AC-76-02
Technical Report
Sources of Variability and Inaccuracies in
Aircraft Gas Turbine Emission Measurements
February 1976
Ho-tice
Technical support reports for regulatory action do not
necessarily represent the fina] EPA decision on regulatory issues.
They are intended to present a technical analysis of an issue and
recommendations resulting from the assumptions and constraints of
that analysis. Agency policy considerations or data received
subsequent to the date of release of this report may alter the
recommendations reached. Readers are cautioned to seek the latest
analysis from EPA before using the information contained herein.
Standards Development and Support Branch
Emission Control Technology Division
Office of Mobile Source Air Pollution Control
Office of Air and Waste Management
U.S. Environmental Protection Agency
-------
Abstract
Variability and inaccuracies in aircraft gas turbine emission
measurements caused by calibration gases, instrument precision, sampling
errors, engine setting precision, and ambient condition effects continue
to be of major concern. Understanding and/or resolution of these factors
is critical to the development of a unified measurement procedure.
This report details and provides analysis of the various factors
contributing to emission measurement inaccuracy. Discussion is presentee
for all major sections and components of the emission sampling system.
PrepSrea* by
Approvec
Project Manager - Aircraft
Approved
Branch Chief, SDS
sion Director, ECTD
Distribution;
D. Alexander
J. DeKany
C. Gray
W. Houtman
G. Austin
E. Stork
M. Williams
G. Kittredge
-------
Table of Contents
Page
1.0 Introduction 1
2.0 Summary and Recommendations 1
3.0 Sample Transfer System Analysis (Gaseous) 2
3.] Sample Line Temperature 2
3.2 Sample Line Material 3
3.3 Sample Line Flow Rate 4
3.4 Sample Line Length 4
4.0 Analytical Sample System Analysis (Gaseous) 5
4.1 Hydrocarbon Analyzer 5
4.2 Carbon Monoxide, Carbon Dioxide Analyzer 5
4.3 Oxides of Nitrogen Analyzer 6
5.0 Sample Probe Analysis (Gaseous) 7
6.0 Smoke Sample System Analysis 9
6.1 Smoke Sample Collection and Analysis 9
6.2 Smoke Transfer System 9
7.0 Effect of Variations in Ambient Conditions 10
7.1 Ambient Condition Effect on NO 11
x
7.2 Ambient Conditions Effect on HC, CO, and Smoke 11
7.3 Magnitude of Variability 12
8.0 Background Concentration Effect 12
9.0 Calibration Gas Accuracy 13
10.0 Variability and Repeatability of Engine Parameters 13
11.0 System Accuracy 14
References 17
Appendix
-------
1.0 Introduction
This report presents an analysis of the sources of inaccuracies and
variability propagated in the determination of the constituent levels of
HC, CO, C09, NOx (NO + N0~), and smoke in aircraft gas turbine engines.
t~ £~
The regulations promulgated by the Environmental Protection Agency
for the "Control of Air Pollution From Aircraft and Aircraft Engines"
(40 CFR Part 87) provide a well accepted procedure for determination of
emission levels. These procedures provide the basis for analysis of the
pollutant levels and are therefore the focal point of this report.
2.0 Summary and Recommendations
A. The sample transfer system for gaseous emissions has been shown
to provide an adequate means of transporting the sample from the
collection probe to the analytical equipment.
B. The analytical system provides analysis of the gaseous con-
stituents consistent with the present state of the art.
C. Further consideration should be given to allow the use of a
dryer in the CO and CO- (ND1R) system but only if valid correction
factor for correction to a wet basis can be provided.
D. The sample probe extracts a representative sample of HC, CO,
C0?, and NOx for all non-afterburning engines except the mixed flow
turbofan engines. Further study is required for the determination
of probe design which provides a representative sample for the
mixed flow turbofan engine.
E. Further study and a possible system modification is recom-
mended for the smoke measurement system. Data indicates that it is
difficult to obtain a representative sample by extraction of a
sample from the gaseous emission probe.
E. To satisfy the calibration gas requirements of the regulations,
there is a need for additional standard reference materials such
as NO in nitrogen, intermediate levels of C02 in air, and higher
concentrations of CO in nitrogen, and propane in air.
G. Ambient condition effects are the largest single source of a
variability in gas turbine emission measurements. The EPA's
Ambient Effect Program is addressing this problem.
H. Error analysis indicates that sample error is secondary only to
ambient conditions in its contribution to the total exhaust gas
measurement uncertainty.
-------
-2-
3.0 Sample Transfer System Analysis (Gaseous)
The sample line must transport the extracted gas sample from the
collection probe to the interface of the analytical instrument without
affecting the concentration level of the constituents analyzed. EPA
promulgated regulations (40 CFR Part 87) require that the sample line
must be:
A. Constructed of either stainless steel or Teflon
B. Heated to 350 °C + 5 °C (302 °F + 9 °F) prior to the flame
ionization detector
C. Heated to 55 °C + 5 °C (131 °F + 9 °F) after the flame ioni-
zation detector
D. In no case shall it be longer than 24.4 meters (80 feet)
E. Capable of transporting the sample from the probe inlet to the
analytical equipment in 2 seconds or less.
F. Of 0.46 to 0.81 centimeters (0.18 to 0.32 inches) inside
diameter.
3.1 Sample Line Temperature
Maintenance of the sample line temperature at the prescribed temp-
erature levels is mandated due to concern for water vapor and hydrocarbon
condensation in" the line. Gas turbine engines operated with up to 8%
water vapor in the exhaust of which about 4% can be attributed to the
combustion reaction of the hydrocarbon fuel. The remainder of the water
vapor is contributed by the background concentration of humidity. A
sample line temperature of 50 C will prevent water vapor condensation
in the sample for all practical operating conditions.
Although only a limited amount of data is available on the
analysis of the condensable hydrocarbon compounds present in the exhaust
of gas turbine engines it supports the need to maintain the sample
temperature at 150 C or greater prior to hydrocarbon analysis. Boiling
point data of the hydrocarbon compounds identified in the exhaust indicates
that this temperature will insure that compounds which have a carbon
number of 13 or less will remain in the vapor state. The partial pressure
effect on the hydrocarbon compounds present in the system which operates
at subatmospheric pressure further insures against condensation because
of the reduced boiling point of the mixture.
(2)
Data on concentration levels of the identified hydrocarbon
compounds is presented in Figure 1 (Appendix). It was collected by..,
cryogenically trapping the hydrocarbon compounds. A second study
was conducted by cryogenically trapping the higher boiling point compounds
and bag sampling the remainder of the noncondensable compounds. Results
of both of these studies indicated that for Jet-A fuel, 95% of the
-------
hydrocarbon compounds present in the exhaust have a carbon number of 13
or less. For other fuels such as DF-2 and JP-4 this percentage was
lower and there would therefore be greater concern for condensation of
hydrocarbon compounds at a sample temperature of 150 C. The maximum
temperature which the sample can be heated to is limited only by practical
and mechanical limits of the system.
A high temperature (150 °C) is further supported by the length of
time required for the system to purge itself of possible contamination
encountered during start up or transient operation of the engine.
Experience has shoxm that a low temperature contaminated system may
require up to 30 minutes to stabilize to background levels. This length
of time is greatly reduced in higher temperature systems.
The procedure used for the determination of the sample (sample
line) temperature is also of extreme importance. It is desirable but
impractical to measure the actual sample temperature in the sample line.
It is also important that the sample lane surface be maintained at the
prescribed temperature and be free of any cold spots where condensation
would be encouraged. It is therefore practical to determine the tempera-
ture of the sample line (see sketch) in a sufficient number of locations
along the line length to ensure against any cold spots. Presently this
is not clearly specified in the EPA regulations.
Thermocouple
Sheild
Sample line wall
thermocouple
Material Stainless
Steel or Teflon
Heating Tape
Insulation
Sample Line Construction
3.2 Sample Line Material
The use of stainless steel or Teflon as the non-reactive sample
line material to be used -for,.transport of the gaseous sample has pre-
viously been established . No evidence has been presented which
establishes concern for these materials inability to maintain the integrity
of the sample during transport to the analysis instruments.
-------
-4-
3.3 Sample Line Flow Rate
The difficulty of transferring the sample from the sample line
inlet to the analysis instruments in the required two seconds or less is
compounded by the requJrement that 20 liters per minute of the sample be
bleed off in the instrument system. Since the maximum flow rate is a
function of the pressure drop in the sample line, systems which require
the maximum line length of 24.4 meters,will necessitate the use of a
high capacity vacuum pump. A study of the capabilities of various
system designs including variations in the sample line length substanti-
ated that these requirements could be met.
The need for the short transfer times is established through
concern that a molecule of exhaust gas should have a minimum residence
time at any point in the sample line to discourage condensation of any
constitutent and deter possible chemical reactions.
Concern primarily from industry that these requirements are severely
restrictive and are unnecessary to insure the integrity of the sample
has not been documented with substantiating evidence. It is therefore
felt that the sample line flow requJrement can be maintained at their
present level.
3.4 Sample Line Length
It has been established that the sample line length should
not exceed a length of 24.4 meters (80 feet), otherwise sample integrity
may be comproiiused. A study conducted for the EPA by Scott Environ-
mental Technology demonstrated that for sample lines up to 80 feet in
length the sample integrity is maintained (Table 1). Lengths longer
than 80 feet were not tested in this study. The need to allow sample
line lengths in excess of 24.4 meters is therefore not warranted except
for possible isolated incidences where lengths 24.4 meters or less are
physically impossible. In this case the burden of insuring the sample
integrity should be placed on the agent seeking the variance from the
regulations.
Table 1.
Emission Measurements Made through a Heated Stainless Steel
Sample at 150° and a bypass flow of 20 L/M
Sample THC* CO CO,, NO
2 x
Line PPM-C PPM % PPM
Length (ft)
80 167 327 1.48 11.7
80 168 329 1.48 11.9
40 173 327 1.48 11.9
40 168 329 1.48 12.1
* The Hydrocarbon Sample Line used was 20 feet shorter than that indicated.
-------
-5-
4.0 Analytical Sample System Analysis (Gaseous)
4.1 Hydrocarbon Analyzer
The Flame lonization Detector (FID) is specified as the instrument
to be used for the determination of the total hydrocarbons present in
the exhaust. This instrument operates on the principle that a pure
hydrogen-air flame produces very little ionization. When a hydrocarbon
compound is introduced into the diffusion flame a chemical ionization is
produced. This ionization potential is proportional to the number of
carbon atoms present in the hydrocarbon molecules. The instrument's
ability to respond in an approximately linear manner to saturated,
unsatuiated and aromatic carbon atoms in a variety of organic structures
provides a realistic method of monitoring the total mass emission of
hydrocarbon compounds.
The reaction chamber and all sections of the analyzer exposed to
the sample are heated to approximately 160 C (320 F). This relative
high temperature is required to assure that minimal condensation of
gaseous hydrocarbon compounds in the sample system occurs. As previously
stated, existing data on the hydrocarbon compounds present in the
exhaust of gas turbine engines indicates that approximately 95% of the
hydrocarbon compounds present have a carbon number of 13 or less (Figure
1). This evidence coupled v?ith the effect of the partial pressure on
the mixture will deter any possibility of hydrocarbon condensation.
The two predominate sources of inaccuracies in this analysis system
are synergism (oxygen interference) and its response (linearity) per
unit carbon atom to all hydrocarbon compounds. It is difficult to
predict .quantitatively the effective interference level contribution
from oxygen in the sample. Excess oxygen in the sample changes the
nature of the flame from a pure diffusion flame to a premixed one with a
higher flame temperature. This higher flame temperature enhances thermal
cracking and rapid oxidation of the hydrocarbon compounds. As previously
stated, the EPA regulations provide a check of the detector response for
the oxygen effect by introducing various concentrations of oxygen into
the sample and determining their effect on measured levels. Maximum
limits of interference are placed on this effect. The instrument's
linear response to hydrocarbon compounds (differential hydrocarbon
effect) is checked and the detector response optimized by introducing a
mixture of propane in air to the burner and adjusting the burner fuel
and background air flow to optimum response.
4.2 Carbon Monoxide, Carbon Dioxide Analyzers
The nondispersive infrared (NDIR) analyzer is specified for the
determination of concentrations of both carbon monoxide (CO) and carbon
dioxide (C0~)• In this instrument the gas being measured is used to
detect itself by comparison to a reference gas. The method of detection
is based on the principle of selective absorption. The infrared energy
-------
-6-
•of the particular wave length of the gas being measured will be absorbed
by that gas and the infrared energy of other wave lengths will be transmitted.
The detector determines this absorption by difference and records it.
Interference in the NDIR measurements may arise if the exhaust
sample contains other species which will absorb radiation at the same
frequencies as the gas being measured. For example the water vapor and
C09 absorption spectra will interfer with that" of CO. The use of a
dryer to eliminate the water vapor interference problem would provide a
means of avoiding part of the problem. This creates an additional
problem in that a correction factor for the conversion of the mass
pollutant level of the constituent from a dry basis to a weL basis would
have to be provided. At the present time a definitive correction factor
for this conversion has not been documented. Since it is the intent of
the regulations to determine the levels of CO and C0~ on a wet basis a
procedure is provided to determine and correct this interference. The
procedure provides for the passing of a known range of carbon dioxide
and water vapor through the CO instrument and observing the response.
If the effect is greater than two percent of the measured carbon monoxide
level all subsequent measurements must be corrected for this interference.
Further consideration should be given to allow the use of a dryer in the
CO and C0? (NDIR) system but only if valid correction factor for correc-
tion to a wet basis can be provided.
One additional problem which can create discrepancies in the determi-
nation of CO and C0~ levels is the presence of condensed water vapor in
the system. A blockage in the system which would affect the instrument
working pressure will effect the instrument calibration (A5" R~0-
1% change in the.instrument calibration). Extreme care must therefore
be taken_to prevent water vapor condensation.
4.3 Oxides of Nitrogen Analyzer
Chemiluminescent analysis is the method of analysis required for
measuring the emission levels of NO and NOx. This analysis technique
utilizes the principle that nitric oxide (NO) reacts with ozone (0~) to
yield nitrogen dioxide (N0~) and oxygen (0?). Approximately 10% of the
N02 is electronically excited. The transmission of the excited N0_ to
the ground state yields light emission. The intensity of the light
emission is proportional to the mass flow rate of NO in the reactor. To
determine the level of N0_ in the exhaust and therefore the total NOx
(NO + N0_) in the sample, conversion of NO- to NO is provided by a
thermal converter. An efficiency check of the converter is provided to
ensure that at least 90% of the NO- is converted to NO.
Condensation of water vapor or hydrocarbon compounds in the chem-
iluminscent instrument system is also a persistent danger to the integrity
of the sample. The two predominate effects are the solubility of NO and
N0 in the condensed hydrocarbon compound (small effect) or water
-------
-7-
vapor, and a water droplet blockage in the system which would effect the
sample flow-rate. One approach to the solution of this problem would be
to remove the objectional constituents in a manner which maintains the
integrity of the NOx sample. This could be accomplished by using a
dryer similar to that described for the alternative NDIR procedure. A
second approach and that required by the EPA Regulations is to heat the
sample to a temperature which is above the dew point for the water vapor
at the system operating pressure. This will prevent the loss of NO and
N0? due to their solubility in water and limit any loss due to hydrocarbon
absorption.
5.0 Sample Probe Analysis (Gaseous)
The sample probe as required by the EPA regulations shall have a
minimum of 12 sampling points either mixed into an integrated sample or
individually sampled with separate analysis for each. If a mixing probe
is used the orifices shall be of equal area. In all cases the probe
shall be designed such that 80% of the total pressure drop of the probe
assembly will be taken at the probe orifice or orifices and it shall
provide a representative sample.
The sample probe is therefore required to collect a representative
sample without bias to any constituent (HC, CO, C0_, NOx and Smoke), and
to transport the sample Lo the sample line unreacted or modified in any
manner. In non-afterburning engines this mechanism is simplified by the
fact that the exhaust gas reaction of the constituents of concern is
complete at the sampling point. The complexity of extracting a represen-
tative sample from the exhaust*, plume of a mixed flow fan engine can be
illustrated from studies conducted which detail the exhaust gas
pollutant levels on concentration isoplots, (Figure 2, 3, A). This data
was collected from a detailed traverse of the engine exhaust. These
figures show the non-uniform characteristics of these constituents and
the difficulty experienced in attempting to design an integrating probe
which would provide a representative sample for all constituents. Also,
this situation is further complicated by changes in concentration profiles
with power setting. The exhaust gas profiles for turbojets and non-
mixed flow turbofan engines are not nearly as complex and are uniform
when compared to the mixed flow turbofan engines. Probe design for
mixed flow engines is more difficult than for the nonmixed flow engines,
in fact, attempts at probe design for mixed flow engines such as the
JT8D and Spey have been unsuccessful, and show discrepancy of about 25%
when compared on a carbon balance basis.
(4)
A comprehensive study was conducted by the EPA to determine if
mass flow weighting of the constituent concentration instead of the area
weighting technique presently required would provide a means of extracting
a representative sample for mixed flow engines. Results of this study
did not provide a satisfactory method of extracting a representative
sample on a mass flow basis.
-------
-8-
A second test for sample representativeness is the comparison of
the total number of carbon atoms in the extracted sample to the total
number calculated from the fuel and air flow (carbon balance). This is
commonly done by comparing the calculated vs the measured fuel to air
ratio (F/A). The F/A ratio for exhaust gas emissions is given in the
simplified formula:
F/A Exhaust
207 - 2
~A + CO
10
CO
104 - C°2
+ —
2 io4
The emission concentrations of typical present day class T2 are:
HC CO C02
Idle 400 450 1.6
Take off 3 10 3.5
where:
CO and HC are in PPM
C02 is in %
As can be seen by the above equation the representativeness of CO
and HC is not really tested because the relative value of the C0_ con-
centration is predominate over the HC and CO concentrations. NOx and
Smoke are not tested in this method for representativeness.
At present the two techniques described above provide the best
method of demonstrating the representativeness of the sample extracted.
Manufacturers of mixed flow turbofan engines (Pratt and Whitney
Aircraft and Rolls Royce) are conducting in-house investigations with
the intent of developing an acceptable probe. It is therefore recom-
mended that these programs be monitored closely to determine if progress
is being made towards the design of a representative sampling technique
for mixed flow turbofan engines. If a solution can not be found there
are two alternatives to consider.
A. Continue to require that they comply with the probe design
requirements of the EPA Regulations (40 CFR Part 87).
B. Require a detailed traverse of the exhaust plume to provide an
acceptable representative sample.
-------
-9-
6.0 Smoke Sample Collection and Analysis
This section presented separately from the gaseous system analysis
because of unique problems which exist with this procedure.
The smoke sample system as specified in the EPA regulation requires
that the smoke sample be extracted from the probe or sample line used in
the gaseous emission analysis system. The sample is then transported
through a stainless steel or copper, heated, (above the dew point) line
with a maximum length of 22.9 meters (75 feet) to the sample collection
system. The sample collection system consists of a filter holder and
sample measuring devices. The filter holder collects a series of 4
stained spot filter samples. From these filter samples the actual smoke
number is determined and reported for each power setting using a re-
flectometer.
6.1 Smoke Sample Collection and Analysis
The procedures and mechanisms used in the collection of the stained
filter sample are generally considered to be acceptable and provide a
representative analysis of the sample presented to it. The only exception
to this is the requirement that 4 stained filter samples shall be collected
at each powei setting. Collection of these 4 one minute samples plus
the time required for the system to stabilize requires approximately 5
minutes of engine running time at each power setting. This procedure
becomes particularly burdensome at the take off power setting where the
required time is often in excess of the manufacturers recommended operating
procedure.
The' four filter samples are required because the determination of
the smoke number is based on a linear least squares fit line of the
reflection density smoke number vs the sample weight per square inch of
filter spot area. Fewer than four filter samples would therefore effect
the accuracy of the smoke number calculated because of its adverse
effect on the least squares line accuracy.
Even though there would be an adverse effect on the calculated
smoke number accuracy, consideration should be given to the reduction in
the number of filter samples required at the take off power setting. A
reduction such as this would have an added advantage of conserving fuel.
6.2 Smoke Transfer System
As described above the smoke sample is to be extracted from the
gaseous emission sample line (or probe) and transported to the smoke
sample collection system. Review of the evidence presented in section 5
indicated that in some cases it is quite difficult to supply a represen-
tative gaseous emission sample. Investigations have shown that for a
mixed flow turbofan engine the 12 point integrating, manifolded probe
-------
-10-
will not yield a representative smoke sample. In this study, the
investigator relied on a multipoint rectiliner traverse of the exhaust
plume to provide a representative sample for certification of the
engine.
There are several factors contributing to the non-representativeness
of the smoke sample.
A. Investigation has shown that the sample line connection
(Tee) where the smoke sample is extracted from the gaseous emissions
sample can have an adverse effect on the smoke level (lowering the
level by as much as 4 smoke numbers when gaseous and smoke samples
are measured simultaneously).
B. If the two emission levels are determined independently with a
12 point sample probe, the requirement that 80% of the pressure
drop in the probe assembly must be taken at the probe orifices,
requires the probe,orifice diameters to be so small that problems
are experienced with orifice blockage from carbon particulates.
C. A sample probe designed to provide a representative sample from
the carbon balance technique will not ensure representativeness of
the smoke sample.
Confusion is added to the smoke sampling technique by the schematic
drawing of the smoke sampling system presented in the regulations 40 CFR
Part 87, Figure 5. It shows a separate and unique sampling probe and is
not interconnected to the gaseous system as shown in the gaseous emissions
system schematic, Figure 3.
To reduce the inherent inaccuracies of this system and to provide
assurance of a representative smoke sample, revisions to the smoke
measurement procedure should be considered. A system such as that
recommended by the SAE in ARP 1179, which requires a multipoint traverse,
would provide a data base. From this data a unique single point or
intergrating smoke probe of less than 12 points could be designed and
representativeness shown prior to smoke certification testing. This
system would not rely upon the gaseous sampling system and avoid its
inherent disadvantages for smoke sampling.
7.0 Effect of Variations in Ambient Conditions
Variations in ambient conditions (te>mperature humidity and pressure)
have been shown to effect the pollutant constituent levels HC, CO, C0_,
and NOx of gas turbine engines. Concern for thisxproblem developed
several years ago when Marchionna and Lipfert identified corre-
lations of emission (NOx) changes to combustor inlet temperature and
humidity. The EPA addressed this problem in the regulations by stating
"Projects are being started to define these relationships so as to allow
introduction of suitable correction factors as soon as possible."
-------
-11-
The program undertaken by the EPA to define these relationships for
HC, CO, NOx, and Smoke is a three task effort. The first task which is
nearing completion is the full scale engine testing under controlled
temperature, humidity, and pressure conditions. The second task which
presently is being contracted is combustor rig testing of engine combustors
under controlled conditions. The third task is the analysis of generated
and existing data to provide determination and formulation of these
effects. Results of this program are expected in January of 1977.
Results Lo date have not indicated whether one set of correction
factors will uniformly apply to all classes and types of engines or
whether a different and unique set will have to be developed for each.
7.1 Ambient Conditions Effect on NOx
Of the various parameters affecting NOx concentrations, combustor
inlet air temperature has been found to cause the largest variation.
The effect of humidity as the second largest with both temperature and
humidity effects following the exponential form:
r-T t .. i\ T.T r A • >. j\ c (Temp, or Humidity)
ET (corrected) - El (indicated) e
The effect of pressure on NOx concentrations is relatively mild compared
to the temperature effect and follows the kinetic reaction prediction of
approximately the square root of the pressure:
EI(corrected) - Elindicated)
NOx p
indicated
Considerable information has been generated for the effects of
ambient conditions on NOx by government and private agencies. Although
this work has been done under uncontrolled conditions it does verify the
forms of the equations given above. The remaining factors to determine
are the constants to be used in the equations and whether a single
equation will apply to all engine types.
7.2 Ambient Condition Effect on HC, CO, and Smoke
The EPA's ambient effects program (described above) is essentially
the first formal effort conducted toward the understanding of the effects
of ambient conditions on HC, CO and Smoke. The effects on HC and CO do
not lend themselves to theoretical chemical kinetic analysis as is the
case for NOx because of the complexity of the reactions. Also since HC
and CO are predominate at low power (Idle) where the combustion efficiency
is low and engine variability is high it has been very difficult to
analyze and statistically treat the existing data to determine the
effects. The effects of ambient conditions on smoke is unknown but is
expected to be small.
-------
-12-
7.3 Magnitude of Variability
The magnitude of the effect of changes in ambient conditions on
exhaust gas constituent emission levels is the largest single sources of
variability in gas turbine engines. In a study conducted for the
EPA under controlled ambient test conditions it was shown that with
increasing temperature there were decreases in hydrocarbons, and carbon
monoxide and increases in the oxides of nitrogen. With increasing
humidity there were increases in hydrocarbons and carbon monoxide and a
decrease in the oxides of nitrogen.
8.0 Background Concentration Effects
Background concentration levels (ambient air qualn ty) of the gaseous
constituents being measured are of concern and do contribute to the
total engine emission level measured. The process of certifying an
engine to a specific emission level should rightly consider these effects
but at the present time a procedure has not been determined. The total
contribution to the emission EPAP level from the background concentrations
is shown below. The background levels used in these calculations were
derived from typical levels experienced during testing at various manufac-
turers. The background levels are:
Hydrocarbon = 3 ppM
Carbon monox]de = .06 vol %
Oxides of nitrogen = .5 ppM
Background Effects on EPAP
_ hC EPAP CO FP.P NOx EPAP
•f.
Engine As read Corrected fiZ As read Correcced fiX As read Corrected C.Z
J78D-11(13) 2.82 2.73 -3.19 16 10 16.65 +3.42 8 99 9 20 +2.34
JT3D-11<13) 2 06 1 95 -5 26 15.60 16 32 +3 32 7 75 7.89 +3 81
/TIN
TPF 331V ' 2.289 2.16 -5 70 16 48 16 85 +2.35 10 07 10 12 + .5.
In the cases presented above the background levels were assumed to pass
through the combustor unreacted and thus contribute directly to the
total Exhaust Emission level.
Determination of the background level contribution for HC and CO to the
total emission level is difficult to access because these constituents are
partially consumed or reacted during the combustion process in the
engine. The determination of this effect would be further complicated
in a mixed flow turbofan engine where only a portion of the background
constituents would pass through the core of the engine.
To study this effect in greater detail and to provide a correction
technique for each engine type, a complex program of uniformally intro-
ducing a known concentration of pollutant at the engine inlet and comparing
-------
-13-
the resultant exhaust gas pollutant level to that of a different (prefer-
ably zero) background level, would have to be run.
At the present time, it is expected that background constituent
levels would have an effect on the total pollutant levels. The certify-
ing agent should be cautioned as to their possible influence. This
effect would be particularly influential where the possibility of recirc-
ulation or contamination from other test cells exist.
9.0 Calibration Gas Accuracy
The EPA regulations require that instrument calibration gases used
for emission measurements have an accuracy of +_ 2% of the true value.
These working gases are third level gases with NBS primary standards (+
1% accuracy) as first level and laboratory reference gases as the second
level. There are several problems related to calibration gases which
effect accuracy:
A. A number of standard reference gases have been developed by the
National Bureau of Standards (NBS) for use in the automotive industry,
some of which are also suitable for gas turbine exhaust measurements.
To satisfy the requirements of the regulations, there is a need for
additional standard reference materials such as NO in nitrogen,
intermediate levels of C0_ in air, higher concentrations of CO in
nitrogen, and propane in air.
B. Difficulty is experienced in maintaining certain reference
gases with confidence of assay, particularly due to instability in
the cylinder. This particularly becomes a serious problem when
attempting to purchase gases of the high purity and accuracy of
assay.
C. The regulations presently specify an accuracy figure without
stating a required confidence interval. To properly specify the
accuracy of a gas a confidence interval should be stated.
10.0 Variability and Repeatability of Engine Parameters
Variation in engine operation such as approach to pox\?er and run to
run changes coupled with inaccuracies in the critical instrumentation
parameter (thrust, or horsepower) do effect the EPA thrust based emission
level parameter (EPAP). Other instrumentation errors effect the calcula-
tions for representativeness by the carbon balance technique. These are
engine air flow and fuel flow. In general these instrumentation errors
include both precision and bias inaccuracies.
The thrust measurement accuracy in terms of a 50,000 pound measure-
ment system might reach 500 pounds (1% of full scale). The rate of
change of NOx emissions with thrust and the uncertainties in the NOx
-------
-14-
measurement itself show that 500 Ibs. thrust uncertainty is a negligible
contributor to EPAP NOx uncertainty. The case of CO and HC, however,
is quite different. In the idle region, the only region contributing
significantly to CO and HC, EPAP, 500 Ibs. uncertainty in thrust setting
will result in a 12% error in carbon monoxide in EPAP and 35% error in
Total Hydro-carbon EPAP. Running several points in the region of idle
and picking off the true idle thrust point from a curve fit would reduce
much of the uncertainty contributing to the EPAP uncertainty.
Analysis of the engine air flow and fuel flow indicate that the
uncertainties at high power are usually on the order of 1- 2 as this is
the design point of the engine. In the region of idle, however, air
f]ow is usually not clearly delineated in the engine testing as it is in
"off-design" operation. The air flow measurement uncertainties of less
than 0.5% may result in 5-10% air flow uncertainties. Fuel flow uncertainty
remains relatively constant throughout the engine operating range and is
insignificant when compared to air flow uncertainty particularly at idle
pox<7er.
The effect of approach to power and run to run variability, can be,
illustrated from a study conducted for the EPA by Scott Research Labs
The approach to power did not have a significant effect on the CO and
NOx emission indicies but did show THC emission indicies to have a heavy
dependence on the approach to power (Figures 5, 6, 7). Run to run
variation showed that the greatest uncertainty also to be in the THC
emission indicies, where the coefficient of variation was 5.37%. The
coefficients of variation for CO and NOx were 1.5% and 3.5%, respectively.
-e
11.0 System Accuracy
Assessing gas turbine emission levels requires a knowledge of the
measurement accuracy. The assessment of emission measurement uncertainty
requires properly combining the uncertainties in both emission measurement
and related gas turbine performance parameters to obtain a total uncertainty.
Emission measurement uncertainty results from calibration gas
analysis uncertainty, instrument precision, sample error, the effects of
ambient factors such as temperature, humidity and pressure and back-
ground pollutant levels. These are illustrated in figure 8. Also con-
tributing to the total uncertainty is the particular constituent being
analyzed and the absolute magnitude of that constituent. The related
gas turbine parameters whose uncertainties have previously been considered
include thrust and fuel-air ratio.
At the present time a comprehensive understanding of the individual
uncertainty and its contribution to the total measurement uncertainty is
poorly understood. This is partially because the emission measurement
-------
-15-
systera, which represents an advanced state of the art in technology, is
a combination of subsystems, associated mechanisms and components which
react and interact to form a composite system of high complexity.
Further detailed analysis of the system and its components contribution
to the total uncertainty is needed to provide a comparative assessment
of variability within the system and from system to system. Knowledge
of within the system uncertainty and its contribution to the promulgated
EPAP level would provide the certifying agent a better understanding of
the systems ability to analyze a given constituent and whether the
system is supplying a representative exhaust gas sample for analysis.
An understanding of system to system uncertainty is important because a
standardized comparative base must be established for verification of
the certifying agent's calculated EPAP level. It is therefore recommended
that consideration be given to conducting a comprehensive study into the
component and system accuracy and its influence on the constituent EPAP
level.
The follox^ing is an assessment of the basic sources of variability,
their approximate range of inaccuracy, and their approximate contribution
to the systems total EPAP inaccuracy. It must be remembered that this
is only an approximate range of inaccuracy and that dependent upon
analysis of the individual systems these levels may vary greatly.
Contribution to Inaccuracy
Range of Contribution to
Inaccuracy - % Total Inaccuracy-%
Sample Error 9
Sample line and Probe 0 to 25
External Influencing Factors
Background Pollutant
levels 0 to 5
Effects of Changes in
Ambient Conditions 0 to 50
Engine Effects 5
Performance Parameters
Thrust 0 to 25
Fuel to Air Ratio 0 to 10
Run to run 0 to 5
Approach to Power 0 to 5
Instrumentation Error
Calibration Gases 0 to 6 1.7
Analytical Instruments 0 to 4 2
Expected System Inaccuracy
(Not considering Ambient
Conditions) 5 to 25 10.4
-------
-16-
As can be seen from the data presented above engine inaccuracies and
inaccuracies created in the analytical system are small when compared to
the sample error. The sample error is largely due to the fact that a
non-uniform concentration of exhaust is sampled at a discrete number of
points.
(9)
Data presented by Dieck , table II, illustrates typical levels of
total precision error associated with the CO, HC, and NOx measurements
for a single point sample for T2, T3, and T4 class engines. (T2 = high
bypass fan engine; T3 = turbofan engine; T4 = mixed flow fan engine).
Total precision error is a combination of sample error and instrument
precision error given by the equation:
2 2 1/2
Precision Error = [ ( Sample error) -I- (Instrument precision error) ]
Table II. Total Precision Error, %
Low Power Low Power High Power
Engine Class CO Precision THC Precision NOx Precision
T2 +32 +79 + 22
T3 +42 +53 + 17
T4 +61 +59 + 64
The precision error is 2S over the 95% confidence interval and it
means, for CO for example, that 95% of the time, single point samples of
CO at low power in the T2 class engine tailpipe will be within +_ 32% of
the mean.
The term usually involved in emission evaluation is the mean of
several measurements. Variation in that mean is to be expected and may
be estimated by a quantity called the precision of the mean. The
precision of the mean may be defined as the precision of the data
divided by the square root of the number of data points sampled:
„ ,. Prec. of data
Prec. of mean =
1/2
(No. of single point samples)
For example, the precision of the mean for the T2 class engine CO
data in Table I would vary by the square root of the number of single
point samples as shown in Figure 9.
The dependence of the precision of the mean on the number of
sampling points is clearly shown as is the relatively minor contribution
of instrument precision to the overall precision of the mean. (Although
both the precision of the mean, and the instrument precision contribut-
ion to the uncertainty decrease as the number of sampled data point
increase, the uncertainty in the average will never be less than the
uncertainty in the calibration gases, called calibration gas bias.)
Except for the magnitude this comparison between instrument and sample
error for CO is generally applicable for all constituents and all engine
testing where data was taken fVPJP various points in the engine exhaust
as previously shown by Nelson
-------
-17-
References
(1) Moses, C. A., and Stavinoha, L. L., "Gas Chromatographic Analysis
of Exhaust Hydrocarbons from a Gas Turbine Combustor," presented at
the Western States Section Meeting of the Combustion Institute,
Palo Alto, California, October, 1975, paper no. 75-17.
(2) Conkle, J. P., Lackey, W. W., Martin, C., and Miller, R. L., "Organic
Compounds in Turbine Combustor Exhaust," USAF School of Aerospace
Medicine, Brooks AFB, Texas, 1975.
(3) Souza, A. F., and Reckner, L. R., "Variability in Aircraft Turbine
Engine Emission Measurements," Environmental Protection Agency Report
No. EPA-460/3-74-006, January 1974.
(4) Souza, A. F., "Further Investigation Into the Causes of Variability
in Aircraft Turbine Engine Emission Measurement," EPA Report, to be
published, Dec., 1975.
(5) "Procedure for the Continuous Sampling and Measurement of Gaseous
Emissions from Aircraft Turbine Engines," APR 1256, Society of Automotive
Engines, January, 1971.
(6) JT8D-17 Certification Program, Letter to the Department of Transportation
from Mr. A. W. Oberg, Pratt and Whitney Aircraft, East Hartford, Conn.,
dated, December 14, 1973.
(7) Marchionma, N., "Effect of Inlet Air Humidity on the Formation of
Oxides of Nitrogen in a Gas Turbine Combustor," NASA Report TM-X-68209,
March 1973.
(8) Lipfert, F. W. , "Correlation of Gas Turbine Emission Data," ASME
Paper 72-GT-60, 1972.
(9) Dieck, R. H., and Elwood, J. H., "The Assessment of Emission
Analysis of Accuracy," Journal of the Air Pollution Control Association,
Volume 25,' No. 8, p. 845, August-1975.
(10) Nelson, A. W., "Exhaust Emission Characteristics of Aircraft Gas
Turbine Engines," ASME, San Francisco, Calif., March 26, 1972, Paper No.
72 GT-75.
(11) Elwood, J. H., Pratt and Whitney Aircraft, East Hartford, Conn.,
Personal Communication.
(12) Matuschak, P., Garrett Airesearch Manufacturing Co. of Arizona,
EPA Report to be published, Jan. 1976.
(13) McAdams, H. T., "Analysis of Aircraft Exhaust Emission Measurements,"
Calspan Corporation, Buffalo, N.Y., CAL NO. NA-5007-K-1&2, Nov. 1971.
-------
-18-
APPENDIX
Figure 1
Organic Compounds in Turbine
Combustor Exhaust Using JP8 Fuel
This results was extracted from data collected and reported by
Dr. James P. Conkle of the USAF School of Aerospace Medicine, Environ-
mental Science Division, Brooks AFB, Texas.
Analysis of Hydrocarbon compounds present in the gas turbine exhaust
of a T-56 combustor using JP8 fuel. JP8 fuel is very similar to commercial
ASTM-D1655-Jet A.
Compound Family
Identified
Total* Carbon Numbers
Up to 8 9 to 13 13 and up
.0358
Paraffins
Olefins 1
Diolef ins
Naphthalenes
Aroma tics
Aldehydes
Alcohols
»..
Ketones
Ethers
Esters
Nitrogen Containing
Peroxides
.6207
.0586
.046
.104
.635
.331
.118
.3258
.0761
.0032
.0119
.0023
.1489
1.0136
.046
.104
.58
.331
.118
.3137
.0761
-
.0119
.0023
.436
.045
-
-
.055
-
-
.0121
-
-
-
-
3.3326
% of Total HC
Carbon Numbers up to 8
Carbon Numbers up to 12
Carbon Numbers 13 and up
2.7455
= 82.4%
= 98.8%
= 1.17%
.0032
.5481
.0390
*PPM as Hexane
-------
-19-
-OCi*,' ^..^
47T^ \>J
co9CO
FIGURE 2
NOZZLE TRAVERSE - P&W JT8D IDLE POWER
CO(ppm)
N0x(ppm)
-------
-20-
THC(ppm)
co2(%)
NOZZLE TRAVERSE -
/
CO(ppn)
\
NO ppm
FIGURE 3
P&W Ji'SD APPROACH POWER
-------
-21-
' V ^s>
/-Y-^ >.
THC(pp:n)
CO(ppm)
co2(%)
NO (ppm)
FIGURE 4
NOZZLE TRAVERSE - P&W JT8D MAXIMUM CONTIGUOUS POWER
-------
-22-
:24
22 .
Effect of
Approach to Power
THC
20
18
f
&
O
16
14 .
10
8 .
6 .
2 .
0
Increasing Power
Decreasing Power
Figure 5
Engine Pressure Ratio
-------
Approach to Power
31
X\
20 ,
\°
NO
\
\
101 \
9 •
•\
\
\
6 '
5
X
2! "*<
i
i \
X
u*x
\o
O Increasing Power \C
.64 °
X Decreasing Power
.4
.3
.2
Figure 6
1.0 1.2 1.4 1.6 1.8 2.0 2.2
Engine Pressure Ratio
-------
20J
Approach to Power
NOx
O
o
18-
16-
14-
12.
/\
o
/
o
10,
6.
o
/o
Q Increasing Power
X Decreasing Power
Figure 7
1.0
1.2
1.4
1.6
1.8
2.0
2.2
Engine Pressure Ratio
-------
Background
Pollution Level
Ambient
Conditions
Sources of Uncertainty
Engine
Parameter
Sample Probe
Sample Line
o o o o
Analytical
System
Calibration
Gases
Figure 8
-------
10-,
Low Power Carbon Monoxide
•rl
tl
cd
OJ
Precision of Mean
for Traverse Data
g
o
M
(U
fM
M
M
W 3
(1)
i-H
P4
2.
Instrument Precision
Calibration
10 20 30 40 50 60 70 80 90
Number of Individually Sampled Points in Tailpipe
Figure 9
Reduction of sampling error with increased sampling
100
------- |