Roessler, Hafner - Modelling and Simulation in Industrial Applications: Applying energy optimization to large scale systems

Modelling and Simulation in Industrial
Applications
Applying Energy Optimization to
Large Scale Systems
DI Matthias Rößler
DI Irene Hafner
dwh simulation services

Current Situation
• Energy Consumption Austria
• Challenges regarding energy efficiency
– no holistic view on production process with respect to resource
consumption
– highly complex matter
– lack of expertise on energy systems in enterprises
– lack of knowledge of possibilities
2.1%
28.7%
32.9%
12.4%
23.9%
Agriculture
Manufacturing
Transport
Service Sector
Private Households
Statistik Austria, 2011
2

Energy Optimization -
Motivation
• Energy consumption in production industry
approx. 40% of total energy consumption
in industrialized nations
• Potential for reduction:
30-65% (depending on sector)
 Increase of energy costs
 Tougher regulations
 Rising ecological awareness
Importance of energy efficiency in the industrial sector
3

Application Projects
Interdisziplinäre Forschung
zur Energieoptimierung in
Fertigungsbetrieben
(interdisciplinary research
for energy optimization in
production facilities)
Balanced Manufactoring
4

5
Interdisziplinäre Forschung
zur Energieoptimierung
in Fertigungsbetrieben
(interdisciplinary research
for energy optimization
in production facilities)

INFO - Approach
Analysis
Modelling
Integrated Simulation
... aiming energy optimization in production facilities
... to achieve economic and ecologic goals
Optimization
Fields of Optimization
Energy
System
Production
System
MachineProcess Building
6

INFO
Partial Model: Machines
Analysis
Modelling
Optimization
Energy
System
Production
System
7

INFO
Partial Model: Machines
• machines
– machine tools
– laser cutters
– ovens
– compressors
• production scenario
– modelling a load profile via SAP data of a
representative production week
• considered energy flows (3 thermal zones)
– electric
– diffuse heat emission
– recoverable heat
8

INFO
Machine Tools
initial approach
• technological
• focus on
modelling
individual tasks
of machine tools
• what is possible
from the
technological
point of view?
Step Back and focus
on 15 minute average
values
• approach from
the opposite
direction
• which values
are required to
generate the
desired output?
data based machine
model
• model based on
measured data
• easily
parameterized
• modular built,
hence flexible
• available
production data
from respective
enterprise are
essential
9
measurements
•AMS (Stiwa): Hermle
C40, C32
•Anger Machining: HCX
BA 1035, HCX BA 1110
•CNC Profi (DMG): DMU
65
•EMCO: Maxxturn 45
•Krause Mauser: Invers
BAZ for Daimler
•Hoerbiger: Stama/MC
334 Twin

initial approach
• technological
• focus on
modelling
individual tasks
of machine tools
from the
technological
point of view?
Step Back and focus
values
• approach from
the opposite
direction
• which values
are required to
generate the
desired output?
data based machine
model
• model based on
measured data
• easily
parameterized
• modular built,
hence flexible
• available
production data
from respective
enterprise are
essential
measurements
C40, C32
BA 1035, HCX BA 1110
65
BAZ for Daimler
334 Twin
INFO
Machine Tools
3,000
5,000
7,000
9,000
11,000
13,000
468 470 472 474 476 478 480 482 484
El.Power[W]
Zeit [s]
Leistung ohne
Werkstück
Leistung mit
Werkstück
slowing-down process
of the approaching
cutting unit
approaching the workpiece
without tool usage
tool usage (drilling)
tool usage (finish drilling) drill move out and approach
tothe next drilling
short move out of the drill
(ejection of chippings)
Power without
workpiece
Power with
workpiece
10

INFO
Machine Tools
initial approach
• technological
• focus on
modelling
individual tasks
of machine tools
from the
technological
point of view?
Step Back and focus
values
• approach from
the opposite
direction
• which values
are required to
generate the
desired output?
data based machine
model
• model based on
measured data
• easily
parameterized
• modular built,
hence flexible
• available
production data
from respective
enterprise are
essential
measurements
C40, C32
BA 1035, HCX BA 1110
65
BAZ for Daimler
334 Twin
location
building
production chain
machine
process
T
O
P
D
O
W
N
B
O
T
T
O
M
U
P
compressor
model
machine tool model
physical background and
measurement
oven and laser
model
11
0
50
100
150
200
Mon Tue Wed Thu Fri Sat Sun Mon
elektrischeLeistunginkW
Maschinenpark Shedhalle Kompressorencompressorsmachines
electricpowerinkW

INFO
Machine Tools
initial approach
• technological
• focus on
modelling
individual tasks
of machine tools
from the
technological
point of view?
Step Back and focus
values
• approach from
the opposite
direction
• which values
are required to
generate the
desired output?
data based machine
model
• model based on
measured data
• easily
parameterized
• modular built,
hence flexible
• available
production data
from respective
enterprise are
essential
measurements
C40, C32
BA 1035, HCX BA 1110
65
BAZ for Daimler
334 Twin
0
20
40
60
80
100
120
140
160
180
200
Mon Tue Wed Thu Fri Sat Sun Mon
electricpowerinkW
Messung Modell
• 25 machine tools in the
production hall
• comparison
measurement/model
measurement model 12

INFO
Partial Model: Building
Analysis
Modelling
Optimization
Energy
System
Production
System
13

INFO – Building III
Output
daylight
dependent
control of
- artificial light
- shading
heat output/
cooling capacity
zone
temperature
Building ModelInput
weather data
waste heat
people/
devices
waste heat
machines
14

INFO
Partial Model: Energy System
Analysis
Modelling
Optimization
Energy
System
Production
System
15

INFO
Energy System I
Output
required
power
•heat
•electricity
•others
CO2
emissions
Energy System ModelInput
weather data
heat output/
cooling capacity
zone
temperatures
waste heat
machines
recoverable
CO2
16

INFO
Energy System II
0
500
1000
1500
10.1. 8.4. 1.8. 10.1. 8.4. 1.8. 10.1. 8.4. 1.8.
Location Cairo
low utilization
high utilization
medium utilization
0
500
1000
1500
10.1. 8.4. 1.8. 10.1. 8.4. 1.8.
10.1. 8.4.
1.8.
EnergyDemand[kWh]
Location Vienna
low utilization
medium utilization
high utilization
0
500
1000
1500
2000
2500
10.1. 8.4. 1.8. 10.1. 8.4. 1.8. 10.1. 8.4. 1.8.
Location Moscow
high utilization
medium utilization
low utilization
Scenario 1:
oil heating
compression chiller
Scenario 2:
Heat pump
absorption chiller with heat
recovery
Scenario 2 heat (cooling)
Scenario 2 electricity (heating)
Scenario 1 heat (heating)
Scenario 1 electricity (cooling)
17

INFO
Overall Simulation
Analysis
Modelling
Optimization
Energy
System
Production
System
18

INFO: Specific Aims
Optimization based on simulation
• increase of energy efficiency
• inclusion of new carriers of energy
• manual comparison of specific scenarios
• no automatic optimization
Formalization of the model structure – reference model
• independent of specific implementation and
simulation environment
component based black-box approach, modularization
• illustration of dynamic dependencies and feedbacks
connection of variables and interface definition
• integration of planning and simulation
19

INFO: Approach
Theoretical Modelling Technical Integration
Goal: integrated dynamic simulation
• overall system not implementable in one simulator
– different modelling approaches
– gravely differing dynamics (time constants)
• several fields of expertise
• dynamic coupling
Coupling of well-established simulation tools
Co-Simulation
20

INFO: Overall Simulation
coupling
framework
economic and
ecologic
evaluation
static input
 temperature
 solar radiation
 waste heat of people and
devices
 electricity consumption
of devices
 energy
consumption
 CO2 emission
machine model
building model
energy system
model
 e.g. waste heat
reuseable/diffuse
 electricity
consumption of
machines
 weather data
 diffuse waste heat
machines
 waste heat of people and
devices
 room temperatures
 air change rate
 heating and cooling
demands
 room temperatures
 air change rate
 heating and cooling
demands
 reuseable waste heat of
machines
 energy consumption
 CO2 emission
21

INFO – Co-Simulation
• cooperative simulation with control of data exchange via
framework
• individual simulators calculate system parts independently
– different solver algorithms
– different time steps
• data exchange between simulators via framework at
previously defined points in time
• different ways of data exchange
– Strong Coupling: iterative data exchange in every step
– Loose Coupling: extrapolation between synchronization references required
…
22

Loose Coupling (Jacobi Type)
System 1
System 2
Jacobi Type:
Model Problem:
Extrapolation of y1 and y2
23
System 1:
System 2:

Loose Coupling (Gauß-Seidl Type)
Gauß-Seidl Type:
System 1
System 2
Extrapolation of y2
Interpolation of y1
24
Model Problem: System 1:
System 2:

Consistency
• consistency error measures the error of the
numeric method in one step
• consistency error in loose coupling co-simulation:
• ODE solver of first order: consistency order
maintained
• solver of higher order: lower consistency order
… consistency error of the method in a mono-simulation
… Lipschitz constant of the “right side“ from
… coefficient from the second characteristic polynomial
25

BCVTB I
• Building Controls Virtual Test Bed
• open-source software platform (developed at Lawrence
Berkeley National Laboratory, University of California)
• middleware for run-time coupling of different simulation
environments
• software components (clients) are executed in parallel
26

BCVTB II
• communication via BSD sockets and network protocol
(inter-process communication)
• Loose Coupling (Jacobi Type) with equidistant time steps
• in INFO: combination of
– MATLAB: data-based models
– EnergyPlus: thermal building simulation
– Dymola: component-based modelling of technical equipment
27

Simulation control framework
BCVTB
Machine Simulation
MATLAB/Excel
Building Simulation
EnergyPlus
Energy System Simulation
Dymola
Post - Processing
MATLAB
28

INFO - Results
• scenarios for different HVAC systems –
performance prediction
• energy performance certificate
• lifecycle cost-benefit analysis
• roadmap for energy efficient production
Energy Efficient
Production
29

Software Tool-Chain, embedded in operational automation
systems:
 BaMa-Optimization: optimization of line operation
regarding the goals energy, time, costs, quality
 optimized operational management strategy
 identification of main potential savings
 BaMa-Prediction: prediction of energy demands of the
whole facility based on production plan, operational
management and prediction data
 BaMa-Monitoring: aggregation and visualisation of
resource demands
BaMa - Goals
31

BaMa - Approach
• Modularisation of the system „production facility“
 partitioning according to energetic reasons
 separation into manageable parts
 systematically approaching the high system complexity
 modular approach allows flexibility
• consistent terminus: „cube“
32

BaMa – Cubes I
 Cubes are clearly confined units  basic modules for system
analysis
 integration of different points of view and system areas
(machines, building services, building, logistics) in one system
 general Cube specification
 Cubes bundle information and resource flows (energy, material,
costs, etc.) within identical balance borders
 transparency und analysis of energy flows
 new modular technology allows optimal connection of the real and
the virtual system
real production facility
machine
building
services
building
logistics
energy, material and
information flow
33

modelling hitherto modelling with Cube approach
 equal system boundaries
 modular, expandable and easy to apply to special areas
in practise
 concurrent consideration of energy flows and material
flows in one system
 overlapping/non-equal system boundaries, hence
redundancies
 different models for energy flow, material flow and
costs
 concurrent consideration of flows not possible
BaMa – Cubes II
34
Mass balance
Energy balance
Time balance
Cost balance
production
machine
production
machine
air
compressor
waste disposal
production
process
Cube
production
machine
Cube
production
machine
Cube
air
compressor
Cube
waste disposal
Mass balance
Energy balance
Time balance
Cost balance
production
process
information and
resource flow

BaMa
Cubes: Interfaces
 cubes have uniformly defined interfaces
 flexibility, modularising, exchangeability
 connections and interactions
between cubes
 material flow
 energy flow
 information flow
 diffuse waste heat, recoverable heat
 CO2 share
 balance equations at cube borders
monitoringdata
controlaction
energyflow
energyflow
 work piece,
baking goods, etc.
 discretized
 footprint (costs, CO2)
material flow material flow
parameters:
 dimensions
 power characteristics
 efficiency
 etc.
 production plan
 operating mode
 control signal
 etc.
 energy demand
 operational state
 etc.
 power: electric, thermal, etc.
 exergy measure
 CO2 share
 work piece,
baking goods, etc.
 updated footprint
35

BaMa Toolchain
• Cubes also help with the description in the simulation environment
• Cubes have a virtual „counterpart“ - based on simulation models and
measured data
• Cube view supports reusability in implementation
control
status
User Interface
BaMa - Virtual Cubes
real production facility
machine
building
services
building
logistics
energy, material
and information flow
virtual system
Virtual cube
machine
Virtual cube
building
services
Virtual cube
building
Virtual cube
logistics
information flow
36

BaMa – Cube Classes
„Cube“
machine,
production process
value-adding
non-value-
adding
building
building hull
thermal zone
energy system,
building services
energy converter
energy storage
energy networks
logistics
transport system
handling system
storage system
38

BaMa – DEV&DESS I
39
products
“flows”
products & “flows”

BaMa – DEV&DESS II
Formalism
• building on systems-theoretical basics
• allows the description of hierarchically structured systems
• DEVS: description of purely event based (and hence time-discrete) systems
• DESS: description of causal continuous systems
• DEV&DESS: suitable for hybrid systems supporting continuous as well as discrete
changes in system states
Implementation
• event scheduling required
• zero-crossing detection for(real) State Events desired
• numerical solving of differential equations can be realised in the model
• data models can be included
40

BaMa – DEV&DESS III
Cube
guarantees
 consistency in the cube description
 technical feasibility
 requirements for sustainable implementation
 scientific acceptance
41

Real Cube
Model
(verbal, conceptual, physical, mathematical)
Formal Cube Description
DEV&DESS Formulation of the Cube
DEV&DESS Implementation of the Cube
= virtual Cube
BaMa – Cube Workflow I
42

Real Cube
BaMa – Cube Workflow II
43

Model
(verbal, conceptual, physical, mathematical)
BaMa – Cube Workflow III
44

BaMa – Cube Workflow IV
45

...
Bedarf el. Leistung (PelB)
Anforderung Entität (Ereq)
Elektrische Leistung (Pel)
Entität (E)
Entität(E)
Abfall (EA)
Umgebungstemperatur (Tu)
Nicht nutzbare Abwärme
(QAW)
Nutzbare Abwärme (Qrec)
Produktionsplan (Pplan)
Heizleistung (PH)
Haltedauer (tB)
Solltemperatur (Tsoll)
Zweipunktregler Hysterese (H)
Volumen Ofen (V)
Wärmedurchgang Ofenwand (UA)
Wärmekapazität Luft (cpL)
Dichte Luft (rhoL)
Abwärmenutzung (eta)
Abfallmenge (alpha)
Parameter:
Zustandsgrößen:
Betriebszustand (p): standby,
aufheizen, warten, halten
Heizzustand (h): on, off
Masse der Entität im Ofen (m)
Wärmekap. der Entität im Ofen (cp)
Temperatur im Ofen (T)
BaMa – Cube Workflow V
46

Name Kürzel Einheit Datentyp Wertebereich
Entität E Entität
Attribut: Masse E.m kg Skalar > 0
Attribut: Temperatur E.T K Skalar > 0
Attribut: Wärmekap. E.cp J/(kg*K) Skalar > 0
Name Kürzel Einheit Datentyp Wertebereich
Entität E Entität
Attribut: Masse E.m kg Skalar > 0
Attribut: Temperatur E.T K Skalar > 0
Attribut: Wärmekap. E.cp J/(kg*K) Skalar > 0
Abfall EA Entität
Attribut: Masse EA.m kg Skalar > 0
Attribut: Temperatur EA.T K Skalar > 0
Attribut: Wärmekap. EA.cp J/(kg*K) Skalar > 0
Materialflüsse
Eingänge:
Ausgänge:
BaMa – Cube Workflow VI
47

Ausgang
• wird nur bei Beendigung des Betriebszustands "halten"
ausgegeben
• Unterscheidung: Entstehung von Abfall
BaMa
Cube Workflow VII
48

DEV&DESS-Implementation of the Cube
= virtual Cube
BaMa
Cube Workflow VIII
49

BaMa - Optimization
• scenario: production plans, operational conditions
(constraints, initial solution)
• optimization selects control variables (production plan)
• target function: evaluating the current simulation results
for the chosen parameters
• selection of new parameters for next simulation run
• iteration to find the most suitable production plan for the
respective scenario within a given time span
Scenario
control variables
optimization
target function
parameters feedback
modified parameters
Simulation
50

BaMa - Optimization
Target Function
• weighing of different criteria:
– on-time delivery, storage
– total energy cost
– throughput time
– idle period
– …
delayed delivery, storage costs (on-time delivery)
total throughput time
total energy: costs – CO2
total number: DESIRED - ACTUAL
lot throughput
weights (adjustable)
51

production
BaMa - Carbon Footprint of
Products (CFP)
 evaluation of environmental sustainability of a product throughout its whole life
cycle
 comparison to other products
 identification of pollution during life cycle
 reduction of pollutant emissions
CO2-footprint of a product
resources utilization disposal
52

CFP from
heating/cooling of
storerooms
BaMa - CFP Method
exemplary tasks at an up-to-date CFP calculation
consideration of
stand-by and setup
times
energy for building
services
energy input of
machines apportioned
to machines
energy for
transport systems
ventilation,
illumination,… of
the building
53

BaMa - Results
• modular approach for high flexibility
• carbon footprint of products
• automated optimization of production plans
• aims: effecitivity regarding
– energy
– costs
– resources
– CFP
• proof of concept with six use cases in several
production facilities from different fields
54

Conclusion
• energy efficiency: increasing need for simulation
based solutions
• two different approaches
– co-simulation
(quasi) arbitrary amount of participating simulators
most suitable software for every partial system
individual solvers/time steps for partial systems
loss of accuracy
– DEV&DESS formalism
monolithic approach (one simulator)
no accuracy loss
need to formalize (adapt model description)
55

Roessler, Hafner - Modelling and Simulation in Industrial Applications: Applying energy optimization to large scale systems

More Related Content

What's hot (6)

Similar to Roessler, Hafner - Modelling and Simulation in Industrial Applications: Applying energy optimization to large scale systems (20)

More from Vienna Data Science Group (20)

Recently uploaded (20)

Roessler, Hafner - Modelling and Simulation in Industrial Applications: Applying energy optimization to large scale systems

Editor's Notes