CALIBRATION & VALIDATION PHILOSOPHY AND PROCEDURES FOR TUNEABLE DIODE LASER ANALYZERS IN PROCESS APPLICATIONS

CALIBRATION & VALIDATION PHILOSOPHY AND
PROCEDURES FOR TUNEABLE DIODE LASER
ANALYZERS IN PROCESS APPLICATIONS

Dr. Janardhan Madabushi David Fahle
Servomex Company Inc Servomex Company Inc
525 Julie Rivers Rd. Suite 185 525 Julie Rivers Rd. Suite 185
Sugar Land, TX 77598, USA Sugar Land, TX 77598, USA

Dr. Christian Heinlein
NEO Monitors AS
Solheimveien 62A, P.O.Box 384
N-1471 Loerenskog, Norway

KEYWORDS

Tuneable Diode Laser, Optical Absorption Spectroscopy, Calibration, Validation

ABSTRACT

Gas analyzers based on Tuneable Diode Laser (TDL) Absorption Spectroscopy are now
widely accepted in the industry as process analyzers because of their reliability and low
maintenance. TDL analyzers offer a non-contact, optical measurement without moving parts
with most manufacturers recommending a calibration frequency of up to one year. This
extended calibration interval has raised skepticism among industry users because it is
contrary to their experience with traditional analyzers, which require more frequent
calibration, supported by most maintenance protocols which call for frequent validation and
calibration. This paper will discuss the calibration and validation philosophy for TDL
analyzers in process applications.

INTRODUCTION

Gas analyzers for industrial applications must have high reliability and availability and
require little maintenance. Analyzers based on tunable diode laser absorption spectroscopy
(TDLAS) meet these requirements, because they offer a non-contact, optical measurement

Session 8.2: – Page 1 –
© Copyright 2010, International Society of Automation. All rights reserved.
ISA 55th Analysis Division Symposium 2010, New Orleans, LA.

without the need for consumables or moving parts. Because of this, TDLAS analyzers can be
applied for true in-situ (cross stack or pipe) measurements for many applications as they do
not require regular access.

TDLAS became feasible when reliable, tunable laser diodes became available as a direct
spin-off from the development of light sources for optical fiber communication. These lasers
have wavelengths in the near infrared (typically from 1300 nm to 1700 nm, but more recently
out to 2700 nm), where many of the process critical gases have (overtone) absorption lines
sufficiently strong to allow measurements of gas concentrations in the low ppm range.

The TDLAS technique has become widely accepted in the chemical and petrochemical
industry by providing in-process measurements of gases such as oxygen (O2), carbon
monoxide (CO) and carbon dioxide (CO2) in hazardous and/or poisonous atmospheres. In
addition to process applications, this technology is also widely implemented for emission
control of pollutants such as hydrogen fluoride (HF), hydrogen chloride (HCl), and ammonia
(NH3) from aluminum smelters, power plants or waste incinerators. This makes it a very
viable tool for use in Continuous Emission Monitoring Systems (CEMS).

The non-contact measurement approach on the TDLAS lends to increased stability of the
measurement, gradually enhanced by improvements of electronic components. Accordingly,
most manufactures of the TDLAS based analyzers suggest a calibration frequency of 6 to 12
months, which greatly contrast to the regular calibration requirements of traditional analyzer
techniques. This paper includes a brief description of measurement principle, a typical TDL
system (Instrumentation), measurement techniques and the calibration and validation
philosophy for TDL analyzers in process applications.

MEASUREMENT PRINCIPLE

Optical absorption spectroscopy for quantification of gas concentration has been used in
industrial applications for more than 60 years. The technique is based on the Beer-Lambert
law:

T = exp(-Sg(f)NL) (1)

T is the transmittance,
S is the absorption line strength,
g(f) is the line shape function,
N is the concentration of absorbing gas molecules,
L is the optical path length.

Concentration N can be calculated by measuring T and knowing S, g(f), and L. The
conventional way of obtaining quantitative information is to compare the measured response
to that generated by a calibration gas with known content of the molecule in question. Unlike
conventional ultra violet (UV) or infrared (IR) spectrographic instruments, laser-based
monitors employ the measurement principle known as ‘single line spectroscopy’, where a
single gas absorption line is chosen in the NIR spectral range and scanned by the single-mode
diode laser. Many detection techniques have been developed for TDLAS and an overview of


some of these can be found in [1 – 3]. The dual modulation (2f) technique is, in our opinion,
the best candidate for an industrial monitor.

INSTRUMENTATION
A typical TDLAS system will consist of the following units:

• Laser module with temperature controlled diode laser, driving circuit and collimating lens
plus window to isolate from the process
• Detector module with pre-amplification circuit, photo diode, and focusing lens plus
window to isolate from the process
• Main electronics with central processing unit (CPU) for signal treatment and calculation
of gas concentration
• Interface electronics with analog and digital input and output connections

Figure 1 shows the mechanical layout of such a monitor installed on a stack.

Transmitter Flanges Receiver
Purge gas Purg e gas
Diode inlet inlet
Laser
Focusing lens Detector

Collimating
lens
Instrument Process
window
gas
Loop cable

FIGURE 1. SCHEMATIC DRAWING OF A TDLAS CROSS STACK
INSTALLATION

MEASUREMENT TECHNIQUES

TDLAS is an emerging technology for process measurement and its implementation for
process measurement has been elaborately described in various publications and
presentations [4 - 6]. To summarize, measurements can be performed by in-situ, bypass loop
or extractive installation, but in all cases the concept remains the same: a laser transmitter and

receiver (detector) unit are both located or fiber optically coupled at the measurement point
with mounting flanges and intermediate purge and alignment units. A continuous flow of
purge gas is utilized to prevent dust from fouling the optical windows, and in case of O2 and
H2O measurements, to displace ambient concentrations of these components in flange nozzles
and analyzer optics. The in-situ approach is illustrated in Figure 1 with installation across
pipes, stacks or ducts, with typical path lengths ranging from 0.5 – 20 meters. This is the
most preferred approach because it eliminates the need for sample conditioning system and is
therefore ideal for fast and real time measurement.

A bypass mode with a slip stream (Figure 2) is employed to increase the optical path length
(OPL), when the process pipe does not have sufficient OPL to obtain the desired
measurement range. An extractive set up (Figure 3) is typically implemented when sample
conditioning is not conducive for a direct in-situ measurement and some sample pre-
conditioning is required to address pressure, temperature or dew point issues.

FIGURE 2. SCHEMATIC OF A TDLAS SYSTEM IN BYPASS LOOP SETUP

CALIBRATION AND VALIDATION
The absorption signal peak shape, formed by amplitude and width, is influenced by a range of
factors including analyte concentration, optical path length (OPL), pressure, temperature and
variation in stream composition created by background gases. The amplitude is influenced
by analyte concentration and OPL, while there are three contributors to the peak width:
natural line broadening, Doppler broadening and collision broadening.

Collision broadening of the absorption signal dominates industrial processes, depending as it
does on the collision frequency of the molecules. Collision broadening directly depends on
the process pressure, temperature and composition, so the accuracy and repeatability of the
measurement is always influenced by these parameters [7].


FIGURE 3. SCHEMATIC OF A TDLAS SYSTEM IN EXTRACTIVE SETUP

In industrial applications, line broadening effects must be taken into account in order to
achieve the required measurement accuracy. For process measurement, the effect of
pressure and temperature is compensated by live temperature and pressure input to the
analyzer. Broadening due to variation in gas composition is addressed by measuring the line
width of the absorption signal.

EXPERIMENTAL:

A typical layout of initial factory calibration is shown in Figure 4. Based on the submitted
process temperature and pressure data from the end user, a pressure dependence curve and a
temperature dependence curve are generated in a controlled environment, using a calibrated
standard gas and the above referenced set up. The pressure and temperature profile is
specific for each transition hence wavelength dependant. This procedure is an important
aspect of final testing, to compensate for pressure and temperature induced line width
broadening.

RESULTS AND DISCUSSIONS:

Figure 5 illustrates the relationship between pressure and detector signal. The signal
increases with density at low pressures, while the line broadening mechanism dominates at
pressures above ≈ 1 bar, hence the signal decreases at P > 1 bar. This profile is wavelength
dependant; as different absorption lines will have different pressure dependence profile. In
addition to this the location of maximum in Fig 5 could be moved towards higher pressure
by applying higher frequency modulation amplitude. In process application pressure can
vary from slightly negative to positive pressure greater than 1 Bar A. Hence each analyzer is
programmed with the respective pressure profile to enable live pressure compensation,
typically controlled by a 4-20mA pressure input. For processes operating at a fixed pressure,
live pressure compensation is not required but the magnitude of the process pressure is
manually entered in the instrument configuration for accurate measurement.

FIGURE 4. TYPICAL SCHEMATIC FOR INITIAL FACTORY CALIBRATION

25
Measured
20 Fitted curve
S n l (a . u its)
ig a rb n

15

10

5

0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Pressure (Bar absolute)

FIGURE 5. PRESSURE DEPENDENCE OF THE ABSORPTION SIGNAL

Figure 6 illustrates the relationship between temperature and detector signal. Temperature
affects absorption line width and absorption strength, and different absorption lines will have
different temperature dependence profile. Each analyzer is programmed with the respective
temperature profile to enable live temperature compensation, typically by a 4-20mA
temperature input.


20

Signal (arb. units) 15

10

5 Measurement
Fitted curve
0
0 50 100 150 200 250
Temperature (C)

FIGURE 6. TEMPERATURE DEPENDENCE OF THE ABSORPTION SIGNAL

Three gas types that occur frequently in industrial processes and have a large influence on the
line width are H2O, CO2 and hydrocarbons such as CH4. The concentration of H2O from
waste incinerators and boilers may typically vary from 10 to 25 vol%, but in some cases it
can be as high as 40 vol%. Figure 7 illustrates the influence of H2O concentration on the peak
amplitude of the 2f signal for HCl at 270 °C: It is clear that variations in gas composition can
influence the measurement and must be compensated for. The broadened peak will decrease
in amplitude and increase in width but overall area under the curve remains the same. Line
width broadening due to variation in gas composition can be compensated by measuring or
calculating the integrated area of the absorption peak.

1
Measured
Fitted curve
0.8
HCl Signal, normalised

0.6

0.4

0.2

0
0 5 10 15 20 25 30 35 40
H2O conc. (%vol.)

FIGURE 7. EFFECT OF H2O CONCENTRATION ON 2f PEAK SIGNAL OF HCl

Peak area is calculated using the peak height and the line width of the 2f signal. This
technique is also known as Line Width Compensation (LWC) [4].

40 %
80 %
100 %

10 %
H2S 2f-signal

0 1 2 3 4 5 6 7
Sample (corresponds to wavelength)
FIGURE 8. VARIATIONS IN 2F SIGNAL FOR H2S AT PERCENT
CONCENTRATIONS

Figure 8 illustrates self-induced band broadening, where large variation in the analyte
concentration can alter the line width of the absorption peak. Measuring the peak amplitude
alone will result in a large systemic error, but measuring the integral 2f signal using LWC
corrects the systemic error in the measurement (Figure 9).

FIELD APPLICATION – CALIBRATION:

Having established the factory calibration procedure for a TDLAS analyzer, the next logical
question is how these analyzers can be calibrated in the field. It is easy to calibrate analyzers
installed in the extractive mode (Figure 3). To calibrate the analyzer when installed in the
in-situ mode and by-pass modes the analyzer is removed from the process and a bench
calibration is performed using a calibration cell and span gas. The calibration is typically
performed at ambient temperature and pressure, which entails resetting the OPL to the length
of the calibration cell, and restoring temperature and pressure to ambient values. Since each
analyzer is programmed with its unique temperature and pressure profile it is feasible to
calibrate the analyzer at ambient temperature and pressure. This answers one of the most
commonly asked questions for calibration: namely, how can an ambient temperature
calibration of the analyzer still read accurately when installed in a process at 400 °C?


120

Measured concentration (%) 100

80

60

40

20 y = 1.0024x + 0.1203
R2 = 0.9996

0
0 20 40 60 80 100 120

H2S concentration in % set with mass flow controllers

FIGURE 9. LWC COMPENSATED DETECTOR RESPONSE FOR H2S

In some instances the analyzer cannot be easily removed from the process. In such instances
it may be possible to rescale the TDLAS response to the expected concentration, if there is
some legitimacy, based on historical data, for the expected reading. While this practice is not
generally recommended, this feature can be used with the acceptance of the risks, when the
analyzer is used mainly to trend the concentration of the analyte and an absolute
measurement is not necessary. Situations like this also bring up the need for in-situ
validation of the analyzer.

FIELD APPLICATION – IN-SITU VALIDATION:

Very few TDLAS analyzers are installed at grade level. Most in-situ installations are up on a
stack or at higher elevations in the plant. Thus it is not very practical to frequently remove
the analyzer from the process to verify its performance. The extended calibration
frequency does help in this regard but most plants have operating procedures that call for
routine validation of the analyzer’s measurement to verify performance. Subsequent
calibration may be initiated based on the results from this validation. The concept of
validation is also important in most CEMS applications [8].

Validation is performed either by comparison against a validated reference method or with
analyte spiking. This section will highlight validation of TDLAS by the analyte spiking; also
know as the bump test method, in process applications.

Once again it is easy to validate analyzers installed in the extractive mode by introducing the
spike gas at the sample inlet. For in-situ applications a different approach is taken. Typically


the spike gas is introduced into a fixed volume gas cell situated outside the process duct but
along the optical path as depicted in Figure 10.

FIGURE 10. TYPICAL OPTION FOR INTRODUCING SPAN GAS FOR
VALIDATION

The gas cell is a stainless steel chamber, typically about 10 cm long, with sealed windows on
both sides (axial), to contain the spike gas. During normal operation a spectroscopically inert
gas is present in the gas cell; i.e. if oxygen is being measured then the inert gas in nitrogen.
Likewise if NH3 is the measured component dry instrument grade air is used as the inert gas.
To validate, spike gas containing a significantly higher concentration of the analyte is
introduced to bump the analyzer response. The analyte concentration in the spike gas is
dictated by the process OPL, process measurement range and the OPL of the validation
chamber.

For example, consider the following situation where, process OPL is 1 m (~ 40”),
measurement is 0 – 6 % O2 and the chamber OPL is 0.01 m. The absorption signal from the
gas cell should always be stronger than the signal from the process in order to achieve a
reliable validation check. The path length of the gas cell is only 0.01 m and the maximum
analyte concentration that can be introduced into the chamber is 100%. The effective
concentration of the spike gas is calculated using equation (2).

CS = [S] x LC (2)
LP


CS is the Effective Spike Concentration,
[S] is the concentration of the spike gas,
LC is the length of the gas cell and
LP is the process pathlength.

Therefore 100% O2 in 0.01 m gas cell corresponds to 1 % O2 for 1 meter process pathlength.
Thus it is obvious that the validation check will not be reliable for this process. However, if
the process OPL is much shorter, 0.2 m (~ 8”), then 100% O2 on 0.01 m OPL corresponds to
5 % O2 for 0.2 meter process pathlength. In this situation it is viable to perform a reliable
validation check using 100% O2 as the spike gas.

In CEMS applications the measurement range is usually much lower. For example if we
consider measurement of ammonia slip in DENOX scrubbers, the typical range is 0 – 20 ppm
NH3 and the process OPL is 2 to 3 meters. To achieve an effective spike concentration of 10
ppm with the 10 cm chamber, the required spike gas concentration will be:

[S] = 10 ppm x 2.5 m = 2500 ppm. (3)
0.01 m

Thus for such applications, if needed, it is possible to use smaller chamber with higher spike
gas concentration. One approach is to use a small sealed glass cuvette containing high
concentration of NH3 as the validation chamber. To initiate validation, the cell is flipped into
the optical path using a servomotor. The cell has a very short OPL of 2 mm and typically
filled with 5% to 10% NH3. This yields an effective spike concentration of 4 ppm to 8 ppm
for a process OPL of 2.5 m. The entire validation assembly can be integrated inside the
receiver unit. This approach eliminates the requirement for a spike gas and inert gas, hence
more economical to the end user.

It is important to realize that the validation is performed along with the process measurement.
The TDLAS analyzer calculates the CS as follows:

CS= RS – RP (4)

RS is response from the spike gas
RP is the response from the process gas just before the initiation of the validation.

Typical equilibration time for the validation will depend on the methodology employed; the
fixed volume gas cell method will have a longer equilibration time as compared to the sealed
glass cuvette method. It is evident from equation (4) that the process should be fairly stable
during this entire duration, to obtain meaningful validation results.

Zero check is initiated when the gas cell is filled with the inert gas or when the sealed glass
cuvette is not in the optical path. It is performed either by subtracting the process reading just
before the initiation of the validation from the current process reading.

CZ = RPC - RP (5)

CZ is the zero check response
RPC is the current response from the process

TDLAS analyzers utilizing the 2f technique for signal processing is immune from baseline
offsets hence do not require a zero calibration [4]. Thus it is also possible to perform a zero
check by momentarily turning off the laser modulation.

This approach of in-situ validation is not yet endorsed by the USEPA for CEMS
measurement. However a preliminary performance specification for ammonia CEMS has
been drafted for further review [8].

CONCLUSIONS

Measurement by TDLAS can be performed by in-situ, bypass loop or extractive installation
hence they are gaining acceptance in the process industry. The in-situ approach is the most
cost effective mode of use because it eliminates the need for a sample conditioning system
and provides real time response. TDLAS analyzers are predominantly employed in this mode
to exploit this beneficial feature. While there are several publications on product development
and its application to the process industry, there is limited information on calibration and
validation of these analyzers in the field.

The non-contact measurement approach and improvements of electronic components lends to
increased stability of the measurement, prompting most manufactures of the TDLAS based
analyzers to suggest a calibration frequency of 6 to 12 months. This is a benefit because the
analyzer has to be dismounted from process for calibration. Built in temperature and pressure
compensation profiles conveniently permit TDLAS analyzers to be calibrated at ambient
temperature and pressure conditions regardless of the actual process measurement conditions.

Despite the extended calibration frequency most users prefer routine validation of the
analyzer’s measurement to verify performance. Fixed volume gas cells or sealed glass
cuvette containing the target gas can be used for validation. These apparatus are used
external to the process stream hence do not interfere with or alter the process chemistry. As a
consequence there is a practical limitation on the size of the gas cell. Feasibility of in-situ
validation is dictated by the process pathlength, process measurement range and the
pathlength of the validation gas cell. CEMS applications typically have measurements in the
ppm or very low % range hence they are an ideal candidate for this validation technique.

ACKNOWLEDGEMENTS

The authors wish to thank Chris Lawrenson (Coda Communications, Poole, UK.) for proof
reading the manuscript.

REFERENCES

[1] Schiff, H.I., Mackay, G.I., Bechara. J.: "The use of tuneable diode laser absorption
spectroscopy for atmospheric measurements", in Chemical Analysis Series, Vol. 17
(Wiley, New York 1994).


[2] Ye, J., Ma, L.S., Hall, J.L.: J. Opt. Soc. Am. B 15, 6 (1998).

[3] Feher, M., Martin, P.A.: Spectrochim. Acta, Part A 51, 1579 (1995).

[4] Linnerud, I., Kaspersen, P., and Jæger, T.: Gas monitoring in the process industry using
diode laser spectroscopy. Applied Physics B 67, 297 – 305 (1998).

[5] Amerov, A., Fiore, R., Langridge, S.: “ Process Gas Analyzer for the Measurement of
Water and Carbon Dioxide Concentrations”, ISA Analysis Division 54th Symposium
Proceedings, Instrumentation, Systems and Automation Society, Houston Texas 2009,
Session 7.2 pp1-14.
[6] Zhou, X., Liu, X., Feitisch, A.: “Advanced TDL Gas Analyzers for Petrochemical
Process Industries”, ISA Analysis Division 53rd Symposium Proceedings,
Instrumentation, Systems and Automation Society, Calgary, Canada 2008, Session 3,
Paper 3, pp1-12.
[7] Hobby, J., Gaskin, I., Kovacich, R., Lopez, M., Alizadeh, B.: “A Comparative Analysis
of the Errors in the Measurement of Oxygen Between in-situ and Extractive Methods”,
ISA Analysis Division 54th Symposium Proceedings, Instrumentation, Systems and
Automation Society, Houston Texas 2009, Session 2.1 pp1-14.
[8] www.epa.gov/ttn/emc/prelim/pps-001.pdf: PPS-001: Preliminary Performance
Specifications for Ammonia Continuous Emission Monitors (CEMs), pp1-18


CALIBRATION & VALIDATION PHILOSOPHY AND PROCEDURES FOR TUNEABLE DIODE LASER ANALYZERS IN PROCESS APPLICATIONS

More Related Content

Similar to CALIBRATION & VALIDATION PHILOSOPHY AND PROCEDURES FOR TUNEABLE DIODE LASER ANALYZERS IN PROCESS APPLICATIONS (20)

More from ISA Interchange (20)

Recently uploaded (20)

CALIBRATION & VALIDATION PHILOSOPHY AND PROCEDURES FOR TUNEABLE DIODE LASER ANALYZERS IN PROCESS APPLICATIONS