Performance Specification 15
1/14/2019
While we have taken steps to ensure the accuracy of this Internet version of the document, it is not the
official version. To see a complete version including any recent edits, visit: https://www. ecfr. sov/csi-
bin/EC FR?page=browse and search under Title 40, Protection of Environment
Performance Specification 15—Performance Specification for Extractive FTIR
Continuous Emissions Monitor Systems in Stationary Sources
1.0 Scope and Application
1.1 Analytes. This performance specification is applicable for measuring all hazardous air
pollutants (HAPs) which absorb in the infrared region and can be quantified using Fourier
Transform Infrared Spectroscopy (FTIR), as long as the performance criteria of this performance
specification are met. This specification is to be used for evaluating FTIR continuous emission
monitoring systems for measuring HAPs regulated under Title III of the 1990 Clean Air Act
Amendments. This specification also applies to the use of FTIR CEMs for measuring other
volatile organic or inorganic species.
1.2 Applicability. A source which can demonstrate that the extractive FTIR system meets the
criteria of this performance specification for each regulated pollutant may use the FTIR system
to continuously monitor for the regulated pollutants.
2.0 Summary of Performance Specification
For compound-specific sampling requirements refer to FTIR sampling methods (e.g., reference
1). For data reduction procedures and requirements refer to the EPA FTIR Protocol (reference 2),
hereafter referred to as the "FTIR Protocol." This specification describes sampling and analytical
procedures for quality assurance. The infrared spectrum of any absorbing compound provides a
distinct signature. The infrared spectrum of a mixture contains the superimposed spectra of each
mixture component. Thus, an FTIR CEM provides the capability to continuously measure
multiple components in a sample using a single analyzer. The number of compounds that can be
speciated in a single spectrum depends, in practice, on the specific compounds present and the
test conditions.
3.0 Definitions
For a list of definitions related to FTIR spectroscopy refer to Appendix A of the FTIR Protocol.
Unless otherwise specified, spectroscopic terms, symbols and equations in this performance
specification are taken from the FTIR Protocol or from documents cited in the Protocol.
Additional definitions are given below.
3.1 FTIR Continuous Emission Monitoring System (FTIR CEM).
3.1.1 FTIR System. Instrument to measure spectra in the mid-infrared spectral region (500 to
4000 cm-1). It contains an infrared source, interferometer, sample gas containment cell, infrared
detector, and computer. The interferometer consists of a beam splitter that divides the beam into
two paths, one path a fixed distance and the other a variable distance. The computer is equipped
with software to run the interferometer and store the raw digitized signal from the detector
1
-------
Performance Specification 15
1/14/2019
(interferogram). The software performs the mathematical conversion (the Fourier transform) of
the interferogram into a spectrum showing the frequency dependent sample absorbance. All
spectral data can be stored on computer media.
3.1.2 Gas Cell. A gas containment cell that can be evacuated. It contains the sample as the
infrared beam passes from the interferometer, through the sample, and to the detector. The gas
cell may have multi-pass mirrors depending on the required detection limit(s) for the application.
3.1.3 Sampling System. Equipment used to extract sample from the test location and transport the
gas to the FTIR analyzer. Sampling system components include probe, heated line, heated non-
reactive pump, gas distribution manifold and valves, flow measurement devices and any sample
conditioning systems.
3.2 Reference CEM. An FTIR CEM, with sampling system, that can be used for comparison
measurements.
3.3 Infrared Band (also Absorbance Band or Band). Collection of lines arising from rotational
transitions superimposed on a vibrational transition. An infrared absorbance band is analyzed to
determine the analyte concentration.
3.4 Sample Analysis. Interpreting infrared band shapes, frequencies, and intensities to obtain
sample component concentrations. This is usually performed by a software routine using a
classical least squares (els), partial least squares (pis), or K- or P- matrix method.
3.5 (Target) Analyte. A compound whose measurement is required, usually to some established
limit of detection and analytical uncertainty.
3.6 Interferant. A compound in the sample matrix whose infrared spectrum overlaps at least part
of an analyte spectrum complicating the analyte measurement. The interferant may not prevent
the analyte measurement, but could increase the analytical uncertainty in the measured
concentration. Reference spectra of interferants are used to distinguish the interferant bands from
the analyte bands. An interferant for one analyte may not be an interferant for other analytes.
3.7 Reference Spectrum. Infrared spectra of an analyte, or interferant, prepared under controlled,
documented, and reproducible laboratory conditions (see section 4.6 of the FTIR Protocol). A
suitable library of reference spectra can be used to measure target analytes in gas samples.
3.8 Calibration Spectrum. Infrared spectrum of a compound suitable for characterizing the FTIR
instrument configuration (Section 4.5 in the FTIR Protocol).
3.9 One hundred percent line. A double beam transmittance spectrum obtained by combining
two successive background single beam spectra. Ideally, this line is equal to 100 percent
transmittance (or zero absorbance) at every point in the spectrum. The zero absorbance line is
used to measure the RMS noise of the system.
2
-------
Performance Specification 15
1/14/2019
3.10 Background Deviation. Any deviation (from 100 percent) in the one hundred percent line
(or from zero absorbance). Deviations greater than ±5 percent in any analytical region are
unacceptable. Such deviations indicate a change in the instrument throughput relative to the
single-beam background.
3.11 Batch Sampling. A gas cell is alternately filled and evacuated. A Spectrum of each filled
cell (one discreet sample) is collected and saved.
3.12 Continuous Sampling. Sample is continuously flowing through a gas cell. Spectra of the
flowing sample are collected at regular intervals.
3.13 Continuous Operation. In continuous operation an FTIR CEM system, without user
intervention, samples flue gas, records spectra of samples, saves the spectra to a disk, analyzes
the spectra for the target analytes, and prints concentrations of target analytes to a computer file.
User intervention is permitted for initial set-up of sampling system, initial calibrations, and
periodic maintenance.
3.14 Sampling Time. In batch sampling—the time required to fill the cell with flue gas. In
continuous sampling—the time required to collect the infrared spectrum of the sample gas.
3.15 PPM-Meters. Sample concentration expressed as the concentration-path length product,
ppm (molar) concentration multiplied by the path length of the FTIR gas cell. Expressing
concentration in these units provides a way to directly compare measurements made using
systems with different optical configurations. Another useful expression is (ppm-meters)/K,
where K is the absolute temperature of the sample in the gas cell.
3.16 CEM Measurement Time Constant. The Time Constant (TC, minutes for one cell volume to
flow through the cell) determines the minimum interval for complete removal of an analyte from
the FTIR cell. It depends on the sampling rate (Rs in Lpm), the FTIR cell volume (Vceii in L) and
the chemical and physical properties of an analyte.
TC =^1L Eq. 1
For example, if the sample flow rate (through the FTIR cell) is 5 Lpm and the cell volume is 7
liters, then TC is equal to 1.4 minutes (0.71 cell volumes per minute). This performance
specification defines 5 * TC as the minimum interval between independent samples.
3.17 Independent Measurement. Two independent measurements are spectra of two independent
samples. Two independent samples are separated by, at least 5 cell volumes. The interval
between independent measurements depends on the cell volume and the sample flow rate
(through the cell). There is no mixing of gas between two independent samples. Alternatively,
estimate the analyte residence time empirically: (1) Fill cell to ambient pressure with a (known
analyte concentration) gas standard, (2) measure the spectrum of the gas standard, (3) purge the
cell with zero gas at the sampling rate and collect a spectrum every minute until the analyte
standard is no longer detected spectroscopically. If the measured time corresponds to less than 5
cell volumes, use 5 * TC as the minimum interval between independent measurements. If the
3
-------
Performance Specification 15
1/14/2019
measured time is greater than 5 * TC, then use this time as the minimum interval between
independent measurements.
3.18 Test Condition. A period of sampling where all process, and sampling conditions, and
emissions remain constant and during which a single sampling technique and a single analytical
program are used. One Run may include results for more than one test condition. Constant
emissions means that the composition of the emissions remains approximately stable so that a
single analytical program is suitable for analyzing all of the sample spectra. A greater than two-
fold change in analyte or interferant concentrations or the appearance of additional compounds in
the emissions, may constitute a new test condition and may require modification of the analytical
program.
3.19 Run. A single Run consists of spectra (one spectrum each) of at least 10 independent
samples over a minimum of one hour. The concentration results from the spectra can be averaged
together to give a run average for each analyte measured in the test run.
4.0 Interferences
Several compounds, including water, carbon monoxide, and carbon dioxide, are known
interferences in the infrared region in which the FTIR instrument operates. Follow the
procedures in the FTIR protocol for subtracting or otherwise dealing with these and other
interferences.
5.0 Safety
The procedures required under this performance specification may involve hazardous materials,
operations, and equipment. This performance specification may not address all of the safety
problems associated with these procedures. It is the responsibility of the user to establish
appropriate safety and health practices and determine the applicable regulatory limitations prior
to performing these procedures. The CEMS users manual and materials recommended by this
performance specification should be consulted for specific precautions to be taken.
6.0 Equipment and Supplies
6.1 Installation of sampling equipment should follow requirements of FTIR test Methods such as
references 1 and 3 and the EPA FTIR Protocol (reference 2). Select test points where the gas
stream composition is representative of the process emissions. If comparing to a reference
method, the probe tips for the FTIR CEM and the RM should be positioned close together using
the same sample port if possible.
6.2 FTIR Specifications. The FTIR CEM must be equipped with reference spectra bracketing the
range of path length-concentrations (absorbance intensities) to be measured for each analyte. The
effective concentration range of the analyzer can be adjusted by changing the path length of the
gas cell or by diluting the sample. The optical configuration of the FTIR system must be such
that maximum absorbance of any target analyte is no greater than 1.0 and the minimum
absorbance of any target analyte is at least 10 times the RMSD noise in the analytical region. For
4
-------
Performance Specification 15
1/14/2019
example, if the measured RMSD in an analytical region is equal to 10 3, then the peak analyte
absorbance is required to be at least 0.01. Adequate measurement of all of the target analytes
may require changing path lengths during a run, conducting separate runs for different analytes,
diluting the sample, or using more than one gas cell.
6.3 Data Storage Requirements. The system must have sufficient capacity to store all data
collected in one week of routine sampling. Data must be stored to a write-protected medium,
such as write-once-read-many (WORM) optical storage medium or to a password protected
remote storage location. A back-up copy of all data can be temporarily saved to the computer
hard drive. The following items must be stored during testing.
• At least one sample interferogram per sampling Run or one interferogram per hour, whichever
is greater. This assumes that no sampling or analytical conditions have changed during the run.
• All sample absorbance spectra (about 12 per hr, 288 per day).
• All background spectra and interferograms (variable, but about 5 per day).
• All CTS spectra and interferograms (at least 2 each 24 hour period).
• Documentation showing a record of resolution, path length, apodization, sampling time,
sampling conditions, and test conditions for all sample, CTS, calibration, and background
spectra.
Using a resolution of 0.5 cm-1, with analytical range of 3500 cm-1, assuming about 65 Kbytes
per spectrum and 130 Kb per interferogram, the storage requirement is about 164 Mb for one
week of continuous sampling. Lower spectral resolution requires less storage capacity. All of the
above data must be stored for at least two weeks. After two weeks, storage requirements include:
(1) all analytical results (calculated concentrations), (2) at least 1 sample spectrum with
corresponding background and sample interferograms for each test condition, (3) CTS and
calibration spectra with at least one interferogram for CTS and all interferograms for
calibrations, (4) a record of analytical input used to produce results, and (5) all other
documentation. These data must be stored according to the requirements of the applicable
regulation.
7.0 Reagents and Standards [Reserved]
8.0 Sample Collection, Preservation, Storage, and Transport [Reserved]
9.0 Quality Control
These procedures shall be used for periodic quarterly or semiannual QA/QC checks on the
operation of the FTIR CEM. Some procedures test only the analytical program and are not
intended as a test of the sampling system.
9.1 Audit Sample. This can serve as a check on both the sampling system and the analytical
program.
5
-------
Performance Specification 15
1/14/2019
9.1.1 Sample Requirements. The audit sample can be a mixture or a single component. It must
contain target analyte(s) at approximately the expected flue gas concentration(s). If possible,
each mixture component concentration should be NIST traceable (±2 percent accuracy). If a
cylinder mixture standard(s) cannot be obtained, then, alternatively, a gas phase standard can be
generated from a condensed phase analyte sample. Audit sample contents and concentrations are
not revealed to the FTIR CEM operator until after successful completion of procedures in 5.3.2.
9.1.2 Test Procedure. Spike the audit sample using the analyte spike procedure in section 11. The
audit sample is measured directly by the FTIR system (undiluted) and then spiked into the
effluent at a known dilution ratio. Measure a series of spiked and unspiked samples using the
same procedures as those used to analyze the stack gas. Analyze the results using sections 12.1
and 12.2. The measured concentration of each analyte must be within ±5 percent of the expected
concentration (plus the uncertainty), i.e., the calculated correction factor must be within 0.93 and
1.07 for an audit with an analyte uncertainty of ±2 percent.
9.2 Audit Spectra. Audit spectra can be used to test the analytical program of the FTIR CEM, but
provide no test of the sampling system.
9.2.1 Definition and Requirements. Audit spectra are absorbance spectra that; (1) have been well
characterized, and (2) contain absorbance bands of target analyte(s) and potential interferants at
intensities equivalent to what is expected in the source effluent. Audit spectra are provided by the
administrator without identifying information. Methods of preparing Audit spectra include; (1)
mathematically adding sample spectra or adding reference and interferant spectra, (2) obtaining
sample spectra of mixtures prepared in the laboratory, or (3) they may be sample spectra
collected previously at a similar source. In the last case it must be demonstrated that the
analytical results are correct and reproducible. A record associated with each Audit spectrum
documents its method of preparation. The documentation must be sufficient to enable an
independent analyst to reproduce the Audit spectra.
9.2.2 Test Procedure. Audit spectra concentrations are measured using the FTIR CEM analytical
program. Analytical results must be within ±5 percent of the certified audit concentration for
each analyte (plus the uncertainty in the audit concentration). If the condition is not met,
demonstrate how the audit spectra are unrepresentative of the sample spectra. If the audit spectra
are representative, modify the FTIR CEM analytical program until the test requirement is met.
Use the new analytical program in subsequent FTIR CEM analyses of effluent samples.
9.3 Submit Spectra For Independent Analysis. This procedure tests only the analytical program
and not the FTIR CEM sampling system. The analyst can submit FTIR CEM spectra for
independent analysis by EPA. Requirements for submission include; (1) three representative
absorbance spectra (and stored interferograms) for each test period to be reviewed, (2)
corresponding CTS spectra, (3) corresponding background spectra and interferograms, (4)
spectra of associated spiked samples if applicable, and (5) analytical results for these sample
spectra. The analyst will also submit documentation of process times and conditions, sampling
conditions associated with each spectrum, file names and sampling times, method of analysis and
reference spectra used, optical configuration of FTIR CEM including cell path length and
temperature, spectral resolution and apodization used for every spectrum. Independent analysis
6
-------
Performance Specification 15
1/14/2019
can also be performed on site in conjunction with the FTIR CEM sampling and analysis. Sample
spectra are stored on the independent analytical system as they are collected by the FTIR CEM
system. The FTIR CEM and the independent analyses are then performed separately. The two
analyses will agree to within ±120 percent for each analyte using the procedure in section 12.3.
This assumes both analytical routines have properly accounted for differences in optical path
length, resolution, and temperature between the sample spectra and the reference spectra.
10.0 Calibration and Standardization
10.1 Calibration Transfer Standards. For CTS requirements see section 4.5 of the FTIR Protocol.
A well characterized absorbance band in the CTS gas is used to measure the path length and line
resolution of the instrument. The CTS measurements made at the beginning of every 24 hour
period must agree to within ±5 percent after correction for differences in pressure.
Verify that the frequency response of the instrument and CTS absorbance intensity are correct by
comparing to other CTS spectra or by referring to the literature.
10.2 Analyte Calibration. If EPA library reference spectra are not available, use calibration
standards to prepare reference spectra according to section 6 of the FTIR Protocol. A suitable set
of analyte reference data includes spectra of at least 2 independent samples at each of at least 2
different concentrations. The concentrations bracket a range that includes the expected analyte
absorbance intensities. The linear fit of the reference analyte band areas must have a fractional
calibration uncertainty (FCU in Appendix F of the FTIR Protocol) of no greater than 10 percent.
For requirements of analyte standards refer to section 4.6 of the FTIR Protocol.
10.3 System Calibration. The calibration standard is introduced at a point on the sampling probe.
The sampling system is purged with the calibration standard to verify that the absorbance
measured in this way is equal to the absorbance in the analyte calibration. Note that the system
calibration gives no indication of the ability of the sampling system to transport the target
analyte(s) under the test conditions.
10.4 Analyte Spike. The target analyte(s) is spiked at the outlet of the sampling probe, upstream
of the particulate filter, and combined with effluent at a ratio of about 1 part spike to 9 parts
effluent. The measured absorbance of the spike is compared to the expected absorbance of the
spike plus the analyte concentration already in the effluent. This measures sampling system bias,
if any, as distinguished from analyzer bias. It is important that spiked sample pass through all of
the sampling system components before analysis.
10.5 Signal-to-Noise Ratio (S/N). The measure of S/N in this performance specification is the
root-mean-square (RMS) noise level as given in Appendix C of the FTIR Protocol. The RMS
noise level of a contiguous segment of a spectrum is defined as the RMS difference (RMSD)
between the n contiguous absorbance values (A;) which form the segment and the mean value
(Am) of that segment.
Eq. 2
7
-------
Performance Specification 15
1/14/2019
A decrease in the S/N may indicate a loss in optical throughput, or detector or interferometer
malfunction.
10.6 Background Deviation. The 100 percent baseline must be between 95 and 105 percent
transmittance (absorbance of 0.02 to -0.02) in every analytical region. When background
deviation exceeds this range, a new background spectrum must be collected using nitrogen or
other zero gas.
10.7 Detector Linearity. Measure the background and CTS at three instrument aperture settings;
one at the aperture setting to be used in the testing, and one each at settings one half and twice
the test aperture setting. Compare the three CTS spectra. CTS band areas should agree to within
the uncertainty of the cylinder standard. If test aperture is the maximum aperture, collect CTS
spectrum at maximum aperture, then close the aperture to reduce the IR through-put by half.
Collect a second background and CTS at the smaller aperture setting and compare the spectra as
above. Instead of changing the aperture neutral density filters can be used to attenuate the
infrared beam. Set up the FTIR system as it will be used in the test measurements. Collect a CTS
spectrum. Use a neutral density filter to attenuate the infrared beam (either immediately after the
source or the interferometer) to approximately Vi its original intensity. Collect a second CTS
spectrum. Use another filter to attenuate the infrared beam to approximately lA its original
intensity. Collect a third background and CTS spectrum. Compare the CTS spectra as above.
Another check on linearity is to observe the single beam background in frequency regions where
the optical configuration is known to have a zero response. Verify that the detector response is
"flat" and equal to zero in these regions. If detector response is not linear, decrease aperture, or
attenuate the infrared beam. Repeat the linearity check until system passes the requirement.
11.0 Analytical Procedure
11.1 Initial Certification. First, perform the evaluation procedures in section 6.0 of the FTIR
Protocol. The performance of an FTIR CEM can be certified upon installation using EPA
Method 301 type validation (40 CFR, Part 63, Appendix A), or by comparison to a reference
Method if one exists for the target analyte(s). Details of each procedure are given below.
Validation testing is used for initial certification upon installation of a new system. Subsequent
performance checks can be performed with more limited analyte spiking. Performance of the
analytical program is checked initially, and periodically as required by EPA, by analyzing audit
spectra or audit gases.
11.1.1 Validation. Use EPA Method 301 type sampling (reference 4, section 5.3 of Method 301)
to validate the FTIR CEM for measuring the target analytes. The analyte spike procedure is as
follows: (1) a known concentration of analyte is mixed with a known concentration of a non-
reactive tracer gas, (2) the undiluted spike gas is sent directly to the FTIR cell and a spectrum of
this sample is collected, (3) pre-heat the spiked gas to at least the sample line temperature, (4)
introduce spike gas at the back of the sample probe upstream of the particulate filter, (5) spiked
effluent is carried through all sampling components downstream of the probe, (6) spike at a ratio
of roughly 1 part spike to 9 parts flue gas (or more dilute), (7) the spike-to-flue gas ratio is
estimated by comparing the spike flow to the total sample flow, and (8) the spike ratio is verified
by comparing the tracer concentration in spiked flue gas to the tracer concentration in undiluted
8
-------
Performance Specification 15
1/14/2019
spike gas. The analyte flue gas concentration is unimportant as long as the spiked component can
be measured and the sample matrix (including interferences) is similar to its composition under
test conditions. Validation can be performed using a single FTIR CEM analyzing sample spectra
collected sequentially. Since flue gas analyte (unspiked) concentrations can vary, it is
recommended that two separate sampling lines (and pumps) are used; one line to carry unspiked
flue gas and the other line to carry spiked flue gas. Even with two sampling lines the variation in
unspiked concentration may be fast compared to the interval between consecutive measurements.
Alternatively, two FTIR CEMs can be operated side-by-side, one measuring spiked sample, the
other unspiked sample. In this arrangement spiked and unspiked measurements can be
synchronized to minimize the affect of temporal variation in the unspiked analyte concentration.
In either sampling arrangement, the interval between measured concentrations used in the
statistical analysis should be, at least, 5 cell volumes (5 * TC in equation 1). A validation run
consists of, at least, 24 independent analytical results, 12 spiked and 12 unspiked samples. See
section 3.17 for definition of an "independent" analytical result. The results are analyzed using
sections 12.1 and 12.2 to determine if the measurements passed the validation requirements.
Several analytes can be spiked and measured in the same sampling run, but a separate statistical
analysis is performed for each analyte. In lieu of 24 independent measurements, averaged results
can be used in the statistical analysis. In this procedure, a series of consecutive spiked
measurements are combined over a sampling period to give a single average result. The related
unspiked measurements are averaged in the same way. The minimum 12 spiked and 12 unspiked
result averages are obtained by averaging measurements over subsequent sampling periods of
equal duration. The averaged results are grouped together and statistically analyzed using section
12.2.
11.1.1.1 Validation with a Single Analyzer and Sampling Line. If one sampling line is used,
connect the sampling system components and purge the entire sampling system and cell with at
least 10 cell volumes of sample gas. Begin sampling by collecting spectra of 2 independent
unspiked samples. Introduce the spike gas into the back of the probe, upstream of the particulate
filter. Allow 10 cell volumes of spiked flue gas to purge the cell and sampling system. Collect
spectra of 2 independent spiked samples. Turn off the spike flow and allow 10 cell volumes of
unspiked flue gas to purge the FTIR cell and sampling system. Repeat this procedure 6 times
until the 24 samples are collected. Spiked and unspiked samples can also be measured in groups
of 4 instead of in pairs. Analyze the results using sections 12.1 and 12.2. If the statistical analysis
passes the validation criteria, then the validation is completed. If the results do not pass the
validation, the cause may be that temporal variations in the analyte sample gas concentration are
fast relative to the interval between measurements. The difficulty may be avoided by: (1)
Averaging the measurements over long sampling periods and using the averaged results in the
statistical analysis, (2) modifying the sampling system to reduce TC by, for example, using a
smaller volume cell or increasing the sample flow rate, (3) using two sample lines (4) use two
analyzers to perform synchronized measurements. This performance specification permits
modifications in the sampling system to minimize TC if the other requirements of the validation
sampling procedure are met.
11.1.1.2 Validation With a Single Analyzer and Two Sampling Lines. An alternative sampling
procedure uses two separate sample lines, one carrying spiked flue gas, the other carrying
unspiked gas. A valve in the gas distribution manifold allows the operator to choose either
9
-------
Performance Specification 15
1/14/2019
sample. A short heated line connects the FTIR cell to the 3-way valve in the manifold. Both
sampling lines are continuously purged. Each sample line has a rotameter and a bypass vent line
after the rotameter, immediately upstream of the valve, so that the spike and unspiked sample
flows can each be continuously monitored. Begin sampling by collecting spectra of 2
independent unspiked samples. Turn the sampling valve to close off the unspiked gas flow and
allow the spiked flue gas to enter the FTIR cell. Isolate and evacuate the cell and fill with the
spiked sample to ambient pressure. (While the evacuated cell is filling, prevent air leaks into the
cell by making sure that the spike sample rotameter always indicates that a portion of the flow is
directed out the by-pass vent.) Open the cell outlet valve to allow spiked sample to continuously
flow through the cell. Measure spectra of 2 independent spiked samples. Repeat this procedure
until at least 24 samples are collected.
11.1.1.3 Synchronized Measurements With Two Analyzers. Use two FTIR analyzers, each with
its own cell, to perform synchronized spiked and unspiked measurements. If possible, use a
similar optical configuration for both systems. The optical configurations are compared by
measuring the same CTS gas with both analyzers. Each FTIR system uses its own sampling
system including a separate sampling probe and sampling line. A common gas distribution
manifold can be used if the samples are never mixed. One sampling system and analyzer
measures spiked effluent. The other sampling system and analyzer measures unspiked flue gas.
The two systems are synchronized so that each measures spectra at approximately the same
times. The sample flow rates are also synchronized so that both sampling rates are approximately
the same (TCi~ TC2 in equation 1). Start both systems at the same time. Collect spectra of at
least 12 independent samples with each (spiked and unspiked) system to obtain the minimum 24
measurements. Analyze the analytical results using sections 12.1 and 12.2. Run averages can be
used in the statistical analysis instead of individual measurements.
11.1.1.4 Compare to a Reference Method (RM). Obtain EPA approval that the method qualifies
as an RM for the analyte(s) and the source to be tested. Follow the published procedures for the
RM in preparing and setting up equipment and sampling system, performing measurements, and
reporting results. Since FTIR CEMS have multicomponent capability, it is possible to perform
more than one RM simultaneously, one for each target analyte. Conduct at least 9 runs where the
FTIR CEM and the RM are sampling simultaneously. Each Run is at least 30 minutes long and
consists of spectra of at least 5 independent FTIR CEM samples and the corresponding RM
measurements. If more than 9 runs are conducted, the analyst may eliminate up to 3 runs from
the analysis if at least 9 runs are used.
11.1.1.4.1 RMs Using Integrated Sampling. Perform the RM and FTIR CEM sampling
simultaneously. The FTIR CEM can measure spectra as frequently as the analyst chooses (and
should obtain measurements as frequently as possible) provided that the measurements include
spectra of at least 5 independent measurements every 30 minutes. Concentration results from all
of the FTIR CEM spectra within a run may be averaged for use in the statistical comparison even
if all of the measurements are not independent. When averaging the FTIR CEM concentrations
within a run, it is permitted to exclude some measurements from the average provided the
minimum of 5 independent measurements every 30 minutes are included: The Run average of
the FTIR CEM measurements depends on both the sample flow rate and the measurement
frequency (MF). The run average of the RM using the integrated sampling method depends
10
-------
Performance Specification 15
1/14/2019
primarily on its sampling rate. If the target analyte concentration fluctuates significantly, the
contribution to the run average of a large fluctuation depends on the sampling rate and
measurement frequency, and on the duration and magnitude of the fluctuation. It is, therefore,
important to carefully select the sampling rate for both the FTIR CEM and the RM and the
measurement frequency for the FTIR CEM. The minimum of 9 run averages can be compared
according to the relative accuracy test procedure in Performance Specification 2 for SO2 and
NOx CEMs (40 CFR, Part 60, App. B).
11.1.1.4.2 RMs Using a Grab Sampling Technique. Synchronize the RM and FTIR CEM
measurements as closely as possible. For a grab sampling RM, record the volume collected and
the exact sampling period for each sample. Synchronize the FTIR CEM so that the FTIR
measures a spectrum of a similar cell volume at the same time as the RM grab sample was
collected. Measure at least five independent samples with both the FTIR CEM and the RM for
each of the minimum nine runs. Compare the run concentration averages by using the relative
accuracy analysis procedure in Performance Specification 2 of appendix B of 40 CFR part 60.
11.1.1.4.3 Continuous Emission Monitors as RMs. If the RM is a CEM, synchronize the
sampling flow rates of the RM and the FTIR CEM. Each run is at least 1 hour long and consists
of at least 10 FTIR CEM measurements and the corresponding 10 RM measurements (or
averages). For the statistical comparison, use the relative accuracy analysis procedure in
Performance Specification 2 of appendix B of 40 CFR part 60. If the RM time constant is
< V2 the FTIR CEM time constant, brief fluctuations in analyte concentrations that are not
adequately measured with the slower FTIR CEM time constant can be excluded from the run
average along with the corresponding RM measurements. However, the FTIR CEM run average
must still include at least 10 measurements over a 1-hour period.
12.0 Calculations and Data Analysis
12.1 Spike Dilution Ratio, Expected Concentration. The Method 301 bias is calculated as
follows.
B=Sm- Mm - CS Eq. 3
Where:
B = Bias at the spike level
Sm = Mean of the observed spiked sample concentrations
Mm = Mean of the observed unspiked sample concentrations
CS = Expected value of the spiked concentration.
The CS is determined by comparing the SF6 tracer concentration in undiluted spike gas to the
SF6 tracer concentrations in the spiked samples;
11
-------
Performance Specification 15
1/14/2019
[ 1
DF = 6, Eq, 4
[^W
The expected concentration (CS) is the measured concentration of the analyte in undiluted spike
gas divided by the dilution factor
cs = i E 5
DF
Where:
[analjdir = The analyte concentration in undiluted spike gas measured directly by filling the FTIR
cell with the spike gas.
If the bias is statistically significant (Section 12.2), Method 301 requires that a correction factor,
CF, be multiplied by the analytical results, and that 0.7
-------
Performance Specification 15
1/14/2019
S71
RSD = — Eq. 8
s
Repeat the calculations in equations 7 and 8 to determine SDU and RSD, respectively, for the
unspiked samples. Calculate the standard deviation of the mean using SDs and SDU from
equation 7.
SD = JSD, 2+ SDa 2 Eq. 9
The t-statistic is calculated as follows to test the bias for statistical significance;
t = Eq. 10
EDM
where the bias, B, and the correction factor, CF, are given in section 12.1. For 11 degrees of
freedom, and a one-tailed distribution, Method 301 requires that t <2.201. If the t-statistic
indicates the bias is statistically significant, then analytical measurements must be multiplied by
the correction factor. There is no limitation on the number of measurements, but there must be at
least 12 independent spiked and 12 independent unspiked measurements. Refer to the t-
distribution (Table 2) at the 95 percent confidence level and appropriate degrees of freedom for
the critical t-value.
13.0 Method Performance [Reserved]
14.0 Pollution Prevention [Reserved]
15.0 Waste Management [Reserved]
16.0 References
1. Method 318, 40 CFR, Part 63, Appendix A (Draft), "Measurement of Gaseous Formaldehyde,
Phenol and Methanol Emissions by FTIR Spectroscopy," EPA Contract No. 68D20163, Work
Assignment 2-18, February, 1995.
2. "EPA Protocol for the Use of Extractive Fourier Transform Infrared (FTIR) Spectrometry in
Analyses of Gaseous Emissions from Stationary Industrial Sources," February, 1995.
3. "Measurement of Gaseous Organic and Inorganic Emissions by Extractive FTIR
Spectroscopy," EPA Contract No. 68-D2-0165, Work Assignment 3-08.
4. "Method 301—Field Validation of Pollutant Measurement Methods from Various Waste
Media," 40 CFR 63, App A.
17.0 Tables, Diagrams, Flowcharts, and Validation Data
13
-------
Performance Specification 15 1/14/2019
Table 1—Arrangement of Validation Measurements for Statistical Analysis
Measurement
(or average)
Time
Spiked
(ppm)
di spiked
Unspiked
(ppm)
di unspiked
1
Si
Ui
2
s2
S2-S1
U2
U2-U1
3
S3
u3
4
s4
S4-S3
u4
U4-U3
5
s5
Us
6
s6
S6-S5
u6
U6-U5
7
s7
u7
8
s8
S8-S7
u8
U8-U7
9
s9
u9
10
Sio
0
1
m
\D
U10
U10-U9
11
Sn
Un
12
Sl2
S12—Sn
U12
U12-U11
Average ->
Srn
Mm
Table 2—t=Values
n-la
t-value
n-la
t-value
n-la
t-value
n-la
t-value
11
2.201
17
2.110
23
2.069
29
2.045
12
2.179
18
2.101
24
2.064
30
2.042
13
2.160
19
2.093
25
2.060
40
2.021
14
2.145
20
2.086
26
2.056
60
2.000
15
2.131
21
2.080
27
2.052
120
1.980
16
2.120
22
2.074
28
2.048
8
1.960
an is the number of independent pairs of measurements (a pair consists of one spiked and its
corresponding unspiked measurement). Either discreet (independent) measurements in a single run, or run
averages can be used.
14
------- |