always designed according to the same safety. This contra- dicts the opinion of many designers that the translation of many planets in one stage is a disadvantage. Only the consistent use of the available space with the largest pos- sible number of planets results in a minimum gear weight. Of course, the stage with the lowest speed but the highest torque determines the overall diameter. The savings in gear weight with the same safety factors are considered. 6.4.5 The problem of load distribution and its control In epicyclic gear units with several planetary gears, the applied load is not distrib- uted quite evenly among the individual power branches. The unavoidable manu- facturing deviations within the prescribed tolerances are responsible for this. This is taken into account in the calculation by the mesh load factor Kγ [1]. There is a large number of patented compensation devices. Some of these compensators oper- ate on a kinematic principle. The other part of the balancers uses the compliance of certain gear elements for load balancing. The importance of some of these factors was fully realized by the famous Lanchesters automotive from UK, who referred to this question in their paper in 1924 [23] on epicyclic gears. They had also grasped one of the constructions when they stated: The success of any mechanism, however, well- conceived, depends finally upon its correctness in detail. Figure 6.23 Weight reduction due to more planets and stages
- 352. Gearbox concepts and design 323 To this end, Ray Hicks proposed in 1964 that the planet gears should be mounted advantageously on sleeves that are connected to the planet pin at one end only (Figure 6.24). A transverse force F and a bending moment M thus act at the end of the pin. The inclination of the pin and the sleeve caused by the tooth forces largely cancel each other out with optimized rigidity, so that the planet gears can yield in the circumferential direction while maintaining their parallel axial position. This pre- vents skewing of the planetary gears, which is one of the main causes of tooth dam- age. The compensation works very simply and robustly. Nevertheless, the choice of compliance to determine a reasonable approximation of the spring constants for a planetary gear set in a planetary spur gear stage is of crucial importance. Especially the extreme loads in the load spectra show the limits of the solution. 6.4.6 The load-sharing measurement The load can be distributed to the individual planet wheels about the deformation (bending) of individual teeth of the ring gear or about the deformation (bending) of the planet gear axles. A positive aspect of the first measuring method is that the tooth load is measured directly, which is actually of interest. However, this measure- ment method requires accessibility to the ring gear, which is not always the case. In addition, there is the high cost of the test arrangement and the evaluation. But the most unpleasant thing is that the load on the planetary gears cannot be determined as a continuous function of the position (angle of rotation). The second measuring method does not have the disadvantages mentioned above. For the loading of the planet wheels, the deformation (bending) of their axes is used, where strain gauges are conveniently used. The most advantageous feature of this method is that the load on individual planetary gears can be determined as a continuous function of posi- tion, which is more useful. This measuring method also has some disadvantages, e.g., that it practically does not determine the tooth load, but the axle load. Strictly speaking, it is not the axis load that is of interest for the load distribution, but the tooth load. In planetary gears with large masses and considerable moments of inertia of the gears, but only at very high speeds (turbo gears), it is not impossible that the difference between the axis load and the tooth load becomes considerable, possibly between the two tooth meshes—external and internal mesh—of a planetary wheel. Figure 6.24 Principal explanation of the flexible pin
- 353. 324 Wind turbine system design By definition, the mesh load factor Kγ is the quotient of the largest occurring load of one of the planet gears, or the power branches, to the average load of the same. Likewise, the deflection or tension of the sleeve/bolt of a planet can be used as an occurring measured value always at the same position. 6.4.7 Microgeometry In drive technology, we speak of torque and this is transmitted as a normal force between the meshing teeth. But even if the teeth are geometrically perfect, the force causes deformation of the teeth, which causes pitch errors. In addition, tooth mesh- ing is also affected by deformations in shafts, bearings, and housings that support the gears (Figure 6.25). It becomes even more difficult when the gear is subjected to external reaction forces due to the variable nature of the wind. All these effects lead to an unacceptable maldistribution of tooth load across the tooth width of the gears. The microgeometry explanations that now follow lead inevitably into the heart of the company’s own secret design assumptions. The tip relief must be so designed that on every tooth pair, as it passes through the path of contact, the load increases as uniformly as possible, and gradually sinks back to zero again, i.e., the desired theoretic trapezoidal load variation along the path of contact is sought. A convenient load diagram is shown in Figure 6.26. In order to attain this load variation AHIE, the flanks of the pinion and wheel are to be tip- relieved. The first tooth contact between the tooth tip of the driven gear and the root of the driving gear occurs in point A according Figure 6.26. A second tooth pair is already in contact in Figure 6.25 Elastic deformation of the HSS pinion
- 354. Gearbox concepts and design 325 point D. Immediately before this tooth constellation, the entire load is transmitted by the second tooth pair, the driving gear moves the amount along the line of action. In order to prevent a shock load, the gear is eased back an amount of tip relief Cαa . Modification methods are made mostly in house and belong to the best- kept secrets of a company. Simplified lead deformation values can be roughly estimated from the MAAG Gear book [24]. Each company uses its own calculation method and implements this in drawing templates. For example, the complex determination can be read in detail in the work of Reference [25]. From standard ISO 21771- 1, the correct designations exchange data information can be taken. Especially when a gearbox fails, these values are often not exchanged. Unfortunately, this causes a great loss of confidence in the industry. However, the prerequisite remains, to cor- rectly study and interpret the tooth contact patterns. Gear design is still handcrafted and requires a large backpack of different experiences. When determining the microgeometry, it is important to keep an overview. Especially in wind turbines, stationary gearboxes with one load step are not found. On a tall slender tower, it vibrates constantly, depending on the tower head mass even more and now the housing must be considered very flexible and the loads all constantly changing. This big challenge needs a lot of calculation and test work and the following support over 20 years. Unfortunately, this is forgotten, although every gearbox should be designed according to machine guidelines 2006/42/EG [26] and must be maintained for the entire life of the gearbox, signed by the manufacturer, not only when it has banged. Therefore, test rigs play a very important role. The calcu- lated tooth load pattern can be easily checked for plausibility by stepwise increasing torque load. Of course, the observation of the gearbox up on the tower must con- tinue, as just mentioned. If, for example, a gearing correction is calculated for 100% nominal load and the gear unit is then loaded with a load spectrum of 50–200% of the nominal load, Figure 6.26 Load variation and profile diagrams for involute spur gears with tip relief
- 355. 326 Wind turbine system design then a different load distribution is produced for each load. As the load increases, the contact pattern of the gearing increases until, at 100% nominal load, there is a load distribution that is constant across the width, which generates a face load factor KHβ close to one for this load level. As the load continues to increase, the contact pat- tern shifts, for example, to the left side of the gearing, resulting in damaging excess loads. To illustrate this, Figure 6.27 shows the displacement of the contact pattern for a load spectrum of 50–200 % of the nominal load. Noise optimization is also the goal of corrected tooth interventions. However, the wrongly determined and ground microgeometries are not always the only cul- prit. Also, helical gears do not always solve the problems. The wrong number of teeth can generate excitations in a gear stage and amplify them in the turbine through natural frequencies. Torsional vibration analyses can help here in the design stage. In planetary stages in particular, the choice of the number of teeth is of decisive importance. Reference is made here to the relevant technical literature [27]. 6.4.8 Absolute, coupling, and relative (rolling) power When examining the question of power in PGTs, the notions derived from the super- position ‘Swamp method’ are all useful. Two partial movements (rotations) transmit their partial powers, e.g., coupling power Pcoup (NKZ in Figure 6.28), which transmits by coupling movement when PGT rotates as a coupling (as a whole) without relative rotation of gears to the carrier. Consequently, this power is assumed to be transmit- ted without internal losses. Relative (rolling) power Prel (NBZ in Figure 6.28) or the power in the mesh that is transmitted by relative movement of the gears with respect to the carrier. This is the power in PGT after inversion (‘pseudo- PGT’with fixed car- rier, we call it also ‘PF’) that causes the meshing losses. These losses (lost power PѰ ) are considered by the basic loss factor Ѱo , respectively, and by the basic efficiency ηo . Of the above follows: PA = Pcoup + Prel . 6.5 Bearings The most commonly used bearing in the wind industry is the rolling bearing. These bearings are selected for their low friction and high load capacity. The reason for this is that roller bearings have a higher capacity than larger bearings and are less expensive. Tapered roller bearings are generally used for combined loads, both radi- ally and axially. Figure 6.27 Load distribution over the tooth face (width)
- 356. Gearbox concepts and design 327 In general, the design criteria for such bearings lead to a finite life, which takes into account the total number of hours at different loads [29]. The heaviest loads are in high- torque, low- speed planetary stages and, in particular, the plan- etary spindles, which must support the sum of the tooth loads from the planetary meshes with the sun and the ring. The most successful arrangement is a pair of preloaded tapered roller bearings, which ensure that there is no risk of skidding under light loads. Tapered roller bearings are used extensively as mechanical components in most auto- moving machinery and must withstand time- varying loads. The influence of the preload required for this type of bearing to prevent pitting and fatigue problems is well known. The aim is to achieve the most even distribution of contact pressure in the inner and outer raceways. To maximize the available space for a better bearing capacity design between the bore and the spindle, especially at low ring/sun ratios, it is helpful to select a fine- tuned module to increase the root diameter that allows the planet bore to enlarge. It also helps to omit roller outer rings from bearings and inte- grate the raceways into the planetary bores. Timken has gone a step further and also integrated the inner races into the planetary spindle and used full- roller, preloaded tapered rollers. All planetary bearings, as well as all other lower- loaded, higher- speed bearings in the secondary trains, require a pressurized lubricant supply. No bearing should be subjected to misalignment, and self- aligning bearings should be avoided. They cannot be effectively preloaded because they have a clearance that can cause skidding at low loads. The selection and arrangement of bearings are determined based on industry experience and the recommendations of IEC 61400- 4. The fatigue strength of the Figure 6.28 Power split and Pcoup, Prel according to Reference [28] in compound two- carrier three- shaft PTGs (Diff and Star stage only). Numbers are from the example in Figure 6.15.
- 357. 328 Wind turbine system design bearings is checked on the basis of the required service life of 20 years for wind turbine gearboxes, which means that the service life of the bearings should be longer than the required service life of 20 years. Bearing life is defined as the length of time, or the number of revolutions, until a fatigue spall of a specific size develops. This spall size, regardless of the size of the bearing, is defined by an area of 6 mm2 . This life depends on many different factors such as loading, speed, lubrication, fitting, setting, operating temperature, contamination, maintenance, many other environmental factors. Due to all these factors, the life of an individual bearing is impossible to predict precisely. Also, bearings that may appear to be identical can exhibit consider- able life scatter when tested under identical conditions. Remember also that sta- tistically the life of multiple rows will always be less than the life of any given row in the system. L10 life is the life that 90% of a group of apparently identical bearings will complete or exceed before the area of spalling reaches the defined size 6 mm2 criterion. If handled, mounted, maintained, lubricated, and used in the right way, the life of your tapered roller bearing will normally reach and even exceed the calculated L10 life. If a sample of apparently identical bearings is run under spe- cific laboratory conditions, 90% of these bearings can be expected to exhibit lives greater than the rated life. Then, only 10 % of the bearings tested would have lived less than this rated life. Figure 6.29 shows new types of plain bearings mounted in high- speed planetary gear units directly flanged to high- speed generators. Without the use of plain bear- ings, this stage would be very difficult to control. Plain bearings are now also being discussed for the new IEC 61400- 4 ED 2 standard. The first wind turbines used plain bearings. In modern large turbines, attempts are being made to switch tentatively to journal bearings. The start- up conditions, the real big unknown, in the build- up of sufficient oil film, lack cal- culation methods. It will be some time before this type of bearing is mastered and accepted in the wind world. Figure 6.29 Example from the wind world: plain bearings of flexible pins
- 358. Gearbox concepts and design 329 6.5.1 Bearing failure mechanisms The failure mechanisms of bearings are quite similar to the failure mechanisms of gears. The failures can be divided into two subgroups—lubrication failures and sur- face fatigue failures. Failure due to surface fatigue is generally progressive and can be divided into micro- pitting and spalling. Typically, the raceway fails first relative to the other components of the bearing under surface fatigue (Figure 6.30). The bearings give an audible sign of pitting. Pitting increases the vibration of the system and can lead to spalling or fracture of the rolling elements. Flaking produces a large amount of abrasion. There are characteristic loading conditions that trigger this mechanism of sur- face fatigue failure in the bearing, including misalignment and reverse axial load- ing. Other parameters that influence bearing failure can be that the bearing is either loosely or tightly connected to the housing, which can cause fretting. A lubrication failure is the result of the lack of the required lubricant film thick- ness necessary to prevent contact between the rolling elements and the raceways. The lack of the required film thickness results in metal- to- metal contact and causes overheating of the bearing. This usually manifests itself in discoloration of the roll- ing elements, rings and cages. In many cases, the high temperatures can also degrade or destroy the lubricant. Overheating leads to loss of hardness of the bearing mate- rial, which eventually leads to failure. In wind turbines, many bearings are operated at low speeds, resulting in wear due to loss of film thickness or complete loss of elastic-hydrodynamic suspension. 6.6 Coupling A shaft- mounted gearbox requires a properly designed and installed torque arm to maintain the alignment of the high- speed coupling. Regular maintenance is also required to keep the high- speed coupling aligned. This calculation takes into account equipment efficiency, availability, and maintenance requirements. For manufactur- ers, this means that the efficiency, quality, and reliability of the equipment must meet the highest standards. The drivetrain plays a central role in this. This applies to Figure 6.30 Different bearing failures occur in wind turbine gearboxes
- 359. 330 Wind turbine system design all fields of application for wind turbines (onshore, offshore, repowering, low wind areas). A distinction is made between couplings with and without overload protec- tion and torque detection as well as rotor and azimuth brakes. As a rule, the wind power coupling is used in combination with a brake disc of up to 1 600 mm diameter and a sensor disc for speed monitoring. In addition, there is an overload system that ensures precise torque limitation even under unfavourable conditions. Slip hubs are ideal for this task. They work with special friction linings that make them stick- slip- free and extremely wear- resistant. The slip clutch is cali- brated at the factory and then disappears into the intermediate piece of the clutch to save space. As soon as the set slip torque is reached, the power flow is limited and the system is protected from generator- side load peaks up to 1 000 times. This is because the slipping torque is reproducible many times, protecting the plant from stress and significantly reducing the amount of service required. 6.7 Mechanical brakes The most commonly used mechanical brake in wind turbines is the disc brake. Although it can also be found on the slowly rotating shaft, in most cases, it is used on the fast- rotating shaft, as the torque load is much lower there. The brake consists of a steel disc that is rigidly attached to the braked shaft. A series of calipers are attached to the frame of the shaft being braked. The calipers exert sufficient force on the disc to stop the shaft. Wind turbine brake calipers are usually fail- safe, meaning that the braking force is applied by springs and the calipers are opened by hydraulics or elec- trical, and the brakes are applied if the hydraulics or electrical fails. Since the brake is a friction device, it can cause a nonlinear torque increase when the turbine stops. There is also the possibility that it introduces additional vibration to the transmis- sion and drivetrain. When modelling the brake, these two phenomena are generally neglected, and a linear torque increase against the rotational motion is represented. This increase peaks at a certain maximum braking torque. The speed at which the peak is reached is of great importance, since it directly determines the magnitude of the torsional forces introduced into the driveline to counteract its inertial behaviour. 6.8 Lubrication system and its design principles Lubrication and cooling systems in gearboxes are very complex. On the one hand, all contact points in the gearing and bearings must be sufficiently lubricated, and, on the other hand, sufficient heat flow must be dissipated. For more information on this problem, please refer to Chapter 8 in this book. Here, we will only briefly discuss the necessary amount of lubrication in the gearing and bearings. This is the experience of the individual gearbox manufacturer and is assessed very differently. A very low gearbox temperature at nominal load should be ensured. Because above all thereby the oil aging can be massively retarded. Gear units run at full load up to 2000 h/a depending on the on- shore location in the country. This means that about three- fourth of the time, the speed and performance are uncertain. Sufficient lubrication
- 360. Gearbox concepts and design 331 must also be provided here. First basic oil quantities were already discussed and specified in the AGMA guidelines. At the turn of the millennium, an approximate oil quantity for lubrication and cooling of 0.15 l/kW installed power, but at least 20 l, was valid. In modern gear unit sizes, this recommendation is of course no longer sufficient. Here, the gearbox designer must evaluate all possible friction points in the gearbox and determine a sufficient oil quantity. This is done, for example, by means of a summary in tabular form (Figure 6.31). Separate mechanical gear pumps bolted to the fast shaft were already installed in the MS3, and the MAAG gearbox also operated with a direct- drive pump recessed in the oil sump inside the gearbox (Figure 6.32). These mechanically driven pumps are not sufficient on their own. Additional electrically controlled pumps are connected (Figure 6.32). The exact oil quantities must be matched. A specially constructed lubrication tube test stand is Figure 6.31 Requirements for oil quantities at all friction points in a gear unit (example) Figure 6.32 Oil tank for dry lube system and pump and filters
- 361. 332 Wind turbine system design used to visually check the injection direction, and by collecting the oil with a bucket and weighing it, the oil quantity can also be easily determined. 6.9 Bolted joints Loads can be applied to bolted joints in a number of different ways, each of which produces unique effects on the joint. These effects result from the way the joint is loaded, as well as how the joint responds to the load. Some of the various load types include tensile, shear, and bending. A tension joint is affected by loads that try to pull the joint apart. The forces on the joint and those on the bolts are roughly parallel to the axes of the bolts. All tensile forces try to stretch and/or separate the joint. The tension load, no matter how small, will add to the stress in the bolt and/or partially relieve the joint. The bolts in a tension joint must act like clamps. The tightening of the bolt and nut produces a ten- sile pre- stress, which is approximately equal to the compressive stress introduced in the joint material. The behaviour and life of the joint depend on how tightly the bolts clamp and how long they can maintain their preload. A proper amount of tensioning of the bolts is vital. With too little clamping force, the joint may loosen. If the joint is exposed to cyclical loads, too little clamping force can shorten the bolt’s fatigue life. Too much clamping force can also cause severe problems. By over- tightening the bolt, one may exceed the proof load of the bolt. Even if the bolt does not fail during assembly, it may later break under the external tensile load. Over- tightening of the bolt can also encourage the advancement of hydrogen embrittlement or stress corrosion cracking. The joint members can also be damaged or warp from too much clamp force. The bolted joint diagram is illustrating elastic bolt elongation and elas- tic joint compression in the axial direction. A joint diagram may help illustrate what happens as we apply our preload and the effects of external loads. Threaded fasteners can clamp materials together only when they are holding with the proper amount of tension. For this to happen, they must be properly tight- ened. To this day, a simple, inexpensive, and effective way to consistently and accu- rately tighten a fastener does not exist. There are a number of tensioning methods that function well enough, but they are both complicated and expensive. Engineers compensate for the inability to consistently and accurately determine bolt tension by massively over- designing joints. This accommodates inaccurate tightening and avoids catastrophic failure. Designers will specify more or larger bolts than needed in order to ensure that the joints are sufficiently clamped together. Fewer or smaller fasteners can be used when bolt preload is accurately and consis- tently controlled. However, current trends are moving away from the use of over- design. Increasing demands on cost, strength- to- weight ratios, product safety, product per- formance, and environmental safety have put pressure on designers, manufacturers, and assemblers to increase design efficiency. This trend has led to the invention of more options for controlling bolt preload.
- 362. Gearbox concepts and design 333 The tightening of a bolt follows a defined sequence of events and causes pre- dictable results within the fastener. If the nut and head of the bolt are firmly seated against non- compressible materials, the torsional action of tightening the assembly stretches the bolt, thereby creating tension in the bolt. In most cases, this tension or preload is required to make a fastening. By controlling torque, turn, or stretch, one can control the buildup of tension. The closer a method is to direct control of tension, the more expensive it will be. Some options for tension controls during assembly are: Torque control, torque and turn control, stretch control and direct tension control. Bolted joints in wind turbines are designed for both mechanical engineering and steel components on the basis of the state- of- the- art calculation and standard regula- tions according to VDI 2230 and current knowledge. Knowledge of the possibilities of assembly of the usually larger nominal diameters must be available and is aware of the necessary care and knows the interrelationships with the design. The tightening of the bolts must be recorded, and VDI/VDE 2862 sheet 2 pro- vides sufficient information to document the tightening process. If high- strength fasteners of grade 10.9 or 12.9 are required to transfer the loads, strict quality pro- cedures must be followed to avoid hydrogen embrittlement. Hydrogen embrittle- ment results in low ductility and can cause nuts to break, washers to crack, and bolt heads to break off. Purchases of high- strength hardware must be source controlled. Substitutes or changes in the coating process must be controlled, and samples must be taken from each batch of fittings manufactured and tested with a load tester. Fittings such as set screws, bolts, nuts, pins, and fittings should be secured inside the gearboxes using typical methods such as anaerobic adhesives. If the housing has a split plane, this must be kept oil- tight. O- rings and sealing compounds must be compatible with the lubricant. Split plane housings must have positive locating devices such as dowel pins. Bearings must not be used for centring split housings. Bolted housing connections between the spider and the mating sur- faces (housing and torque arm) in the gear unit require special design considerations to avoid relative movements. The connection must be designed to prevent fretting corrosion. The annular space must be designed so that the connections remain oil- tight. The joint must be capable of supporting the maximum operating load due to friction under the intended bolt tension with an adequate safety margin. If the friction is not sufficient, the joint shall be provided with sufficiently solid pins to carry the extreme load without overstressing the housing material by compression at the pin face. The contribution of the friction of the connection to the pin capacity is not to be considered in this calculation. The design of the solid pins must ensure an oil- tight fit. Hardened washers are to be used under bolt heads for all torque- transmitting connections and with all fasteners of grade 10.9 and higher. 6.10 Pitch tube The pitch tube is arranged coaxially to a rotation axis of the planetary stage. The pitch tube extends through the entire gearbox in the axis and provides access to the
- 363. 334 Wind turbine system design hub with power and communication lines for blade pitch control. It also prevents oil leakage from the gearbox and is itself also used with a second tube to convey oil to rotating parts (Figure 6.33). This simplifies the lubrication of many hard- to- reach lubrication points. Greatest care must be taken to secure the pitch tube against twisting. The slip rings with a corresponding weight at the end are the reason that insignificant welds fatigue due to bending fatigue failure. Seemingly unimportant, especially large leaks or sheared- off anti- rotation devices often give the reason for the trouble. If this part has to be replaced in the nacelle, this always causes a lot of assembly work. 6.11 Repair work An initially strongly neglected area in wind power gearboxes, repair and main- tenance, is increasingly coming into focus. The required 20- year service life was almost never achieved in the gearbox class from 1 MW. Initially, gearbox manufac- turers did not care much if one out of 50 gearboxes failed. Over time, the number of failed gear units increased dramatically. Many new repair companies were able to position themselves in the market. There are countless replacements in the nacelle on the high- speed shaft. Disassembly is relatively quick with suitable personnel. Of course, it always needs appropriate special tools. The bearing exchange is mostly confronted with the fact that the bearing designations are not available and the special bearings are available only from the original manufacturer. Good relations with the bearing manufactur- ers are very important. Quickly, incorrectly fitted bearings are again the reason for nasty rework. Famous are the large number of early gearbox failures. One type of bear- ing was very conspicuous and sometimes failed very early. The spherical roller bearing, very inexpensive to procure at half the price of an equal pair of tapered rollers, was very quickly cited as the evil. The assumption that spherical roller bearings can support large axial forces as well as radial forces led to some incor- rect placements in bearing positions in wind turbine gearboxes. Upgrades were made in gearboxes (there were up to four upgrades per turbine type) to improve Figure 6.33 Pitch tube with supplementary oil supply
- 364. Gearbox concepts and design 335 the conspicuous gears. In the past, spherical roller bearings were dispensed with entirely, paired tapered roller bearings were installed in their place (Figure 6.34), and additional lubricating oil pumps were installed. It is the transient loads (every- thing that does not run ‘normally’ is labelled transient) that are responsible for difficult lubrication conditions in cold, slow- running wind turbines. In addition, there were major failures in maintaining the quality of the gearing parts. In the event of problems with the gearbox, rotor shaft, or rotor bearings, the rotor blades must usually be removed and the generator, gearbox, and rotor shaft removed as a unit. The total weight of the assembly must be within the lifting capacity of nor- mally available cranes and ground conditions. For large WTGs, this can be a major obstacle to maintenance. For example, repair and maintenance of components in WTGs with integrated gearbox systems may require removal and replacement of the rotor blades, which can be significantly delayed and cause considerable down- time in high winds. 6.12 Standards for load gear units in the drivetrain Proven national and international standards exist to verify the load- bearing capac- ity of the individual components of a gear unit. In the 1980s, systematic damage occurred to gearboxes in the first major wind farms in the United States, although there was sufficient dimensioning of the individual components in accordance with the relevant standards. For this reason, standardized regulations were required to prove the service life specifically for gearboxes used in WTGs. The first ‘regulation’ to appear in the United States in 1997 was the AGMA/ AWEA 921- A97 information sheet [30]. Since information sheets have no legal relevance, it was decided in 1999 to revise the information sheet and convert it into Figure 6.34 Modified HSS shaft
- 365. 336 Wind turbine system design an American standard. Thus, the first standard published in 2004 was the American standard ANSI/AGMA/AWEA 6006- A03 [31]. At present, IEC 61400- 4 ‘Design requirements for gearboxes’ [32] Edition 2 is being prepared at the international level in a joint IEC/ISO working group. The load capacity of helical gears is verified on the basis of ISO 6336 [19]. However, the increased incidence of gear damage, particularly in the early days of wind turbines, has shown that there is too much room for interpretation in the application of ISO 6336 or the previous versions for many factors for gears in wind turbines. For this reason, specific provisions for the application of ISO 6336 have been included in the wind power standards to narrow this scope for interpretation. The failure rate of gearboxes that meet the boundary conditions of AGMA 6006, for example, is significantly lower than that of gearboxes designed in accordance with ISO 6336 using the scope for interpretation. The load- carrying capacity of bearings is generally verified by means of the ‘basic dynamic load rating’ and ‘modified rating life’ approaches described in ISO 281 [33]. The increased incidence of bearing damage, particularly in recent years, has shown that these methods alone are by no means suitable for life verification and can only be used for prediction purposes. In the newer wind power standards, there- fore, the ‘extended contact analysis’ according to ISO 281 Annex B or according to the procedures of the bearing manufacturers is provided as a service life verification taking into account the maximum contact pressure. Furthermore, the wind power standards contain tables for bearing selection, since many of the bearing failures listed above were due to incorrect bearing selection. A junior gear designer started 30 years ago with the DIN 3990 [4], lists with characteristic values, or vice versa with a gear manufacturer his specific basic train- ing. At the moment, a multitude of gear software is available that a young engineer has to master. In addition, they are crushed by an enormous burden of standards in the wind sector. Hand on the heart, who reads and understands standards? Everyone adorns themselves to know standards, although certain standards are sometimes contradictory in themselves. For this purpose, a simple understandable gear design was deliberately pointed out in this chapter. In the future, the gearbox unit will no longer be considered in isolation but as part of the entire drivetrain, whose operational safety can no longer be described solely on the basis of the load capacity verifications of the individual components. Operational safety will therefore increasingly be assessed on the basis of the results of dynamic simulations of the entire powertrain. The approach to such simulations is thus an important point in the further development of standards and guidelines in the field of wind turbines. 6.13 Gearbox design methodology For the 7.5- MW offshore wind turbine of Fraunhofer Institute for Wind Energy Systems IWES, the main gearbox shall be designed exemplarily with the KMAAG cri- terion, recorded and compared and checked after with ISO 6336 [19].
- 366. Gearbox concepts and design 337 The specifications of the wind turbine can be found in IWES Wind Turbine IWT- 7.5- 164 Rev 4- Wind Energy Report [34]. The geometric design of the gearbox is based on the GGS in- house method and international standard IEC 61400- 4 [32]. All gearbox components, i.e., gears, bearings and shafts, are calculated only with rated mechanical power due to the limited scope in this book. To determine the maximum safety, the approach to write a simple spreadsheet that will do the calculations in a matter of seconds and check the layout at the same time is the fastest way. The selection of the gearbox layout is based on the explana- tions in this chapter. Experience shows that very few parameters are needed to start the work. A good and experienced gear engineer makes their own gears and filing the teeth by hand. This is a somewhat old English method, but it is all the more valid today, as software tools suggest how easy it is to construct a gearbox. Instead, with this proven quick method, it is basically still highly contemporary. As a special fea- ture, a gearbox layout version is selected with two output shafts. That is, two smaller generators can be used, which split the input power of the rotor for a generator in half. Especially the geometrically necessary manufacturing data are not very interest- ing at the beginning. Much is determined by the standards. The designer starts the design with proven data, which he has collected from successful experience gearbox designs. Load and operating conditions, such as PA - input power (A), TB -working load (B), T1 - torque on the sun gear, n1- rotation speed of sun gear, are the basic data. For this, minimum geometrical data, such as m-module, aw -operating centre distance, z1 , z2 , and z3 —number of teeth, b1 , b2 , b3 —tooth face widths, are selected and sketched. For the kinematic calculation in the GGS torque–speed spreadsheet, practical equations are used. With seven formulas, the speeds are easy to calculate. The balanced gearing in the individual stages with a very high number of planets (up to 8) is the result of a long iteration process involving drawings and calcula- tions. If too much stiffness is chosen in the flexible pin, a high uncertainty remains with respect to the Kγ mesh load factor. And the reasonable ratio value of the torque per weight of the stage remains unused. Moreover, the planetary bearing decisively determines the whole system from inside to outside. A balanced epicyclic stage requires to consist of well- balanced gearing, flexible pin and bearing load- carrying capacity. The design starts with a consideration of the gear forces in the PGT stage. This load determines the size (flexible pin in this case) of the planetary stage. It enables a reasonable choice of stiffness in the epicyclic system. It is flexible enough to compensate for manufacturing errors of the carrier plate and provide an additional reserve for load operation. With the equations shown in Figure 6.35, the required input variables from the specification are a great help to keep a tabular overview in a simple sketch and table like in Figure 6.36. In this stage of the design process, calculation is still possible without a gear program. Building up the sizes of the gear stages volumetrically based on the permissible Hertzian pressure at the contact point is simple and fast. The KMAAG empirical values automatically include limits; with these values, the bend- ing in the tooth root (tooth fracture), the pressure on the tooth flank (pitting), and
- 367. 338 Wind turbine system design via a simple control function also the flash temperature criterion for large modules are controlled. In slowly, high- torque- loaded rotating planetary gears, the mesh con- tact must be using the flash temperature method derived by Prof. Blok [35]. Later, further optimization steps are pending, such as noise- reduced gearings by the small choice of the module and possibly smaller pressure angle (17.5°) for so- called high Figure 6.35 Kinematic relationships in GGS design calculation spreadsheet Figure 6.36 Torque–speed spreadsheet GGS
- 368. Gearbox concepts and design 339 contact ratio gearings, which use transverse contact ratio above 2.0. The asymmetric GGS flexible pin for PTGs with multi- planet solutions requires less space but must be carefully checked against the permissible stresses and stiffnesses with the bound- ary conditions. From the many possibilities we have learned from past concepts, we select something that allows minimum weight and maximum safety factors. The determination of the torque ratio of the compound three- stage PTG is done with the torque method [10, 36]. The known and very useful Wolf’s symbol [9] is used, however, with a few additions. The shafts with the aligned torques, i.e., the sun gear shaft 1 and the annu- lus gear shaft 3 are marked with single lines (Figure 6.37). Their thickness is based on the value of the torques, where T1 T3 . The carrier shaft S is marked with double lines. The carrier S has the greatest, however, opposite torque, which equals that of the other two torques in absolute terms: TS = T1 + T3 (6.1) A torque ratio t is defined (formed) based on the aligned ideal torques T1 and T3 (without gear loss): t = T3 T1 = | Z3 Z1 | +1 (6.2) which proves to be very advantageous for the analysis of precisely intricately assem- bled multi- carrier planet gears. By starting with a torque +1 (unity) with one of the sun gear shafts, which is, however, not a condition and any positive or negative value can be assumed. In the present case, we start with the output torque TB of the sun shaft 7 (Figure 6.38) of the third gear (III) unit, i.e., with the torque method [10, 36]: TB T7 = +1 (6.3) Proceed step by step using the individual torque ratios tI , tII and tIII of the three gear trains I, II and III. The sequence of torque calculation is marked by the green circled numbers (Figure 6.38). The following must be observed during this procedure. The two free coupling shafts, i.e., the ring gear shaft 3 with the sun gear shaft 4 and the sun gear shaft 1 with carrier shaft SIII , do not have an outward connection. Two equal but opposite torques, i.e., with different algebraic signs, therefore act on their ends. The drive input torque TA and the reaction torque TC of the connected linkage shafts—i.e., the carrier shaft SI with annulus gear shaft 6 and the carrier shaft SII with Figure 6.37 Elementary (one- carrier) planetary gear
- 369. 340 Wind turbine system design annulus gear shaft 9 result from the sum of the torques of the corresponding gear train shafts. If the ideal torques of the assembled gear are at hand (Figure 6.38), it is easy to calculate the transfer ratio i with the following formula using the drive torque TA and the output torque TB : i = TB TA = +1 35.006 = + 1 35.006 1 35 (6.4) The easy- to- handle torque method makes the extensive calculation work accord- ing to ‘Willis’s method’ superfluous, which method is not immune to possible calculation errors. The ‘Kutzbach method’ is ruled out from the beginning, espe- cially for complicated compound multi- carrier PTGs on the one hand because of the inaccuracy of the graphical method and on the other hand because of the impossible accumulation of lines. Another advantage of the torque method is the possibility to check the correctness of the calculation based on the equilibrium conditions of ideal torques. In the present case, the check confirms the correctness of the calculations: P Ti = TA + TB + TC = 35.006 + 1 + 34.006 = 0 (6.5) Figure 6.39 illustrates the procedure for determining the efficiency η. The flow directions of the shaft powers are shown (red) and of the rolling powers (green). The sequence for determining the real torques is the same as for determining the ideal torques. To determine the efficiency of the compound three- carrier PTG, the real torques of all three partial gear units are required, with which the internal gear losses are to be taken into account in the calculation. For this, one needs on the one hand the three static efficiencies ηoI , ηoII , ηoIII of the three partial gearboxes and on the other hand the flow direction of the three rolling powers PWI , PWII and PWIII also in the three partial gear trains, which are responsible for the dominant tooth meshing losses. There are two possibilities for the flow direction of the rolling powers PWI , PWII and PWIII in each gear train. In the example of the gear train ‘I’, these possibilities are flow direction from sun gear 1 to ring gear 3 or flow direction from ring gear 3 to sun gear 1. The rolling power PWI is transmitted from sun gear 1 to ring gear 3 when the directions of the torque T1 and the angular velocity ω1 are the same, with respect to the carrier SI coincide or vice versa. Figure 6.38 Determination of the transmission ratio i using the ideal torques
- 370. Gearbox concepts and design 341 For both cases, the following expressions result in the real torque T3 of ring gear 3: T3 = oI tI T1 (6.6) in the first case and T3 = tI T1 oI (6.7) in the second case. The determination of the flow direction of the rolling power PW is in principle not always straightforward and easy to determine. In such cases, the practical trial procedures according to Seeliger [21] can be used. In the present case, the determi- nation of the flow directions of the three rolling powers PWI , PWII and PWIII does not cause any difficulties, since the carrier SII is fixed and ring gearIII 9 is also fixed. To determine the stand efficiencies ηoI , ηoII , ηoIII of the individual PGT, the very simple formula of Förster [37], which primarily takes the dominant impact of the number of teeth into account, is used here for lack of space. This formula related to the first gear unit is as follows: oI = 1 0.15 1 z1 + 1 z2 + 0.2 1 z2 1 ˇ ˇz3 ˇ ˇ !# (6.8) Further partial gear losses are usually taken into account across the board with a certain percentage. In the present case, the calculation of with the concrete number of teeth yields the following numerical value for the individual stationary efficiency of the three single- carrier PGTs is: oI = 98.7%; oII = 98.8%; oIII = 98.5% (6.9) With the flow directions of the three rolling powers PWI , PWII and PWIII shown, there is the following relationship between the torques of sun gear shafts 1, 4 and 7 and the corresponding ring gear shafts 3, 6 and 9: Figure 6.39 Verification of the efficiency using real torque
- 371. 342 Wind turbine system design T3 = oI tI T1 T6 = oII tII T4 T9 = oIII tIII T7 (6.10) The real torques are used to determine the torque conversion µ of the following: = T 0 B T 0 A = +1 36.073 = 1 36.073 (6.11) The efficiency η is then calculated as follows: = i = 1 36.073 +1 35.006 = + 35.006 36.073 = 0.9704 97% (6.12) This is the efficiency of only the compound three- carrier planetary gear. Also, when determining the efficiency η, it is possible to check the correctness of the calculation, similar to the ideal torques: P T 0 i = T 0 A + T 0 B + T 0 C = 36.073 + 1 + 35.073 = 0 (6.13) A similar efficiency calculation has been performed in Appendix D1.2 in VDI 2157:2012. The conclusion confirms the advantages of the minimized weight design over several stages independently. 6.13.1 Oil quantities and power losses Subsequent calculations of oil quantities and power losses are based much on empir- ical values. Only a precise calibration on the test field can confirm the empirical values. After all, the pipe losses from the distribution tubes must be observed pre- cisely. The efficiency of the drivetrain must be given the greatest importance and care. Although little attention is paid to this subject in the development phase, the interest is greater as soon as the design has to stand out against the competition in the market. The estimated compilation already shown in Figure 6.31 indicates an approximate oil flow of 723 l/min calculated with the simple orifice equation. With this empirical formula, the oil quantity can be estimated. The diameter of a mini- mum nozzle, the size of the main bearings, and the decision to use a dry sump con- tributed to the relatively large oil volume for the main bearing and in consequence the total amount of 723 l/min. How far this quantity can still be reduced, the test run will check trough observation of the temperature distribution in the gearbox. The power loss of about 200–225 kW of the 7.5 MW gearbox solution from gearing and bearing losses is as well an indication. 6.13.2 Calculation of gearing according to ISO 6336 standard (Part 1–6) A number of standards are available for the recalculation of the gearing and its assess- ment by independent third parties in the wind world. It is important that a sensible
- 372. Gearbox concepts and design 343 selection is made. For a long time, not all requirements complement each other and can be processed. Many things simply contradict each other and have to be rejected in excep- tion lists. In the end, the designer bears the responsibility and no certification company participates in the certified damage. The design of gears is based on the ISO 6336 [19] series of design codes. The material properties and heat treatment processes are defined in ISO 6336 Part 5. The bending fatigue strength and the fatigue strength of the gear surface are checked using the safety factors recommended in IEC 61400- 4 (Table 6.3). Before the ISO recalculation of each individual tooth contact, the material char- acteristics must be determined. Here we adhere to ISO 6336- 5. Minimum quality requirements for gear components: Q5 external, Q7 internal. All gears and torque transmitting shafts will be delivered with certification ISO10474- 3.1C to be in compliance with ME- quality per ISO6336- 5. The certifica- tion includes the following documentation: • • surface hardness Eht550 and Eht400 determined on a representative test piece (see ISO 6336- 5 for definition) • • surface roughness • • grinding temper etch inspection per ISO 14104 • • magnetic particle inspection • • inspection of the microgeometry ISO 6336 [19] (all parts) consists of international standards, technical specifi- cations (TS) and technical reports (TR) under the general title: calculation of load capacity of spur and helical gears. International standards contain calculation methods that are based on widely accepted practices and have been validated. TS contain calculation methods that are still subject to further development. TR contain data that are informative, such as example calculations. The formulas of the ISO 6336 series are intended to create a uniformly acceptable method for calculating the load capacity of cylindrical gears with straight or helical involutes. Several methods for calculating the load capacity as well as for calculating various factors are permitted. The instructions in ISO 6336 are therefore complex, but also flexible. The formulas include the major factors cur- rently known to affect gear damage covered by the ISO 6336 series. The formulas are designed to allow the addition of new factors as new knowledge emerges in the future. Table 6.3 The architecture of the can be roughly sketched consisting of a bottom sensor layer a middle network layer, and a top application layer Material Allowable stress for surface Allowable bending stress Sun, planet:18CrNiMo7- 6 +HH σHlim = 1 500 N/mm², σFlim = 500 N/mm² Ring gear: 34CrNiMo6, nitrated σHlim = 1 000 N/mm², σFlim = 370 N/mm²
- 373. 344 Wind turbine system design In the above (Figure 6.40), the selected gearbox is shown schematically and the rough dimensions of the gearings are determined by KMAAG for all four stages: If this range is exceeded, the calculated results must be confirmed by experi- ence. Design considerations to avoid fractures originating from stress increases in the tooth flank, chipping at the tooth tip and fractures of the gear blank by the web or hub, must be analysed using general methods of machine design. To calculate the safety factors of the gear meshing, the load factors given in the standard according to ISO 6336- 1 method C have been taken into account, i.e., very conservative values (Table 6.4). Figure 6.40 GGS design for the 7.5- MW gearbox [38] Table 6.4 K- factors for ISO 6336 calculation proposed in GGS design Use parameters applied by GGS Name Value Application factor KA for nominal torque KA 1.30 Dynamic factor KV 1.05 Transverse load factor (root stress) KFα 1.00 Transverse load factor (contact stress) KHα 1.00 Face load factor (root stress) KFß 1.15 Face load factor (contact stress) KHß 1.15 Mesh load factor (eight planets) KγGGS 1.23 Mesh load factor (six planets) KγGGS 1.16 Life factor at 10E10 cycles YNT/ZNT 0.85 Pitting safety factor SH2 1.56 (SH 1.25) Bending safety factor SF 1.56
- 374. Gearbox concepts and design 345 The safety factors of gears for given load- duration distribution can be deter- mined by an iterative procedure, as described in reference ISO 6336 Part 6, whereby the calculation of these application factors forces extremely complicated proce- dures, which must be scrutinized closely. As always, there are programs behind the loads that generate synthetic loads. Developing a scientific method from these loads, how high an application factor should be, usually raises new questions. And so, you can spend days on calculations. Ultimately, all load criteria are subject to arbitrary fixed limits of material pairings and manufacturing errors, which include a number of ‘ignorance factors’. Ten such factors, with a 5% increase in each, would reduce the permissible load by 40%! The ISO 6336 series contains procedures based on tests and theoretical studies to which each procedure refers. The procedures are validated for the following: • • normal working pressure angle from 15° to 25°; • • reference helix angle up to 30°; • • transverse contact ratio from 1.0 to 2.5. The procedures in the ISO 6336 series provide design formulas for calcu- lating the load carrying capacity with regard to various failure modes such as pitting, tooth root fracture, tooth flank fracture, scuffing, and micro- pitting. At pitch speeds below 1 m/s, the load- carrying capacity of the gear is often lim- ited by abrasive wear. The procedures described in Parts 1–19 of the ISO 6336 series address fatigue analysis for gear design. The procedures described in Parts 20–29 of the ISO 6336 series relate primarily to the tribological behaviour of the lubricated flank contact. Parts 30–39 of the ISO 6336 series contain example calculations. The ISO 6336 series allows for the inclusion of new parts under appropriate numbers to reflect knowledge gained in the future. This document and the other parts of the ISO 6336 series provide a coherent system of proce- dures for calculating the load- carrying capacity of external or internal gears. ISO series 6336 is intended to facilitate the application of future knowledge and developments and the exchange of information from experience. The influenc- ing factors presented in these methods provide a method for predicting the risk of damage based on industry experience and experimentation. They may not be scientifically accurate. Therefore, calculation methods from one part of the ISO 6336 series are not applicable in another part of the ISO 6336 series unless spe- cifically referenced. Part 1 presents the basic principles, an introduction, and the general factors influencing the calculation of the load- carrying capacity of spur and helical gears. Together with the other documents in the ISO 6336 series, it provides a method by which different gear designs can be compared. It is not intended to ensure the performance of assembled gear systems. It is not intended for use by the general public. Instead, it is intended for the experienced gear designer who is able to select reasonable values for the factors in these formulas based on knowledge of similar designs and knowledge of the effects of the items discussed (Table 6.5).
- 375. 346 Wind turbine system design 6.14 Future prospects Gearboxes will maintain an important role in the wind industry. In 2013, GGS pre- sented a 15 MW mono off- shore wind turbine in Munich, Germany [39, 40]. The drivetrain with a novel gearbox system with high ratio 400 used a six- way power split at the output for six generators. Much different, but with a much lower cost, the 2021 idea comes to the fore. The power split of the newly proposed GGS multirotor (MR) is already pulled out five times on the lattice tower and five relatively small 3 MW machines are attached to a large low- cost lattice tower structure. As read at the beginning of Kleinhenz 1937, the system of generation of electricity moves in this order of magnitude 15 MW and consequently again with multiple generators. This turbine proposal is a real alternative to the high- cost increasingly large turbines. The presented GGS monoturbine 15 MW, 2013, in Munich comes to an esti- mated tower head mass of about 1 000 tons, a single drivetrain of an MR, however, only 130 tons. Figure 6.41 compares these two different GGS systems. The off- shore 15 MW machine nearly 10 years ago planned to improve serviceability with a spa- cious nacelle design. The now envisioned MR machine of the same power size is built with 5 x 3 MW machines and with 12 superfast generators in each nacelle. The maintainability is again much easier for the total of 60 distributed superfast Table 6.5 As a matter limitation of space, only results from single Diff_Stage I, listed Particulars of toothing Sun Planet Annulus Power kW 7 500 Speed rpm 100.4 −96.9 −21.6 Centre distance mm 688 Number of teeth PF-type 59 55 −169 Pitch circle diameter dw mm 711.623 663.377 −2 038.377 Helix angle ° 0 Normal module mm 12.00 Normal pressure angle ° 20.00 Active face width mm 390 385 380 Addendum circle da mm 738.240 684.893 −2 012.026 Dedendum circle df mm 680.640 627.293 −2 066.026 Transverse contact ratio [eps_α] 1.73, 1.88 Safety factor for contact stress at operating pitch circle [SH] 1.44, 1.52/2.54, 1.63 Backlash mm 0.55
- 376. Gearbox concepts and design 347 generators. The generator units weigh only 130 kg. A rotating five- arm lattice star and additional tiltable single rotor stars ensure that assembly, repair, and disassem- bly are always easy without the need for large cranes. This MR machine is specially designed for high alpine locations. 6.15 Conclusion Gear units for wind turbines must ensure maximum reliability over a period of around 20 years and withstand high dynamic loads. At the same time, lightweight construction and cost minimization are required. These requirements can only be met by a well- thought- out design, high- quality materials, high manufacturing qual- ity and maintenance. To design a reliable and lightweight gearbox, it is necessary to describe the loads acting on the gearbox as accurately as possible. Chapter 6 showed how the transient torque/speed characteristics of a wind turbine affect the volume/ weight of the drivetrain and the advantages of using planetary compound gearboxes. It also emphasizes the importance of isolating the gearbox from the parasitic forces acting on the rotor arm from the turbine. The volumetric concept facilitates the synthesis of the design of gearboxes instead of an analytical/iterative approach. It helps optimize the overall size and weight of gearboxes by applying lower ratios in the high- torque, low- speed indi- vidual stages, especially when planetary stages are involved. The presented robust KMAAG calculation method is freely available in the literature. The original patents of Hicks flexible pin have expired, so the technology is in the public domain at least for the Hicks patent. Designing and constructing with classic manual formulas also result in lightweight and reliable gearboxes. The development phase can be started very quickly with the creation of the correct models. Simulation tools, on Figure 6.41 GGS multirotor (MR)
- 377. 348 Wind turbine system design the other hand, must always be evaluated skeptically. Furthermore, test runs remain necessary for practical control and verification. Unfortunately, modern simulation tools are displacing the classic simple methods. But greatest care is required, beware of the operator! The most important literature references for the basics are listed in Table 6.6. References [1] [ANSI/AGMA 6123- B06] Design Manual for Enclosed Epicyclic Gear Drives. [2] Hertz H.G.W., Gesammelte Werke B., Band I. ‘Miscellaneous’. [Verlag: Leipzig] 1895. [3] ‘IEC 61400- 4:2012(en) wind turbines — part 4: design requirements for wind turbine gearboxes’ in [IEC 61400- 1] International Electrotechnical Commission. Geneva, Switzerland; 2012. [4] Stoeckicht W.G. ‘Some advantages of planetary gears’. The American Society of Naval Engineers Inc. 1948. Available from https://doi.org/10.1111/j.1559- 3584.1948.tb02762.x [5] Buckingham, E. Manual of gear design. Industrial Press; 1935. [6] Available from http://eolienne.cavey.org/en/commentaires.php [7] Förster H. ‘Zur Berechnung des Wirkungsgrades von Planetengetrieben, Konstruktion’.1969, vol. 5, pp. 165–178. [8] Berger G. Windenergieanlagen effizienter gemacht. H.6/2005 S.35- 37; [9] Wolf A. Die Grundgesetze der Umlaufgetriebe. Germany: Friedr. Vieweg Sohn, Braunschweig; 1958. Table 6.6 Short recap of relevant literature for the basics of gear design # Standards used Remark 11 MAAG Taschenbuch Ausgabe 1985 Rough design 6 ISO 6336: Calculation of load capacity of spur and helical gears (2019) Part 1: Basic Principles, Introduction and General Influence Factors (1996) Part 2: Calculation of Surface Durability (Pitting) (1996) Part 3: Calculation of Tooth Root Bending Strength (1996) Part 5: Strength and Quality of Materials (2003) Single tooth contact…recalculation 1 ISO 9084 and AGMA 6123 Kγ values 36 IEC 61400–4 Safety Factors ISO 1328–1 Accuracy Grade
- 378. Gearbox concepts and design 349 [10] Kril A., Dimitar Petkov K. ‘Planetary gear trains, theory, calculations and design, load capacity and durability, manufacturing and control, application’. CRC Press, Boca Raton, Florida. 2017. [11] Barenhorst Fet al. ‘New drive train concept with multiple high speedgenera- tor’. J. Phys.: Conf. Ser. 2016. [12] Giger U., Fox G.P. (eds.) ‘Leistungsverzweigte Planetengetriebe in Windenergieanlagen mit flexibler Planetenlagerung’.2003. [13] Hehenberger G. Electro- mechanical differential drives for wind energy con- verters. SET Sustainable Energy Technologies GmbH; [14] Lanchester F.W., Lanchester G.H. ‘DEWI- magazin’.1923, p. 605. [15] Merritt H.E. Gears, London: Pittman Sons. 1955. [16] Shigley J.E. Mechanical engineering design. United states: McGraw- Hill; 1963. [17] Jura Ir.G.J., Rademakers Amdrijvingen B.V. Optmalisatie van overbrengin- gen voor windturbines. Rotterdam: Overdruk uit Aandrijfetechniek; 1983. [18] Available from https:// buch- der- synergie. de. © 2007 - 2022 achmed A: W. Khammas Lager [19] ‘Deutsches Institut für Normung, Berlin: Germany’. Calculation of Load Capacity of Cylindrical Gears’[DIN 3990:1987]. 1987. [20] Muller H.W. Epicyclic drive trains. Detroit, MI: Wayne state University Press; 1985. [21] Seeliger K. ‘Das einfache Planetengetriebe. Antriebstechnik’. 1964, pp. 216–221. [22] Berechnungsgrundlagen ‘Planetengetriebe- Begriffe’. Symbole. 2012. [23] 2011. ‘Bestimmung von Verzahnungskorrekturen und Lagerkräften in Planetengetrieben für Lastkollektive’. [Dipl.-Ing. Mohamed Zeyed Sfar]. Dissertation, Bochum [24] ‘MAAG- Taschenbuch: Berechnung und Herstellung von Verzahnungen in Theorie und Praxis — 2’. Erw. u. Erg. Aufl., Maag- Zahnräder AG, Zürich. 1985. [25] Guo Y., Keller J. Combined Effects of Gravity, Bending Moment, Bearing Clearance, and Input Torque on Wind Turbine Planetary Gear Load Sharing. Michigan: National Renewable Energy Laboratory, W. LaCava University of Massachusetts, to be presented at the American Gear Manufacturers Association (AGMA) Fall Technical Meeting Dearborn; 2012. [26] ‘Richtlinie 2006/42/EG (Maschinenrichtlinie)’.2006. [27] Hähnel T. Auslegung von Maschinenelementen dynamisch hochbelasteter Antriebe mittels Messung und Simulation; Fakultät Maschinenwesen der Technischen Universitäten Dresden, 2009. [28] Giger U. Entwicklung und Erprobung eines neuartigen Antriebstranges für die Kompaktwindturbine Falcon; DMK, Dresden, 2009. [29] ‘Rolling bearings - dynamic load ratings and rating life. Geneva, Switzerland’. [ISO 281:2007] International Organization for Standardization. 2007. [30] Kleinhenz F. ‘Das Gross- Windkraftwerk MAN- Kleinhenz, Bericht Nr.6 der RAW’. 1943.
- 379. 350 Wind turbine system design [31] ‘Design and Specification of Gearboxes for Wind turbines. Alexandria, Virgina’. [ANSI/AGMA/AWEA 6006- A03] American Gear Manufacturers Association. 2004. [32] IWES wind turbine IWT- 7.5- 164 rev 4. Bremerhaven: Fraunhofer- IWES. 2018. Available from https://doi.org/10.24406/IWES-N-518562 [33] Rolling bearings — dynamic load ratings and rating life [ISO 281:2007] International Organization for Standardization; Geneva, Switzerland, 2007. [34] Giger U., Arnaudov K. International Conference on Gears; Sofia, Bulgaria, 2013. pp. 101. [35] Blok H. ‘Les températures de surface dans des conditions de graissage sous pression extrème’ in Congr. mondial du petrole, 2me congr. paris blitztemperatur; [36] Arnaudov K., Giger U. ‘High efficiency high torque gearboxfor multi megawatt wind turbines’. Presented at SCIENTIFIC PROCEEDINGS VIII INTERNATIONAL CONGRESS MACHINES, TECHNOLОGIES, MATERIALS; Bulgaria:, 2011. [37] ‘All parts, calculation of capacity of spur and helical gears’. [ISO 6336:2019] International Organization for Standardization, Geneva, Switzerland. 2019. [38] Giger U. Transmission for a Wind Power Installation, Torque Supports, and Wind Power Installation. [WO 2008/104258 A1]. [39] Available from https://www.aramis.admin.ch/Texte/?ProjectID=36954 [40] Giger U., Kleinhansl S., Schulte H. ‘Design study of multi- rotor and multi- generator wind turbine with lattice tower—a mechatronic approach’.Applied Sciences. 2021, vol. 11(22), p. 11043.
- 380. 1 HAWE Hydraulik SE, Aschheim, Germany 2 bewind GmbH, Rendsburg, Germany 3 windwise GmbH, Münster, Germany Chapter 7 Hydraulic systems and lubrication systems Andreas Nocker1 , Arved Hildebrandt1 , Christian Bulligk2 , and Daniel von dem Berge3 The hydraulic systems and the lubrication systems belong to the so- called sub- systems of a wind turbine. There is no doubt that these systems have a considerable influence on the lifetime and service intervals of far more expensive wind turbine components such as the bearings, rotor blades, gearbox, tower, or foundation. Due to the low costs of the systems, compared to the above- mentioned core components, these systems are often underestimated in their influence on the profitability of a wind turbine over the complete lifetime. Therefore, the design of the sub- systems frequently enters quite late in the design process of the wind turbine. The components used and the design of these subsystems also have a decisive influence on the overall efficiency of the wind turbine. The following section will give an overview of the different systems and the most relevant variants and layouts that there are in the market. 7.1 Hydraulic systems When speaking of hydraulic applications in wind turbines, it usually refers to assem- blies for brake control (rotor and yaw) and for changing the blade setting (pitch con- trol) (see also Chapter 3).Another aspect to be considered is the use of hydraulics for permanently installed handling cranes, mechanisms for opening the nacelle and the mechanical blocking of the rotor during maintenance work via a rotor lock. The following section will give an overview about the core components of a hydraulic system in a wind turbine. Subsequently an exemplary layout will be done for a hydraulic power pack supplying the brake systems. Hydraulic systems for the pitch system actuation will be explained in a separate section in this chapter since the functionalities and the system setup differs from the brake system hydraulics.
- 381. 352 Wind turbine system design 7.1.1 Main Components Hydraulic systems in general consist of a hydraulic pump, motor, reservoir, valve control, and peripheral equipment and so it is the case for hydraulic power packs in a wind turbine. Motor/pump combination The electric motor drives the pump enabling it to pump the oil from the reservoir to the consumer. Electric motors are usually asynchronous motors for 50 Hz or 60 Hz and for 3- phase voltages of the specific country where the turbine is set up. The motor is connected via a coupling or directly to the hydraulic pump. The direct connection has the advantage that it does not cause any efficiency losses. Pumps are chosen based on their specific advantages and features. External gear pumps are applied up to a working pressure of 200–220 bar, internal gear pumps are applied if there is a special request for low noise level and radial piston pumps are used for higher pressure levels and if a robust design is the priority. The motor is assembled on top of the cover plate of the reservoir, or it can be immersed in the tank together with the pump. The second option gives the advantage of a quasi- capsulated system and shows advantages when the system is used at low ambient temperatures since the efficiency losses heat up the oil and keep it at an acceptable temperature without any additional heater. Pressure valves As the name suggests, pressure valves (Figure 7.1) influence the pressure. They limit it to a maximum value (pressure limiting valves) and release the oil finally back to the tank when exceeding a defined pressure level. Or when used as a pressure reduc- ing valve (also called pressure control valve) they reduce it from high inlet pressure to a defined outlet pressure level, which is suitable for the consuming component. While the pressure limiting valve is leakage- free anyway, as long as closed, a pressure reducing valve (Figure 7.2) may be a spool design or leakage- free seated design. But as explained later, in the overall design it is to be combined with an Figure 7.1 Hydraulic symbols for pressure limiting valve and pressure reducing valve (HAWE Hydraulik SE)
- 382. Hydraulic systems and lubrication systems 353 accumulator charging operation to avoid unnecessary restarts of the power pack and to ensure a long- term safety position of the brake in emergency cases. Check valves Check valves (Figure 7.3) are used to control the oil flow direction. They are open in one direction and closed in the other direction. They are comparable to the component of an electric diode. They are leakage- free when closed due to their seated design using a ball or a cone as closing element. Figure 7.2 Pressure reducing valve Type CDK (HAWE Hydraulik SE) Figure 7.3 Hydraulic symbol for a check valve (HAWE Hydraulik SE)
- 383. 354 Wind turbine system design Figure 7.5 Hydraulic symbols for 3/2- way seated valve and 4/3- way seated valve (HAWE Hydraulik SE) Flow valves Load-independent flow regulators, load dependent throttle valves and fix orifices belong to the flow valves which are influencing the volume flow inside the hydraulic line (Figure 7.4). All these components can be described as a “bottleneck” that reduces the line/pipe/hose diameter and lead to an oil flow difference before and after the com- ponent. In case the line profile must be adjustable, a flow regulator or throttle valve can be used which allows to manually increase or decrease the line cross section. A flow regulator gives a constant flow independent from load and oil viscosity, while a simple and cheap throttle does not. In order to exclude certain sections completely from the oil supply a manual closing valve or drain valve can be implemented. Directional valves Directional valves are one of the most common parts in hydraulics. They allow the fluid to flow into different paths. The path can be controlled by changing the valve position, thus enabling to supply different consumers. They are named by the number of connections and the number of possible switching positions. For exam- ple, 3/2- way valve or 4/3- way valve (Figure 7.5). Most valves are directly operated electronically by the turbine controller that energizes a solenoid (usually 24VDC) which is mounted directly onto the valve. For certain functions also a manual operation must be enabled or a combination of both. For instance, in the case of a power outage of the turbine or in case of other failures the solenoid valves can be equipped with an additional manual override. Despite the solenoid actuation also hydraulically operated valves are in use. Figure 7.4 Hydraulic symbols for throttle valve and orifice (HAWE Hydraulik SE)
- 384. Hydraulic systems and lubrication systems 355 The directional valves in general can be distinguished between the directional seated valve (Figure 7.6) and the directional spool valve. The directional seated valve technology offers many advantages over direc- tional spool valves. A spool valve is inherently prone to leakage. The keeping of a particular position of the consumer over a long time must be realized with additional check valves. Since oil cannot leak from directional seated valves in the blocked position, there is no need to use additional check valves. Spool valves can also clog if they remain in one switching position for a long time. The annular clearance of the spool valve piston in the housing can be obstructed by dirt particles contained in the oil. This leads to increased stick- slip effects and greater hysteresis or – in the worst case – avoid the switching back into the desired position by the included spring. Secondary measures such as, e.g. dither frequencies superimposed on the control signal or an optimized mechanical design can eliminate these disadvantages, but on the other hand, they cause higher demands and additional effort and costs. Also, a higher oil purity involves more expensive and more complex filtration measures. The directional seated valve offers considerably higher switching reliability than the directional spool valve since clogging with dirt particles circulating in the oil does not occur. These are flushed out repeatedly when the valve opens. Anyway, there are several requirements that the brakes must stay safely closed after a shut- off for a certain time period until maintenance can be done [1]. Also, for this, leakage- free seated valves in combination with an accumulator charging opera- tion is the optimal choice. Cartridge valves The most common design of hydraulic valves is the cartridge design (Figure 7.7). A cartridge valve is a valve on which the housing is in the form of a cartridge, where the port openings are aligned with the ports in the cavity of the control block (sec- tion 7.1.3 Manifold/control block) and which, therefore, can be assembled by simply screwing it into the control block. Unfortunately, the cavities are different when talking about different hydraulics manufacturers and so the valves, even when hav- ing the same function, are not simply interchangeable. Nevertheless, with cartridge valves the most compact and flexible control block hardware solutions can be build up easily. Figure 7.6 Directional seated valve, type NBVP 16 (HAWE Hydraulik SE)
- 385. 356 Wind turbine system design Figure 7.8 Hydraulic symbol for an accumulator and assembly example in a power pack (HAWE Hydraulik SE) Plate- mounted valves Another option is to use so- called plate- mounted valves. These are valves within an own manifold which can be assembled to a sub- plate or sub- manifold by screws. The interface between the blocks is arbitrary, but there are also standardized ones to make these valves interchangeable even when they are from different manufactur- ers. This design is mainly used for directional spool or seated valves. 7.1.2 Hydraulic auxiliaries Accumulator An accumulator (Figure 7.8) is capable of storing an oil volume within a certain pressure range. Inside the accumulator, the oil is kept under pressure by means of a kind of spring. This is not a mechanical spring but a rubber element (valid for diaphragm and bladder accumulators) or a piston (valid for piston accumulators). This “spring” is pressurized on one side by nitrogen gas. When now pushing oil into the accumulator it is pressurized by the compression of the gas and can be taken out when needed later on to support the hydraulic pump to increase oil flow or in emer- gency cases during grid loss to supply oil volume for the actuation of a consumer. While the diaphragm is standard at power packs for brake control, the piston accumulator is the choice for pitch control systems. The bladder accumulator is best Figure 7.7 Plate mounted valves (left) and cartridge valves (right) (HAWE Hydraulik SE)
- 386. Hydraulic systems and lubrication systems 357 for accumulator volumes bigger than 3.5 liter as the diaphragm accumulators are only available in smaller sizes. For further explanations about the accumulator operation and pressurization, see section 7.2.4.3 Pitch accumulators. Hand pump Normally all power packs in wind turbines are equipped with a hand pump (Figure 7.9) that allows to build up pressure by manual actuation in case of a malfunc- tion of the power pack motor/pump or in case of a power outage of the turbine. In some cases, power packs that only supply the maintenance brake the electrically driven pump is completely skipped and the power pack is only equipped with a hand pump. It has to be considered, that the pressure build- up via the hand pump is quite physically demanding since it takes a number of strokes until the pressure is built up and the higher the pressure rises the more demanding it gets. Another factor is that it is time extensive to operate the power pack via the hand pump. For the overall busi- ness case, the labor time that will be required over to a total lifetime of the turbine and the resulting service turbine down times must be considered. Pressure switches and sensors To monitor and control the operation of a power pack, pressure switches and sensors are implemented. These components monitor pressure levels digital or analog and the turbine controller uses them to control the hydraulic system. This includes, for instance, switching on/off the power pack, checking the backup pres- sure in the accumulator(s), checking if brakes are released and controlling if the turbine can be started and many others. Pressure gauge/ measuring points To allow the service staff to check the pressure level and characteristics on site even without an electrical power supply, the power packs can be equipped with additional optical pressure gauges and measuring points also for electronic measure- ment equipment. Figure 7.9 Hydraulic symbol for a hand pump and HAWE hand pump, type CH (HAWE Hydraulik SE)
- 387. 358 Wind turbine system design Figure 7.10 Mono- block (HAWE Hydraulik SE) 7.1.3 Manifold / control block The question now is how to connect all these hydraulic components enabling them to build up a complex hydraulic system and to execute a specific function. To do that with tubes or hoses is not a practical way. Only in the case of a rework or subsequent modification of the system this would be used to avoid the exchange of more com- plex and expensive other components. To mount and connect the above- mentioned peripheral components an additional component is required – the manifold. A mani- fold is a steel or aluminum block which has several bores which serve as pipe for the oil and which allows to mount the peripheral components into or onto it, usually by screw- in or plate mounting via screws. It can be distinguished between two types of manifolds. It can either be a so- called mono- block (Figure 7.10) which is a customized control block developed especially for the individual turbine. In case of a malfunction, the control block, or in some cases even the complete power pack would have to be exchanged. Due to the customer- specific fabrication of the block, servicing results in high costs, long delivery times of replacement components and increased work expenditure. In contrast with the mono- block, with a functional module design, the control is separated into several elements which can be exchanged or modified individually and therefore avoid above- mentioned disadvantages.
- 388. Hydraulic systems and lubrication systems 359 Functional module Also, the functional module philosophy (Figure 7.11) is based on a unit designed for plate mounting. The difference to the above- mentioned design is that it is not a single block, but a subassembly covering a complete function control. Regardless of whether yaw brake, main shaft brake, rotor- lock cylinders, or others, all functions are covered by an individual module. If a malfunction occurs, the faulty module is easy to identify from its marking and can be exchanged after undoing only a few screws. There are no pressure hoses which need to be disassembled from the consumers. The compact size of the modules allows transportation without a crane or any other lifting equipment. Standard modules that can be used for different wind turbine sizes and different manufacturers guarantee low- cost spare parts with short- term availability. 7.1.4 Centralized and decentralized systems Hydraulic systems in a wind turbine can be distinguished between centralized and decentralized systems. A centralized system consists of a hydraulic power pack that supplies various consumers with one power pack, in a decentralized system one power pack supplies only one consumer. The decision for one of these configura- tions must be based on the following factors: Technical requirements The power pack must be able to supply all consumers in a safe and reliable man- ner. First, it must be clear which consumers will have to be served simultaneously. For this, it is important to know which amount of oil must be moved at which time. Furthermore, it must be considered that especially brake systems play a significant role in the safety chain and depending on the safety concept of the wind turbine it can be necessary to operate them at the same time. Figure 7.11 Functional modules (HAWE Hydraulik SE)
- 389. 360 Wind turbine system design For instance, it can be required to apply the yaw brake and the rotor brake in parallel. This is usually the case in an emergency situation. In case of a power out- age of the wind turbine, the rotor and yaw brake systems might have to be supplied completely by the accumulator volume since the pump cannot be operated anymore. Even though some of the above- mentioned situations are not very likely to happen it is clearly stated in the design guidelines and standards that the worst case must be considered for all components that are part of the safety chain. This also includes the hydraulic systems. One must be aware that a single failure in the design of the hydraulic system can lead to harm to persons. Distance between power pack and consumer The distance between the power pack and the consumer affects the system perfor- mance. Especially for big turbines with large distances, it can make sense to divide the supply of the consumers between several power packs. In general, it is also possible to supply consumers over wide distances, but this can influence the costs of the power pack so that a decision for a decentralized system can be more cost- efficient. Safety philosophy As already mentioned above, depending on the overall turbine concept, the brake systems can play a significant role in the safety chain. Here the two possible solutions offer different arguments. A centralized system usually includes fewer components. The lower number of components also reduces the possibility of fail- ure. On the other hand, in case of an outage of a single power pack in a decentralized system the other power packs can still be operated. Price competitiveness Since the number of components is lower for a centralized system it can be assumed that the initial costs will also be lower here. Also, the installation and main- tenance costs for the overall lifetime of a wind turbine for a single power pack in a centralized system can be assumed to be lower. On the other hand, as mentioned above, the cost of a hydraulic power pack depends highly on the required power and the specified functionalities. In case a special power pack has to be developed con- taining only a few standardized components it can be cheaper to switch to a higher number of small, standardized power packs in a decentralized system. 7.1.5 How to engineer a hydraulic power pack In the following section an exemplary layout of a power pack for wind turbines will be made, based on an imaginary customer specification. The power pack shall sup- ply the rotor brake and the yaw brakes. The design is based on a so- called “compact” power pack (Figure 7.12) that is especially suitable for a decentralized configuration. It forms a compact unit
- 390. Hydraulic systems and lubrication systems 361 comprising a submerged electric motor and hydraulic pump in a tank, needs only a small installation space and builds a quasi- sealed entity. The valve control is attached directly to this unit via a connection block without the need for any addi- tional external assembly material. Rotor brake characteristic The rotor brake system consists of one rotor brake that is mounted on the back of the gearbox. It shall be operated at a nominal pressure of 54 bar. The rotor brake is an active brake, which means that it is activated when pres- sure is applied. In normal operation, the brake is free of pressure. The rotor brake shall be activatable in two ways, either via the turbine controller or via a manual override. To protect the gearbox from damage a smooth pressure build- up shall be enabled. Figure 7.12 Compact power pack for brake applications (HAWE Hydraulik SE)
- 391. 362 Wind turbine system design It can be assumed that there will be very few switching operations ( ~500) in 20 years’ lifetime. Anyway, the brake is often used as a holding brake only and not as a functional brake to stop the rotor movement. Yaw brake characteristic The yaw brake system is made up of several hydraulically activated brake cali- pers which act on a brake disk which is a ring at the top end of the tower. The yaw brake is operated on three pressure levels: 1. Holding pressure of 200 bar: This is the pressure to be applied in case the nacelle is in the desired position toward the wind. 2. Low pressure of 15 bar for yawing: When the nacelle position needs to be adjusted toward the wind, the brakes need to be operated at a lower pressure so the yaw motors can turn the nacelle. This procedure may happen 2–3 times per 10 minutes and therefore approximately 1.5 million times in the turbine’s lifetime. 3. Free of pressure: Wind turbines have a cable unwinding procedure. The pro- cedure is required when the nacelle has been rotated into one direction for a cer- tain number of rotations. This leads to a twist in the cables. Once the permitted number of rotations of the nacelle has been exceeded the turbine has to unwind the cables. For this procedure, the yaw brakes shall be free of pressure allow- ing a fast unwinding of the cable. This status is also used for flushing the lines, tubes, and fittings with clean oil by running the power pack pump continuously as usually an oil exchange is not given in the brake circuit. The power pack shall furthermore comprise an accumulator that enables activa- tion of the yaw brake system in the case of a power outage of the turbine, avoiding further movement of the nacelle, and to reduce ON- time of the motor- pump combi- nation. It must also comprise a manually operated hand pump enabling an operation on site in case of a malfunction of the turbine. Exemplary layout For the above- mentioned specifications, the hydraulic schematic (Figure 7.13) below has been developed. It follows a modular design approach and consists of a basic module and two separate modules for the rotor brake and the yaw brakes. The three main elements are easy to recognize: the compact motor/pump subas- sembly (1) in a tank with a usable oil volume of approximately 6–8 liters and the modules for the rotor brake (3) and yaw brake (2). The manually operated hand pump is used for emergency operation of the brakes in case of grid loss or malfunc- tion of the turbine. This example is a constant pressure system based on an accumulator charging operation. The pump is activated directly by the turbine controller which switches the power supply based on the input signal of the pressure sensor. Once the accu- mulator is filled the pressure in the system is rising. When the desired pressure is
- 392. Hydraulic systems and lubrication systems 363 Figure 7.13 Hydraulic schematic for exemplary layout (HAWE Hydraulik SE) reached the sensor indicates it and gives a signal to the turbine controller and the power supply for the pump is stopped. The maximum pressure level of 200 bar is mainly determined by the consumer with the maximum pressure level. These are usually the yaw brakes, which is also the case here.
- 393. 364 Wind turbine system design Rotor brake actuation: In normal operation, the brake is released via the 3/2- way valve 3.1 to the tank. Pressure reducing valve 3.2 is located in the supply line to the valve and reduces the pressure to a lower level than for the yaw brake. The pres- sure relief valve 3.3 has a safety function protecting against exceeding the maximum brake pressure. The check valve 3.4 and the small diaphragm accumulator 3.5 are responsible for the controlled build- up of pressure on the brake. The small actuator in the line will be filled when the rotor brake will be applied. Hence, not the full oil volume will go to the rotor brake. Instead for a short period, the oil volume will be shared between the accumulator and the rotor brake. Once the accumulator is filled the full oil volume will reach the rotor brake and the complete pressure will be built up. By modifying the prefilling pressure and volume of the accumulator, the brake curve can be influenced. By using this assembly, it is possible to create a smooth but still dynamic pressure increase and to follow any specified braking characteris- tics defined by the turbine manufacturer. Pressure switch 3.6 controls the complete relieving of the brake pressure after the braking process to make sure the brake is not applied anymore when starting the turbine again. Yaw brake actuation: The yaw brake calipers are controlled via a 3/2- way seated valve 2.1. In normal turbine operation with no diagonal wind flow, the valve is in the position that it is pushed into by the spring. This is the so- called idle posi- tion when the solenoid is de- activated. The brake calipers are under full pressure. When the nacelle is swiveled, the valve 2.1 is activated. The pressure is reduced to the level set at the pressure relief valve 2.3 and the nacelle can be realigned by the yaw drive motors. The 2/2- way valve 2.2 is used for exchanging oil and flushing the brake lines during maintenance work and is closed with no flow in normal operation. Further options: rotor lock and nacelle roof opening Other hydraulic functions not shown in this schematic are the optional rotor lock cylinder control and the nacelle roof opening device. A rotor locking unit ensures that the rotor of a wind turbine remains securely at standstill during servicing or repair. The blocking cylinder for the rotor only extends or retracts when it is deliberately actuated by the service engineer via the rotor locking module on the hydraulic power pack. For fail- safe operation, a “dou- ble safety” or “two- hand- operation” mechanism is required. This feature is mainly applied by a 4/3- way valve which actuates the double- acting rotor lock cylinder. An additional, optional safety precaution is realized by connecting an interme- diate plate with a 2/2- way shut- off valve in the pump port of the rotor locking mod- ule. As long as this valve remains closed, there is no pressure applied to either the rotor locking module or the rotor blocking cylinder. This provides double security. The opening is implemented manually with or without detent or electrically. Easy and comfortable exchange of big and heavy components like the gearbox or the generator by an external mobile crane is possible if the nacelle has the option of a hydraulic roof opening. Also, this device is hydraulically operated. Cylinders, driven by 4/3- way directional valves are responsible for opening and closing one- half of the nacelle roof. If the wind forces are too high for the mechanical structure
- 394. Hydraulic systems and lubrication systems 365 additionally assembled over center valves, keeping the roof normally leakage free in its position, open and release oil to the tank and limit the pressure in the cylinders to a permissible maximum value. 7.2 Hydraulic pitch systems The pitch system of a wind turbine allows to set the blade angle to the present wind conditions to optimize the power output and to limit the component stress to an acceptable level. This section shows, after pointing out the general differences of electrical and hydraulic systems, especially the elements of the hydraulic pitch control, their special features and how to choose suitable ones. Based on exemplary hydraulic schematics the function is explained. 7.2.1 History After an exhibition in Husum in the early 1980s, the chairman of the Danish Hydraulic Association went back to Denmark and told the members, among other things, to focus on hydraulically driven solutions and systems for wind turbines. Especially the companies Islef Hagen (today named PMC) and Industri Consult A/S (later named AVN) put huge efforts into this application field. Until mid- 1990s, the usually used wind turbine type was a stall- regulated machine with less than 1 MW nominal power. The market saw the first turbine with a hydraulic pitch control around 1987 which was a Vestas V27. In 1997, another turbine manu- facturer (Bonus Energy A/S) started investigating hydraulic control systems for the adjustment of the blades. It was a 1 MW turbine for an onshore plant in Japan. The big breakthrough for hydraulic pitch control emerged years later (2001) in the rise of the Middelgrunden project just off the coast of Copenhagen consisting of 20 pieces of 2 MW wind turbines. In 2002, a 2.3 MW turbine with hydraulic pitch control came into production and a 3.6 MW followed in 2005. After 2005 hydraulic pitch control took the lead for blade control, especially for offshore turbines. The development moves on and today we see turbines with hydraulic pitch con- trol with a nominal power of 13 MW. 7.2.2 Pitch control One undesirable property of the wind is that it frequently changes its direction and strength. But designers and operators of wind turbines know how to help them- selves: pitch drives (also called blade angle adjustment systems) not only ensure an optimal power flow but they are also responsible for the safety of the entire turbine in high winds and storms. Figure 7.14 shows a typical arrangement of hydraulic pitch system components inside the nacelle. With the use of active pitch systems, it has been possible to reduce the mechani- cal loads on wind turbines, increase their service life and at the same time increase the energy yield. The pitch system ensures that the rotor blades are always in the best aerodynamic position so that the generator always operates at the optimal operating
- 395. 366 Wind turbine system design Figure 7.14 Hydraulic pitch system in a wind turbine hub (HAWE Hydraulik SE) Figure 7.15 Pitch control depending on wind speed (HAWE Hydraulik SE) point. Since it also acts as a brake on the drivetrain in the event of a fault, the pitch drive is one of the most important system components in a wind turbine. The follow- ing operating states can be distinguished (see Figure 7.15): • • In very light winds (0–4 m/s = lower than Cut- in wind speed), the wind force is too low to overcome the mechanical losses and the inherent friction of the drivetrain. The blades are then in the so- called flag position (pitch angle 90°). The wind turbine is at a standstill.
- 396. Hydraulic systems and lubrication systems 367 • • At low to the medium wind (4–14 m/s = between Cut- in wind speed and rated wind speed), the wind turbine rotates and produces electricity. The pitch angle is then 0°. As much of the wind power as possible is converted into mechani- cal energy until the turbine produces its maximum power at nominal wind speed. • • Above the rated wind speed (14–16 m/s), the turbine must be limited more and more in its power output. It is then “pitched”. The pitch angle of the rotor blades increases with the wind speed from 0° to 30°. The lift force of the wind to the blades is influenced in such a way that the power output of the wind turbine remains constant at the level of the rated power. During a storm (from 22 to 25 m/s) the wind is too strong so that the wind turbine must be switched off to avoid damage. This is the so- called Cut- out wind speed. The pitch angle is then 90° again. This position must be reached under all circumstances to secure the turbine. This is an essential part of the safety concept of the turbine. Pitch adjustment variants In today’s market, there is a “battle” between the advocates of the electric and the hydraulic pitch solution. Both options have advantages as well as disadvantages and failures mean loss of production and extra service costs. In order to help decision- makers to choose the right system from the beginning of the design phase, these two systems are compared in the following text and rel- evant differences are shown in detail. Figure 7.16 System setup of electrical and hydraulic pitch systems (HAWE Hydraulik SE)
- 397. 368 Wind turbine system design a. Simplicity The hydraulic pitch system is a simple system and easy to understand for the production staff and the service technicians. Therefore, service and maintenance are quite simple whereas the electric pitch system involves more components and higher complexity. In the case of an emergency shutdown, all components in line must operate correctly to perform in case of a shutdown. In the hydraulic pitch system, only one valve must be opened and the hydraulic pressure will do the shutdown and bring the blades in flag position. Figure 7.16 shows how easy a hydraulic assembly can be expressed com- pared to an electric circuit and it also explains quite well how many compo- nents must be operated in an electric pitch system to shut down the turbine. b. Total cost of ownership (TCO) As major components must be replaced only twice or less in the entire life of a wind turbine the TCO of a hydraulic pitch solution is very low compared to an electric pitch system where especially the batteries fre- quently need a replacement. c. Service intervals In addition to the regular service intervals, condition monitoring- based service is of high importance in the wind turbine sector to prevent unplanned downtimes. Hydraulic components offer standard solutions for condition monitoring by additional pressure sensors, position transducers, filter monitoring switches and optional online particle counting in the oil. The hydraulic pitch system is relatively simple to design for longer service intervals. As most of the service on hydraulics is the hydraulic filter replacement, it is relatively easy to increase the service interval by installing larger filters and keep the necessary oil cleanliness. This allows the components work properly in the allowed range. Another ser- vice aspect to be kept in mind is the gas pressure inside the accumula- tors. This is essential for an emergency case to bring the blades in flag position but can easily be monitored by a pressure sensor. On the other side especially these two arguments can also be seen as the advantages of the electric pitch system since a high degree of cleanliness is necessary when opening the hydraulic circuit and the charging of the accu- mulators – if necessary – with a nitrogen bottle in the hub is not an easy job. Also, the possible leaking of a hydraulic system speaks for the electric system. d. Fast response The hydraulic pitch system has a fast response time that will minimize the potential overload on the wind turbine. It is easy to imagine a situation
- 398. Hydraulic systems and lubrication systems 369 in which the pitch mechanism is about to pitch the blade in the opposite direction of the emergency stop direction while a failure occurs. In this situation response time is critical. On an electric pitch system, the pitch motors will first have to ramp down, turn the direction and then turn the blade toward the stopping position. This can take up to one second which can be critical for the rotation speed of the rotor. On a properly designed hydraulic pitch system, the cylinder can stop the movement in less than 100 milliseconds and move the blades in the opposite direction fast and accurate. Therefore, it is easy to avoid overspeed of the rotor and the resulting overload on the drivetrain and the wind turbine. e. Low- temperature influence The hydraulic pitch system is only slightly influenced by a low- temperature environment. When carefully selecting the right sealings inside the com- ponents (valves, cylinders, and accumulators) and when using a proper hydraulic pressure fluid with a low- temperature coefficient, an operation at temperatures down to −30°C is not a problem for a hydraulic solution. For an electric solution, this is a more serious problem due to the applied batter- ies that are losing a great extent of their power at low temperatures. f. Power availability The necessary torque to turn the blade is proportional to the pressure in the hydraulic system. The pressure over the necessary stroke can easily be adjusted and it depends mainly on the size and the pre- charging level of the accumulators installed. The hydraulic pitch system therefore usually offers a higher perfor- mance reserve. The hydraulic accumulators normally do not reduce the available power as much as a battery does during a 90° turn of the blade. Therefore, hydraulics normally offers more power (torque) over the entire turn process of the blade. g. Need for lubrication In an electric pitch system, lubrication of the pitch gear is required. This is difficult due to the small movements of the blades which do not allow adequate lubrication of the affected teeth. Furthermore, it has to be con- sidered that the gears have a certain backlash by nature. A hydraulic pitch system does not include any gears that will wear and eventually need replacement therefore a lubrication is superfluous. h. Quick start up after emergency stop It only takes a few minutes to recharge the hydraulic accumulators whereas it takes much longer time to recharge the batteries of an electric system after an emergency stop.
- 399. 370 Wind turbine system design Figure 7.17 Pitch cylinder with attached block (HAWE Hydraulik SE) i. Hydraulic always leaks Hydraulics always bare the risk of leakage, but by selecting the right components including fittings, tubes and hoses, by taking care, espe- cially at the assembly (correct tools, correct torques and skilled staff) and by performing the necessary maintenance work, a hydraulic system is leak free for years of operation. In general, you can say: the more often and more powerful pitch move- ments are required, the more hydraulics is suited. 7.2.3 Hydraulic pitch adjustment systems 7.2.3.1 Systems with 4/3-way proportional valve Usually, hydraulic cylinders with integrated position sensors will be used for this purpose. They are generally turn- key units together with directly mounted control blocks (Figure 7.17). Modern 4/3- way proportional valves are used at the control manifold, usually driven either via an interface in the nacelle or via the BUS system of the wind tur- bine. The problem with these directional spool valves is the design- related internal leakage. Therefore, it is a must to add leakage- free seated valves in the control as the set pitch angles have to be maintained over prolonged periods also during grid- loss conditions when the main power pack cannot be started and accumulators cannot be re-charged. Figure 7.18 shows an exemplary pitch system layout with a 4/3- way proportional valve.
- 400. Hydraulic systems and lubrication systems 371 Figure 7.18 Schematic for hydraulic pitch system with 4/3- way proportional valve (HAWE Hydraulik SE)
- 401. 372 Wind turbine system design How it works: Two cylinders 109A and 109B are acting on one blade. They are controlledbya4/3- wayproportionalspoolvalve102with2/2- wayvalves109and119 at its inlet and outlet. Usually, the movement is only about fractions of a degree. In the static condition when the wind is constant in force and direction, the 2/2- way seated valves 109 and 119 in front and behind the 4/3- valve can be de- activated/ closed. But this is not always the case and depends on the pitch angle control strat- egy for the wind turbine. In combination with the activated/closed valves 116 and 120 the 4/3- way valve is then isolated, and the position of the cylinders is kept leak- age free also for longer time periods. The system is designed to work in differential mode (see section 7.2.3.3 Differential circuity) which means that the backflow from the cylinder rod side is added to the pump flow to move the cylinder OUT. Only in emergency case/grid loss the valves 120 and 116 switch into idle posi- tion/open and the 3 accumulators (106 A–C) push the cylinders out with maximal force and turn the blades into the flag position. For additional safety, valve 103, in a different mechanical design, is redundant to valve 120. The system is completed by two ball valves to isolate the cylinders, if neces- sary. Also, there is a manually actuated bypass valve 117 to discharge the accu- mulators and a pressure relief valve 101 to limit the maximal hydraulic system pressure. 7.2.3.2 Systems with 2/2-way proportional seated valve Another design of the pitch control system illustrated here is more compact and bet- ter priced without sacrificing the existing functionality. The solution is based on two cost- efficient proportional 2/2- way directional seated valves (Figure 7.19) featuring a zero- leakage idle position. The circuitry (see Figure 7.20) shows the same functionality as the above- described solution with 4/3- way proportional spool valves but additionally features much reduced spatial requirement and increased flexibility as cartridge valves are used. Furthermore, the zero- leakage design of these directional seated valves makes the use of additional blocking valves superfluous. The set angle of the rotor blade is thus reliably maintained over a longer period of time. In some cases, the size of the applied pressure accumulators can be reduced while maintaining the energy reserves to turn the rotor blade out of the wind in an emergency case. The integration of additional functionalities, such as the mechani- cal blade fixing (see the small cylinder on the right top of the schematic), redundant valve arrangement which is required in the DNV/GL specific approvals and flow control valves to limit the oil flow is possible in this design without any problems. Another advantage of using seated valves instead of spool valves is that they do not show a decrease in accuracy over their lifetime. The decrease in accuracy is usually caused by fluid contamination getting trapped in the gap between the spool and bore, causing a significant stick- slip effect and increasing the hystere- sis. Counter measures like a dither overlaid on the control signal and an optimized mechanical design may overcome this drawback but will on the other hand cause
- 402. Hydraulic systems and lubrication systems 373 Figure 7.19 Pitch control block (left) and proportional seated valves (right) for hydraulic pitch application (HAWE Hydraulik SE) Figure 7.20 Schematic for hydraulic pitch system with 2/2- way proportional seated valves (HAWE Hydraulik SE) much higher service costs due to required high- level filtration. The contamination level of the fluid in this application must be held below 17/15/12 (according to ISO 4406) whereas the contamination level of 20/18/15 is sufficient for directional seated valves.
- 403. 374 Wind turbine system design Furthermore, seated valves are in many cases rated for higher pressures (up to 400 bar) while 4/3- way directional spool valves are usually rated for maximum 315–350 bar. This makes the cartridge solution superior as it represents a high safety margin regarding functionality and service life. A restriction of the 2/2- way valves is the available current/flow curves. They are usually not as linear as these of the 4/3- way valves. Nevertheless, they are able to fulfill the requirements of an accurate flow during small adjusting movements and also give maximal flow in emergency cases by using a buckled characteristic (see Figure 7.21). 7.2.3.3 Differential circuity The arrangement in a so- called differential circuitry employs the outflow from the cylinder rod side to additionally feed it to the inflow side (piston side) to the volume that is supplied from the pump and the accumulator. Figure 7.22 shows the succes- sive operation conditions (pitching IN, pitching OUT and cylinder moving OUT at grid loss/emergency case). This gives the possibility to design the pump or accumu- lator smaller and allows the necessary power pack to be significantly cheaper if the resulting reduced forces when moving the cylinder out are acceptable. In general, the service friendliness of the illustrated design in case of malfunc- tions is superior. In case of a malfunction, as the directional seated valves are car- tridge design and one valve is for IN and another separate valve is for OUT, must be replaced only for the one side where the malfunction really takes place. Whereas the 4/3- way directional spool valve must be replaced as a complete unit. Figure 7.21 Current flow curves for different valves (HAWE Hydraulik SE)
- 404. Hydraulic systems and lubrication systems 375 7.2.4 How to engineer a hydraulic pitch system The key components which need to be properly chosen for a hydraulic pitch system are the cylinder(s), the pump/motor assembly and the tank size, the accumulator(s) and the pitch- valve. 7.2.4.1 Cylinders A cylinder is characterized by three main dimensions. Diameter of the rod, the diam- eter of the bore and the stroke. The main input data from the wind turbine manufacturer are the forces or the torques which are required to turn the blade against the wind over the blade angle. The geometric design of the blade adjustment mechanism is needed to get the necessary stroke of the cylinder. To get the diameters of the rod and the bore, we need the forces and calculate the necessary pressurized area at the cylinder via the formula 7.1 [2]: A = F p (7.1) Certain limits must not be exceeded regarding the system pressure which is related to the pump design (see Table 7.1) and the limits of the other circuit components. A reasonable maximal system pressure is 350 bar or less. If the diameters and the stroke of the cylinder are fixed, a buckling calculation based on the mechanical installation situation in the hub of the turbine has to be performed and the dimensioning has to be approved. 7.2.4.2 Pump/motor assembly and tank size As we have already seen during the dimensioning of the cylinder when fixing the maximum working pressure of the pitch system, a proper choice of the components always depends on the rest of the integrated components. Figure 7.22 Functionality of differential circuity: left, pitch IN; middle, pitching OUT; right, OUT at grid loss/emergency case (HAWE Hydraulik SE)
- 405. 376 Wind turbine system design When now talking about a suitable motor/pump combination used in the pitch control power pack in the nacelle, we additionally need to fix the size, respectively the oil flow Q of the pump. To do so, another input from the wind turbine manufacturer is required: the dynamic of the cylinder. What is the maximal velocity v for moving the cylinder IN and/or OUT? This value, together with the cylinder area A, gives us the necessary oil flow of the pump as per formula 7.2 [2]: Q = v A (7.2) It must be mentioned here that the pump usually is not chosen based on the maximal flow which has to arrive at the cylinders as the accumulator normally supports the pump. Therefore, the pump and the electric motor can be downsized. By the following equation 7.3 [2], we can calculate the necessary power P of the electric motor which is driving the pump: P = Q p (7.3) Here the factor p is the nominal pressure to be built up. A reasonable value for the efficiency ŋ of a “state of the art” pump is 0.75–0.85. The maximum delivered oil flow by the pump also gives us an appropriate tank size T which is calculated via formula 7.4: T = 3...5Q (7.4) The resulting value is additionally strongly influenced by the heat balance of the power pack and by the number and the size of the accumulator(s). 7.2.4.3 Pitch accumulators The tasks of the accumulator(s) in a hydraulic pitch system are as follows: • • Ensure “back- up” energy in a grid loss case to turn the blades in flag position and bring the turbine to stop. • • Support the pump when flow peaks are necessary to fulfill reaction time to adjust the blades quickly to changed wind conditions. Table 7.1 Pump type overview Pump design Maximum pressure (bar) Special feature External gear pump 210 bar Low costs Internal gear pump 250 bar Low noise level Radial piston pump 700 bar Long lifetime Axial piston pump 350 bar Variable flow
- 406. Hydraulic systems and lubrication systems 377 • • Supply oil for small blade adjustments and this way avoid a start of the power pack at every blade movement. As the accumulators are assembled in the rotating hub of the turbine, piston accumulators are the correct choice as these show the highest reliability when used under such conditions. Accumulators are characterized by four values, the size, the pre- charge pressure p0 , the lower limit of the working pressure p1 , and the upper limit p2 (Figure 7.23). The pressure p1 usually needs to be slightly higher than the maximum pressure that is required to turn the blades under highest wind loads when the accumulator fulfills an emergency actuation. Between p1 and p2 the accumulator is storing an amount of oil pressurizing the other side of the piston which is pre- charged by nitrogen gas to the p0 level. The stored amount of oil is depending on the difference of p0 to p1 and p2 , the difference between p1 and p2 and of course also on the size of the accumulator and the actual nitrogen gas temperature. The combination of different functional requirements of the turbine manufacturers will determine the required oil volume to be stored in the accumulator(s). These requirements can be how often and how far it should be possible to move the pitch cylinder IN or OUT without switching on the power pack pump, a predefined backup volume, or a particular safety philosophy in grid- loss cases. To calculate an appropriate size of an accumulator and to fix the pressure values described above, free of charge calculation programs of accumulator manufacturers are available. It is important to use one that also considers the temperature influence. It also has to be kept in mind that piston accumulators (Figure 7.24) are no light- weight components and two or three smaller accumulators may be the better choice than one big accumulator when thinking of the first assembly or carrying out maintenance work on site. Figure 7.23 Accumulator pressure and volume levels (HAWE Hydraulik SE)
- 407. 378 Wind turbine system design Figure 7.24 Piston accumulator (HAWE Hydraulik SE) 7.2.4.4 Pitch valve The pitch valve is, without doubt, one of the devices exposed to the harshest oper- ating environment in the rotation wind turbine hub. For each blade axis, one pitch valve is required and a failure of only one of the three valves will force the wind turbine out of service. The cost of a replacement of a pitch control valve is secondary to the cost of the service personnel and the loss of energy generation. Checking only the basic parameters like the maximal flow and working pressure will not be sufficient. It is essential when choosing such a valve to have a detailed look at several other technical properties also. Pitch valves – 4/3- way proportional spool valves with a position sensor and optional on- board electronics in standardized nominal size NG6 or NG10 – are spe- cially designed for fast- response closed loop controls (see Figure 7.25). Furthermore, they provide perfect pre- conditions for economic operation in demanding environ- ments (temperature and vibrations) which are present in wind turbine hubs. The valve precision is the basis for high accuracy in pitch control. The optional on- board electronics is usually separately protected by a robust and sealed enclosure. These valves show a high level of dynamics combined with high flow rates, linear or pro- gressive flow characteristics for precise pitching and low hysteresis for a predictable control of blade position.
- 408. Hydraulic systems and lubrication systems 379 Usually, the control valves receive an input signal from the main wind turbine controller via Profibus or CAN bus based on the monitoring of the generator output and other monitored data. The valve flow and performance specifications have to be matched to the system requirements to be compatible with the existing control parameters and to co- exist with the valves on the other axis. The pitch control valve must be capable of withstanding ambient temperatures, ranging from −30°C to 50–60°C in the turbine hub and should be designed to com- ply with the standards for protection against dirt, dust, and moisture [3]. 7.2.5 Outlook The future of the design of hydraulic pitch controls will be driven by cost reductions on the components side, optimization in manufacturing and production processes and reduction of assembly effort. Some of their optimizations can be achieved by the application of “modularization” approaches, breaking down the big system into smaller sub-systems. One example is shown below. Such a “pitch module” represents a complete unit to control the setting the angle for one blade. It is mounted into the bearing between the hub and the blade. The main power pack to supply the energy (pressure and flow) or to charge the blade accumulators remains in the nacelle and the oil is transferred to the hub via a rotary transmission unit. Figure 7.25 Pitch valve (HAWE Hydraulik SE)
- 409. 380 Wind turbine system design By doing that several advantages are achieved: • • Ready- to- run product • • Simplified solution reducing Total Cost of Ownership • • Achieving a standardized, scalable solution • • Implementation of a simplified and efficient supply chain with a reduction of complexity • • Unique solution for the replacement of the existing pitch system • • Applicable to hydraulic systems as well as electric pitch system • • Easy installation directly onto the hub on site • • Easy access for maintenance • • Reduced complexity in logistics set- up 7.3 Automatic lubrication system for bearings The automatic lubrication system of a wind turbine supplies different slewing bear- ings in the turbine with a lubricant. This section summarizes the differentiation of the common automatic lubrication concepts and designs and focuses on an exem- plarily simplified layout of a standardized automatic lubrication system for a yaw bearing raceway and gearing application. This layout is based on the yaw bearing design for the IWES wind turbine IWT- 7.5- 164, which is presented in Chapter 4 of this book. 7.3.1 Fundamentals Several components, especially the larger slewing bearings of the wind turbine, nor- mally have to be lubricated with grease. From the beginning of the wind industry up to the early 21st century, the larger bearings of a wind turbine, e.g. blade bearings, yaw bearings, main bearings and generator bearings were lubricated manually. This means that the service technician lubricated the bearings in a half- yearly up to the yearly interval with a hand pump. Thereby it was necessary due to personal safety reasons for the service technician, that the bearings were not turning. With the development of new and larger wind turbines, the bearing sizes and thus the re- lubrication amount also increase. In parallel, the re-lubrication intervals were shortened on the recommendation of the bearing suppliers. This effect also rises the service costs. Together with the positive effect of shorter re- lubrication intervals and lubrication during the operation of the turbine, respectively, the bear- ings, which ensures a better quality level of the lubricant due to a better and more effective distribution of the lubricant in the bearings and an associated simple method of re- lubrication under certain parameters of the bearings, as well as the reduction of service costs, the automatic lubrication systems are increasingly being used in wind turbines. Meanwhile, the wind turbines are mostly equipped with automatic lubrication systems, where a lubrication pump supplies via a distributor system the grease to the lubrication inlets at the bearings. The excess lubricant escapes through specially
- 410. Hydraulic systems and lubrication systems 381 designed boreholes in the bearing or via one of the sealing lips and is collected with connected containers or a grease pan. Basically, two kind of automatic grease lubrication systems can be differenti- ated, the progressive distributor lubrication system and the injector lubrication sys- tem. Table 7.2 shows the major pros and cons of the different systems. The differentiation of these systems is based on the distributor type, the progres- sive distributor or the injector. With the progressive distributor all connected lubrication points are lubricated one after the other. With this system, it must be taken into account that due to the inherent resistance of the progressive distributor, residual pressures remain between the pump and distributor during the non- operation phase, which can lead to the lubricant bleeding out in the case of sensitive lubricants. This means that the oil separates from the lubricant and the remaining soap with the included additives could block the distributor. These problems usually do not occur, if the right type of grease, suitable for usage in a progressive distributor and regular lubrication activity are taken into account or previously checked. In the meantime, the lubricant manufacturers have reacted to this problem and developed lubricants with significantly more stable Table 7.2 Advantages and disadvantages of the common automatic grease lubrication systems Version Advantages Disadvantages Progressive distributor lubrication system • Simple and robust construction • More economical than injector systems • Good monitoring of the entire lubrication system up to and including the distributor • Different kind of distributor sizes and combinations (number of outlets and volume size of outlets) available • Remaining pressure between pump and distributor during non- operation phase • Grease type must be released for the usage in a progressive distributor • Passive piston activity within the distributor and thus subsequent delivery of the amount of lubricant at each outlet Injector lubrication system • Wide range of lubrication types suitable for this application • Simultaneously dispending of the lubricant from the injectors • Almost complete pressure reduction in non- operation phase • Error- prone due to more required components • More expensive than progressive distributor systems
- 411. 382 Wind turbine system design properties. The robustness of this progressive distributor system and the relatively low costs are the main advantages in opposite to the injector system. The injector system consists of a variable number and size of lubrication cyl- inders, which are mounted on a block or segments. At the end of the longest lubri- cation line, a pressure sensor or manometer is installed, so that the pressure in the system can be observed. There are two functional versions of the injector systems available, which differ in “direct” and “indirect” lubrication. With the “direct” lubrication, the delivery pistons in the injectors are moved by pump pressure. When all the pistons in the injectors have moved to the end position, the pistons block the outlet of the injectors and the pressure in the mainline rises to the set value of the pressure switch at the end of the system. The pump switches off via the signal from the pressure switch and a solenoid valve switch over so that the main line is relieved. During relief, the pistons in the injectors are pressed into their initial position by an integrated spring. During this process, the lubricant for the next lubrication cycle is shifted within the injectors. With the “indirect” lubrication, the lubricant moves the control piston in the injec- tors, which clears the line to the filling chamber of the injector and fills it up. When all the chambers in the injectors are filled, the pressure in the main line increases. The pump is switched off by means of a pressure switch at the end of the main line and a directional valve is switched back so that the line can be relieved. When the pressure is relieved, the control pistons in the injectors move back first and clear the line for the lubricant from the chamber to the lubrication point. The lubricant is delivered to the lubrication point by the spring pressure of the piston spring in the filling chamber. The main advantage of this system is that the grease is not permanently under pressure in the lining. On the opposite, such a system is more expensive and perhaps also more error prone. In the following sections, the different components of an automatic lubrication system and a simplified layout are presented based on the progressive distributor lubrication system. 7.3.2 Components of an automatic lubrication system Generally, the automatic lubrication system consists of a lubrication pump, with or without an integrated control unit, with or without follower piston, some pump ele- ments, a tubing kit, one or more distributors and depending on bearing raceway or teeth lubrication, also one or more lubrication pinions. 7.3.2.1 Lubrication pump The lubrication pump (Figures 7.26 and 7.27) consists of a pump body and a metal- lic or acrylic glass tank. The metallic tank is used when a very big grease reservoir is necessary. The acrylic glass tank is used for grease reservoirs up to approximately 30 liter. It has the advantage for the service technician, that the grease filling level is easily visible during maintenance. An electric motor (AC or DC) drives via a gearbox an eccentric disc. The pistons of the pump elements are moved by this eccentric turning disc and thus sucking the
- 412. Hydraulic systems and lubrication systems 383 Figure 7.26 Example of lubrication pump with connected distributor (Groeneveld- BEKA) Figure 7.27 Example of lubrication pump with follower piston (Groeneveld- BEKA)
- 413. 384 Wind turbine system design grease out of the tank and pressing it into a tube toward the distributor. A check valve included in the pump element ensures, that the grease is not sucked back into the tank. An impeller inside the tank forces the grease from the tank toward the pump elements. Therefore, the lubrication pumps have to be mounted in a vertical position. For usage in the rotor hub as an automatic lubrication system for blade bearing raceway or teeth, the tank is equipped with a spring- loaded follower piston instead of the impeller. This ensures that even in the upside- down position of the lubrication pump, the grease inside the tank is pressed in the direction of the suction area of the pump elements. The pump can be either controlled and activated by the wind turbine control- ler, or by a built- in controller unit, based on a simplified pump and break time. The WT- controlled lubrication pump has the advantage that, depending on the grade of software implementation, the lubrication system can be activated depending on the turbine or bearing condition- oriented. This means that the lubrication system can be activated in defined parameters (e.g. the surrounding temperature, bearing turning speed, amount of necessary lubrication cycles per month, week, or day). The lubrication pump can be equipped with a grease level sensor, which can give a pre- warning or a warning signal when the tank is nearly empty. To refill the grease reservoir during the maintenance of the WT, the pump body of the lubrication pump is often equipped with a connector. 7.3.2.2 Progressive distributor The progressive distributor (Figures 7.28 and 7.29) can be either a kind of mono block with a defined number of grease outlet ports and one size of grease outlet volume per port or it could be a block consisting of a variable number and size of segments which can have a different size of grease outlet volume per port. The distribution blocks or segments consist of several channels and pistons. The outputs can be bridged or combined individually. The size of the piston volume in the segment design can be selected in certain size increments. A lubrication system can be composed of main distributors and subsequent sub- distributors (Figure 7.29). Figure 7.28 Progressive distributor with eight outlets (Groeneveld- BEKA)
- 414. Hydraulic systems and lubrication systems 385 By pumping the grease into the inlet port on the distributor, the pistons in the connected channels of the distributor are moved and thus deliver their lubricant one after the other through the outlet ports. When the last piston has pushed out its grease volume, one lubrication cycle is over. If one of the outputs on the distributor is blocked by any fault or problem, the entire distributor blocks. The cycles can be counted on the distributor with a proximity switch, which is often named as cycle switch. With the information of the volume size of the distributor outlets and the number of counted lubrication cycles, the exact amount of lubricant for the individual lubrication points can be determined. The implementation of a monitoring sensor is very easy, so that the faults like a blockage of the distributor can be observed. 7.3.2.3 Lubrication tubing The lubrication tubes connect the lubrication pump with the distributor and/or sub- distributors and the distributor with the lubrication points or the lubrication pinion. In most cases, a high- pressure hose is used for the connection between the lubrica- tion pump and the distributors, and a polyamide- tube (PA) is used for the connection between the distributor and lubrication point. The tubing length can be estimated according to the 3D model of the turbine. When installing the first lubrication sys- tem, the final tubing length has to be measured, to implement this information in the final part list. 7.3.2.4 Lubrication pinion In the case of yaw and blade bearing teeth lubrication, one or more lubrication pinions (Figure 7.30) are typically used to apply the grease onto the teeth of the bearings or the gearbox pinions. In most cases, aluminium, aluminium- foam- combinations, or rubber- based materials are used for the pinion. The lubrication pinions are designed in that way, that the grease is pushed out at one defined position, ideally at the teeth mesh with the bearing. The gear module, as well as the width of the lubrication Figure 7.29 Progressive distributor with connected cycle switch (Groeneveld- BEKA)
- 415. 386 Wind turbine system design pinion should be adapted to the teeth of the bearings or gearbox pinions so that the application of grease onto the teeth is ensured over the complete width of the teeth contact zone between gearbox pinion and bearing teeth. Most lubrication pinions apply the grease to the tooth flanks of their counterpart. The initial lubrication of the bearing and gearbox pinion teeth should be applied manually so that the complete areas of the teeth flanks are sufficiently covered with a grease film. A lubrication pinion could only ensure an adequate re- lubrication but not the initial lubrication. The correct or necessary re- lubrication amount depends on different factors like: • • number of teeth, • • tooth width, • • amount of lubrication pinions, which should be used, • • frequency and duration of the rotational bearing activity, and is estimated in most cases based on knowledge and experience. It is rec- ommended to validate the homogeneity of the lubrication distribution and the re- lubrication amount during the prototype test. 7.3.2.5 Collection of old grease Depending on the application different methods are used to collect the old/used grease. For the main, generator, and yaw bearing, often a grease pan or canister is used to collect the old, flushed- out grease. At the blade bearings, grease collecting bottles or canisters are used to collect the old grease from the raceway. Thereby the volume of the bottles should be large enough to collect the grease during the mainte- nance interval. It should be noted that due to the stiffness of the connecting structure and the bearing itself, the deformation of the bearing rings, load zones inside the bearing, and the size and number of the grease outlet ports at the bearing ring, the bottles are not filled equally. Typically, there are two areas, 180° in opposite to each other, where the old grease is squeezed into the bottles. At 90° offset to this position, Figure 7.30 Lubrication pinion with mounting bracket and two lubrication inlet ports (Groeneveld- BEKA)
- 416. Hydraulic systems and lubrication systems 387 barely any lubricant comes to the bottles. This effect has to be taken into account for the size of the collection bottles or containers. To avoid, that the internal back pressure in the connection of the collecting bottles becomes too high and the grease is pressed out over the sealing system, the connection borehole at the bearing should be as big as possible (typically M16 × 1.5 mm) – of course without negatively affecting the bearing design – and angular fit- tings, as well as a long tube from the outlet port at the bearing up to the bottle should be avoided. Also, the bottle has to be equipped with a ventilation, so that the air inside the bottle can exhaust when the grease comes to the bottle. 7.3.3 Simplified exemplary design of an automatic lubrication system In the following steps, a simplified design of an automatic lubrication system for a yaw bearing raceway and gearing system is shown. Generally, the design is driven by the following points: • • The necessary re- lubrication quantities for raceway or teeth lubrication; • • Grease type for bearing raceway or bearing teeth; • • Maintenance interval, the tank of the lubrication pump should be refilled (half- yearly, yearly, or for a longer period); • • Number of lubrication ports at the bearing/number of lubrication pinions. With these points, the design of the lubrication system can also be adapted to the other bearing components. Following assumptions and yaw bearing specifications are made for the exem- plary design (Table 7.3). Table 7.3 Assumptions and boundary condition for the automatic lubrication system design Type Value Type of bearing Number of lubrication points Size of lubrication points Type of grease for raceway Re- lubrication quantity raceway* Gear module Teeth width Type of grease for bearing teeth Re- lubrication quantity teeth* Number of lubrication pinions Maintenance interval Yaw bearing and double- row ball bearing 2 × 14, equally spaced M10 × 1 NLGI class 1 (e.g. Klüberplex BEM 41- 141) 7 kg = 7.78 liter† 16 mm 160 mm NLGI [4] class 1 (e.g. Klüberplex AG 11- 461) ~4 kg = 4.44 liter† 4 6 months‡ *Amount per year. † Liter quantity already calculated with grease density of ~0.9 g/cm³. ‡ +1 month tolerance.
- 417. 388 Wind turbine system design 7.3.3.1 Design of lubrication pump tank volume The bearing supplier usually makes a recommendation on the type of grease and annual re- lubrication quantity. Depending on maintenance intervals the grease lubri- cant quantity can be scaled. The tank volume can be chosen according to the use of the lubrication pump for different bearings or as a single pump unit for the yaw bearing. The lubrication system suppliers offer tank volumes in approximately 5- liter steps. In this example, two lubrication pumps are used for the yaw bearing application, one for the raceway lubrication and one for the yaw teeth lubrication. 7.3.3.2 Tank volume – yaw raceway lubrication VTank = yearly amount/12 month maintenance interval, incl. tolerance (7.5) VTank = 7.78 lit r e /12 7 = 4.54 lit r e For a “6+1 months” maintenance interval a 5-liter tank can be chosen. 7.3.3.3 Tank volume – yaw teeth lubrication In the case of our example, it is very easy, because the smallest standardized tank volumes of the most lubrication system suppliers are in the range of 4–5 liter (except of cartridges with an equipped battery pump unit). For this example, also a 5-liter tank is chosen. Due to this fact, it is predictable, that the yaw teeth lubrication pump could be refilled only once a year and not half- yearly. 7.3.3.4 Progressive distributor layout – yaw raceway lubrication In total 28 lubrication points are located at the yaw bearing, where per raceway 14 lubrication points are equally spaced on the circumference of the inner ring. Each lubrication point should be supplied with the same amount of grease per lubrication cycle. Two distributors, each with 7 piston segments and 14 output ports can be used for this application. For better detectability and for more flexibility for the control of the lubrication system, the progressive distributor should be equipped with a cycle switch, which can detect the piston position and thus whether a full complete lubri- cation cycle is run through. 7.3.3.5 Progressive distributor layout – yaw teeth lubrication For this example, a distributor with four or eight output ports is necessary, depending on the width and of the design of the lubrication pinion (see fol- lowing section). The distributor should be equipped with a cycle switch to give feedback about the lubrication cycle numbers and thus the grease amount which is applied to the teeth.
- 418. Hydraulic systems and lubrication systems 389 7.3.3.6 Lubrication pinion layout - yaw teeth lubrication The lubrication pinion runs as a free- running pinion in the teeth of the yaw system, that means, it is driven by the rotation of the yaw system. The gear module and the width of the lubrication pinion must be chosen depending on the yaw teeth design. To ensure, that the grease will be applied onto the yaw teeth over the complete width, the lubrication pinion should have the same or a slightly larger width than the yaw teeth. The lubrication pinion is equipped with one or two lubrication feeding ports, depending on the tooth width. This ensures a better distribution of the grease over the tooth width. Four lubrication pinions are foreseen for the design of the IWT- 7.5- 164 yaw system. Two of them can be positioned at the bearing teeth in front of the first yaw drive pinions at the front side and the other two lubrication pinions can be positioned at one of the middle yaw drives on each side of the yaw system. The lubricating pat- tern (sufficient re- lubrication quantity and distribution on the bearing teeth) should be checked during the validation period. If the distribution or quantity of lubrication is insufficient here, the system would have to be adjusted with more lubrication pin- ions and a larger distributor. 7.3.3.7 Collecting old grease It makes sense to mount a grease pan or grease collecting bottles or containers onto or in the area of the bearing, which catches the old, squeezed out lubricant. Otherwise, the system can become heavily soiled by the lubricant. In addition, such an uncontrolled draining of lubricant also poses a risk of slipping for the service personnel who maintain the turbine. In the case of the yaw bearing, which is often designed with a single sealing lip on the upper sealing side of the bearing, a grease collecting pan circumferentially around the bearing should be installed. By this grease pan, the squeezed- out lubri- cant from the bearing raceway as well as the used grease from the bearing teeth can be collected by the same pan. 7.3.4 Schematic overview and final clarifications Figures 7.31 and 7.32 show an example schematic overview of the lubrication sys- tem for the yaw bearing raceway and an example schematic overview of the lubrica- tion system for the yaw teeth lubrication. In this example, the yaw bearing lubrication system consists of a lubrication pump with two pump elements and a sensor for an empty tank alarm. Each of the pump elements fed a progressive distributor with 14 outlet ports, which are con- nected via lubrication tubings to the grease inlet connection ports at the yaw bearing. Each progressive distributor is also equipped with a cycle switch. In this example, the yaw bearing teeth lubrication system consists of a lubrica- tion pump with one pump element and a sensor for an empty tank alarm. The pump element feeds a progressive distributor with eight outlet ports. Each two ports are connected with a lubrication pinion.
- 419. 390 Wind turbine system design Figure 7.31 Schematic overview of yaw bearing raceway lubrication system (Groeneveld- BEKA) Figure 7.32 Schematic overview of yaw bearing teeth lubrication system (Groeneveld- BEKA)
- 420. Hydraulic systems and lubrication systems 391 7.3.4.1 Final clarifications 7.3.4.1.1 Control software Depending on using a built- in controller in the pump, which is driven by a pump and break interval activities, or if the pump is controlled via the turbine controller, the programming has to be written, tested, and implemented. 7.3.4.1.2 Electrical connection The electrical connection of the lubrication pump and the sensors (e.g. empty tank alarm, cycle switch sensors) has to be clarified, so that it is in line with the electri- cal setup, respectively with the supply voltage for those components in the turbine. Also, the plugs, cable types and length must be chosen. 7.3.4.1.3 Mechanical connection In general, most lubrication system suppliers can offer a mechanical adaption of their lubrication system by using special designed brackets based on 3D model design. Of course, those mechanical connection points could also be provided by the turbine manufacturers themselves. Particularly when connecting larger lubrication pumps to the hub, a detailed analysis of the connection points must be taken into account, so that the tank body gets no defects due to little deformations of the hub and/or the hub internal structure. References [1] IEC standard 61400- 1:2019- 02, Wind energy generation systems – part 1: Design requirements. Geneva, Switzerland: International Electrotechnical Commission; 2018. [2] Gieck K., Reiner G. Technische Formelsammlung. 2005, vol. 31. [3] ‘DIN EN 60529; VDE 0470- 1: 2014- 09, Schutzarten durch Gehäuse (ipcode) (IEC 60529:1989 + A1:1999 + A2:2013)’. Deutsche Institut für Normung. 2019. [4] ‘DIN 51818:1981- 12’. [NLGI grades] Lubricants; Consistency Classification of Lubricating Greases. 1981.
- 421. This page intentionally left blank
- 422. 1 Hydac International GmbH, Sulzbach/Saar, Germany Chapter 8 Cooling systems concepts and designs Ernst- Wilhelm Langhoff1 The main cooling system of a wind turbine is responsible for the complete tempera- ture management of the drivetrain system. 8.1 Introduction This section is intended to briefly summarize the essential basics of the different cooling and sub- systems and their designs in a wind turbine. The main function of the cooling fluids, mainly water- glycol, the gear oil and transformer oil and sometimes oil within the generator is to transport the losses, respectively, the heat of the main drivetrain components to the ambient atmosphere. Smaller, older turbines, like the Lagerwey LW 18/80 or the Tacke TW 600, were completely cooled directly by the ambient air. Normally, there are three temperature range designs (environmental temperatures) in wind turbine market, depending on the site where the turbine shall be installed: • • NCV – the normal climate version, from −20°C to +40°C • • HCV – the hot climate version, from −10°C to +45°C • • CCV – the cold climate version, from −40° to +45°C These are the ranges for the environmental temperatures, which the turbine is exposed to in general. The electrical lube pumps start at a defined internal coolant temperature level (e.g., +5°C), with or without a preheating phase. The mechanical lube pumps usually start with the first revolution of the wind rotor. The following main drivetrain components need to be lubricated and cooled due to extensive generation of heat by their internal losses. Some of these components are optional (e.g., the gearbox) depending on the drivetrain concept of the WT (wind turbine).
- 423. 394 Wind turbine system design • • Gearbox (‘classic’ Geared Drive, Hybrid- Drive) • • Generator (IG, DFIG, EESG, PMSG) for geared DTs or Direct- Drives • • Main converter (full- power, partial- power converter) • • Main transformer (oil- type,dry- type transformer) 8.2 Gearbox Within geared drivetrain concepts, the gearbox is the main mechanical linkage between the wind rotor with the main shaft and the generator (refer to chapter 5), usually it step- up the slow wind rotor revolutions for the generator, e.g., from input nominal speeds from 8 to 25 rpm to 1,500 or 1,800 rpm or 300 to 600 rpm in case of Hybrid- Drive DT configurations. Thus, in Spain the gearbox is called “multiplica- tor” and these naming gives a better understanding for its main function in the wind turbine. Historically, we find only spur gears in older and smaller turbines like the Lagerwey LW 18/80, the Tacke TW 600 or the Dewind D4 (Figure 8.1). Turbines with a higher rated power from 600 KW upwards utilize additional planetary gear input stages instead. Typical representatives of those early turbine utilizing planetary gear stages with big numbers in the market is the NEG Micon NM 750kW or the 1.5 MW class from Tacke/Enron/GE (Figure 8.2) as well as the corresponding Micon and Vestas types. All of these common gearboxes at that time had a so- called wet sump lube system, refer to Figure 8.3. It takes some time with the development of larger and powerful multi- megawatt turbines that dry sump lubrication (Figure 8.4) became more and more popular. In addition, this principle offered a much higher supply reli- ability of oil as a lubricant and coolant for modern gears with extremely high torque density, even under critical operating conditions (temperature, tower vibration, etc.). Figure 8.1 DeWind D4 gearbox, here already with a multigear upgrade, © Multigear GmbH, 2020
- 424. Cooling systems concepts and designs 395 In such a dry sump lube system the oil is not directly stored in the gearbox, it accu- mulates below (the gearbox) in a separate steel reservoir. The mechanical transmissions within the gearbox cause relevant power losses, and thus most of these losses have to be transferred to and dissipated by the gear oil. The typical maximum range of entire losses is between 2% and 3.5% of the Figure 8.2 Winergy/Flender, gearbox for a GE 1.5 turbine, © Flender GmbH, 2022 Figure 8.3 IEC 61400- 4 standard for wet sump lube systems, refer to [1]
- 425. 396 Wind turbine system design rated performance. Generally, highly viscous oil types are used as gear oil, with a viscosity ranging from 220 to 460 cst, mostly we see today ISO VG 320 as a mineral oil or PAO (Poly- Alfa Olefin). The PAO has better temperature performance than mineral oils, with less viscosity at low temperatures and higher viscosity at higher temperatures. In wind turbine classes from 600 kW up to 3.5 MW, so- called direct cooling systems are a typical feature, that means, oil to air cooling devices and mostly the cooler is located inside the nacelle, refer to Figures 8.2 and 8.3. This also means that a corresponding exchange of air between the interior of the nacelle and the ambient air must be ensured to avoid extensive heating within the nacelle. The normal posi- tion of these kinds of forced air coolers is right above the gearbox. However, within larger turbines in the multi- megawatt range typically combined systems are applied, which means these kinds of systems have an oil/water- glycol- heat exchanger (inside the nacelle, near the gearbox) and the cooler (as a passive heat exchanging device) is mounted on the roof of the nacelle. Sometimes the cooling system of the generator is implemented in this cooling circle, too. Within the gearbox, the tooth contacts and the bearings need to be lubricated by gear oil and additionally the gear oil is transporting the heat from these compo- nents to the cooler and heat exchanger, respectively. A mechanical and/or electri- cal gear pump feeds the complete lube system. In larger turbines, there are several pumps, sometimes a mechanical for idling and one or more electrical pumps for normal, productive operation range. An important note here is that, nowadays the oil flow is more calculated for the need for cooling and not for the need of lubrication. As a concrete example and to explain the combined functionality (lubrica- tion, cooling, filtration), let’s have a deeper look into the wind turbine gearbox lubrication- cooling- filtration- system made by the Hydac Cooling GmbH originally for a 1.5 MW turbine (Figure 8.2). The main components are the motor pump group, the Hydac two- stage filter element and the oil/air cooler. In the manifold below the filter housing a thermal valve, type TB 45, is also located. Figure 8.4 Typical dry sump lube system unit, © Hydac International GmbH, 2020
- 426. Cooling systems concepts and designs 397 The way this sub- system works is as follows: The flow path to the cooler is open all the time. The bypass way to the gearbox will be continuously controlled by a temperature- sensitive element by this thermal valve (refer to Figure 8.3). At about 58°C to 60°C the way to the gearbox is com- pletely closed. As a side note, the same technology is also used in automotive cool- ing systems, but of course with different temperature settings. 8.2.1 Filtration Besides cooling and lubrication, the filtration of the gear oil in wind turbine gear- boxes is essential. The life time of the bearings today is not only calculated by the load, but there is also an additional input by the cleanliness about the lube oil (e.g., according ISO 281, Figure 8.5). In addition to the lubrication, the oil also has the task of keeping the particles in suspension and transporting them to the filter system. These particles are collected in the lube filter and will be removed from the lube system by the next filter element exchange (service operation). The target for the general oil cleanliness is during operation −17/15, −/16/14 according to ISO 4406 and it is already known that with wire mesh filter technology it is not possible to reach any defined ISO classes, because the filtration rate is not fine enough. Initial contamination means the particles result from the production and assembling process. So nearly all gearbox manufacturers have their own test benches for flushing, cooling and filtration units with particle counting according to the ISO code 4406. A sufficient filtration of the gear oil in wind turbine gearboxes during operation is also essential for the overall DT reliability. As already men- tioned, the lifetime of the gearbox bearings is not only calculated for the specific design load cases (DLCs) but also their corresponding time shares in the anticipated service life of the system, there is an additional input factor regarding cleanliness for the lube oil (refer to [2, 3]). In the beginning of wind turbine gearbox technologies the filtration rate had been at about 50 microns up to 100 microns nominal, this was realized by wire mash filter elements. The present design is about 10 microns and made of glass fiber [4, 5]. In lube systems made by the HYDAC group there are two- stage filter elements (Figure 8.6). The two- stage filter elements have the advantage that, if the bypass valve opens during Figure 8.5 General definitions for cleanliness and bearing life time calculation
- 427. 398 Wind turbine system design cold start or when the filter element is blocked by dirt or oil aging products, the oil has to pass the 50 microns security stage anyhow. 8.3 Generator Nowadays, all kinds of generators for wind turbines regardless of the dedicated type of generator are forced air- air cooled generators, liquid- cooled generators or com- bined systems with an air to water and water to air heat exchanger. In most cases, these are closed systems to avoid contamination inside the generator, which reduces the insulating ability and leads to degradation of active generator material. In general, generators are divided into the type- classes of synchronous and asynchronous machines. The most commonly used generator type in WTs is the DFIG, which belongs to the class of asynchronous machines. On the other hand, the use of synchronous generators within Direct- Drive drivetrains is currently without alternative. For the largest wind turbine platforms (on- and offshore) medium speed drivetrains become increasingly popular, these also use liquid cooled permanent magnet excited synchronous machines. The following Figures 8.7, 8.8 and 8.9 show some of the common principles for generator cooling systems; however, no claim is made to completeness, there are many different variants. Figure 8.6 Hydac Filtertechnik GmbH coaxial two stage filter element for 10 microns and 50 microns and a 4 bar bypass valve
- 428. Cooling systems concepts and designs 399 Figure 8.7 Principle of a generator cooling system for a jacket cooled generator (water- cooling with top mounted forced air heat exchanger) Figure 8.8 Hydac Cooling GmbH generator air to air cooling system (example) Figure 8.9 Principle of a standard generator cooling system with an air to water and water to air cooling system
- 429. 400 Wind turbine system design 8.4 Main converter In the case of liquid- cooled main converter concepts it is recommended to use vacuum- soldered coolers for the power modules of the converter, which do not have any problems with the solder- forming flux that is known for CAB (controlled atmos- phere brazing) soldered coolers. Since not all material pairings at the installation are known exactly, there is less potential for faults with vacuum- soldered coolers in terms of corrosion and silting. Once the converters have become silted up, it is really not easy to flush them out again. Because there are very few rinsing media that are compatible with the used water- glycols mixtures. In case of such a fault, the relevant equipment has to be transported up the tower, because usually the main converters are not located at the bottom of the tower or in an external cabinet outside next to the tower. The common trend that can be observed for new WT platforms is to install all electrical equipment modular within the nacelle (Enercons new E2 design variants, Vestas EnVentus™ platform). There are only a few companies in Europe and Germany qualified for the service work described, usually they are also carrying out oil changes on the wind turbines. Since the converters are not only used in wind turbines but also in large PV systems, refer to Figure 8.10, operators have the same service issues there. Figure 8.10 Cooling system (heat exchangers, water/glycol- air) of a 500 KW PV park in NRW, Germany
- 430. Cooling systems concepts and designs 401 In order to apparently simplify the service issues, semi- open cooling systems are often used instead of closed cooling systems (Figure 8.11). Semi- open systems have the advantage that they do not have to be hydraulically pre- tensioned and therefore do not require an additional expansion tank, since the vol- ume compensation takes place in the tank. But with the disadvantage that the tank, e.g., has to be preloaded with 0.5 bar, which then corresponds to the maximum operating pressure and must be additionally located at the highest point of the system. Another disadvantage is the connection to the atmosphere via the pre- stressed cover, which can lead to oxygen inclusions and thus to corrosion if there is a malfunction. A closed system design helps to avoid all these critical points with only slight disadvantages. System comparison: So, actually hydraulically preloaded systems are only a little more work- intensive during commissioning, the reason for this is that the system has to be filled carefully with a low volume flow rate, so that only small air pockets occur. A rule of thumb states that the maximum filling volume flow may only be as large as the sum of the venting capacity of all individual venting valves. Depending on the sys- tem, typically up to five vent valves are installed, and each vent valve has a venting capacity of around 5 l/min. This means that, in such a case, the system should be filled with a maximum flow rate of 25 l/min. Hydraulically pre- stressed systems also have the advantage, to ensure that there is cooling liquid at all points in the system, if filled correctly and the medium has no or a minimized contact to the ambient air respectively. In consequence, if the system has also been properly vented, it can be operated almost trouble- free for a long period of time. The expansion valves or so- called active mixing valves in such systems are usually used to control the temperature in the cooling circuits. The disadvantage of active mixing valves is that they require external energy. On the other hand, they also have an advantage that their power losses during the run- up phase and can be used in a targeted manner to heat up the system more quickly. Some electrical sys- tems have a lower efficiency in the lower power range than at nominal power and Figure 8.11 Open, semi- closed respective semi- open and closed coolant systems
- 431. 402 Wind turbine system design thus the gear oil, e.g., can be heated up with these energy loss at low load before increasing drivetrain output power. The technological alternative to active mixing valves is so- called (passive) expansion valves. They are inexpensive and are actually sufficient for many appli- cations. It is also known as wax valves, and they have a service life of between 2 and 4 years and can be replaced inexpensively after this time. Of course, the service life of the valves also depends on the maintenance of the cooling medium, this means, to use the coolant media as recommended by the cooling system manufacturer, only ready- mixed products and never to mix media from different manufacturers. The TB 25 in the water- glycol cooling circuits or the TB 45 in the oil cir- cuits (cooling and lubrication) are passively controlled (by temperature, due to expansion of a phase- change medium, e.g., wax) valves that decide to return cold medium to the cooling component and warm/hot medium to the cooler (Figure 8.12). The control behavior is around 10 K, so that a TB 25, e.g., switch over the medium flow at approximately 35°C from the cooled component towards the cooler. The Hydac group has more than 100,000 water glycol systems in the field, equipped with expansion valves and with reasonable maintenance they run almost trouble- free. However, thermal valves are also wearing parts, they reach their antici- pated service life, with a well- kept cooling medium. A tip from personal experience, very “economical” OEMs sometimes mixed the cooling medium themselves, i.e. concentrate with water and thus, with an unknown mineral content which can results in very expensive consequences. Which mean, all components, including the converter, suffered corrosion dam- age after a short time and had to be replaced. On the other hand, a lot doesn’t always help a lot. Thus a too high concentration of glycol in the system causes corrosion in the system and pitting corrosion can occur within the cooler, refer to Figure 8.13. Figure 8.12 Function principle of a thermo bypass- valve
- 432. Cooling systems concepts and designs 403 8.5 Main transformer The main source of heat generation in a transformer is the copper loss, the so- called I2 R losses. Of course, there are also other physical effects that generate losses and by that heat within the transformer, such as eddy current and hysteresis losses within the core mate- rial, but the contribution from I2 R losses is dominant by far. If this heat is not properly transferred to the ambient air, the internal temperature of the transformer will continu- ously rise, which can damage the winding insulation and the liquid insulating medium (oil) of the transformer. Therefore, it is essential to control the internal temperature to reduce the thermal degradation of its insulation system and thus ensure the service life of the transformer. The cooling system of transformer shall improve its internal heat dis- sipation capability. In general, there are different transformer cooling methods available for transformers. For modern wind turbines up to roughly 5 MW dry- type transform- ers are common, usually these ones are active air- cooled. Then, due to a controlled air exchange between the nacelle and ambient air sufficient cooling is provided. For high- end turbines (on- and offshore) mainly above 5 MWoil- type transformers more and more become a standard. These ones are more compact (high power density) and generally have advanced cooling capabilities. A passive oil to air is the simplest transformer cooling system for oil- type transformers. The natural convection flow of the internal hot oil (within the transformer housing) is used for cooling. Here, the principle is to use the natural convective circulation of oil inside the entire transformer, this means the hot oil flows into the upper part of the transformer housing, and cold oil circulates back in these areas. The heat dissipation to the atmosphere works by natural conduction through radiators at the outside transformer housing surface. This Figure 8.13 Pitting corrosion in the cooling element due to too high glycol concentration within the coolant (water- glycol mixture)
- 433. 404 Wind turbine system design optimized surface structure is known as the radiator (with fins) of the trans- former. In this way, the oil in the transformer housing and tank, respectively circulates constantly. The heat dissipation capability of this basic cooling system can be increased by applying forced airflow to this dissipating surface area, referred to as the radia- tor bank. The heat transfer to the ambient air can be realized just by cooling fins and fans directly at the transformer housing, by means of an additional oil- air heat exchanger, which can be placed near the transformer or outside the nacelle or an oil to liquid (water- glycol) heat exchanger integrated in the entire WT cooling system (Figure 8.14). The internal heat dissipation rate of the oil- type transformer can be further increased if the internal oil circulation is actively supported by oil circulation pumps. This forced oil circulation can be used in its simple form or more specifically and more efficiently by guiding the oil internally through predetermined paths in or near the transformer winding. 8.6 Essential questions for cooling system design Some general assumptions must be made for the design of the main cooling sys- tem or product requirements (e.g., climate zone) of the WT must be specified, respectively. Furthermore, detailed technical information about the DT compo- nents to be cooled is already required for a rough preliminary design. In the fol- lowing, there are some essential questions summed up, which the system designer Figure 8.14 Oil- type transformer (principle)
- 434. Cooling systems concepts and designs 405 must answer in order to enable experts on cooling systems to start with a dedicated design variant. • • Maximum ambient temperature of the components during operation • • Maximum ambient temperature of the components during standstill • • Minimum ambient temperature of the components during operation • • Minimum ambient temperature of the components during standstill • • Maximum supplied air temperature at the cooler • • Maximum oil temperature for the gearbox • • Maximum water glycol temperature the main coolant circuit (and/or the con- verter coolant circuit if separated) • • The lowest inlet temperature is of interest, especially for the converter • • Maximum ambient humidity (rel., abs.) • • Maximum working altitude (above sea level) • • Minimum air density for the specified cooling capacity at 25°C, 0 m MASL (metres above sea level) 8.7 Example – cooling design for IWT-7.5-164 variant The IWT- 7.5- 164 is the generic WT, already introduced and mentioned in the other chapters (e.g. chapter 1, 2) of this book, which is preferably used here as an IP- free design basis. The design of its drivetrain is not fundamentally fixed. Direct Drive and DT with geared transmission are possible. Here, the suggestion from Chapter 7 of a power split gearbox is taken up for the exemplary design of a IWT cooling sys- tem. A gearbox with a dual power split on the output side with high- speed generators was designed in detail there (Figure 8.15). Thus this special variant is now used for a corresponding cooling system design in this subchapter. Figure 8.15 IWT- 7.5- 164 gearbox with two high- speed output shafts for generator
- 435. 406 Wind turbine system design As a general design decision, a dry sump lube system unit with an oil tank of 2,500 liter capacity is chosen (Figure 8.16). The unit is equipped with two electrical pumps and one additional heater pump, for CCV application, but not active in this example. The technical data for the installed oil pump- units are as follows: • • Motor Dahlander type, two speed 750 rpm/1500 rpm • • 140/280 l/min flow for each pump at 12 bars operating pressure, thus it results a maximum flow- rate of about 560 l/min in the entire system • • Dahlander motor with the pump is able to start at +5°C with 140 l/min flow, at about 45°C the unit is able to work with 280 l/min flow The corresponding filter group (refer to Figure 8.17) is equipped with: • • Internal thermos- valves, check and pressure relief valves for electrical pumps • • Three Hydac two stage filter elements type: 2600 R 010 BN4HX/-V- B4- KE50 in each housing The standard dirt holding capacity is at about 980 gram ISO MTD test dust for each filter element. Nevertheless, it should be taken into account that metal chips and particles will have a higher weight. There is an additional mechanical lube circle for idle operation (or grid- loss situ- ations), where the electrical pumps are not working. This circuit is installed to ensure, that the lubrication circuit is functional in any case, e.g., during the start- up phase or when there is a grid failure. The mechanical pumps are usually also referred to as “idle pumps.” Because they supply the gearbox with “some” lubrication even when the WT is de- energized. However, the connection to the oil sump is often a little bit tricky, espe- cially the suction line and the pumps must be suitable for left- right rotation. Figure 8.16 Dry- sump lube system for the IWT- 7.5- 164 gearbox
- 436. Cooling systems concepts and designs 407 In this example, the mechanical pump is equipped with an extra filter, here again with a two- stage element from Hydac Filtertechnik GmbH, with filter fineness of 10 µ absolute and 50 µ nominal (Figure 8.18). Since the pump is the first pump to start up and the two electrical ones are only switched on from +5°C, this circuit is also equipped with a particle sensor. In this case, it’s the Hydac MCS 1500 (Figure 8.18), it counts 200 µ ferromagnetic (Fe) and 550 µ non- ferromagnetic particles. Further technical data are: The oil circuit of the gearbox (for heat dissipation, lubrication and filtering) is connected in turn to a main water/glycol cooling circuit to transfer the heat (Figure 8.19), the gearbox losses respectively to another medium, to transport it to fluid/ambient air heat exchangers (passive coolers) in order to finally release the heat to the ambient air outside the nacelle. Other main drivetrain components Figure 8.18 Hydac Filter Kit with particle counter MCS 1500 and valve manifold Figure 8.17 Integrated filter group with valves, manifolds and filter elements
- 437. 408 Wind turbine system design like the generators, the main converters and sometime the transformer can be connected to this water/glycol circuit in a direct way or indirect by additional heat exchangers (Figure 8.20). In some cases those components are equipped with separated cooling circuits and coolers independent from the main (gearbox) coolant circuit. In our example for the generic IWT, we integrate the cooling of the two 3.6 MW generators into the main water- glycol cooling circuit of the gearbox (Figure 8.20). The converters, which are also liquid-cooled, require a lower cool- ing water temperature and, as described later, will be equipped with their own separate cooling circuit. This main water- glycol cooling circuit for the external heat exchanger is designed for a flow rate of 380 l/min at 4.5 bar. The diagram in Figure 8.21 shows the corresponding design point of the coolant pump unit. Water/glycol (ethylene glycol) with a mixing ratio of 60:40 is selected as the coolant within this circuit. For the complete (estimated) losses of the gearbox and the generator of roughly 420 KW (@ nominal operation), 3 passive roof top cooling elements are necessary, refer to Figure 8.22. Here, passive means, these coolers have no electrical fan units and thus cooling is mainly dependent from the natural inflow wind speed. For technical data of the passive roof top cooler (heat exchnager) refer to Figure 8.22. It is not unusual for the main converters to be equipped with a separate cool- ing circuit, provided they are also water- cooled (water- glycol). There can be many reasons for this, e.g., smaller required volume flows, different temperature levels or a solution that has already been integrated by the supplier. In principle, however, there is nothing to be said against integrating the main converters into the main cooling circuit with the generator and gearbox too. For our example for the IWT, such a separated cooling circuit will be designed. The main converter units consist of two time back to back converter each with 4MVA capacity for the two generators. Figure 8.19 Plate heat exchanger (fluid/fluid; oil/water- glycol)
- 438. Cooling systems concepts and designs 409 Figure 8.20 Schematic of the complete lube and cooling system in the IWT nacelle
- 439. 410 Wind turbine system design Their efficiency during nominal operation is assumed to be roughly 98.5%. The Figures 8.23 and 8.24 show cooler and coolant circuit. Both converters are coupled to the grid via a multi- winding transformer and a common filter. An oil- type transformer (efficiency approximately 99%) with built- on oil/air heat exchangers and forced air cooling is to be used in the IWT. Figure 8.21 Pump curve for the main coolant (water- glycol) circuit and pump unit Figure 8.22 Three of these passive cooling elements are installed for a sufficient cooling on the roof top of the nacelle of the IWT
- 440. Cooling systems concepts and designs 411 Figure 8.23 Selected cooler or heat exchanger, type: Hydac AC- LN12S/1.3/F/B1 Figure 8.24 Schematic of the converter cooling system
- 441. 412 Wind turbine system design Requirement for converter cooling system: Total power losses (converter): ~ 115 KW Flow- rate: 260 l/min Ambient temperature: 30°C Maximum coolant inlet temperature: 52°C (converter cooling plate inlet) Result of the calculation Coolant temperature inlet: 50.95°C Delta temperature inlet/outlet: 1.05°C Pressure drop coolant: 74.36 mbar Air outlet temperature: 40.91°C Necessary air flow- rate: 33,500 m³/h 8.8 Experiences In general, it is strictly recommended to always use high quality water glycol mixtures, preferably already mixed by the manufacturer, as well as never to mix different brands. Furthermore, OEMs and service providers should only use the brands that are normally approved and listed by the manufacturer of the cooling system or its components. Sometimes it makes sense or is necessary to flush the water/glycol coolant sys- tem. During the assembling and flushing process, it is recommended to install a temporary filter with about 100 µ filtration rate, but not finer, because this could physically destroy the coolant liquid. When designing the hose and tubing of the cooling system, care should be taken to ensure that the flow resistance is as low as possible to achieve a low pressure drop (pressure difference between in- and outlet) in the system. As a consequence, there will be a significant loss of flow- rate, which can cause lack of sufficient cooling performance. The reason for that is that the dedicated coolant pumps are usually centrifugal pumps with a specific behavior, this means relatively high flow but less pressure generating capability. This results in the simple rule for the entire coolant system, “low pressure- high flow, high pressure- low flow” with high nonlinearity and pressure sensitivity, refer to Figure 8.25. But problems or faults occur not only during design, assembly or commission- ing, but also when granting licenses and the usual demand for the so- called “local content” in that wind energy business. Just to explain possible influences and their criticalities of the gearbox oil cooling system as clear as possible, it shall be explained here on a real example. A well- known 1.5 MW wind turbine was manufactured at a different location due to local content production requirements. Then, after some time of operation, the gearbox suddenly burst due to overheating. Temperatures up to 80°C were measured in the oil sump (here a wet lube sump). What had happened and what had changed? In the original design, the convert- ers were liquid- cooled and located in the tower base. The tower tier plates were made of waterproof multiplex plates. The new production site resulted in a new mix of components. The multiplex panels became gratings and the liquid- cooled converter became an air- cooled one and additionally the nacelle was even better
- 442. Cooling systems concepts and designs 413 thermally insulated and the existing openings in the nacelle were closed, due to problems arose due to pollen and other external influences. This combination led to situations where the gearbox oil cooler didn’t get enough cold air from outside the nacelle. Even worse, at the sites of the WTs, it wasn’t really cold there either, in summer temperatures of up to 40°C prevailed at typical sites there. In addition to the new design details described above, the heated air from the converters was blown upwards in the tower by a large fan. Thus, the delta T, oil to ambient temperature was no longer 30 K (like assumed in the original design) but only less than 25 K and so the system simply lacked on cooling capacity. The original 45 kW heat dissipa- tion capacity at 100 l/min. oil flow rate shrank to 33 kW to a maximum of 37 kW. The difference seems to be small but the effect was fatal. The diagram in Figure 8.25 shows the typical behavior as an example. At 1.90 bar a typical coolant pump will produce a flow of about 120 l/min. At 2.50 bar you’ll get only 95 l/min. Often an additional pressure drop in the system is caused by the wrong assembling, wrong diameters and elbow fittings. Rectangular elbow fittings cause even more pressure drop than wide bending radius elbow fittings, made like a tube. Figure 8.25 Typical pump curve for a converter coolant (water- glycol) circuit.
- 443. 414 Wind turbine system design In general, the absolute length of a hose is not so important, more important is the inner diameter. As an example, a one meter hose at a flow- rate of 100 l/min with an inner diameter of 31 mm will produce 0.04 bars pressure- drop. On the other hand the same hose length with an inner diameter of only 16 mm, will produce already 0.36 bars pressure drop. Especially the fittings and the connections to the generator or the converters are often the critical throttle points in the coolant circuits. Sometimes it can be noticed, that the channels within the cooling plates of the converters (for power modules base plate cooling) are not well designed (in terms of pressure losses) as well as the connection fittings between the converter parts or the cooling plates, respectively. Figure 8.26 shows a proper design of a cooling plate for converter power modules. For a practical design optimization or assessment of shortcomings it is obvi- ous from a practical point of view, to have as at optimum the design data, the test bench data and the field data too. Only the consideration and comparison of all this data can confirm the correct design of the system, or not, in the worst case, or make an assessment of a system already in operation possible at all. In general, the same design procedure has to be done for the main lube system of the gearbox. Currently, vibration sensors are sometimes installed on the transmissions systems, which then indicate changes (condition monitoring systems, for more details refer to vol. 2) during operation. Due to the fact, that more and more planetary stages are designed with plain bearings, a different particle measurement technique is required. Today, 70/100/200 µ and 400 µ Fe particles are measured, which are usually detected in the gearbox oil in conjunction with failure events, if state- of- the- art measuring technol- ogy is installed. However, in the event of insufficient lubrication or assembly errors or other destructive influences, plain bearings form much smaller chips/particles, they are in the range of 1 to 5 µ and/or even smaller. In order to detect those, it is best to use optical particle counters or a combination of both, see Figure 8.27. Figure 8.26 Cooling plate design for low pressure loss optimization, Hydac International GmbH, 2022
- 444. Cooling systems concepts and designs 415 References [1] ‘Wind turbines – part 4: design requirements for wind turbine gearboxes’. [IEC 61400- 1] International Electrotechnical Commission, Geneva, Switzerland. 2012. [2] 'Rolling bearings – dynamic load ratings and rating life’. [ISO 281] International Organization for Standardization, Geneva, Switzerland. 2007. [3] Dean J.A. Lange’s handbook of chemistry. New York: MacGraw- Hill Inc.; 1990. [4] “Hydraulic fluid power – filters – multi- pass method for evaluating filtration performance of a filter element”. [ISO 16889] International Organization for Standardization, Geneva, Switzerland. 2008. [5] ‘Hydraulic fluid power – filter elements – verification of material compatibil- ity with fluids’. [ISO 2943] International Organization for Standardization, Geneva, Switzerland. 1998. Figure 8.27 Features of a compact, complete ‘smart’gearbox cooling system (gearbox wet sump lube system) with sensor package, © Hydac International GmbH, 2022
- 445. This page intentionally left blank
- 446. 1 Fraunhofer Institute for Wind Energy Systems IWES (IWES), Bremerhaven, Germany Chapter 9 Validation, verification, and full- scale testing Hans Kyling1 , Anna Wegner1 , Karsten Behnke1 , Malo Rosemeier1 , and Alexandros Antoniou1 This chapter deals with the basic ideas behind, methodologies used and derived activities to prove product compliance with stakeholder’s expectations. 9.1 Introduction For a beginning, the terms used in the chapter title need some explanation. Dependent on the field of expertise, the industry and the people involved you will get a broad set of definitions for validation and verification. In this chapter, validation is defined in accordance with guideline 2206 of Verein Deutscher Ingenieure (VDI) [1] and cited from the Guide to the Project Management Body of Knowledge (PMBOK) [2] as the assurance that a product, service, or system meets the needs of the customer and other identified stakeholders. It often involves acceptance and suitability with external customers. Whereas verification is often an internal process and understood as the process to assess whether a product, service, or system complies with a regu- lation, requirement, or specification. 9.2 Validation and verification strategy A validation and verification process should accompany any professional product development process. Some of the advantages coming along with a verified and validated product are or can be: • • Less risk for the manufacturer and user • • Shorter time- to- market (less intensive field testing) • • Better conditions for product financing • • Better conditions for insurances linked to the product • • Higher product reliability and availability time
- 447. 418 Wind turbine system design • • Create customer confidence • • Added value for product advertising The validation and verification process and the derived activities require expen- diture in the lower two digit per cent range of the costs for a new turbine develop- ment. It is challenging to find a good trade- off between gained benefits and their related costs. At this stage, the validation and verification strategy comes into play, which is understood as the (company) specific approach to deliver the best trade- off between the advantages resulting from validation and verification activities and the linked costs. It is specific to a company or even a smaller unit as the priorities of the exem- plary advantages given above will vary. A broadly applicable mindset that can be helpful in developing a validation and verification strategy is, e.g., the Golden Circle [3]. Here, the starting question is why validation and verification activities should be conducted and thus will help to prioritize expected outcomes. The next ques- tion is how the strategy should be implemented. A proven four- step procedure was developed within the European Union (EU) funded project ReaLCoE [4] consisting of following steps: 1. Initialization 2. Detailing 3. Evaluation and decision 4. Implementation The first step of the procedure, Initialization, starts with the collection of infor- mation on the to be developed wind turbine (WT) and on past experiences such as design description, field experiences, and solved and open challenges. With this knowledge, a qualitative risk evaluation can be implemented for all subsystems and components. A list of all mandatory (“must”) validation and verification activities is developed taking into account the certification needs. This list will be comple- mented by a list of “can” activities that help to further mitigate the risk. By the end of this phase, a complete and quantifiable list of validation and verification activities based on the potential risks is defined, and a first review regarding the overall bud- get, timing, and technical feasibility can be performed (see Figure 9.1). The next step, Detailing, is an iterative process to elaborate verification and validation specifications. These contain information as the validation and verifica- tion aim (functional testing, end of line testing, field measurements, and simulation report), needed input, requirements, relevant interfaces, and explicit success criteria. At the same time, the viability of the approach is investigated. For this, aspects as budget, time, technical feasibility, as well as, e.g., the availability of any involved test rigs have to be elaborated. In the third step, Evaluation and Decision, the list of validation and verification needs is assessed, and the decision on which activities will be performed is taken. Therefore, the identified risks and their proposed mitigation activities have to be
- 448. Validation, verification, and full- scale testing 419 compared with the linked boundary conditions of technical feasibility, test rig avail- ability, costs, and time, as shown in Figure 9.2. The last step, Implementation, starts after the detailed planning of the validation and verification efforts is concluded. The validation efforts are prepared, conducted, and evaluated. The benefit of the presented approach for a validation and verification strategy is that the decision on the conducted activities is a transparent risk- based systematic approach that converges into a complete list of validation needs. Due to the system- atic approach, it is assured that no aspect is ignored, even though at first glance it might have seemed to be of no importance. Figure 9.1 Illustration of the Initialization step (© Fraunhofer IWES/Kyling) Figure 9.2 Illustration of the step Evaluation and Decision (© Fraunhofer IWES/ Kyling)
- 449. 420 Wind turbine system design 9.3 Purpose of testing The validation and verification activities defined in the section before can be, e.g., analytical proofs, virtual experiments (simulations), hardware tests, or a hybrid form. In the following, we will deal only with real hardware tests. The specific aim of conducting experiments using real hardware can be versatile ranging from simple functional, over model validation and robustness to fatigue tests. The so- called bath- tub curve (Figure 9.3) is a representation to help understand the different reasons for failure and thus test aims. Obviously, one major aim of product development should be to minimize or at least manage actively the overall number of failures. The con- tinuous line in Figure 9.3 represents the overall failure occurrence during a product’s life and has the form of a bathtub. This line results from adding up the three main types of failures. The early “infant mortality” or “teething failures” can be reduced, e.g., by functional tests. To minimize the constant failures, e.g., model validation is a very constructive test type. The failures due to wear out can be addressed by robustness or fatigue testing. Not only the expected failure mode and phase of the product lifetime for a failure to occur are relevant information, when planning the required test activity, but also the maturity of the product and its underlying technology. For this pur- pose, the concept of the technology readiness level (TRL) was introduced in the 1970s at NASA [5]. The European Commission (EC) defines the TRLs, as given in Table 9.1. A product development based on a predecessor can use partially the results from earlier validation and verification activity. That is why complex systems such as airplanes and cars are usually developed in product families. Figure 9.3 A typical bathtub curve (product failure rate vs. time) and its composing portions (© Fraunhofer IWES/Kyling, source Wikipedia. com)
- 450. Validation, verification, and full- scale testing 421 9.4 Product development using the V-Model The V- Model as explained in Reference [1] and shown in Figure 9.4 can be used to derive or to structure the validation and verification activities derived in a separate procedure as introduced in section 9.2. On the left half of the V- Model, the product is decomposed from top- level system requirements down to material- level require- ments. On the right half of the V- Model, the design properties on different integra- tion levels are assured by means of validation and verification and integrated into the overall product property fulfillment. To keep it simple, Figure 9.4 shows only one test activity on each integration level, but in general, there will be many activities on each level. Table 9.1 Definition of TRLs according to EC [6] TRL Definition 1 Basic principles observed 2 Technology concept formulated 3 Experimental proof of concept 4 Technology validated in lab 5 Technology validated in relevant environment 6 Technology demonstrated in relevant environment 7 System prototype demonstration in operational environment 8 System complete and qualified 9 Actual system proven in operational environment Figure 9.4 V- Model with examples from application to a WT (© Fraunhofer IWES/Kyling)
- 451. 422 Wind turbine system design The following chapters are structured and named in accordance with Figure 9.4. Different exemplary validation and verification activities, devices under test or spec- imen, and the involved test infrastructure are presented and discussed. If applicable, the certification relevance of the presented activities is explained. Where possible and in alignment with the other chapters, the Fraunhofer IWES Wind Turbine (IWT) 7.5 generic model [7] shall be used as an example. 9.5 Full-system testing 9.5.1 Certification measurements The full system of a modern multi- megawatt WT can only be tested in the field. This test presents the final comprehensive check of the complete turbine, with the goal to verify the design assessment and is an important step within the certification process of the turbine. For WTs, the certification measurements follow the standards of the series 61400 of the International Electrotechnical Commission (IEC). The individual parts most relevant for the full turbine testing comprise IEC 61400- 11 “acoustic noise measurement techniques” [8], IEC 61400- 12- 1 “power performance measure- ments of electricity producing wind turbines” [9], IEC 61400- 21 “measurement and assessment of electrical characteristics” [10], and IEC 61400- 13 “measurement of mechanical loads” [11]. The latter one is the relevant standard for the structural test- ing of the turbine. Details of this standard and the application on load validation are presented in chapter 1 of this book. Whereas for some components (e.g. a rotor blade test according to IEC 61400- 23 “full- scale structural testing of rotor blades” [12]), a component test beforehand is mandatory, for others (e.g. the nacelle as a whole) this is not the case. Testing components both in field and on the test bench provide the additional opportunity to validate test procedures. One example of the direct comparison of a nacelle test with the field tests is given in Reference [13]. Compared with test bench applications, the field test poses additional challenges on the measurements. First of them is the external environment with its variable and non- influenceable conditions. Other than during laboratory component tests, during the field test, external conditions cannot be influenced but have to be assessed as precise as possible in order to correctly interpret the data obtained during the field test. Also, surrounding obstacles influence the measurements in the field. Prior to any field tests a site evaluation must be performed to assess the site- specific conditions. Details of the site evaluation are given in the standard IEC 61400- 12- 1 [9]. All obstacles, especially surrounding WTs, are assessed, whether the wind conditions at the WT site or the position of the meteorological mast are influenced by an individual obstacle or not. For certification measurements, all wind directions, which are influenced by obstacles or other WTs, have to be identified. Data acquired during times with wind direction originating from the disturbed sec- tor will be excluded from the analysis to obtain an undisturbed measurement sector.
- 452. Validation, verification, and full- scale testing 423 Using a hub height meteorological mast, the incoming wind speed and wind direction can be quantified at hub height or along a profile, mostly covering the height of the lower half of the rotor. Another possibility to quantify the incoming wind speed presents the use of a remote sensing device like LiDAR (light detec- tion and ranging) or the less frequently used SoDAR (sound detection and ranging) devices. Those methods cover the total rotor height, but just as the meteorological mast, they only give wind speed values at specific points along a profile. However, presently the use of a remote sensing device is only allowed for power performance measurements when it is accompanied by a mast reaching to the height of the lower rotor tip. Also, the turbulence quantification is less accurate using a remote sensing device. Usually, prototype certification is performed at specified test sites, where a meteorological mast is available, therefore the use of a remote sensing device plays a minor role in certification processes. The wind measurement is usually only possible at one specific point, where the best feasible location of the mast or remote sensing device is identified. The best spot for this location, given by the IEC 61400- 12- 1, is 2.5 times of the rotor diam- eter in front of the WT with an acceptable range between 2 and 4 times of the rotor diameter (see Figure 9.5). A third method to quantify the incoming wind speed is the use of a nacelle- mounted anemometer together with a nacelle transfer function that is calculated Figure 9.5 Measurement points and set up of a WT field measurement (© Fraunhofer IWES/Anna Wegner)
- 453. 424 Wind turbine system design beforehand. This method has the large drawback that the nacelle anemometers mea- sure the disturbed wind field by the rotor. Other environmental conditions that are typically quantified during a field test are wind direction, precipitation, temperature, humidity, and air pressure. When using a hub height mast, the latter three can be measured at hub height, otherwise they are measured typically at 2 m or 10 m above ground. The duration of the field tests also depends on the environmental conditions at the site. To complete the certification measurement of a prototype, a capture matrix, which is specified in the respective standards, has to be filled. To this end, each dataset of a 10- minute time interval is sorted according to the corresponding prevail- ing wind speed into the respective wind speed interval (so- called bins). The capture matrix covers wind speed bins from cut- in wind speed to rated wind speed. For load measurements also, certain turbulence intensities are covered by the capture matrix. This procedure ensures that the measurements cover the main operating range of the WT. Another challenge is the time synchronization. A measurement system on a WT often consists of different sub- systems, which comprise the data collection of a certain part in the turbine, e.g., hub with blades, nacelle, and tower sub- systems. To ensure a correct analysis of the data, the time stamps of all sub- systems have to be synchronized to one common system or to each other. To this end, all data are typically referenced to the same time stamp, which can be acquired using a Global Positioning System (GPS) time stamp. Besides prototype certification, a full- system testing can be used to provide a proof of the applicability of the test bench results to full- turbine operation. In order to use the results from the test bench, a link between test bench and field measure- ments has to be established to identify comparable test situations. The main refer- ence parameter of the field measurements is the incoming wind speed. However, the wind speed is not available for test bench measurements. The electrical power would be one approach to overcome this gap. Eustorgi et al. [13] used the rotor speed to identify comparable datasets from the field test and the test bench experiments. All approaches present a simplification and have limitations, e.g., neglect the wind pro- file over the complete rotor. The use of full- system testing for load model validation is explained in Chapter 1. 9.5.2 Measurements on the yaw system In addition to the aforementioned certification relevant works during the full- system testing in the field, a lot of further validation activity is typically conducted. As an example, the often- neglected yaw system will be discussed in the following lines regarding field validation to give an impression of the versatile questions the engi- neers seek to answer and the high level of detail hidden behind all components con- stituting a WT. The yaw system is explained at length in Chapter 4. The components of the yaw system are validated to a certain extent depending on the underlying strat- egy of the WT manufacturer. Typical tests on the yaw bearing are proper functional- ity of the sealing (also under harsh environmental conditions) and measurement of
- 454. Validation, verification, and full- scale testing 425 the unloaded breakaway and dynamic torque. For the yaw drive including the gear- box, usually the torsional stiffness and backlash are determined, and extreme load, fatigue, and harsh temperature trials are conducted. During the prototype testing in the field, it is possible for the first time to validate the whole yaw system under real conditions. This circumstance makes it understandable that there is a high demand for functional tests and tuning of controller parameters. The measurement of the applied torque (more sophisticated but also more difficult to realize) or the electrical power consumption of the yaw drives is an example for such tests. In Figure 9.6, the average electrical power imbalances of all four yaw drives of a 2.X MW WT during the start- up of a yaw movement are depicted. It can be seen that the loading of the yaw drives is asymmetric, and it takes roughly 2 s to distribute the torque load homogeneously. This is probably resulting from different torsional backlashes of the yaw drives. Figure 9.7 depicts the electrical power consumption during the yaw movement ramp- up phase as mentioned before of the same four yaw drives over the acting yaw moment measured on the main shaft. Depending on the acting yaw moment from the rotor and the desired rotational direction, the yaw drives will act as a motor or generator. Again, the power respectively torque is not equally divided among the yaw drives. Such measurements help to improve the system understanding and to derive improvement measures. The breakaway and dynamic torque can be determined under realistic conditions and compared to the design loads. With this information a re- assessment regarding the expected lifetime can be performed. Usually, checks for a proper lubrication (amount and distribution) of the yaw gearing are done as well. It will be also checked, if the yaw brakes are able to withstand the designed yaw moments. Depending on the WT concept and its maturity, the knowledge of the Figure 9.6 Exemplary mean electrical power imbalances of the yaw drives during rotation start- up (© Fraunhofer IWES)
- 455. 426 Wind turbine system design WT manufacturer, and the involved suppliers, the list of typical activities for field validation of the yaw system turns out shorter or longer. 9.6 Integration testing On the integration level of the V- Model (as shown in Figure 9.4), a practical very relevant validation and verification activity is nacelle testing. First, the involved infrastructure is introduced, followed by the test requirements regarding the load capabilities. Subsequently, it is drafted how a nacelle validation campaign in the laboratory is projected and what resources are needed. 9.6.1 System test benches WTs experience complex dynamic load situations during their service life due to the aerodynamic forces acting on the rotor. In addition to the torques desired for energy conversion to drive the generator, the so- called parasitic rotor loads also act on the drivetrain: shear and lateral forces as well as bending moments. Gravitational load- ing is also a notable influence here. Modern system test benches for the investiga- tion and validation of WT drivetrains in the laboratory partly have the possibility to apply parasitic loads to the drivetrain in addition to the torque load. Since the para- sitic rotor loads of modern multi- megawatt WTs are forces and bending moments in the order of several MN and MNm, respectively, the load introduction systems are complex and very costly technical equipment. For the practical implementation of load introduction on WT system test rigs, two main different technical concepts Figure 9.7 Power consumption of the yaw drives over acting yaw moment (© Fraunhofer IWES)
- 456. Validation, verification, and full- scale testing 427 have emerged. On the one hand, the parasitic loads can be transferred to the test specimen by means of hydrostatic power actuators via a rotating disk located on the drive shaft. Such systems are used, e.g., on the test rigs of the Center for Wind Power Drives (CWD) in Aachen, Germany [14] or Clemson University in Charleston, USA [15]. On the other hand, load application systems have been developed in which hydraulic cylinders apply forces to a steel structure that is connected to the drive shaft via a roller bearing. Such a load application system is used in the Dynamic Nacelle Testing Laboratory (DyNaLab), the nacelle test rig of Fraunhofer IWES in Bremerhaven (see Figure 9.8). The following list gives a very brief summary of key features of the DyNaLab: • • Parasitic load application: application of up to 20 MNm bending moment and 2 MN thrust and shear forces, • • Nominal torque: 8.6 MNm (electrically excited synchronous motors in tandem configuration) and overload torque: 13 MNm • • Artificial grid with 44 MVA installed inverter power • • Measurement system: more than 600 synchronous, high resolution and fre- quency measuring channels. Due to the missing rotor and tower in the laboratory, the nacelle has different system characteristics on the test stand compared to the field. In order to simulate real conditions (at least regarding the torsional DOF) in the laboratory, a so- called Figure 9.8 DyNaLab nacelle test bench (© IDOM)
- 457. 428 Wind turbine system design hardware- in- the- loop operation mode was developed. A real- time WT model is simulating for any wind input the resulting load and motion on a defined interface (e.g. hub flange). The test bench is used to simulate this input accordingly, and the specimen including the turbine controller will react as in the field (e.g. adjustment of the pitch angle or generator torque). This reaction is fed back to the real- time WT model and, thus, results in an updated input to the specimen from the test bench. The development of such a real- time capable WT specific model is complex and needs to be planned for prior to testing. A system test bench as explained above cannot only be used to test entire nacelles but also drivetrains or direct drive generators as the EcoSwing high- temperature super- conductor generator [16] shown in Figure 9.9 as the requirements are very similar. 9.6.2 Test requirements Once the turbine developer has taken the decision to test the nacelle as part of their individual and project- specific validation plan, a suited test bench needs to be selected and its availability contractually secured. In order to do so, the major aims of the test campaign need to be clarified (see section 9.3) as it influences strongly the required functionalities and loads of the test bench. For the Fraunhofer IWT 7.5 generic WT, the extreme and damage equivalent loads (DEL) given in Table 9.2 are assumed on the hub interface of the drivetrain. Comparing this load set with the operation envelope of the DyNaLab nacelle test bench leads to the conclusion that the extreme loads cannot be applied as needed. Consequently, using this system test bench design extreme load tests cannot be a part of the validation campaign. Whereas the calculated DEL are entirely covered. Figure 9.9 EcoSwing HTS direct driven generator on DyNaLab nacelle test bench (© Fraunhofer IWES)
- 458. Validation, verification, and full- scale testing 429 The logistics of the specimen and its integrability into the test bench are first and essential topics to work on to clarify the feasibility of test setup. 9.6.3 Projecting a nacelle test campaign As a nacelle, drivetrain or direct drive generator test campaign is a complex and costly activity; this section shall provide an overview of the typical scope, costs, and timing of such a project. The project can be separated into four phases: pre- liminary engineering, assembly and commissioning, test conduction, and disas- sembly. Each of these phases is considered in a dedicated section. 9.6.3.1 Preliminary engineering During this initial project phase all the preparational work to assemble the specimen and conduct the intended tests is done. Figure 9.10 depicts a proven work breakdown structure (WBS) for this project phase. The critical path through this project phase is typically defined by the fol- lowing sequence. In section 9.3, the importance of a focus and the purpose of the test campaign were mentioned. During the preliminary engineering phase, the test plan as an aggregation of all individual tests is defined. This is a nec- essary input to the mechanical adaption design, e.g., a hub adaption will look differently and be more expensive depending on the possibility to transfer para- sitic extreme loads or not. Once the design is frozen, procurement and manu- facturing can start. This project phase usually takes 9–12 months and involves different experts from both involved parties. 9.6.3.2 Assembly, commissioning, and disassembly The assembly starts according to a detailed logistics plan that might involve transport permits, contracting of mobile cranes, and self- propelled modular transporters. In case of the DyNaLab, the logistics were considered from early planning phases on with appropriate attention resulting in a laboratory location near a heavy- duty quay for big offshore WT and highway access for onshore Table 9.2 Test load envelope Degree of freedom Extreme load (kN/kNm) DEL (N ≅ 107 and m = 4) (kN/kNm) Fx(~gravitational direction) 3,000 2,000 Fy(lateral direction 3,000 2,000 Fz(thrust direction) 3,000 450 Mx(bending) 30,000 7,000 My(bending) 30,000 7,000 Mz(torque) 15,000 400
- 459. 430 Wind turbine system design WT. As depicted in Figure 9.11 a gantry crane completes the logistics concept to enable heavy- duty lifting of specimen weighting up to 420 t without external support. The specimen is pre- assembled lifted into the final test position on top of the tower adaption and bolted to the test bench hub flange. Figure 9.10 WBS of a preliminary engineering project (© Fraunhofer IWES/ Kyling) Figure 9.11 Assembly of the direct driven generator in the EcoSwing project (© Fraunhofer IWES/Kyling)
- 460. Validation, verification, and full- scale testing 431 Scaffolding is installed to facilitate access to the specimen. The specimen is integrated into the cooling system of the laboratory, and the power cables are connected to the facility’s switching gear. Control cables are installed to con- nect the specimen and the test bench with each other. Electrical and mechani- cal sensor installation concludes the assembly phase. During commissioning, safety checks are performed and control functions are tested. All auxiliaries (e.g. cooling, greasing, brakes) of the specimen are checked for proper function. The instrumentation and its underlying data acquisition and storage systems are checked for proper function and plausibility. All the mentioned activities of this phase last roughly a month depending on the specifics of the test campaign. The disassembly follows after concluding the test conduction and usually takes less time than the assembly. 9.6.3.3 Test conduction As explained in section 9.3, the purpose of testing is specific to the project. Figure 9.12 lists some individual tests that can be conducted during a nacelle test campaign. The coarse separation into the two fields mechanical and elec- trical tests is a structural simplification. For each of these individual tests, a dedicated test specification is evolved defining, e.g., success criteria, needed instrumentation, and test mode to be used. The tests are controlled by a team usually with the presence of the (turbine) manufacturer on- site due to the com- plexity and development character of the setup. Dependent on the focus and the variety of tests to be executed, this core project phase lasts typically 3–12 months. Figure 9.12 Extract of the scope of testing of a nacelle test campaign (© Fraunhofer IWES/Kyling)
- 461. 432 Wind turbine system design 9.7 Sub-system testing On the sub- system level of the V- Model (as shown in Figure 9.4), a practical very relevant validation and verification activity is gearbox testing. In addition, valida- tion examples of a brake system will be presented. 9.7.1 Gearbox The international standard IEC 61400- 4 [17] gives guidance on a meaningful valida- tion approach for WT gearboxes and states mandatory tests required for certifica- tion. Based on the demands of the gearbox specification, the following stakeholders are to be involved during the planning and conduction of the validation process: tur- bine manufacturer, gearbox manufacturer, bearing manufacturer, lubricant supplier, and certification body. Test criteria and requirements are developed during a design failure mode and effect analysis and conclude in an overall test plan. This test plan includes a mandatory unit prototype test of the gearbox, some tests of the gearbox as an integrated part of the WT as well as requirements for serial production accept- ance testing. The more of the identified validation activities can be performed on a gearbox test bench under repeatable and reproducible conditions the better. Each individual test requires a specific test description containing amongst other purposes and objectives, acceptance/rejection criteria, environmental conditions, and a list of physical quantities to be measured. The prototype test results are processed to define the parameters used for series production acceptance tests. A list of a minimal scope of testing is given in the standard and includes, e.g., low torque application until oil cleanliness requirements are met, torque application in a minimum of four load steps up to nominal torque preferably under rated speed, measurement of actual load sharing for planetary or other split path gear meshes at each load step and heat runs under nominal conditions to check for thermal stability. Major WT gearbox manufacturers dispose of their own test rigs in order to carry out development, certification- relevant, and production- related gearbox tests internally [18]. These gearbox test benches are usually designed in a so- called back- to- back configuration and feature only the rotational DOF as shown in Figures 9.13 and 9.14 in case of a mechanically closed loop setup. The back- to- back configuration consists of two gearboxes mechanically con- nected on the low- speed side (LSS), a driving engine acting on the high- speed side (HSS), and in case of a mechanical power circulation, a torque- tensioning system might be included as well. The torque- tensioning system is needed to simulate rated torque and power conditions while the drive only has to supply the torque to surpass losses. Alternatively, the test setup shown in Figures 9.15 and 9.16 can be applied. In this case, the second HSS shaft is connected to a generator. The power is circu- lated electrically and consequently requires both machines to operate at the required power range of the gearbox test, but again only power losses have to be fed to the system under static operation. Both variants have in common that differing from the WT field condition of the gearbox the LSS is not directly driven. The reason for this
- 462. Validation, verification, and full- scale testing 433 is that assuming the same rated power a slow- speed drive with high torque is consid- erably more expensive than a high- speed drive with little torque. This circumstance is also the reason for incorporating gearboxes into WT drivetrains. The back- to- back setup comes unfortunately with a few drawbacks. The extra gearbox of the test bench introduces, next to some proper dynamics, additional backlash and flexibility in the control loop. This reduces the dynamic range of the test rig as compared to a direct drive system. Furthermore, the additional gearbox adds vibration and noise, which might negatively affect the behavior of the test gear- box and hinder a straightforward interpretation of test results. In the described test setup, only the torsional DOF is studied. Any potential influence of parasitic loads Figure 9.13 Schematic illustration of mechanical back- to- back gearbox test bench (© Fraunhofer IWES/Kyling) Figure 9.14 14 MW mechanical back- to- back test bench (© Winergy)
- 463. 434 Wind turbine system design (bending moments, shear forces, or thrust) as they might occur during WT field operation is neglected. In most of the rigs, the torque is applied under quasi- static circumstances, dynamic effects as they occur, e.g., during fault ride through events are neglected. A good reaction to the aforementioned drawbacks is to focus the gearbox additionally in system tests as part of the drivetrain on system test benches such as the nacelle test bench at Fraunhofer IWES. In the case of the Fraunhofer IWT 7.5 generic WT, the rated torque is in the range of 7.2 MNm at a rated speed of 10 rpm. As this generic WT is direct driven, there is no need for any gearbox. Gearbox manufacturers use end- of- line test benches as a means of qual- ity control. For end- of- line tests, the scope of testing and thus its duration are reduced to the necessary. Complexity and costs of the involved test benches are reduced accordingly. Figure 9.15 Schematic illustration of electrical back- to- back gearbox test bench (© Fraunhofer IWES/Kyling) Figure 9.16 Winergy 17 MW test bench in Voerde (© Winergy)
- 464. Validation, verification, and full- scale testing 435 9.7.2 Brake system The rotor brake system of a multi- megawatt WT is not designed to stop the rotor from normal operation. This would require brake systems with very high braking torques that are compared to state- of- the- art solutions overdesigned. The state- of- the- art solution consists of two aspects: first, the pitch systems reduces the driving torque and decelerates the rotor, and then second, the rotor brake is activated to bring the system to a full halt. As a key part of the safety chain of a WT, the primary brake system must be designed redundantly. In the case of the multi- megawatt, WT is achieved by using for each blade an independent pitch system. Small WTs (1 MW) use a collective pitch system and thus require a second brake system that is able to stop the turbine in case of an emergency. This requirement is usually solved by designing a rotor brake as shown in Figure 9.17 that provides the needed braking torque. For such a small WT brake system, an exemplary validation activity was con- ducted at Fraunhofer IWES. The purpose of that project was to validate the under- lying design and control model as well as to ensure the desired functionality. The test setup is depicted in Figure 9.17 and consists of a driving engine, a coupling equipped with a calibrated torque sensor, and the small WT brake system (speci- men). In general, two types of tests were conducted with different parameters (e.g. used brake pad material and speed). In the first run the specimen was accelerated to the desired speed, the driving engine was deactivated and the brake was activated. Figure 9.17 Test setup of a small WT brake system (© Fraunhofer IWES/Kyling)
- 465. 436 Wind turbine system design The acting brake torque as well as the time to reach a standstill was measured (see Figure 9.18) and used to validate the model. In the second type of test, the driving engine was controlled to maintain the speed, while the brake was activated for a determined period. Figure 9.19 depicts a test run in which the brake torque decreased significantly over time. Such a result gives good proof of the necessity and value of validation. The best fit of brake pads for the needed application can be chosen, and it can be assured that the braking torque is sufficient to halt the rotor in the required period. 9.8 Component testing On the component level of the V- Model (as shown in Figure 9.4), a practical very relevant validation and verification activity is the full- scale blade testing (FST). In addition, validation examples of the main shaft and pitch bearings will be presented. 9.8.1 Main shaft The main shaft supports the WT rotor. As a major structural component, it transmits on the one hand the parasitic bending moments, shear forces, and thrust via an appro- priate bearing configuration into the main frame. On the other hand, it transmits the Figure 9.18 Measured braking torque during transient brake test (© Fraunhofer IWES/Kyling)
- 466. Validation, verification, and full- scale testing 437 torque into a gearbox or in the case of a direct driven WT, as the Fraunhofer IWT 7.5 generic, directly into the generator. The main shaft was focused on two research projects of Fraunhofer IWES [19, 20]. The major aims were to validate existing models used for main shaft design and during the certification process and compare component lifetime acquired based on material with full- scale tests. Consequently, with a validated design concept, lightweight design toward material savings can be driven. The purpose in this case is fatigue testing (as described in section 9.3). This leads directly to the question of how to fatigue test the main shaft. When analyzing all design load cases and their time distribution during a WT lifetime, it can be found (see Table 9.3) that the bend- ing loads are dominant for the fatigue driven damage to the main shaft. Table 9.3 Load DOF and corresponding WT life time damage sum Load Damage sum [%] FXR (vertical shear force) 0.0 FYR (lateral shear force) 0.0 FZR (thrust) 0.0 MXR (bending moment) 49.7 MYR (bending moment) 50.3 MZR (torque) 0.0 Figure 9.19 Plot of a test run under driving torque (© Fraunhofer IWES/ Kyling)
- 467. 438 Wind turbine system design In Figure 9.20, a general stepwise process to develop a test setup is illustrated in the example of the main shaft as described above. It starts with clarifying the purpose of the test and defining accordingly the major aims. During the simplification step, all real operation conditions of the speci- men are checked for their relevance for the defined aims. The best trade- off between realistic and failure mode relevant testing and corresponding effort and costs is to be aligned. In the next step, the detailed design of the test setup (i.e. specimen and test bench) is developed and carried out. The process is concluded by defining a detailed test plan that summarizes, e.g., test conditions, the procedure for the test conduction and states success criteria. The test bench developed for the main shaft fatigue testing is shown in Figure 9.21. The damage- relevant bending moment load cycles are achieved by rotating the specimen while applying a quasi- static shear force on the load lever. The drive in this setup lets the specimen rotate at the desired speed up to 4 times faster than during normal turbine operation. The necessary power results majorly from friction losses. In field operation, the bearings are typically grease- lubricated. In this application, they are oil- lubricated, which enables efficient heat transfer and thus higher operational speeds. For the test purpose, the gearbox is only of interest for its support function of the main shaft and is consequently substituted by the gearbox dummy. In the case of the Fraunhofer IWT 7.5, the damage equivalent bending load assuming 2 million load cycles is roughly 10 MNm. Figure 9.22 depicts the cross- sectional view of a fatigue- tested cast- iron main shaft that reached the end of its lifetime. A set of fatigue cracks started from the outside to grow toward the inner side of the main shaft. Once the remaining cross- sectional area was too weak to withstand the test load an overload breakage split the main shaft into two pieces. Figure 9.20 Test development process (© Fraunhofer IWES/Kyling)
- 468. Validation, verification, and full- scale testing 439 9.8.2 Pitch bearing Next to the other parts of a pitch system, the pitch bearing is probably the most criti- cal part. A failure of a bearing means that it is not possible to turn the blade anymore. Hence, it is also a risk for the turbine in case of an emergency. For an emergency stop, Figure 9.21 Main shaft test bench (© Fraunhofer IWES/Kyling) Figure 9.22 Global isolated view of fracture surface of cast- iron main shaft (left) detail view of fatigue crack growth marks (right) (© Fraunhofer IWES/Kyling)
- 469. 440 Wind turbine system design at least two of the three blades must turn into a feathering position (cf. Chapter 3). In fact, a failure of the pitch actuator has the same risk as a damaged pitch bear- ing, but the bearing has another important function. It connects the blade and hub. Besides the turbine’s dysfunction, a failure of the pitch bearing also comes with the risk that the blade is falling down. Both aspects underline the importance of pitch bearing and its validation. A pitch bearing has different failure modes, which can be distinguished based on their origin. For the raceways, it is possible to differentiate between surface-