EPA-460/3-77-019
September 1977
DETERMINATION OF EFFECTS
OF AMBIENT CONDITIONS
ON AIRCRAFT ENGINE EMISSIONS
DATA ANALYSIS AND CORRECTION
FACTOR GENERATION
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Air and Waste Management
Office of Mobile Source Air Pollution Control
Emission Control Technology Division
Ann Arbor, Michigan 48105'
-------
EPA-460/3-77-019
DETERMINATION OF EFFECTS
OF AMBIENT CONDITIONS
ON AIRCRAFT ENGINE EMISSIONS
DATA ANALYSIS AND CORRECTION
FACTOR GENERATION
by
Paul J. Donovan, William R. Fairchild, and Kenneth W. Graves
Calspan Corporation
P.O. Box 235
Buffalo, N.Y. 14221
Contract No. 68-03*2159
EPA Project Officer: Thomas Cackette
Prepared for
ENVIRONMENTAL PROTECTION AGENCY
Office of Air and Waste Management
Office of Mobile Source Air Pollution Control
Emission Control Technology Division
Ami Arbor, Michigan 48105
September 1977
-------
This report is issued by the Environmental Protection Agency to report
technical data of interest to a limited number of readers. Copies are
available free of charge to Federal employees, current contractors and
grantees, and nonprofit organizations - in limited quantities - from the
Library Services Office (MD-35), Research Triangle Park, North Carolina
27711; or, for a fee, from the National Technical Information Service,
5285 Port Royal Road, Springfield, Virginia 22161.
This report was furnished to the Environmental Protection Agency by
the Calspan Corporation, P.O. Box 235, Buffalo, N.Y., in fulfillment
of Contract No. 68-03-2159. The contents of this report are reproduced
herein as received from Calspan Corporation. The opinions, findings,
and conclusions expressed are those of the author and not necessarily
those of the Environmental Protection Agency. Mention of company or
product names is not to be considered as an endorsement by the Environ-
mental Protection Agency.
Publication No. EPA-460/3-77-019
u
-------
ABSTRACT
This report presents a set of correction factors for variations
in turbine aircraft HC, CO, NOX and smoke emissions due to non-standard
day ambient temperature, pressure and humidity developed for the United
States Environmental Protection Agency, Ann Arbor, Michigan, under EPA
Contract No. 68-03-2159. These correction factors are based on data from
three EPA-sponsored full-scale engine tests, two EPA-sponsored combustor rig
tests, and additional data solicited from industry sources. Key corre-
lating parameters in this analysis were combustor inlet temperature,
combustor inlet pressure, and ambient humidity. The correction factors
have been developed using a multiple least squares regression analysis
approach using functional emissions models based upon theoretical con-
siderations and an extensive review of current ambient effects literature.
Emphasis has been placed upon relating correction factor coefficients
within a general engine class to various operating characteristics of each
individual engine.
iii
-------
TABLE OF CONTENTS
Section Title Page No.
ABSTRACT iii
LIST OF TABLES vii
LIST OF FIGURES ix
LIST OF APPENDICES xiii
1 SUMMARY AND CONCLUSIONS 1~1
2 INTRODUCTION 2-1
2.1 EPA-Sponsored Data 2-2
2.2 Contributed Data 2-3
2.3 Engine and Test Descriptions 2-4
2.3.1 TPE331-5-251 2-5
2.3.2 GTCP85-98CK APU 2-6
2.3.3 ALF502 Full Scale Engine 2-7
2.3.4 ALF502 Combustor Rig 2-7
2.3.5 CFM56 Combustor Rig 2-8
3 DEFINITIONS 3-1
4 SUMMARY OF AMBIENT EFFECTS LITERATURE 4-1
4.1 Oxides of Nitrogen 4-2
4.2 Hydrocarbons 4-5
4.3 Carbon Monoxide 4-7
4.4 Smoke Number 4-8
5 CORRECTION FACTOR APPROACH 5-1
5.1 Theoretical versus Empirical Emissions Models . . . 5-1
5.2 Correlating Parameters - Ambient versus
Combustor Inlet 5-1
5.3 Simple versus Complex Emissions Models 5-4
5.4 Data Variability 5-7
5.4.1 Test-to-Test Variability 5-8
5.4.2 Engine-to-Engine Variability . 5-9
5.5 Regression Models and Correction Factors 5-9
6 INDIVIDUAL ENGINE EMISSION MODELS 6-1
v
-------
TABLE OF CONTENTS (Cont.)
Section Title Page No.
6.1 CFM56 Combustor Rig - Raw Data Plots 6-1
6.2 CFM56 Combustor Rig - Summary Statistics by Mode . . 6-2
6.3 CFMS6 Combustor Rig - Regression Summary 6-2
7 GENERALIZED CORRECTION FACTORS . . 7-1
7.1 Oxides of Nitrogen 7-4
7.1.1 Temperature Correction Factor 7-4
7.1.2 Humidity Correction Factor ... 7-8
7.2 Hydrocarbons 7-9
7.3 Carbon Monoxide 7-11
7.4 Smoke Number 7-12
7.5 Correction Factor Sensitivity Analysis 7-14
7.6 Correction Factor Performance Analysis ....... 7-17
8 REFERENCES 8-1
9 TABLES . . . 9-1
10 FIGURES 10-1
vi
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
LIST OF TABLES
Title Page No.
Engine/Combustor Rig Summary - EPA-Sponsored Data 9-3
Industry Contributed Data - Selected Experimental
Design Features 9-4
Ambient Test Conditions - TPE331-5-251 and GTCP85-98CK APU. 9-5
TPE331-5-251 Engine Load Conditions 9-6
GTCP85-98CK APU Engine Load Conditions 9-7
Ambient Test Conditions - ALF502 Full Scale Engine .... 9-8
Ambient Test Conditions - ALF502 Combustor Rig 9-9
Ambient Test Conditions - CFM56 Combustor Rig 9-10
CFM56 Combustor Rig Replicate Analysis 9-11
TPE331-5-251 Full Scale Engine Replicate Analysis 9-12
Test-to-Test Variability - Mean Idle Coefficient of
Variation by Engine 9-13
CFM56 Combustor Rig - Summary Statistics by Mode 9-14
CFM56 Combustor Rig - HC Regression Summary 9-15
CFM56 Combustor Rig - CO Regression Summary ....... 9-16
CFM56 Combustor Rig - NOX Regression Summary . 9-17
CFM56 Combustor Rig - Smoke Regression Summary 9-18
NOX Correction Factor Coefficients - Tabular Summary . . . 9-19
NOX Correction Factor Coefficients - Summary Statistics . . 9-20
HC Correction Factor Coefficients - Tabular Summary ... 9-21
HC Correction Factor Coefficients - SuTMft&ry Statistics . . 9-22
CO Correction Factor Coefficients - Tabular Summary . . . 9-23
vii
-------
LIST OF TABLES (cont.)
Table No. Title Page No.
22 CO Correction Factor Coefficients - Summary Statistics . 9-24
23 Smoke Number Correction Factor Coefficients - Tabular
Summary 9-25
24 Smoke Number Correction Factor Coefficients -
Summary Statistics 9-26
CI NOX Regression Coefficients vs. Model Formulation -
GE CFM56 Combustor Rig, Idle to Takeoff C-7
C2 NOX Regression Coefficients vs. Model Formulation -
Pratt & Whitney JT9D-7A Pilot Lot Data, Idle to Takeoff . C-8
D1 ALFS02 Rig-Engine Correlation Analysis -
Regression Summary . D-6
D2 ALF502 Rig-Engine Correlation Analysis -
Coefficient Confidence Bound Summary D-7
viii
-------
LIST OF FIGURES
Figure No. Title Page No.
1 Variation of Oxides of Nitrogen (NOX) EI with Ambient
Temperature - ALF502 Combustor Rig, Approach Mode . .
2 Data Overview - EPA Aircraft Emissions Ambient
Effects Program
3 Burner Inlet Temperature versus Ambient Humidity -
TPE331-5-251
4 Effect of Engine Pressure Ratio and Equivalence Ratio
on NOX Temperature and Humidity Correction Factors
(from Blazowski) 10-6
5 Analytically Predicted Ambient Temperature Correction
Factors for Hydrocarbons (from Blazowski and Marzewski) . . 10-7
6 HC EI versus CO EI - CFM56 Combustor Rig @ Idle
and 1.5 Idle 10-8
7 Natural Log of HC EI versus Natural Log of CO EI -
CFM56 Combustor Rig @ Idle and 1.5 Idle 10-9
8 General Correction Factor Development Overview 10-10
9 Combustor Inlet Temperature versus Ambient Temperature -
CFM56 Combustor Rig at Idle 10-11
10 Natural Log of HC EI versus Combustor Inlet Temperature -
CFM56 Combustor Rig at Idle (Pressure and Humidity Fixed) . 10-12
11 Natural Log of CO EI versus Combustor Inlet Temperature -
CFM56 Combustor Rig at Idle (Pressure and Humidity Fixed) • 10-13
12 Natural Log of NOX EI versus Combustor Inlet Temperature -
CFM56 Combustor Rig at Climb (Pressure and Humidity Fixed). 10-14
13 Natural Log of NOX EI versus Ambient Humidity -
CFM56 Combustor Rig at Climb (Pressure and Temperature
Fixed) 10-15
14 Ambient Effects Program - Regression Model Summary .... 10-16
15 Replicate Analysis - Coefficient of Variation for CFM56
Combustor Rig HC Idle Data 10-17
ix
10-3
10-4
10-5
-------
LIST OF FIGURES (cont.)
Figure No. Title Page No.
16 Correction Factor Structure 10-18
17 HC EI versus Corabustor Inlet Temperature - CFM56
Combustor Rig, Idle to Takeoff . . 10-19
18 Natural Log HC EI versus Combustor Inlet Temperature -
CFM56 Combustor Rig, Idle to Takeoff 10-20
19 CO EI versus Combustor Inlet Temperature -
CFM56 Combustor Rig, Idle to Takeoff 10-21
20 Natural Log CO EI versus Combustor Inlet Temnerarure -
CFM56 Combustor Rig, Idle to Takeoff 10-22
21 NOX EI versus Combustor Inlet Temperature
CFM56 Combustor Rig, Idle to Takeoff 10-23
22 Natural Log of NOX EI versus Cumuustor Inlet Temperature -
CFM56 Combustor Rig, Idle to Takeoff 10-24
23 NOX EI versus Ambient Humidity - CFM56 Combustor -Rigj
Idle to Takeoff . . . . . . . . . . . . . . . . . . . . . 10-25
24 Natural Log of NOX EI versus Ambient Humidity -
CFM56 Combustor Rig, Idle to Takeoff 10-26
25 Smoke Number versus Combustor Inlet Pressure -
CFM56 Combustor Rig, Idle to Takeoff 10-27
26 Natural Log of Smoke Number versus Combustor Inlet Pressure -
CFM56 Combustor Rig, Idle to Takeoff 10-28
27 Regression Coefficient Summary - CFM56 Cbmbustor Rig . . . 10-29
28 Correction Factor Coefficient versus Engine Operating
Parameter - Development Approach . ... 10-30
29 Correction Factor Nomenclature ....... 10-31
30 T3MEAN versus Rated Pressure Ratio PR -
Ambient Effects Data Base 10-32
31 NOX Combustor Inlet Temperature Coefficient T3C0EF
versus Combustor Inlet Pressure P3MEAN . . . . . . . . . . 10-33
x
-------
LIST OF FIGURES (cont.)
Figure No. Title Page No.
32 NOX Emissions Index as a Function of Compressor Discharge
Temperature - Production Engines (Reference 13) 10-34
33 NOX Temperature Coefficient versus Combust or Inlet
Pressure P3MEAN - Confidence Bounds 10-35
34 NOX Combustor Inlet. Temperature Coefficient T3C0EF
versus Rated Engine Pressure Ratio PR 10-36
35 NOX Combustor Inlet Temperature Coefficient T3C0EF
versus Mean NOX Emission Index NOXMEAN 10-37
36 Mean NOX Combustor Inlet Temperature T3MEAN versus
Mean NOX Emission Index NOXMEAN 10-38
37 NOX Emission Index versus Combustor Inlet Temperature -
Uncontrolled and Controlled Engines 10-39
38 HC Combustor Inlet Temperature Coefficient T3C0EF
versus Engine Idle Pressure Ratio IPR . 10-40
39 CO Combustor Inlet Temperature Coefficient T3COEF
versus Engine Idle Pressure Ratio IPR 10-41
40 Smoke Number Combustor Inlet Pressure Coefficient P3C0EF
versus Rated Engine Pressure Ratio 10-42
41 Smoke Number Combustor Inlet Pressure Coefficient P3C0EF
versus Mean Smoke Number SMKMEAN 10-43
42 NOX Temperature Correction Factors 10-44
43 NOX Humidity Correction Factors 10-45
44 HC Temperature Correction Factors 10-46
45 CO Temperature Correction Factors 10-47
46 Smoke Pressure Correction Factors 10-48
47 Correction Factor Sensitivity Analysis -
NOX Temperature Correction 10-49
48 Correction Factor Sensitivity Analysis -
NOX Humidity Correction 10-50
xi
-------
LIST OF FIGURES (cont.)
Figure No. Title Page No.
49 Correction Factor Sensitivity Analysis -
HC Temperature Correction . . . 10-51
50 Correction Factor Sensitivity Analysis -
CO Temperature Correction • • • ....... 10-52
51 Correction Factor Sensitivity Analysis -
Smoke Pressure Correction 10-53
52 NOX Ambient Effects Correction Summary 10-54
53 HC Ambient Effects Correction Summary .... 10-55
54 CO Ambient Effects Correction Summary 10-56
55 Smoke Ambient Effects Correction Summary 10-5 7
56 Hyp.otjietical Emission Index Correction ..... . . . . 10-58
CI CFM56 Variance Map - Equation: LNN0X=f(T3) C-9
C2 CFM56 Variance Map - Equation: LNN0X=f(T3, HAMB) . , . . . C-10
D1 Graphical Interpretation t>f ALF502 Rig-Engine Correlation . . D-8
F1 Uncorrected HC Emissions Index versus Combustor Inlet
Temperature - CFM56. Combustor Rig, Idle , F-3
F2 Corrected HC Emissions Index versus Combustor Inlet
Temperature - CFMS6 Combustor Rig, Idle . . F-4
F3 Uncorrected CO Emissions Index versus Combustor Inlet
Temperature - CFM56 Combustor Rig, Idle and 1.5 Idle .... F-5
F4 Corrected CO Emissions Index versus Combustor Inlet.
Temperature - CFM56 Combustor Rig, Idle and 1.5 Idle .... F-6
F5 Uncorrected NOX Emissions Index versus Combustor Inlet
Temperature - CFM56 Combustor Rig, Idle to Takeoff . . . . F-7
F6 Corrected NOX Emissions Index.versus Combustor Inlet
Temperature - CFM56 Combustor Rig, Idle to Takeoff .... F-8
G1 Measured Fuel Flow versus Ambient Pressure -
CFM56 Combustor Rig, Idle to Takeoff G-3
G2 Corrected Fuel Flow versus Ambient Pressure -
CFM56 Combustor Rig, Idle to Takeoff G-4
xii
-------
LIST OF APPENDICES
Appendix No. Title Page No.
A DATA BASE SUMMARY A-l
B COMBUSTOR INLET CONDITIONS RELATED TO
AMBIENT CONDITIONS B-l
C MULTICOLLINEARITY AND AIRCRAFT EMISSIONS DATA ... C-l
D COMBUSTOR RIG - FULL-SCALE ENGINE CORRELATION ... D-l
E COMPARISON OF TWO LINEAR BIVARIATE
REGRESSION LINES E-l
F SAMPLE EMISSION INDEX CORRECTION F-l
G COMPUTATION OF EPA EMISSION PARAMETER (EPAP)
FROM EMISSION INDEX AND SPECIFIC FUEL CONSUMPTION . G-l
xiii
-------
SUMMARY AND CONCLUSIONS
The effect of variations in ambient temperature, pressure
and humidity on hydrocarbon (HC), carbon monoxide (CO) and
oxides of nitrogen (NOX) emission indices as well as. on EPA
smoke number (SMOKE) have been investigated and a set of
general purpose correction factors developed.
The key correlating parameters in this analysis were combustor
inlet temperature, combustor inlet pressure, and ambient
specific humidity.
A regression analysis approach guided by a review of current
ambient effects literature and theory indicates that the
following emissions models can be used to predict the variation
in aircraft emissions due to non-standard day ambient conditions:
Natural Log HC EI = f(Combustor Inlet Temperature)
Natural Log CO EI = f(Combustor Inlet Temperature)
Natural Log NOX EI = f(Combustor Inlet Temperature, Ambient Humidity)
Natural Log Smoke Number = f(Natural Log Combustor Inlet Pressure)
While various alternative correlating variables could have been
chosen, it was felt that the above variables adequately describe
the major portion of aircraft emissions variation due to ambient
conditions required by a general correction factor system.
The functional forms of the proposed emission index correction
factors are:
-------
HC = HC » bT3C0EFhC * *T3REF " T3MEAS>
hcCORRECTED hcMEASUR ED * 0
co _ co #aT3COEFCO * (T3REF •T3MEAS*
CORRECTED MEASURED 6
NOX NOX # eT3 coefnox *'T3ref T3meas'»eHAMBC0EFNox *'hambref -HAMBmeas)
NOXCORRECTED nuxMEASURED 8 8
( P3pc c \ P3COEFSMOKE
SMOKEcorrected= SMOKEmeasured »lJ
where T3 = Combustor Inlet Temperature
P3 = Combustor Inlet Pressure
T3C0EF = Combustor Inlet Temperature Coefficient
HAMBCOEF = Ambient Specific Humidity Coefficient
P3C0EF = Combustor Inlet Pressure Coefficient
Ref = Reference Value
Meas = Measured Value
6)
The temperature, pressure and humidity correction coefficients
for a particular engine are determined using the rated (PR)
and idle (IPR) pressure ratios:
T3C0EF
HC
T3C0EF
CO
T3COEFnox
HAMBCOEF
NOX
= f(IPR)
= f(IPR)
= f(PR)
= CONSTANT = -19.0
PSCOEF-,,^ = CONSTANT = -1.0
SMOKE
Applicable Modes
IDLE
IDLE
IDLE TO TAKEOFF
IDLE TO TAKEOFF
IDLE TO TAKEOFF
1-2
-------
7) The correction coefficients for newer technology low emissions
designs frequently vary from older design engines. In general,
newer design engines exhibit decreased NOX temperature sensitivity
(see Section 4 for discussion of "sensitivity") and increased HC
and CO temperature sensitivity when compared to older technology
engines of similar idle and rated pressure ratios.
8) The NOX combustor inlet temperature coefficient (T3C0EF^q^) was
found to increase approximately linearly with rated engine
pressure ratio.
HC
9) The HC and CO combustor inlet temperature coefficients (T3C0EF
and T3C0EF^,q) were found to increase in a negative direction
with idle pressure ratio.
10) The high variability in measured smoke data makes the application
of a general purpose smoke correction factor tenuous. While
reasonably consistent trends were found between the pressure
correction coefficients for various engines, only a small
reduction in variability was observed when measured smoke data
were corrected to standard day conditions.
11) A comparison of uncorrected and corrected emissions data
indicates the proposed correction factor approach should
provide the following expected reduction in emission index
variability:
Expected Reduction
Emission Index Variability
Pollutant Percent
NOX EI 70
HC EI 58
CO EI 63
SMOKE NUMBER 37
1-3
-------
2. INTRODUCTION
The Clean Air Act Amendments of 1970 specified that the Federal
Aviation Administration (FAA) promulgate regulations enforcing aircraft
engine standards established by the Environmental Protection Agency (EPA).
During the period since the enactment of this legislation, EPA has estab-
lished standards for hydrocarbons (HC), carbon monoxide (CO), oxides of
nitrogen (NOX), and smoke number applicable to variety of both new and
in-use aircraft engines. A description of these standards and required test
procedures is provided in Reference 1 and subsequent amendments.
In recognition of the influence of ambient conditions on aircraft
engine emissions, EPA has conducted a program aimed at evaluating these
effects and at formulating correction factors for adjusting measured
emissions to standard day conditions (ambient temperature = 59°F, ambient
pressure = 29.92 in Hg, and specific ambient humidity = 0.00634 lb H^O/lb
dry air). The need for suitable ambient effects correction factors is
demonstrated by Figure 1 which presents the variation of oxides of nitrogen
(NOX) emission index for the ALF502 combustor rig operating at approach power
versus ambient temperature at a fixed ambient pressure of 29.92 in Hg
and a fixed specific ambient humidity of .0037 lb H20/lb dry air. As seen
in this figure, NOX emissions increase from 4.9 lb NOX/1000 lb fuel at an
ambient temperature of 19 °F to 6.1 lb NOX/1000 lb fuel at an ambient
temperature of 85 °F. This 25% variation difference in NOX emission index,
due solely to ambient temperature, clearly points out the desirability of
developing ambient effects correction factors so that engine certification
tests can be performed under a variety of nonstandard day conditions
and the results compared to EPA emissions standards based on engine perform-
ance under standard day ambient conditions. This report summarizes the
development of such a set of ambient effects correction factors.
In this report, "standard" and "reference" day conditions are used inter-
changeably to indicate ambient temperature = 59°F, ambient pressure = 29.92
in Hg, and specific ambient humidity * 0.00634 lb H^O/lb dry air.
2-1
-------
2 .1 EPA Sponsored Data
The ambient effects program was divided into two main tasks --
data collection and data analysis. Figure 2 presents an overview of these
two tasks. The data collection task of the program can be subdivided into
four phases:
o Full Scale Engine Tests -- Controlled Ambient Conditions
o Full Scale Engine Tests -- Uncontrolled Ambient Conditions
o Combustor Rig Tests -- Controlled Ambient Conditions
o Contributed Test Data from Industry
The first phase involved the testing of two small aircraft engines
operating under controlled ambient conditions. These engines, tested by
AiResearch in Phoenix, Arizona, under contract to EPA, were the P2 class
TPE331-5-251 turboprop and the APU class GTCP85-98CK. A complete descrip-
tion of this portion of the test program can be found in the test contractor's
final reports, References 2 and 3. By testing these engines under controlled
ambient conditions, it is possible to explore the effects of a wide range of
ambient conditions not practically available by relying solely on naturally
occurring ambient variation.
The second phase of the data collection process, on the other hand,
involved testing the T1 class ALF502 full scale engine under uncontrolled
ambient conditions. These tests, also sponsored by EPA, were performed by
AVCO Lycoming Division in Stratford, Connecticut. When testing under
uncontrolled ambient conditions, ambient effects are deduced from emission
response to naturally occurring changes in ambient conditions over a period
of time. The results of limited CFM56 full scale engine testing by General
Electric, scheduled as part of this phase of the test program, were not
available for inclusion in the analysis encompassed by this report due to
test scheduling difficulties. The omission of this small quantity of data
is not expected to alter the findings of this report. Instead, it was
intended to (1) supplement the limited smoke data currently available,
(2) provide an additional source of data to validate combustor rig/engine
correlations, and (3) provide additional large jet engine emissions data.
2-2
-------
The third phase of the data collection process was performed in con-
junction with phase two outlined above. During these test programs, the T1
class ALF502 and T2 class CFM56 combustor rigs (the ALF502 was previously
tested as a full scale engine in Phase 2) were tested under a wide range of
controlled ambient conditions. The main objectives of this portion of the
program were (1) to provide a data source useful in establishing rig/engine
correlating factors, and (2) to provide an extensive ambient effects data
base on small (ALF502) and large (CFM56) gas turbine engines. The ALF502
combustor rig tests were performed by AVCO Lycoming and the CFM56 combustor
rig tests were performed by General Electric in Cincinnati, Ohio. Both these
test programs were performed under contract to EPA. Final contractors'
reports have not been issued as of the publication date of this report.
Table 1 summarizes the EPA sponsored test programs. As shown in
this table, the EPA sponsored data base represents over 900 tests on
(1) a turboprop, (2) an auxiliary power unit, (3) a small jet engine, and
(4) a large jet engine. These engines are representative of many engines
currently in use in general and commercial aviation today. Also presented in
this table is a comparison of emission levels of engines in the EPA sponsored
test program with 1979 emissions standards. As demonstrated by this portion
of Table 1, the test engines are representative of designs which are below,
or relatively close, to 1979 emissions standards.
2.2 CONTRIBUTED DATA
In order to further generalize the data base used to develop ambient
effects correction factors, EPA solicited test data from a variety of engine
manufacturers and government agencies. The purpose of this contributed data
was to supplement the EPA sponsored data base in order to provide wider input
into the correction factor development process. Appendix A briefly summarizes
the complete ambient effects program data base. Shown in this appendix are
the engine, EPA class, number of tests, data source, availability of smoke
data, power modes, and general comments. Approximately 2000 emissions tests
on 30 engines were received during the data solicitation phase of the program.
For the most part, the contributed data consists of emissions tests on several
different full scale engines of the same model type operating under uncontrolled
ambient conditions.
2-3
-------
The Pratt § Whitney JT8D and JT9D pilot lot emissions tests are representa-
tive of this type of data. In addition, uncontrolled ambient conditions were
used to extensively examine the ambient response over a period of several
months of an individual engine such as the TF30. Finally, the contributed
data includes parametric tests on combustor rigs where temperature, pressure
or humidity are varied independently in a controlled fashion in order
to investigate the independent effects of each ambient variable. Frequently,
parametric test points do not correspond to a normal engine operating mode,
such as idle or takeoff. As a result, care must be exercised when comparing
the results of analyses using parametric experiments with data gathered while
an engine is functioning in a normal operating mode. Parametric rig data,
however, is particularly useful in evaluating the impact of changing engine
operating parameters (e.g., idle pressure ratio) on correction factor
coefficients. For instance, the T56 combustor rig was operated in a
controlled environment at four simulated idle pressure ratios (IPR = 2, 3, 4, 5).
An examination of the HC, CO and NOX correction factor coefficients determined
at each idle pressure ratio provides valuable information in assessing the
impact of this particular engine operating parameter on the magnitude of
these correction factor coefficients. Table 2 briefly summarizes the salient
features of the three basic experimental designs included in the contributed
data base.
2.3 ENGINE AND TEST DESCRIPTIONS
In this section, a general description of the test matrix for each
engine in the EPA sponsored data base will be provided. In addition, any
deviation from normal engine operation resulting from either test facility
limitations (e.g., maximum airflow), test engine limitations (e.g., maximum
turbine temperature), or test procedures (e.g., incorrect parameter adjust-
ment) will be noted where these deviations could be expected to influence
the determination of ambient effects. The test description data for each
engine will include four basic types of information:
o Engine Description
o Ambient Test Conditions
o Engine Power Conditions
o Data Anomalies
2-4
-------
2.3.1 TPE331-5-251
The TPE331 series turboprop is a single-shaft gas turbine engine
that operates at essentially constant rotor speed over a range of flight
power settings. At ground idle and taxi, however, operation is reduced to
65 percent of rated engine speed to minimize noise and fuel consumption.
Taxi-idle power is 5 percent of the takeoff rating. Accordingly, for
purposes of this emission measurement program, the idle, 1.5 times idle,
and 2 times idle, test conditions were performed at 65 percent rated engine
rpm. All other power settings were operated at 100 percent rated speed.
In conformance with EPA emission standards for P2 Class (turboprop) aircraft
engines, the power settings for approach, climbout, and takeoff, are at 30, 90
and 100 percent of rated shaft horsepower. The turboprop engine cruise condi-
tion is limited by exhaust gas temperature and varies as a function of
altitude, ambient temperature, and flight speed. As related to static test,
conditions, cruise power would be 85 percent of the takeoff-power rating.
An intermediate power setting of approximately 70 percent of rated takeoff-
power was selected to represent the cruise operating condition. The normal cruise
setting is only 5 percent less power than climbout and the intermediate power
setting provided a more favorable test point distribution for data interpolation.
The TPE331 is the prime propulsion engine for business aircraft such as the
Turbo Commander 690, and Swearingen Metro Liner. The turboprop combustor
system, commonly used for engines in the 500 to 1000 eshp range, is a
reverse-flow annular configuration. The fuel injection system is composed
of 5 radial start and 10 axial main simplex fuel injectors staged through a
single flow divider.
The engine inlet ambient conditions for the TPE331 are shown in
Table 3. This table also represents the ambient test matrix for the GTCP85-98
APU. Turboprop engine emissions, for each engine inlet ambient condition,
were measured for gaseous and smoke emissions at each power setting
described in Table 4. Two replicates or a total of three data points
were provided at each power/ambient test condition.
2-5
-------
Several comments should be made concerning the humidity control system
used to simulate the varying ambient humidities required by the test matrix.
The humidification technique employed used the injection of fine water droplets
immediately upstream of the engine inlet. While this approach assured an
accurate measurement of water introduced into the engine, it was not a true
simulation of humidity since the liquid water had to absorb heat from the
engine to reach a vapor state. This process lowered the compressor discharge
temperature as "humidity" increased. In addition, total flow through the
engine was slightly increased and the air specific heat changed to a small extent.
The decrease in compressor discharge temperature (also referred to as burner
inlet temperature) with increasing humidity is illustrated for the seven TPE331
test modes in Figure 3. The magnitude of this temperature change (approximately
5-10 percent over the range of simulated humidities), however, does not substan-
tially alter either the nature of the functional relationships between the
TPE331-5-251 emission indices and the correction factor correlating parameters
(Section 5), or the magnitude of these relationships (Section 7). In addition,
the NOX temperature and humidity coefficients for the TPE331-5-251 (Table 17)
are consistent with the general coefficient trends observed for other engines
in the test program.
2.3.2 GTCP85-98CK APU
The GTCP83 series pneumatic and shaft power gas turbine engine is
designed for ground and on-board aircraft auxiliary power. The single-shaft
engine provides mechanical shaft power driving aircraft accessories such as an
alternator when airborne, provides pneumatic bleed-air power for starting main
aircraft engines and for aircraft cabin air-conditioning when on the ground,
and provides cooling air for engine lubricating oil and for the engine enclosure
at all times. Typical aircraft applications for the 85 Series APUs include the
Boeing 727 and 737 and the Douglas DC-9 commercial jetliners. The APU combus-
tion system consists of a single-can combustor with a dual orifice fuel injector,
oriented tangentially relative to the engine axis. This is a typical combustor
configuration for the 200-350 eshp size APU.
The ambient conditions at which the engine was tested were identical to
those presented in Table 3 for the TPE331-5-251. APU engine emissions, for each
engine inlet ambient condition, were measured for gaseous and smoke emissions at
each power setting described in Table 5. Two replicates or a total of three
2-6
-------
data points were provided at each power/ambient test condition. The same
humidity control limitations encountered during the testing of the
TPH331-5-251 are also applicable to the APU test program.
2.3.3 ALF502 Full Scale Engine
The ALF502 is a two-shaft, high bypass ratio, geared turbofan
engine in the 6500 lb thrust class. This engine is aimed primarily at the
commercial light transport and executive market. The combustor is a folded
annular atomizing burner with the turbine parts packaged concentrically
within it. This arrangement provides a shorter, compact engine design
and permits reduced casing temperatures.
Limited emissions tests were performed on an ALF502 full scale engine.
These tests were performed under uncontrolled ambient, conditions at seven
different thrust levels and seven ambient test conditions. Table 6 summarizes
the ambient test matrix. As shown in this table, only a limited range of
ambient conditions are present in this data.source.
2.3.4 ALF502 Combustor Rig
In addition to the full scale ALF502 engine tests discussed in
Section 2.3.3, tests were performed under controlled ambient conditions
on an ALF502 combustor rig. Table 7 summarizes the ambient test matrix and
power levels run during this phase of testing. Only the three low power
levels (idle, 1.5 idle, and approach) were used during these tests. Certain
anomalies exist in the test data on the ALF502 comhustor rig which have a
bearing on how ambient effects were deduced from this data. These anomalies
are manifested in part by higher observed HC and CO emission indices at 1.5
idle than at idle. These findings were traced by AVCO to an eleven percent
difference in airflow at idle and 1.5 times idle between the engine and rig
tests. Although AVCO could not determine conclusively the reason for this
variation, it is probable that an error in calculating the rig points was
the cause. The likely error involved a subtraction of twice the proper amount
2-7
-------
of cooling air from the total core flow at idle and the lack of subtraction
of any cooling air from the 1.5 times idle point. As a result of this
discrepancy, the ALF502 combustor rig idle and 1.5 idle data sets were
generally analyzed separately. Airflow levels at approach power agreed with
those used during the full scale engine testing.
2.3.5 CFM56 Combustor Rig
The CFM56 is a two-shaft turbofan, high bypass ratio (6:1), high
pressure ratio (25) engine in the 22,000 lb thrust class. Although this
engine has not yet been certified for use, the proposed applications include
the BAC-111, DC-9, 727, 7N7, 7X7, 707 and DC-8. The combustor type is a
straight flow annular design.
Extensive tests under controlled ambient conditions were performed
by General Electric in Cincinnati on a single CFM56 combustor rig. Table 8
presents a summary of the ambient effects test matrix which included 22
ambient conditions at each of 5 power settings. Of these 110 different test
conditions, three could not be reached due to limitations on the air heater
temperature. These were at simulated take-off conditions at 105 °F ambient
temperature and 175 grains per pound humidity, and were well outside the
operating limits of the CFM56 engine. Thus, instead of the planned 220
readings (2 replications each of 110 test conditions), there were 214 readings
for which data were obtained. In addition, the highest power conditions,
climbout and takeoff, were run at reduced pressure due to facility limitations.
2-8
-------
3. DEFINITIONS
A list of abbreviations used throughout this report is presented
below:
Abbreviation Meaning Units
TAMB Ambient Temperature Deg F
PAMB Ambient Pressure IN HG
HAMB Ambient Specific Humidity lb t^O/lb Dry Air
P3 Combustor Inlet Pressure PSIA
T3 Combustor Inlet Temperature Deg F
HC HC Emission Index lb HC/1000 lb Fuel
CO CO Emission Index lb C0/1000 lb Fuel
NOX NOX Emission Index lb NOX/IOOO lb Fuel
SMOKE EPA Smoke Number
LN Natural Logarithm (Prefix)
e.g. LNHC »¦ NATURAL LOG HC EI
T3C0EF T3 Temperature Coefficient
P3C0EF P3 Pressure Coefficient
HAMBCOEF Ambient Humidity Coefficient
PTHRUST Percent Rated Thrust
FA Fuel Air Ratio (overall)
3-1
-------
4.
SUMMARY OF AMBIENT EFFECTS LITERATURE
Emissions from aircraft turbine engines are affected by the ambient
levels of pressure, temperature, and humidity. Consequently, certification
of such engines for compliance with established standards, if it is to be
meaningful, must properly account for the effects of local test conditions
on measured emissions. The obvious approach would be to determine correction
factors which could be applied to the measured data to reduce it to an agreed
upon set of standard conditions. This section of the report summarizes
selected correction factor techniques representative of those found in the
ambient effects literature.
The purpose of including these summaries, which are not intended to
fully cover all the research performed to date, is twofold. First, they
provide examples of parameters which other researchers have found useful
in correlating emission variation with changes in ambient conditions.
Second, they provide guidance (theoretical in most cases) on how correction
factors change with macroscopic engine operating variables such as pressure
ratio. For instance, theoretical analysis proves useful in addressing
questions such as:
o Should the HC, CO or NOX emission index temperature
sensitivity change with rated or idle engine pressure ratio?
o Should the NOX emission index humidity sensitivity
change with rated engine pressure ratio?
An important distinction should be made concerning the use of the
word "sensitivity." In the above context, sensitivity refers not to the
magnitude of a particular emission index but, rather, to how the rate of
change of that emission index changes with respect to a given correlating
parameter. Thus, while it is well known that the magnitude of NOX emission
levels increase with increasing combustor inlet temperature (T3), sensitivicy
analysis asks the question "How does the rate of change of NOX emission
levels change with temperature?" It is this concept of rate of change which
4-1
-------
is important in analyzing the changes in emission levels induced by variation
in ambient conditions.
From a historical viewpoint, a considerable body of data was collected
on aircraft turbine engine exhaust emissions in engine test cells under uncon-
trolled ambient conditions during the early 1970's. Because of the narrow
range of variability in local pressure, temperature and humidity during these
tests and the large variability in the test data, the effects of ambient condi-
tions on emissions could not be quantitatively determined (Reference 4). More
recently, emissions test data have been collected under controlled conditions
where pressures, temperatures and humidities can be regulated over a broad
range. Such tests have been conducted using complete turbine engines
as well as turbine combustor rigs. The latter configuration utilizes only the
combustor (or a single combustor can) and offers many practical advantages
(e.g., cost, simplicity, ease of parameter variation) over the use of a
complete turbine engine. It is necessary, however, to verify the representa-
tiveness of combustor rig data and the ability to translate the data to the
case of a complete engine. In the following, summaries of selected research
on each regulated pollutant, NOX, HC, CO and Smoke are presented.
4.1 Oxides of Nitrogen
The response of oxides of nitrogen (NOX) emission levels to changing
ambient conditions is unquestionably the best understood of the four regulated
pollutants. Numerous NOX correction factor techniques appear in the ambient
effects literature. Several representative approaches are outlined below.
A key consideration in selecting the techniques presented in this report is
their usefulness in predicting (1) how correction factor sensitivity changes
with engine operating variables and (2) the functional relationships between
NOX emission levels and engine operating parameters.
The relationship between NOX correction factors and engine operating
parameters has been investigated by Blazowski and his co-workers at AFAPL
(Reference 5). This analytical work on NOX emissions is based principally
4-2
-------
on the now well known fact (originally documented by Lipfert using plotted
data, Reference 6) that these emissions correlate well with combustor inlet
temperature. Blazowski therefore proceeds to determine how combustor inlet
temperature is influenced by ambient factors (temperature, pressure and
humidity) and engine operating parameters (pressure ratio, compressor effi-
ciency, combustor residency time, equivalence ratio, etc.).
Assuming ambient air temperature and pressure conditions at the
compressor inlet (dry air), together with a range of compressor pressure
ratios and compressor efficiencies, a set of corresponding combustor inlet
temperatures (T3) and pressures (P3) were obtained (a methodology for per-
forming this calculation is given in Appendix B). From these inputs,
the adiabatic flame temperature was calculated and this, in turn,
permitted computation of the equilibrium concentrations of the significant
combustion products. Using the equilibrium concentrations of 0, N and N^,
the concentration of NO was computed from a relation based on the Zeldovich
mechanism. This relation involves reaction rate constants dependent on
temperature. NO concentrations were also dependent on the time the gas
remained in the combustor.
Using empirical data on a T56 combustor rig, Blazowski solved his
relations for residency time and found it to be substantially constant at a
value of 0.6 msec. In addition, he found his assumed equivalence ratio,
^ = 0.9, in the NO formation zone to be valid. Using this mathematical model,
NOX emission indexes were calculated for a wide range of combustor inlet
temperatures. These results were in general agreement with published test data.
Blazowski postulated, however, that a consideration of combustion efficiency
variations may be required to achieve better prediction of idle NOX variation.
Figure 4a illustrates Blazowski's NOX temperature correction factors
as a function of engine pressure ratio (PR) and equivalence ratio (0). A key
finding illustrated by this figure is that, for a constant equivalence ratio,
NOX temperature sensitivity increases with increasing pressure ratio. This
increased temperature is manifested by an increase in the temperature correc-
tion factor (C^) slope as pressure ratio increases.
4-3
-------
Humidity correction factors, C^, were also computed and graphically
depicted in Figure 4b. This figure shows the strong dependence of NOX emissions
on specific humidity. Presence of water vapor changes the specific heat of
air in such a way that reduced flame temperatures result and, hence, reduced
emissions. Higher pressure ratios result in higher combustor inlet tempera-
tures (other factors being equal) and, therefore, higher emissions. An
equivalence ratio of 0.9 is considered representative of current day engines,
and corresponds to a favorable combination of temperature and available oxygen
radicals for the formation of NO (future designs of engines are expected to
operate at 0 = 0.6). A noteworthy feature of Figure 4b is the theoretically
predicted variation in the NOX humidity correction factor with pressure ratio.
As pressure ratio increases, an increased NOX humidity sensitivity is predicted.
Blazowski hypothesized that a potential cause for this changing NOX sensitivity
is that low pressure ratios have less NOX suppression with humidity because of
the large portion of the emission which is prompt NO. This contribution is less
sensitive to temperature changes than kinetically formed NO. Prompt NO is produced
primarily in the initial stages of combustion and represents an empirically derived
contribution to total NOX formation which attempts to account for the difference
between NOX levels predicted by kinetic equilibrium analysis and those actually
observed.
An alternative NOX emissions model has been developed at Pratt fj Whitney
by Sarli et al (Reference 7). This model is based on plug flow kinetic analysis
and is summarized below:
NOX EI „ , = NOX HI * CF
corrected measured
where:
CF =
P3
ref
P3
meas
1/2
,0.00313 (T3ref-T3meas) g19 (HAMB_-0 .00634)
meas
Verification of this equation was performed using JT9D production
engine data (the JT9D pilot lot data used as part of the present analysis
and outlined in Appendix A) as well as JT9D altitude data. An important
feature of this model is the use of combustor inlet temperature and pressure
4-4
-------
(as opposed to ambient temperature and pressure) as the NOX correlating
parameters. In addition, the Sarli analysis predicts the following func-
tional dependence between NOX emissions and combustor inlet parameters.
NOX EI P3 or LN(NOX EI) ©C LN(P3)
T3
NOX EI e or LN(NOX EI) c*. T3
HAMR
NOX EI ©C e or LN(NOX EI) c< HAMB
where LN( ) indicates the natural logarithm.
These theoretically predicted NOX models were extensively inves-
tigated using empirical data during the development of the correction factor
system proposed in this report. Sections 5, 6 and 7 summarize and discuss
this analysis.
4.2 HYDROCARBONS
Considerably less research than on NOX has been performed on
analyzing the formation of hydrocarbon emissions in aircraft turbine engines.
Three factors however (Reference 8) are generally agreed to have major influ-
ence on the formation of unburnt hydrocarbons from idling aircraft engines.
First, low combustor inlet air temperatures cause quenching to occur
thus terminating combustion before its completion. Second, low fuel air
ratios (fuel lean combustion) results in low equivalence ratios in the
primary zone of the combustor thus reducing burning intensity. Finally,
low fuel and air flows at low power idle operation result in poor fuel
atomization and distribution. An increasing amount of research is being
reported in the literature which explores both the influence of engine
operating parameters and the comparative significance of ambient temperature,
pressure, and humidity on HC emissions. The results of two representative
approaches are outlined below.
4-5
-------
Blazowski and Marzewski (Reference 9) have studied the influence
of ambient conditions on idle HC emissions using a T56 single combustor rig.
With the aid of an analytical global reaction model of hydrocarbon consumption,
they developed HC correction factors which compared favorably with their T56
model validation test data. Figure 5 illustrates the correction factors
developed with this analytical model for engine idle pressure ratios in the
range 2 to 5. A significant feature of this figure is the predicted increase
in hydrocarbon temperature sensitivity as engine idle pressure ratio increases.
This increased sensitivity is manifested by an increase in the slope of the
correction factor curve. Lastly, a small ambient pressure correction was
predicted.
An alternative approach to HC emission correction is typified by the
results presented by Sarli, et (Reference 7) where an empirical correction of
the form
P3
HC EI . , = HC EI . * —
corrected measured p
* ref
was proposed. Two features of this approach are noteworthy. First, HC
emissions are correlated with combustor inlet rather than ambient test
conditions. Second, is the basic simplicity of the technique. This use of
a simplified HC correction factor is predicated less on the grounds that it
fully explains all the variation in HC emissions but, rather, as other
researchers have found, that it represents a reasonable approach to explaining
a significant portion of the variation in highly variable hydrocarbon emissions.
A basic HC model of similar structure but with combustor inlet
temperature as the independent variable served as a guideline in the devel-
opment of the correction factors proposed in this report.
4-6
-------
4.3 CARBON MONOXIDE
The behavior of carbon monoxide emissions is closely related in
many aspects to that of unburnt hydrocarbons discussed in Section 4.2,
In particular, the influence of engine idle pressure ratio is especially
significant because combustion at higher pressures enhances the CO—*-C0?
reaction rate. In addition, the higher combustor inlet temperatures
associated with this higher pressure increase the fuel droplet evaporation
rate, thereby enhancing hydrocarbon reactions. Finally, the higher pressure
into the combustor permits a greater pressure drop across the fuel injectors
which can be used to shatter fuel drops. As a result, carbon monoxide
emissions have been found to sharply decrease with increasing combustor
inlet temperatures and pressures. The ratio of CO to HC emissions,on the
other hand, tends to increase as the efficiency level increases (e.g., as
higher power levels are approached). This finding is consistent with the
chemical kinetics of combustion reactions which predict that HC compounds
should be consumed faster than CO with the result that as gas turbine effi-
ciency is increased, any remaining equilibrium products of combustion tend
to exist mainly as CO. This phenomenon is demonstrated in Figure 6
where the HC emission indexes for CFM56 combustor rig idle and 1.5 idle data
are plotted versus the corresponding CO emission indices. Figure 7 presents
the same data in logarithmic form and suggests a linear relationship between
LN(CO) and LN(HC) or, alternatively, that the HC emission index is propor-
tional to the CO emission index raised to a power.
This interrelationship between HC and CO suggests the possibility of
a parallel correction factor structure for HC and CO. An empirical correction
factor used by Pratt § Whitney and discussed in Reference 7 demonstrates the
use of such an approach. This reference postulates a CO correction factor
of the form:
P3
CO EI . , = CO EI , * D_meaS
corrected measured *^ref
4-7
-------
This model, analogous to the HC model presented in Section 4.2, re-
emphasizes the basic tradeoff between complex emissions models which are
typically optimized for a particular engine design configuration and
simpler models which attempt only to explain the major portion of emission
variation.
4.4 SMOKE NUMBER
Smoke emissions represent the visible segment of particulate
emissions from gas turbines which result when particles of carbonaceous
material formed during the combustion process agglomerate to sizes that
are in the range of the wavelength of visible light (0.45 to 0.65 micro-
meters) . These carbonaceous particles are caused by the incomplete combus-
tion of hydrocarbon fuels and the thermal cracking of fuel in hot regions
of the combustor with locally insufficient oxygen. One of the major
influences on smoke emissions has been found to be fuel selection (Reference
10) .
Only a limited body of data (mostly empirical) currently exists
on the response of visible smoke emissions to ambient conditions. The
results of several test programs are summarized below.
As part of their pilot lot test program, Pratt f, Whitney analyzed
the smoke emissions from 18 JT3D engines tested in 1973 and 1974. A
moderate temperature effect was found with EPA smoke number decreasing as
ambient temperature increased. A similar analysis on nine JT8D pilot lot
engines tested from 1973 to 1976 revealed a similar temperature sensitivity
with the JT8D EPA smoke number decreasing from approximately 27 at 35 Deg. F
to around 18 at 85 Deg. F. The data scatter (+5 SN) exhibited by these
data, however, makes the determination of a strong temperature effect diffi-
cult. Similar pilot lot JT9D test data did not reveal significant ambient
effects.
An alternative approach to the assessment of smoke emissions is
reflected in rig/engine correlation factors used by General Electric, where:
4-8
-------
\ 1.5
Smoke Number
Smoke Number
*
engine
corrected
measured
rig
In this model formulation, smoke number is assumed to increase as combustor
inlet pressure increases. Considering the high positive correlation between
combustor inlet temperatures and pressures, the Pratt 5 Whitney and General
blectric findings appear contradictory if one assumes that combustor inlet
temperatures and pressures are the only determinant of smoke production.
The small magnitude of the corrections predicted by both manufacturers'
approaches and the high degree of variability present in smoke emissions
however suggests that the choice of an ambient effects smoke model of broad
applicability is open to debate and that the major sources of smoke variabi lit)'
rest in other areas.
4-9
-------
5.
CORRECTION FACTOR APPROACH
This section of the report outlines the general approach used to
develop the correction factors for aircraft engine emissions variations due
to non-standard day temperature, pressure and humidity proposed in this report.
Three major considerations were addressed before the design of this correction
factor system was undertaken. These considerations were:
o Use of Theoretically vs. Empirically Derived Correction Factors
o Choice of Correlating Variables - Ambient vs. Combustor Inlet
o Model Structure - Simple vs. Complex
A brief discussion of each of these topics is presented below.
5.1 THEORETICAL VS. EMPIRICAL EMISSIONS MODELS
The first question addressed in selecting a correction factor
development approach was whether a theoretical or an empirical approach
should be used. Ambient effects literature as outlined in Section 4
provides reasonably successful examples of both theoretical and empirical
correction factors. In general, for a single engine the performance of
both approaches is comparable, with NOX correction factors reasonably well
defined, HC and CO corrections considerably less refined, and Smoke Number
corrections relatively unknown. An empirical approach using regression
models based on current theoretical literature was selected for the study
described in this report. Figure 8 outlines the general steps used to
develop this general correction factor approach.
5.2 CORRELATING PARAMETERS - AMBIENT VS. COMBUSTOR INLET
In the review of current literature outlined in Section 4, two sets
of variables were commonly used to correlate emission response to varying
temperature, pressure, and humidity. These two sets of variables are:
5-1
-------
o Ambient Conditions
o ambient temperature (TAMB)
o ambient pressure (PAMB)
o ambient humidity (HAMB)
o Combustor Inlet Conditions
o combustor inlet temperature (T3)
o combustor inlet pressure (P3)
o combustor inlet humidity
(assumed equal to ambient humidity)
Combustor inlet parameters were chosen as the correlating parameters for
this study. The reasons for this choice are summarized below.
The major objective of a correction factor system is the ability
to relate emissions levels measured under non-standard ambient temperature,
pressure and humidity to those which would have been measured under standard
day conditions. Combustor inlet temperature and pressure can be directly
related to ambient temperature and pressure. This relationship is presented
in Appendix B. For a given engine, ambient conditions, engine pressure ratio
and compressor efficiency determine the corresponding combustor inlet condi-
tions. This direct relationship between ambient and combustor inlet conditions
is also illustrated in Figure 9 where combustor inlet temperature is plotted
versus ambient temperature for a subset of CFM56 combustor rig data. The
data presented in this figure were taken at idle power and at a constant
ambient pressure of 29.31 in Hg.
In addition, combustor inlet pressure, temperature, and humidity
adequately predict emission levels. This observation has been clearly demon-
strated in the ambient effects literature (References 5, 6, 7, and others).
This correlation is further demonstrated in Figures 10 to 12 where the natural
logarithms of CFM56 combustor rig HC, CO, and NOX emission indices (lb
pollutant/1000 lb fuel) are plotted versus combustor inlet temperature. The
HC and CO data in these figures were taken at idle power while the NOX emission
5-2
-------
data corresponds to climb power. Ambient humidity and pressure were held
constant during this phase of testing. Figure 13 presents NOX climb data
versus ambient humidity for fixed ambient (also combustor inlet) temperature
and pressure. For the purposes of this report, ambient and combustor inlet
specific humidity are assumed equal. The natural logarithm of the emission
indices was selected as a correlating parameter over the emission indices
themselves because of the linear relationship which exists between the emission
index logarithm and combustor inlet temperature, and because of the theoretical
dependence discussed in Section 4.1.
In the above, the relationship between ambient and combustor inlet
conditions has been summarized and the ability to correlate aircraft emission
levels with these combustor inlet parameters has also been demonstrated for
a sample data set. Neither of these observations alone, however, provides a
sufficient rationale for selecting combustor inlet over ambient parameters as the
basis for a general correction factor system. The primary reason for this
choice is that the use of combustor inlet parameters opens the possibility
for a more direct comparison of correction factors developed for engines of
differing pressure ratios, compressor officiencies,classes, etc. Since
combustor inlet conditions are the primary determinant of ultimate emission
levels, their use as the correlating parameter avoids confounding effects
due to compressor differences between engines and permits data from combustor
rigs to be directly incorporated into the search for a generalized correction
factor system. It should be noted, however, that the use of combustor inlet
parameters as the only determinant of emission levels does not directly admit
the possibility that changes within the combustor itself (changes which do
not affect combustor inlet pressure and temperature) can effect emission levels.
These internal combustor modifications are encorporated into the correction
factor system through the use of a "technology"parameter which permits engines
of more advanced designs to use different correction factor coefficients.
This variation in correction factor is discussed fully in Section 7.
5-3
-------
5.3 SIMPLE VS. COMPLEX EMISSIONS MODELS
Figure 14 presents the emissions models selected as the basis for
the general correction factor system proposed in this report. These models
are reformulated in terms of the emission indices themselves and summarized
below:
AMBIENT EFFECTS PROGRAM REGRESSION MODEL SUMMARY
a * T3
HC EI = aQ * e
b * T3
CO EI = fc>0 * e
c * T3 c * HAMB
NOX EI = c * e * e
o
dl
SMOKE NUMBER = dQ * P3
where aQ, a^ bQ, b^ Cq, c^ c^ dQ, dj
are coefficients determined from regression analysis.
T3 = Combustor Inlet Temperature (Deg. F)
P3 = Combustor Inlet Pressure (Psia)
HAMB = Ambient Specific Humidity (lb l^O/lb Air)
The models were selected after an extensive examination of a variety of
regression models which were applied to both the five EPA sponsored test
engines and the approximately 30 engines in the industry contributed data base.
A variety of reasons led to the final selection of these models; several of
the more important are summarized below.
An examination of Figure 14 reveals that in each proposed emissions
model, either the combustor inlet temperature (T3) or the natural logarithm
of combustor inlet pressure (LNP3), but never both, appear in the regression
5-4
-------
model. The primary reason for this aspect of model formulation rests in
the high degree of interdependence between combustor inlet temperature
and pressure. The correlation coefficient between T3 and P3 provides a useful
measure of this interdependence. The range of the correlation coefficient
^ is (-1 £ ^ ^ + 1). A value of ^> = 0 indicates that T3 and P3 are not
linearly related. A value of that approaches (-1) reflects a very close
inverse linear relationship. As ^ approaches (+1), a positive linear
relationship between T3 and P3 is indicated. For the test data encompassed
by this study, the correlation between combustor inlet temperature and
combustor inlet pressure was typically in the range between 0.95 and 0.98.
As a result, a knowledge of either T3 or P3 permitted the other to be deter-
mined. Similarly, a knowledge of emission variation with respect to either
T3 or P3 alone was sufficient to determine the major influence of ambient
variations. In fact, the inclusion of both T3 and P3 in a regression analysis
can lead to a variety of analytical problems commonly referred to as multi-
col linearity . These problems make the determination of reliable and stable
correction factor coefficients impossible. A basic description of multi-
col linearity and illustrations of this problem in aircraft emissions data
is presented in Appendix C. Combustor inlet temperature was selected as
the primary correlating variable for the HC, CO and NOX models since it
provided a slightly better fit to the observed data. On the other hand,
combustor inlet pressure represented the best predictor of visible smoke
emi ssions.
Further examination of Figure 14 emphasizes the relative simplicity
of the basic models selected. While more complicated emissions models
involving multiple independent variables (e.g., inclusion of fuel air ratio
for HC and CO, ambient humidity for CO, etc.) were found to give a small
improvement in the ability to explain emissions variations for selected
engines, neither the magnitude of this improvement nor its general applica-
bility warranted the use of more terms in the regression models selected.
Instead, it was felt that the development of a general empirical correction
factor scheme required the least complicated emissions models in order to
gain correction coefficient stability and to demonstrate general trends.
5-5
-------
The overall performance of the emissions models selected is remarkably
good considering the number and diversity of engines in the test data
analyzed. A common statistic used to evaluate regression models is the
2
coefficient of determination or R . This statistic expresses the fraction
of the variation in a given set of test data explained by the regression model
2
chosen. An R value of 0, for instance, would indicate that 0 percent of the
variation in emission levels was explained by the emission model selected.
A value of 1 on the other hand, says that 100% of the variation in emission
2
levels was explained by the regression model. The average R values for the
approximately 30 regressions examined for each pollutant are summarized below.
AMBIENT EFFECTS REGRESSION MODELS _
MEAN COEFFICIENT OF DETERMINATION (R )
(Sample Sizepf 30 Regressions/Pollutant)
2
Model Mean R
LNHC = f(T3) 0.82
LNC0 = f(T3) 0.86
LNN0X = f(T3,HAMB) 0.91
LNSMK = f(LNP3) 0.60
These performance figures should be judged in light of the inherent
variability in the test data. This comparison is presented in Section 5.4.
The minimal increase in model performance which typically can be
achieved by including additional terms in the regression model selected is
demonstrated by the TF30 carbon monoxide regression statistics below. Shown
2
in this summary is the increase in the coefficient of determination (R )
which is achieved by adding new terms to the CO regression model.
5-6
-------
TF30 CO REGRESSION MODEL SUMMARY
LNCO=f( )
Variables in Model
T3
+ LNP3
+ HAMB
+ P3
+ FA
R
.9820
.9 896
.9911
.9937
.9938
Increase in R
. 0076
.0015
.0026
.0001
5.4 DATA VARIABILITY
An important consideration in assessing the performance of the
ambient effects emissions models proposed in the preceding section is the
variability observed in aircraft emissions data. A typical question which
might be asked is: How does the variability in measured emissions data compare
to the magnitude of a correction factor for a particular pollutant? For
instance, the application of a 5 percent correction factor to data with an
observed variability of 20 percent is of dubious value.
Variability can be usefully divided into two components:
o Test-to-Test Variability
o Engine-to-Engine Variability
Test-to-test variability is manifested by the fact that replicate
tests on a particular engine ostensibly operating under identical test condi-
tions do not give the same results. This aspect of emissions variability is
due in part to such factors as instrumentation errors, slight variation in
engine power setting, test site, test sequence, etc. An extensive discussion
of many of the causes of test-to-test variability in aircraft engine emission
measurements is provided in References 11 and 12. Engine-to-engine varia-
bility on the other hand provides an additional source of variation in emission
data. This phenomenon occurs when several engines of a particular model are
tested. A discussion of each source of variability is given in the remainder
of this section.
5-7
-------
5.4.1 Test-to-Test Variability
Three test programs were used to assess test-to-test variability.
First, each of the 112 test points for the TPE331-5-251 was repeated twice for
a total of 3 measurements/test point. Second, each of the 107 test points for
the CFM56 combustor rig was repeated once for a total of 2 measurements/test
point. Finally, each of the 7 idle tests for the ALF502 full scale engine was
repeated once for a total of 2 measurements/idle test point.
A measure of repeatability can be obtained as follows:
Consider a particular engine tested in a given mode and compute the mean
XR (k = 1,...,N) of the replicate measurements. Then compute the variance
of the replicates as follows:
A 9 U-- " *
oi2 = z: —
i=i N-'
where X^ is the i*^ replicate measurement for the k1"'1 test point and N is
the number of replicate measurements (2 for CFM56, 3 for TPF.331-5-251, 2 for
ALF502 engine).
A
The coefficient of variation or represents a useful measure
of the relative repeatability of the emissions measurements.
If the coefficients of variation for each test point in a given mode
are computed and averaged, a measure of the average repeatability of emissions
measurements for a given pollutant and mode can be obtained. Figure 15 illus-
trates this process for the 22 CFM56 combustor rig idle 1IC test points where
a mean coefficient of variation of .2232 was determined. This parameter
indicates that for HC CFM56 rig measurements the test-to-test variability CT"^
is approximately 22% of the mean HC idle emission index. Tables 9 and 10
summarize the mode-by-mode coefficient of variation analysis for the CFM56
and TPE331-5-251 engines. In both these tables, high thrust HC and CO values
were omitted because of the very low emission levels at these thrust levels.
Of particular note is the high degree of repeatability (approximately 3-4%)
obtained for NOX emission levels. Conversely, the repeatability for CFM56
smoke measurements was found to be around 40%. It should be noted that the
5-8
-------
above analysis does not include the engine-to-engine variability which can
only be estimated by measurements on multiple engines.
A similar analysis was undertaken on the AVCO ALF502 full scale
engine data using the two idle measurements presented for each test. These
two idle settings represent the same nominal thrust level except that the
first was taken just after start-up (TAXI-OUT) and the second after descending
from full power (TAXI-IN). As such, this ALF502 data provides a measure of
the test-to-test variability in this data. Table 11 presents a summary of
the mean IDLE coefficient of variation for each of the four pollutants.
Similar data presented in Tables 9 and 10 for the CFM56 and TPE331-5-251 is
repeated for comparison.
5.4.2 Engine-to-Engine Variability
An evaluation of engine-to-engine emissions variability requires that
different engines of the same model be tested under identical test conditions:
Unfortunately, insufficient data was available in the ambient effects data base
to adequately evaluate this aspect of emissions variability. In most instances,
the test data consisted of multiple tests on a single engine or combustor rig.
In those cases such as the Pratt f, Whitney JT8D and JT9D pilot lot data
where multiple engines were tested, varying ambient conditions for each engine
limited the ability to segregate the effects of engine-to-engine variability.
Some guidance in this area is provided by Reference 4 which concluded that
engine-to-engine variability is on the same order of magnitude as test-to-test
variability.
5.5 REGRESSION MODELS AND CORRECTION FACTORS
Figure 14 summarized the general regression models used to assess
emission response to ambient effects. From these equations, correction
factors can be derived as shown in Figure 16. The specific equation presented
in this example has been derived from the CFM56 combustor rig NOX emissions
data discussed fully in Section 6. The correction factor (CF) is defined as:
5-9
-------
EMISSIONS AT STANDARD DAY CONDITIONS
MEASURED EMISSIONS
The corrected emissions indices at standard day conditions are then given as
EMISSIONS, . = EMISSIONS, * CF
(corrected) (measured)
5-10
-------
6. INDIVIDUAL ENGINE CORRECTION FACTORS
This section presents a comprehensive example which illustrates the
methodology used to develop correction factors for a single engine. The CFM56
Rig data was chosen for this example because it was sampled under controlled
ambient conditions and it adequately represents the problems encountered in
analyzing aircraft engine emissions data.
6.1 CFM56 Combustor Rig - Raw Data Plots
Figures 17 through 26 present CFM56 combustor rig data plots for the
four pollutants and selected explanatory variables. Each pollutant and the
natural log of that pollutant are presented to demonstrate the general pattern
of the data and the effect of the log transformation upon the data. Figures 18
(LN(HC) vs T3), 20 (LN(CO) vs T3), and 22 (LN(NOX) vs T3) are of particular
interest. The purpose of the log transformation is to make the data amenable
to linear regression techniques. Ideally, the proper transformation would
orient the data in a straight line when the dependent variable (say LN(NOX))
is plotted against the appropriate explanatory variable. In Figure 22, it is
evident that the log transformation on the dependent variable works quite well.
The remaining nonlinearities in this data plot are due primarily to humidity
variation. A comparison of this plot with those for HC and CO (Figures 18 and
20) reveals a linearization of the data in only the portion of the curve where
T3 is less than 5U0°. This portion of the curve represents the data for the
idle and 1.5 idle modes and is the data used in developing the HC and CO
correction factors. The other data points where T3 is greater than 500° were
not used in developing the correction factors because the levels of HC and CO
in this range were considered to be an insignificant contribution to the LTO
cycle. Thus the HC and CO correction factors developed in this report are only
applicable to the idle modes. No IIC or CO correction is applied to the higher
powered modes. Figures 23 and 24 show the decrease in the NOX levels as the
ambient humidity increases. The data for all five test modes are presented in
this plot. Figures 25 and 26 show a relatively large data scatter in the
Smoke vs P3 data with perhaps a slight increase in smoke levels as P3 increases.
The low smoke numbers for the CFM56 rig data are partially the result of facility
6-1
-------
limitations which necessitated running the rig tests at reduced P3 levels.
As a result of these limitations, conditions favorable to high smoke produc-
tion were not included in the test matrix.
6.2 CFM56 COMBUSTOR RIG - SUMMARY STATISTICS BY MODE
Summary statistics by mode for the CFM56 combustor rig data are given
in Table 12 . This table demonstrates the response of the dependent and
independent variables to changes in thrust. The constant values for HC in
CLIMB and TAKEOFF modes are the result of nominally setting the HC values
to .01 for small values of HC. Of particular interest in this table is the
coefficient of variation (C.V.) or the ratio of one standard deviation in the
data to the mean value of a particular data set. This statistic provides a
measure of the relative scatter in a given set of data. Since the percent
rated thrust (PTHRUST) is essentially constant in these mode-by-mode statistics,
the coefficient of variation provides a good measure of the relative variation
in the emissions data which is primarily due to ambient variation. Several of
the more important modal coefficients of variation are summarized below.
CFM56 EMISSION INDEX MODAL COEFFICIENT OF VARIATION (C.V.)
Pollutant
Mode
C.V. %
HC EI
IDLE
98
CO EI
IDLE
25
NOX EI
TAKEOFF
19
SMOKE
TAKEOFF
31
6.3 CMF56 COMBUSTOR RIG - REGRESSION SUMMARY
Section 5 discussed the rationale for selecting the general form of
the regression equations. Figure 27 summarizes the equations derived for
the CFM56 Combustor Rig data. Tables 13 through 16 present selected
regression statistics for these equations. It is useful here to explain some
of the statistics presented in the regression summary.
6-2
-------
We can begin by identifying the specific equation or model used in
the least squares regression approach. (The reader is reminded that because
the natural logs of the emissions are used instead of emissions themselves,
only the log data will have the least squares properties such as the sum of
the residuals equal to zero.) For example, the form of the equation chosen
for the NOX model is given as:
LN(NOX) = aQ + a1 T3 + HAMB
where aQ, a^, a2 are the least squares parameter estimates, T3 is the
combustor inlet temperature in degrees F, and HAMB is the specific humidity
in lb H^O/lb dry air. Thus in Table 15 we find under the heading "ESTIMATE"
the values 0.46742286, 0.00250325 and -20.70200748 for a^, a^ and a^,
respectively. It is possible to perform a statistical test to determine if
the parameter estimates are significantly different from zero. Obviously,
if we cannot say that the parameter estimates for the T3 and HAMB coefficients
are different from zero, we have been unable to establish a statistically valid
model for the relationship we assumed between LN(NOX) and the explanatory
variables T3 and HAMB. The statistical test on the parameter estimates is
carried out as follows:
i) Select a significance level (typically .05 or .01),
that is, the probability of rejecting the hypothesis
that the parameter equals zero when in fact it actually
is zero.
ii) If the value under the heading "PR ^ |t| " corresponding
to the parameter of interest is less than the significance
level, we will reject the hypothesis that the parameter
equals zero. The | T | represents the absolute value of
a t-statistic with appropriate degrees of freedom.
In Table 15 , the value under the heading "PR"^ jT |" corresponding
to the T3 parameter is 0.0001. Assuming a level of significance of .05
we will reject the hypothesis that the T3 parameter equals zero.
6-3
-------
Testing the parameter estimates is also referred to as testing for significance
of regression. Another method of testing for significance of regression is
to use the information given under the heading "PR y F". This is referred
to in the literature as an F-test for significance of regression. In a
manner analogous to that for the parameter estimates, if the value under the
heading "PR ^ F" is less than the significance level chosen, we reject the
hypothesis that the regression is not significant. While it is important
to know whether or not a parameter estimate is different from zero, it is also
useful to have an idea of how much we might expect that parameter estimate to
change if we were to calculate it for another set of CFM56 rig data taken
under identical conditions. An approximate measure of the variation in a
parameter estimate can be found by taking the estimate plus or minus two times
its standard error. In Table 15 then we find that our T3 estimate will vary
over the range 0.00250325 + 2 (.00002079) or from 0.002462 to 0.002544.
The information presented under the heading "SUM OF SQUARES" provides
insight into how well the model fits the data. Suffice it to say that the
CORRECTED TOTAL SUM OF SQUARES is a measure of the overall variation in the
data. This total variation can be segregated into two parts: the first is
the MODEL SUM OF SQUARES and the second is the ERROR SUM OF SQUARES. The
MODEL SUM OF SQUARES is a measure of the portion of the total variation that
can be attributed to the regression equation we have chosen. The ERROR SUM
OF SQUARES is a measure of that portion of the total variation that is not
accounted for by the regression equation. In general, we would like to see
as much of the total variation in the data attributed to the model chosen.
Thus if we have selected a reasonable model:
LN(NOX) = aQ + ax T3 + a2 HAMB
we would expect to reduce the "ERROR SUM OF SQUARES" to some minimal value,
but never to zero because NOX formation is most probably a function to a
limited extent of variables other than combustor inlet temperature and ambient
humidity. As noted earlier, we would like to see the MODEL SUM OF SQUARES
approach the value of the CORRECTED TOTAL SUM OF SQUARES which will conse-
quently reduce the ERROR SUM OF SQUARES. We can write the relationship between
the various sums of squares in equation form:
6-4
-------
CORRECTED TOTAL SUM OF SQUARES = MODEL SUM OF SQUARES
+ ERROR SUM OF SQUARES
Another measure of how well the model fits the data is to examine the ratio:
R2 = (MODEL SUM OF SQUARES) / ("CORRECTED TOTAL SUM OF SQUARES)
2
noting that R (R - SQUARE) will approach unity for a "good" fit. Referring
again to Table 15 , we see that a significant portion of the CORRECTED TOTAL
SUM OF SQUARES (86.73) is allocated to the MODEL SUM OF SQUARES (85.55)
indicating that the model fits the data quite well. In fact, under the
heading "R-SQUARE" we see that 98.64 percent of the total variation in the
LN(NOX) data is attributable to the regression model we have chosen (e.g.,
variation in T3 and HAMB).
The statistic provided under the heading "C.V." expresses the variation
in the data as a percentage of the mean of the data. This measure is called
the coefficient of variation:
0~
C.V. = * 100 percent
where (J~ s the standard deviation of the data
M- - the mean of the data
Thus we see in Table 15 that the coefficient of variation for the idle to
takeoff LN(NOX) data equals 3.8478 percent. It is interesting to compare this
degree of variability in the LN(NOX) data to that presented in Section 6.2 for
the untransformed NOX data where a coefficient of variation of 19% was found
for the takeoff mode only. This reduction in variability is one of the
properties of the logarithmic transformation which linearizes and rescales the
test data by "compressing" the higher valued emission indices.
Comparing the regressions in Tables 13 to 16 we see that the best fit
is with the LN(NOX) data -- this is reflected in both the MEAN SQUARE and
the R-SQUARE values. The major reasons for this are (1) the effect of the
log transformation in linearizing the data, and (2) the relatively low level
of variation in the LN(NOX) data as reflected in the small coefficient of
6-5
-------
variation. The regression fits for LN(HC) and LN(CO) are still quite acceptable.
Note that the error bounds on the parameter estimates (+ 2 * STD ERROR OF
ESTIMATE) are relatively small). Only for the LN(SMOKE) data do the regres-
sion statistics indicate a poor fit of the data. In Table 16 we see that
only 4.99 percent (R-SQUARE) of the variation in the data is attributable
to the model chosen. The parameter estimate for LN(P3) is significant,
indicating a relationship between LN(P3) and LN(SMOKE), but the reasons for
the poor R-SQUARE value are due to (1) the large amount of variation inherent
in the LN(SMOKE) data as demonstrated by a large (66.39 percent) coefficient
of variation, and (2) the low values of smoke number measured. Table 12, for
instance, indicates the mean takeoff power smoke number was only 4.12. Due
to the above limitations in the CFM56 smoke data, alternative data sources
such as JT8D, JT9D, RB211-22B, SPEY 511 and the ALF502 were used to develop
the smoke model, LN(SMOKE) = f(LN(P3)) which was chosen as the most represen-
tative predictor of variation in smoke number due to ambient and combustor
inlet conditions.
In summation, this section of the report has presented a fairly
comprehensive example illustrating the development and the performance of
the emissions models developed for the CFM56 combustor rig. The intent of
this section has been to illustrate, using extensive data from a single
combustor, selected important facets of the data analysis process which are
representative of many of the problems encountered and techniques applied
to the entire ambient effects data base. Of particular importance in terms
of overall representativeness are:
o The relative degree of variability in HC, CO, NOX
and smoke emissions
NOX CO SMOKE HC
(Least Variable ------- Most Variable)
o The emission response to combustor inlet conditions
as expressed by the functional emissions models chosen.
(See Figure 14.)
o The comparative degree of fit by the regression models for
the four pollutants.
NOX CO HC SMOKE
(Best Fit -------- Poorest Fit)
6*6
-------
7.0 GENERALIZED CORRECTION FACTORS
This section of the report summarizes the results of the generalized
correction factor study for each of the four regulated pollutants - HC, CO,
NOX, and EPA Smoke Number. Figure 28 illustrates the four basic steps used
to develop these correction factors. First, regression analyses using the
functional models discussed in Section 5.3 were performed on test data for
each engine in the data base. Section 6 provides a detailed example of how
these regressions were performed and an assessment of how well they typically
predict emission response to changing combustor inlet conditions. For each
engine (assuming data for each pollutant was available), five regression
coefficients were determined.
Pollutant
HC
CO
NOX
SMOKE
Coefficient
T3COEF
T3COEF
T3COEF
HAMBCOEF
P3C0EF
Description
HC combustor inlet temperature coefficient
CO combustor inlet temperature coefficient
NOX combustor inlet temperature coefficient
NOX ambient humidity coefficient
Smoke combustor inlet pressure coefficient
The NOX combustor inlet temperature and ambient humidity coefficients,
for example, are derived from a regression analysis of the form:
LN(NOX) = CONSTANT + T3C0EF * T3 + HAMBCOEF * I IAMB, or
NOX - e(C0NSTANT) * (T3C0EF*T3) * (HAMBCOEF * HAMB)
where
LN(NOX)
CONSTANT
T3
HAMB
the natural logarithm of the NOX emission index
constant term in regression
combustor inlet temperature (Deg. F)
ambient humidity (lb H^O/lb day Air)
7-1
-------
The T3C0EF for NOX,therefore,provides a measure of the increase in the natural
logarithm of the NOX emission index which should result from a one Deg. F
increase in combustor inlet temperature. The effect of such a one degree
change on the emission index itself is computed as exp (T3C0EF). Similar
analyses are applicable to both the HC and CO combustor inlet temperature
coefficients and the NOX ambient humidity coefficient.
The smoke combustor inlet pressure coefficient P3C0EF, on the other
hand, developed using a model of the form:
LN(Smoke Number) = CONSTANT + P3C0EF * LN(P3)
or
C 1 M U CONSTANT * ,P3C0EF
Smoke Number = e * P3
where
LN(Smoke Number) = the natural logarithm of the EPA smoke number
CONSTANT = constant term in regression
P3 = combustor inlet pressure (PSIA)
The P3C0EF for smoke,therefore,provides a measure of the increase in the
natural logarithm of the smoke number which should result from a one unit
increase in the logarithm of combustor inlet pressure.
The final step in the development of the proposed correction factors
was to summarize salient engine operating parameters for each engine and
relate these parameters to the combustor inlet regression coefficients
determined above. Of particular interest in this phase of analysis were the
rated engine pressure ratio (PR) and the idle pressure ratio (IPR). Although
the usefulness of a variety of other engine parameters such as rated thrust
and bypass ratio were analyzed, better correlation was found between the
combustor coefficients and rated and idle pressure ratio. Primary outputs
of this phase of the analysis are plots illustrating the variation in a
particular coefficient as a function of rated or idle pressure ratio.
7-2
-------
Each point plotted in these figures represents the results of one regression
analysis on a given engine. The number of data points used to determine the
regression line for each engine varied from approximately 10 to around 200.
It is important to remember that the coefficients plotted in these figures
represent the temperature, humidity or pressure sensitivity (slope) of a
particular emission index or smoke number as a function of PR or IPR and not
the magnitude of that emission index. Thus, while it is well known that the
NOX emission index generally increases with rated engine pressure ratio,
a plot of the NOX temperature coefficient T3C0EF versus engine pressure ratio
indicates hov the slope of the NOX EI vs PR curve changes with rated pressure
ratio. This slope is indicative of the response of a particular pollutant
to the small changes in combustor inlet parameters commonly introduced by
variations in ambient test conditions.
In the following, the results of the general correction factor
analysis are summarized for each pollutant. Six basic types of information
are presented:
o Correction Factor Coefficients in Tabular Form
o Correction Factor Coefficients Plotted versus Engine
Operating Parameter?
o Proposed Generalized Correction Factors
o Theoretical Background for Proposed Correction Techniques
o Relationship between Newer Technology Engines and
Proposed Correction Techniques
o Sensitivity and Error Analysis
Figure 29 provides the nomenclature used throughout the presentation
of this summary. A word of explanation is in order with regard to two
parameters, T3MEAN and P3MEAN, which are used in this presentation. As stated
earlier in the introduction to this section, the rated engine pressure ratio
(PR) and idle pressure ratio (IPR) are used to correlate the correction factor
coefficients with engine operating parameters. Since selected engines in the
7-3
-------
data base were not tested while operating in close proximity to "rated"
standard day test conditions (e.g., the RB211-22B altitude data was run
at a simulated 10,700 M pressure), two alternative parameters, T3MEAN and
P3MEAN were introduced which are analogous to rated engine T3 and P3
but which also account for anomalies in the test data due to specifics
of how the engines were tested. For example, the RB211-22B rated pressure
ratio is 25. When this engine is operating in close proximity to standard
day rated conditions, a P3MEAN of approximately 350 PSIA is observed.
The same engine tested under high altitude conditions, however, exhibits
a P3MEAN of approximately 260 PSIA. Similarly, a T3MEAN value of around
900 Deg. F is found for normal engine operation while a T3MEAN of 850 Deg.F
is observed during altitude testing. In short, the use of T3MEAN and P3MEAN
facilitates the correlation between correction factor coefficient and engine
operating parameter by eliminating some of the confounding effects attributable
to the fact that the engines analyzed were tested under a relatively diverse
set of experimental designs. Figure 30 illustrates the relationship between
T3MEAN and rated pressure ratio (PR) for the engines in the test data base.
7.1 Oxides of Nitrogen
Table 17 presents a summary of the N0X emission index correction
factor coefficients developed for the model LN(N0X EI) = f(T3,HAMB). A
description of the variables listed in this table was given in Figure 29,
Summary statistics for these coefficients are provided in Table 18. The
temperature and humidity correction factors developed from these data are
presented and discussed below.
7.1.1 Temperature Correction Factor
In order to assess the potential influence of engine operating
parameters on the NOX combustor inlet temperature coefficient T3C0EF, this
coefficient was plotted versus P3MEAN. Figure 31 illustrates the results of
this analysis. Each point in this figure represents the T3 regression
coefficient determined for a given engine. In selected cases, this plotted
7-4
-------
point represents the regression results for only a single operating mode
(e.g., CFM56 Idle). As shown in this figure, the calculated combustor
inlet temperature coefficients (T3C0EF) rise linearly with P3MEAN. As
combustor inlet pressure increases, therefore, the natural logarithm of
the NOX emission index becomes more sensitive to changes in combustor inlet
temperature.
A useful approach to analyzing the data presented in Figure 31
is to compare the results illustrated to those developed by Lipfert (Refer-
ence 6) and summarized in Figure 32 (Reference 13") . In this figure, the
NOX emission index for a variety of production engines is plotted versus
compressor discharge temperature (T3). The slope of the Lipfert curve in
log space represents the combustor inlet temperature coefficient T3C0EF.
T3C0EF =
LN(NOX EI)T- - LN(NOX EI) T?
2 _1
T32 - T3j
where T32^ T3j Deg. F.
In the Lipfert analysis, a constant NOX combustor inlet temperature
coefficient of approximately 0.00385 was determined. This constant value of
T3C0EF is plotted as a horizontal line in Figure 31 and corresponds to
approximately the mean value of the combustor inlet temperature coefficients
determined in the present study.
In order to assess whether the observed NOX T3COEF trend is statis-
tically significant (not necessarily significant from an engineering standpoint)
the 95% upper and lower confidence bounds on the computed combustor inlet
temperature coefficients were computed as follows:
Let: T3C0EF = NOX combustor inlet temperature coefficient
t = t statistic at .05 significance level
(*2 for sample sizes greater than 30)
T3STD = T3C0EF standard error
UCB = Upper Confidence Bound
LCB = Lower Confidence Bound
-------
Then:
UCB = T3C0EF + t Q5 * T3STD
LCB = T3C0EF - t * T3STD
A confidence interval was constructed of the form:
LCB
-------
basis for the proposed NOX temperature correction factor. A least squares
equation was determined which relates T3C0EF to rated engine pressure ratio
(PR). This equation is summarized below and plotted on Figure 34.
NOX T3C0EF = .001735 + .000107 * PR
R2 = 0.77
Standard Errors of Regression Coefficients
Intercept = .000139
PR = .000010
The proposed NOX temperature correction factor is therefore a function of
rated engine pressure ratio. In equation form, this temperature correction
factor (C , ) can be expressed as:
1 NOX
T3C0EF * (T3 . - T3„ )
CT, = e Reference Meas
NOX
where
T3C0EF = .00175 + .000107 * PR
Figure 35 illustrates the variation in T3C0EF as a function of
NOX MEAN (see Figure 29 for definition). As shown in this figure, the
computed temperature coefficient rises uniformly with mean NOX levels. The
only exception to this trend is the highly temperature sensitive SPEY 511,
an older design engine, with high HC and CO emission levels but comparatively
low NOX emissions. Also of note in this figure is the JT9D-7 Vorbix
which employs an advanced technology staged combustor to reduce emissions.
A comparison of the JT9D-7 Vorbix temperature coefficient with those of the
conventional JT9D-7A and JT9D-7F engines reveals that the newer technology
Vorbix exhibits a decreased NOX combustor inlet temperature sensitivity.
An alternative method of presenting this finding is illustrated in Figure 36
7-7
-------
where T3MEAN is plotted versus NOXMEAN. As seen in this figure, two basic
engine groupings appear which roughly correspond to the levels of technology
employed to control NOX emissions.
The first groups represents newer technology engines such as the
JT9D-7 Vorbix, the CFM56 combustor rig, and the T63-A-5A advanced combustor
engines. A primary objective of an NOX emission control strategy is the
ability to operate at a given combustor inlet temperature and pressure
and still reduce NOX emission levels. In simplified terms, the modification
required to achieve this objective can involve lowering the NOX emission
index vs combustor inlet temperature curve as shown in Figure 37. Two
features on this curve are important. First, for a given combustor inlet
temperature (T3) controlled engines exhibit a lower NOX emission index.
Second, and more importantly, to achieve this reduced emission index, the
temperature sensitivity (slope of the NOX vs T3 curve) must also be decreased
This phenomenon is manifested in Figures 35 and 36 for the JT9D-7 Vorbix
and JT9D-7 conventional engines. Additional insights into the reduced NOX
temperature sensitivity of new technology engines is provided in Reference 14
The second group of engines in Figure 36 are representative, for
the most part, of conventional engine designs with limited NOX emission
controls. These engines typically exhibit a higher NOX temperature sensitivi
for a given NOX emission level than the newer technology engines.
7.1.2 Humidity Correction Factor
Ambient humidity variations have a significant impact of NOX
emission levels. This observation was illustrated in Figure 1 where a
25 percent change in the ALF502 NOX emission index was induced by changes
in ambient humidity. As ambient humidity increased, NOX emission levels
decreased, primarily because of the resulting reduced combustor inlet
temperatures.
In developing a NOX EI humidity correction factor, attempts were
made to determine the influence of engine operating parameters on the
7-8
-------
ambient humidity regression coefficient HAMBCOEF presented in Table 17.
Theoretical insights provided by the research of Blazowski (Reference 5
and summarized in Section 4.1 and Figure 4) indicate that the NOX humidity
coefficient should increase with rated engine pressure ratio. No evidence
of this phenomenon could be found in the test data analyzed. Instead,
it is felt that the mean value of the observed humidity coefficients provides
a useful guideline in the choice of a humidity correction coefficient.
This mean value was found to be -18.3 +8.0 (95% confidence level) for the
10 engines for which adequate humidity variation permitted computation of a
humidity coefficient. This predicted range in the NOX humidity coefficient
is comparable to that found by Shaw (Reference 15) of -22 ±8 and is close to
the frequently quoted value of -19. The proposed NOX humidity correction
factor is therefore:
HAMBCOEF * (HAMBn _ - HAMB.. )
C.,.lin = e Reference Meas
hambNOx
where
HAMBCOEF = 18.3+8.0
HAMBn r = .00634 lb ILO/lb Dry Air
Reference 2 }
7. 2 HYDROCARBONS
Table 19 presents a summary of the HC emission index correction
factor coefficients developed using the model LN(HC EI) = f(T3). A descrip-
tion of the variables listed in this table was given in Figure 29. Summary
statistics for these coefficients are provided in Table 20. The proposed HC
temperature correction factor developed from these data is presented and
discussed below.
Efforts were made to correlate the HC combustor inlet temperature
T3C0EF with a variety of engine operating parameters. The most successful
7-9
-------
of these engine correlation parameters was the Idle Pressure Ratio (IPR) .
Figure 38 presents a plot of the HC combustor inlet temperature coefficient
versus engine idle pressure ratio. Two features of this figure are of
importance in developing a generalized correction factor for hydrocarbons.
First, as idle pressure ratio increases the HC temperature sensitivity
as reflected in T3C0EF increases in a negative direction. Since the HC
combustor inlet temperature is defined as:
LN(HC EI) - LN(IIC HI)
2 1
T3C0EF = —_
T3 - T3
2 1
where ^2^ ^1 ^ *
the observed negative increase in T3C0EF indicates that higher idle pressure
ratio engines exhibit larger decreases in HC emission levels for a given
increase in combustor inlet pressure. This observation is consistent with
the theoretical analysis presented by Blazowski et al (Reference 9 and
summarized in Section 4.2 and Figure 5) where he demonstrates that a larger
HC temperature correction factor is required for higher idle pressure ratio
engines than for lower pressure ratio engines.
The second feature of Figure 38 of interest is the relationship
between the coefficients determined for older technology engines with those
of newer technology engines such as the JT9D-7 Vorbix and the CFM56 combustor
rig. By an argument analogous to that presented in Section 7.1.1 where
decreasing NOX temperature sensitivity was hypothesized for newer engines,
an increasing HC temperature sensitivity would be expected for newer
technology engines. Since hydrocarbon emissions are inversely proportional
to combustor inlet temperature, a low HC emission engine designed to operate
at a given idle pressure ratio could be expected to show more temperature
sensitivity than an older technology engine. An examination of Figure 38
reveals that the JT9D-7 Vorbix and CFM56 combustor rig do, in fact, demonstrate
high combustor inlet temperature sensitivity.
7-10
-------
The data presented in Figure 38 form the basis for the proposed HC
temperature correction factor. A line was drawn which represents the general
coefficient trend between the HC temperature coefficient T3C0EF and idle
pressure ratio (IPR). This line is shown in Figure 38. The proposed HC
temperature correction factor is, therefore, a function of engine idle
pressure ratio and can be expressed as:
„ T3C0EF * (T3n _ - T3W )
C = e Reference Meas
1 HC
where T3C0EF is determined from Figure 38 and the appropriate idle pressure
ratio.
7.3 CARBON MONOXIDE
An analysis similar to that outlined in Section 7.2 for hydrocarbons
was also performed for carbon monoxide (CO). Table 21 presents a summary
of the CO emission index correction factor coefficients developed using the
LN(CO EI) = f(T3). A description of the variables listed in this table was
given in Figure 29. Summary statistics for these coefficients are provided
in Table 22. The proposed CO correction factor is outlined below.
Figure 39 presents a plot of the CO combustor inlet temperature
coefficient T3C0EF versus rated engine idle pressure ratio. Comments similar
to those presented in Section 7.2 for HC are generally applicable to data
presented in this figure. First, CO temperature sensitivity increases with
rated engine pressure ratio. Second, newer technology engines tend to
exhibit more temperature sensitivity for a given IPR. This observation is
demonstrated by the high degree of CO temperature dependence found for the
JT9D-7 Vorbix. It should be pointed out, however, that HC and CO correction
factor coefficients do not always parallel each other. This observation is
demonstrated by the relationship between the CFM56 combustor rig HC coefficient
7-11
-------
which is quite large and the CO coefficient which falls on the general trend
line for engines of comparable idle pressure ratio.
The data presented in Figure 39 form the basis for the proposed CO
temperature correction factor. A line was drawn which represents the general
coefficient trend between the CO combustor inlet temperature coefficient
T3C0EF and idle pressure ratio IPR. This equation is plotted in Figure 39.
The proposed CO temperature correction factor is, therefore, a function of
engine idle pressure ratio. In equation form, this temperature correction
factor (C,^ ) can be expressed as
T3C0EF * (T3 . - T3.. )
C = e Reference Meas
CO
where T3C0EF is determined from Figure 39 and the appropriate idle pressure
ratio.
7.4 SMOKE NUMBER
Table 23 presents a summary of the EPA smoke number correction
factor coefficients developed using the model LN (Smoke Number) = f(LN(P3)j.
A description of the variables in this table is given in Figure 29. Summary
statistics for these coefficients are provided in Table 24. The proposed
smoke number correction factor is outlined below.
Efforts were made to find a suitable engine operating parameter
which correlated with the smoke number combustor inlet pressure coefficient
P3C0EF. Figure 40 presents a plot of this pressure coefficient versus rated
engine pressure ratio (PR). Although a general trend of decreasing pressure
coefficient with increasing rated pressure ratio is evidenced, this trend is
7-12
-------
for the most part the result of the smoke coefficient for a single engine,
the ALF502 which exhibits a high smoke coefficient. The remaining engines
in this figure represent large gas turbines (Class T2, T4) and have smoke
coefficients in the range 0.5 to 1.5 with a mean P3C0EF of approximately 1.0.
These figures can be compared to the G.E. smoke coefficient of 1.5 which was
discussed in Section 4.4.
inlet pressure coefficient is provided in Figure 41 where the P3COEF is plotted
versus SMKMEAN or the mean maximum power smoke number. As mean smoke emissions
rise, combustor inlet pressure sensitivity also rises. Of note in this
The discussion above and the data in Figures 40 and 41 provide the
basis for the proposed smoke correction factor. Using the smoke pressure
coefficient which is defined in general terms as:
An additional facet of the behavior for the smoke number combustor
figure are the smoke levels exhibited by the JT9D-7 Vorbix.
P3C0EF
LN(Smoke Number)p^ - LN(Smoke Number)p^
LN(P32) - LN(P3p
where
P32 y P3X PSIA .
The smoke number correction factor (Cp^
Smoke
) can be expressed as
C.
P3
Smoke
P3C0EF
Meas
where
P3C0EF
1.0.
7-13
-------
7.5 CORRECTION FACTOR SENSITIVITY ANALYSIS
In this section of the report, the sensitivity of each proposed
correction factor will be analyzed in light of a variety of sources of error.
Several of the more important considerations in analyzing these sources of
error in ambient effects data correction are listed below:
o Test-to-Test Variability
o Engine-to-Engine Variability
o Model Formulation Error
o T3, P3, HAMB Coefficient Selection Error
Test-to-test and engine-to-engine variability were previously
discussed in Sections 5.4.1 and 5.4.2. Model formulation error represents
the inaccuracy introduced into the correction factor process because of the
use of relatively basic emissions models as outlined in Figure 14. The errors
introduced by these fundamental models can be assessed by examining mean
2
RSQ (R ) values presented in coefficient summary Tables 18, 20, 22 and 24.
2 •
R or the coefficient of determination provides a measure of the percent of
the variation in a given set of data explained by the regression model
2
selected. For the regressions performed in the present study, the mean R
values are summarized below. Also included in this summary is the estimated
test-to-test variability. This test-to-test variability provides a reference
against which to judge the overall performance of the regression models
selected.
MODEL FORMULATION ERROR ASSESSMENT
Pollutant Model Percent Variation Explained Estimated
R2 * 100 Test-to-Test
MIN
MAX
MEAN
Variability
HC
LNHC = f(T3)
21
99
82
16
CO
LNCO = f(T3)
17
99
86
4
NOX
LNNOX = f(T3,HAMB)
73
99
91
3
SMOKE
LNSMK = f(LNP3)
5
96
60
33
7-14
-------
A quantitative discussion of coefficient selection error is
presented below. Correction factors for El's are given as follows:
For example, let:
P^fflFF
Smoke Number = AO * P3
where:
Smoke Number = EPA Smoke Number
P3 = Combustor Inlet Pressure (PSIA)
The smoke pressure correction factor would then be:
CF
P3
P3 \ P3C0EF
Reference
P3
Measured
where:
^Reference = Combustor Inlet Pressure at standard
day conditions
P3
Measured = Combustor Inlet Pressure as measured
under non-reference conditions
In a similar manner, a temperature correction factor can be
generated:
NOX EI = AO * e
T3C0EF * T3
where:
T3 = Combustor Inlet Temperature (Deg. F)
Then:
rv T3C0EF * (T3 - - T3W ,)
CF^j = e Reference . Measured
An ambient humidity correction factor is generated in a manner
similar to that for T3.
7-15
-------
HAMBCOEF * (HAMB _ - HAMB., ,)
CF,,.= e Reference Measured'
HAMB
A sensitivity analysis can be performed on the correction factors
by selecting values for the correction coefficients and allowing the inlet
(P3, T3) and ambient (HAMB) conditions to vary over ranges typically exper-
ienced by production engines. Plots of correction factors as a function
of inlet and ambient conditions are given in Figures 42 through 46. For
example, Figure 42 demonstrates NOX combustor inlet temperature correction
sensitivity. The values selected for the parameter T3C0EF cover the range
of T3 coefficients found for the various engine regressions done in this
study. T3n _ -T3., , represents the difference between reference
7 Reference Measured '
and measured combustor inlet temperatures which typically occur when testing
engines at other than standard day conditions. Thus if T3,n _ , -
6 (Reference)
T3,„ = -100, one would find a range of temperature correction factors
(Measured) 1
of (0.905 - 0.607) corresponding to a range of (0.001 - 0.005) for the T3
coefficient.
Figure 47 shows the percent error in the NOX temperature correction
factor as a result of using an incorrect T3 coefficient. In this case, it
is assumed that the true value of T3C0EF is .002, but a value of .004 was
used to calculate the correction factor. The range of T3 coefficients for
NOX regressions for the data used in this study is approximately .001 to .005.
If we had picked a single T3 coefficient (to use for all engines), say .003,
then the maximum difference between the "true" coefficient and .003 would be
.002. This plot, then, represents the maximum error due to incorrect tempera-
ture coefficient selection one might expect to find if a general model of the
fT3ff)FF * T3")
form NOX EI = AO *e J "is used for a group of engines.
Figures 48 through 51 give similar plots for HC and CO temperature
correction, NOX humidity correction, and SMOKE pressure correction.
7-16
-------
7.6 CORRECTION FACTOR PERFORMANCE ANALYSIS
Figures 52 through 55 indicate representative performance figures for
both the single engine correction factor coefficients and the general trend
line coefficients. Data are presented in these tables for the CFM56 combustor
rig and the ALF502 combustor rig. The reduction in emissions variability
achieved for these engines is comparable to that found when multiple tests are
performed on a single engine. In other words, the major sources of variability
in the emissions measurements for data presented in these tables are ambient
conditions and test-to-test variability. The specific performance figures used
to assess the correction factor performance are the uncorrected data coefficient
of variation or
CY - Q """un corrected
uncorrected uncorrected
and the corrected data coefficient of variation
CV - ^ orrected
corrected corrected
where = the standard deviation of the emissions data
= the mean of the emissions data
Ideally, since the correction factor process attempts to reduce
O . , as much as possible, CV . , will approach 0 for perfect
w corrected r corrected v
correction to standard day conditions. The primary limitations on the
ability to achieve this ideal correction have been discussed in previous
sections and are amplified below. First, the inherent test-to-test variability
in emissions data which is typically on the order of magnitude of 5% for NOX,
10% to 20% for HC and CO, and 40% for SMOKE provides a lower bound on the
achievable reduction in emissions variation. Second, the use of simplified
emissions models provides a further limitation on the correction factor performance.
Finally, the correction coefficient selection error introduces a final major
limitation on the correction process. This coefficient selection error is illus-
trated by the differences in performance between data correction using a correction
coefficient developed for a specific engine and data correction using a general
trend line coefficient.
7-17
-------
The effect of these limitations on the emissions correction process
can be further illustrated by Figure 56 where a hypothetical emission index Y
(e.g., NOX EI) is plotted versus a typical correction factor correlating
parameter X (e.g., T3). As shown in this figure, the uncorrected or measured
emission data increases linearly with the correlating parameter. This
phenomenon can be expressed in equation form as Y = a + bX where a and b
are regression coefficients. The principal effect of the data correction
process is to transform this emissions data so that the corrected emission
index is independent of the correlating parameter. Graphically, the ideal
correction should transform the corrected data onto a perfect horizontal line.
(See Appendix F for sample emission index corrections.) The use of simplified
emissions models and test-to-test variability, however, provide limitations on
how much reduction in variability can be practically achieved. The nature
and extent of these limitations can be evaluated in quantitative terms
as follows:
i^ uncorrected emission index measurement
i**1 correlating parameter measurement
number of measurements
variance of uncorrected emission indices
iyi - y)2
N - 1
variance of the uncorrected emissions indices
around the computed emission index line
(residual variance)
- v)2
i
N - 2
let y.
Then
cry
2
O V
yx =
7-18
-------
where: Y = Estimate of the emission index y. computed
from the regression equation
Y = a + bx.
1
y = Mean uncorrected emission index
A
The relationship between (3 and CP can be shown to be:
r y yx
cr /7~7 £4
y \i N-2
Correlation coefficient between y and x
Coefficient of determination (see Section 5.3)
Provides a measure of the variance in the
emissions data not accounted for by the
regression model chosen and principally includes
the effects due to test-to-test variability and
the use of simplified emissions models. The
effect of coefficient selection error provides
an additional source of variability in this
analysis. An assessment of the magnitude of
this error component was presented in Section 7.5
and Figures 47 to 51.
The residual variance component as shown in the lower half of Figure
56 will remain even after the emissions data is corrected to reference condi-
tions. The relationship between CVuncorrected and CVcorrected (assuming
uncorrected ~ -^corrected
, N is large, ^"^y_^G"corrected)
can be expressed as
^corrected ~ V 1 ~ R ^uncorrected
In order to assess the implications of the above relationship on the correc-
2
tion factor process, the mean, maximum and minimum R values presented in
Section 7.5 for the entire emissions data base were used to compute the range
cr
yx
where R
R2
2
7-19
-------
of expected reduction in emission index variability for each of the four
pollutants.
Percent Reduction =
^uncorrected ^corrected
CV
100'
uncorrected
(1
-
R ) * 100%
Pollutant
Mean R
Statistic
Percent Reduction
Emission Index
Variability
MIN
MAX
MEAN
NOX EI
.91
48
90
70
HC EI
.82
11
90
58
CO El
.86
9
90
63
SMOKE NUMBER
.60
3
80
37
A comparison of the above expected percent reduction in emission
index variability with that actually computed for the CFM56 and ALF502 combustor
rigs for NOX, HC, CO and Smoke in Figures 52 to 55 reveals the following.
First, the CFM56 and ALF502 NOX, HC and CO emissions corrections for the
specific engine coefficients (NOX = 68%, HC = 54%, CO = 54%) and the NOX
and HC general coefficients (NOX = 65%, HC = 53%) provide reductions in
emission index variability close to the mean percent reduction expected for
each pollutant. Second, correcting the CFM56 rig CO data using the general
coefficient provides substantially less reduction in emission index variability
(e.g., at idle, 24% reduction vs 52% reduction) than use of the coefficient
developed specifically for this rig. The principal reason for this occurrence
as seen in Figure 39 is the difference between the specific engine CFM56 CO
temperature correction coefficient and that predicted by the general trend line
Finally, the ALF502 smoke number corrections (5-9% reduction in variability)
perform significantly poorer than the mean expected percent reduction for smoke
number emissions, 37%. The small and highly variable smoke numbers (due
7-20
-------
primarily to the availability of only low power data) measured for this engine
provide an explanation of the performance of the smoke correction on this
particular data set. In general, a reduction in smoke number variability
on the order of 30-40 percent can be expected.
7-21
-------
8.
REFERENCES
1) "Control of Air Pollution from Aircraft and Aircraft Engines,"
Federal Register, Volume 38, No. 136, Part II, July 17, 1973.
2) Slogar, G.A., Determination of Effects of Ambient Conditions
on Aircraft Engine Emissions Engine Testing - GTCP 85 APU,
TPE331 Turboprop, Volume 1, EPA Report 460/3-76-009-a,
March 1976.
3) Slogar, G.A. and Holder, R.G., Determination of Effects of
Ambient Conditions on Aircraft Engine Emissions Engine Testing -
GTCP85 APU, TPE331 Turboprop, Volume 2, EPA Report 460/3-76-009-b,
March 1976.
4) McAdams, H.T., Analysis of Aircraft Exhaust Emission Measurements -
Statistics, EPA Report NA5007-K-2.
5) Blazowski, W.S., Walsh, D.E. and Mach, K.D., "Operating and
Ambient Condition Influences on Aircraft Gas Turbine NOX Emissions"
Journal of Aircraft, Vol. 12, No. 2, February 1975, pp. 113.
6) Lipfert, F.W., "Correlation of Gas Turbine Emissions Data,"
ASME Paper 72-GT-60, 1972.
7) Sarli, V.J., Eiler, D.C., and Marshal, R.L., "Effects of Operating
Variables on Gaseous Emissions," Paper presented at Air Pollution
Control Association (APCA) Specialty Conference on Air Pollution
Measurement Accuracy as it Relates to Regulation Compliance,
New Orleans, October 1975.
8) Durkee, K.R., Noble, E.A., and Jenkens, R., Standards Support and
Environmental Impact Statement - An Investigation of the Best
Systems of Emission Reduction for Stationary Gas Turbines, EPA
Office of Air Quality Planning and Standards Preliminary Report,
July 1976, pp. 3-60.
9) Blazowski, W.A. and Marzewski, J.W., Ambient Correction Factors
for Aircraft Gas Turbine Idle Emissions, Interim Technical Report
for the Period January 1975-March 1976, Technical Report AFAPL-TR.
10) Op. Cit., Reference 8, pp. 3-55.
11) Souza, A.F., Reckner, Louis, R., Variability in Aircraft Turbine
Engine Emissions Measurements, EPA Report No. 460/3/74-006,
January 1974.
12) Souza, A.F., Further Investigation Into the Causes of Variability
in Aircraft Turbine Emission Measurement, EPA Report No.
460/3-75-011, November 1975.
8-1
-------
13) Munt, R. and Danielson, E., Aircraft Technology Assessment -
Status of the Gas Turbine Program, EPA Division of Emission
Control Technology Report, December 1976.
14) Op. Cit, Reference 13, pp. 80.
15) Shaw, H., "The Effect of Water on Nitric Oxide Production in
Gas Turbines," ASME Paper 75-GT-70, 1975.
8-2
-------
TABLES
-------
Table 1
ENGINE/COMBUSTOR RIG SUMMARY - EPA SPONSORED DATA
ENGINE/RIG
TESTS^
EPA
CLASS
TEST
CONTRACTOR
TEST
TYPE
RATED
POWER
TPE331-5-251
336
P2
AiResearch
Control led
f 2)
706 SHP
GTCP85-98CK
240
APU
AiResearch
Controlled
260 HP*-3-1
ALF502 Engine
56
T1
AVCO-Lycoming
Uncontrolled
6,500 lb thrust
ALF502 Rig
66
T1
AVCO-Lycoming
Controlled
6,500 lb thrust
CFM56 Rig
214
912
T2
General
Electric
Controlled
22,000 lb thrust
^Test is defined as a given thrust/ambient condition combination
(e.g., Idle represents one test, 1.5 Idle represents a second).
Derated from 840 SHP.
(3)
Maximum
power operation (45.7 SHP + 214.3 Bleed-Air HP).
EPAP
SMOKE
ENGINE
HC
CO
NOX
NUMBER
TPE331-5-251 1979 Standard
4.9
26.8
12.9
47
MeasuredA
3.6
12.8
8.9
15
GTCP85-98CK
ALF502
CFM56
(lb/1000 hp-hr/LTO)
1979 Standard
Measured®
(lb/1000 hp-hr)
1979 Standard
Measured^
(lb/1000 lbf/hr/LTO)
1979 Standard
Measured^*
(lb/1000 lbf-hr/LTO)
0.4
0.2
1.6
1.5
0.8
1.7
5.0
7.5
9.4
12.5
4.3
12.8
3.0
6.4
3.7
3.0
3.0
4.7
LTO = Landing Takeoff Cycle
Taxi
None
34
23
23
Idle operation with primary atomizers only, Reference 2
D
Reference 2
C ALF502 Test Data - TAMB=63°F, PAMB=29.88 in Hg, HAMB=0.0088 lb H20/lb Air
^ PFRT baseline engine reported in Reference 13; no smoke number reported
9-3
-------
Table 2
INDUSTRY CONTRIBUTED DATA - SELECTED EXPERIMENTAL DESIGN FEATURES
Single Engine
o Generally uncontrolled ambient conditions (e.g., TF30).
o Provides a measure of test-to-test variability if
replicates included.
o Test points correspond to normal engine operation (e.g.,
Idle, Takeoff).
Multiple Engine
o Generally uncontrolled ambient conditions
(e.g., Pratt $ Whitney JT8D and JT9D Pilot Lot Data).
o Provides a measure of engine-to-engine variability.
o Test points correspond to normal engine operation
(e.g., Idle, Takeoff).
Parametric Tests
o Always controlled ambient conditions
(e.g., T56 Rig and T63-A-5A Rig).
o Independent pressure, temperature or humidity variation.
o Combustor rigs only.
o Test points generally do not correspond to normal engine
operation (e.g., Idle, Takeoff).
9-4
-------
Table 3
AMBIENT TEST CONDITIONS
TPE331-5-251 AND GTCP85-98CK APU
Temperature
Humidity
Barometric Pressure
Test
Point
°C
°F
Grams 1^0/
Kilogram Air
Grains H2O/
Pound Air
Mm
Hcj
Inches Hq
1
-7.0
19
1.0
7.0
2
-7.0
19
2.0
14.0
3
4.0
39
2.0
14.0
4
4.0
39
5.0
35,0
5
15.0
59
2.0
14.0
6
15.0
59
5.0
35.0
7
15.0
59
7.5
52. 5
Hold constant at
8
15.0
59
10.0
70.0
standard condition as
9
33.0
91
10.0
70.0
specified in
the Unit-
10
33.0
91
15.0
105.0
ed
States Standard
11
33.0
91
20.0
140.0
Atmosphere, 1962 for
12
33.0
91
25.0
175. 0
the test cell altitude.
13
15.0
59
7.5
52. 5
580
22
.83
14
15.0
59
7.5
52.5
650
25
.98
15
15.0
59
7.5
52. 5
740
29
.13
16
15.0
59
7.5
52.5
500
19
.69
9-5
-------
Table 4
TPE 331-5-251 ENGINE LOAD CONDITIONS
(CORRECTED TO 59°F AND 29.92 IN. HG ABS)
Condition
Load
Percent of
Rated Power
Corrected
Total Shaft
Horsepower
1*
(Taxi) Idle
5
35
2*
1.5 x Idle
7.5
53
3*
2.0 x Idle-
10.0
76
4
Approach
30.0
212
5
Cruipe
70.0
494
6
Climb-Out
90.0
635
7 * *
Takeoff
100.0
706
~Engine operation on primary fuel atomizers only.
**A derated takeoff Joad of 706 shp//06 was necessary to protect
the engine from turbine interstage overtemperature.
9-6
-------
Table 5
GTCP85-98CK APU ENGINE LOAD CONDITIONS
(CORRECTED TO 59<>F AND 29.92 IN. HG ABS)
Condition
Load
Horsepower
Shaft
Bleed-Air
Total
1
Idle
-
-
-
2
Pure Shaft Power
79.5
0
79.5
3
Rated Pov.'er
198.6
0
198.6
4
Pure Bleed
-
245.0
245.0
5
Combinat.i
45.7
214.3
260.0
9-7
-------
Table 6
AMBIENT TEST CONDITIONS - ALF502 FULL-SCALE ENGINE
UNCONTROLLED AMBIENT CONDITIONS
Ambient Ambient Ambient
Temperature Pressure Specific Humidity
°F in 11G Lb Water/Lb Dry Air
77 30.00 .01445
71 30.00 .01632
76 29.80 .01530
78 29.75 .01770
73 29.70 .01725
78 29.75 .01582
63 29.85 .0089
7 Power Levels for Each Ambient Test Condition
1) IDLE
a) Taxi-Out
b) Taxi-In
2) 1.5 IDLE
3) Approach (30% rated)
4) 40% Rated
5) 50% Rated
6) CLIMB (90% Rated)
7) TAKEOFF (100% Rated)
9-8
-------
Table 7
AMBIENT TEST CONDITIONS - ALF502 COMBUSTOR RIG
AMBIENT AMBIENT AMBIENT
TEST' TEMPERATURE PRESSURE SPECIFIC HUMIDITY
°F in HG Grains H^O/lb Air
0 59 29.92 44 (Reference Day)
1 19 25.98 25
2 19 29.92 25
3 19 52.28 25
4 59 29.92 25
5 39 29.92 25
6 59 25.98 52.5
7 59 29.92 25
8 59 29.92 52.5
9 59 29.92 70
10 59 52.28 52.5
11 85 29.92 25
12 85 29.92 52.5
13 85 29.92 105
14 85 29.92 175
15 105 25.98 25
1(> 105 25.98 175
17 105 29.92 25
18 105 29.92 52.5
19 105 29.92 105
20 105 29.92 175
21 105 32.28 25
22 105 52.28 175
3 Power Levels for Each Ambient Test Condition
1) IDLE 5°o Rated
2) 1.5 IDLE 7.5% Rated
3) APPROACH 30°. Rated
9-9
-------
Table 8
AMBIENT EFFECTS TEST CONDITIONS - CFM56 COMBUSTOR RIG
Temper;
t: u i'i'
Hi>!!> Ldi t y
Test
Grams HjjO I I I'
C'l nj'us H O
Point
°C
°F
J' i 1 c i m A ,i r
I'rr l'n'ind A
1
-7
19
2
34
2
4
39
2
34
3
15
59
2
14
*
29. 5
ii 5
2
1/1
5
40
:i o.'i
2
14
6
4
3 S>
5
35
7
IS
59
' 7, 5
t.:-:. 5
8
29.5
85
7.5
5:',.;.
9
40. r.
,lo;»
7.5
r>:>.r,
10
in
59
30
70
1).
?¦<.). 5
85
3 5
105
12
40. 5
105
35
105
13
29. 5
IS5
25
375
14
40. 5
3 05
25
3 75
15
-7
19
2
14
1G
15
59
7.5
52.5
17
40.5
3 05
2
14
18
40. :>
3 05
25
175
19
-7
3 9
2
14
20
15
59
7.5
52.5
21
40.5
3 05
2
34
22
40.5
3 05
25
175
22 - Test
point:;
- given
above.
5 - Po'A'l'l
sett in
¦<\ - id!
e, '}¦}/'/
idle, .' i ] > ] > j' t >: n ¦ 1) j
l'n
Kl'a, Inches II j,
Ho'lU constant as
;;p<;e. i .Tiecl in the
lli.iUii Slates
S rd A tm« >. ;;.!u i'C
1902 J or the I i'S I
Cell Alt.itm.li.*
V
10H.32
32.28
3 on.32
32.28
10'.). 32
32.28
100.32
32.28
US
25.98
till
25.98
fcf<
25 .91!
88
25.98
1 - Hojj I i eat i on oJ rr.'ich I.!¦ poin/ - )\uh jjoinl will Un.•>•*»./he run
twice.
214 - Total measurements (22 x 5 x 2 = 220) Less 6 unrealizable takeoff points.
9-10
-------
Table 9
CFM56 COMBUSTOR RIG REPLICATE ANALYSIS
MEAN REPLICATION COEFFICIENT OF VARIATION BY MODE
2 REPLICATES/TEST POINT, 22 TEST POT NTS/MODE
Coefficient of Variation for Test Point =
where = mean value of the 2 replicates
A
cr* = standard deviation of the 2 replicates
Pollutant
MODE
(% Rated)
HC
EI
CO
EI
NOX
EI
SMOKE
IDLE
(5)
.2232
.0365
.0401
.4467
1.5 IDLE
(9)
. 1872
.0393
.0277
.3676
APPROACH
C30)
.1941
.1413
.0355
.5743
CLIMBOUT
(85)
-
-
. 0336
.2656
TAKEOFF
(100)
-
-
.0351
.2186
TOTAL
. 201S
.0724
.0344
. 3746
9-11
-------
Table 10
TPE331-5-251 FULL SCALE ENGINE REPLICATE ANALYSIS
3 REPLICATES/TEST POINT, 16 TEST POINTS/MODE
Coefficient of Variation for Test Point =
Vx,
where X< = mean value of the 3 replicates
= standard deviation of the 3 replicates
Pol lutant
MODE
(% Rated)
IDLE
I§)
1.5 IDLE
(7.5)
2.0 IDLE
(10)
APPROACH
(30)
CRUISE
(70)
CLIMBOUT
(90)
TAKEOFF
(100)
HC
EI
. 1086
. 0843
.0969
. 1510
CO
EI
. 0363
. 0276
.0325
.0522
NOX
EI
.0365
. 0340
. 0302
.0223
.0176
.0181
.0159
SMOKE
. 1430
1594
. 1710
. 0649
.0556
.0605
.0584
TOTAL
1102
. 0372
.0249
. 1018
9-12
-------
Table 11
TEST TO TEST VARIABILITY
MEAN IDLE COEFFICIENT OF VARIATION BY ENGINE
MEAN COEFFICIENT OF VARIATION
NUMBER
NUMBER
HC
CO
NOX
ENGINE
TESTS
REPLICATES
El
El
El
SMOKE
TPE331-5-251 FULL SCALE
16
3
.1086
.0363
.0365
.1430
CFM56 COMBUSTOR RIG
22
2
.2232
.0365
.0401
.4467
ALF502 FULL SCALE
7
2
.0771
.0315
.0070
.3646
WEIGHTED AVERAGE
45
.1597
.0357
.0338
.3259
9-13
-------
Table 12
CFM56 COMBUSTOR RIG - SUMMARY STATISTICS BY MODE
IDLE
VARIABLE
H
MEAN
STANDARD
MINIMUM
MAXIMUM
VARIANCE
STD ERROR
o
<
DEVIATION
VALUE
VALUE
OF MEAN
PTHRUST
44
5.70414091
0.10116185
5.5132002(0
5.8SP20030
0.01023372
0.01525072
1.773
P3
44
48.33031313
3.16793993
42.36 700000
53.553»0000
10.03584371
0.47758492
6.551
T3
44
363.67498864
43.97538157
292.0J0.5T.SrmC.?'
427. 7BrBBMSe
1933.92213617
6.62968897
11.896
FA
44
0.0!575455
73950
,7.01 41.3330
ii.E\ -sbcen
0.00000062
0.0JP311904
5.012
CO
44
25.68772727
6.40534544
l s. 82.er000.orer
44. B8 3: *000/7
42.06620402
0.97777814
25.249
HC
44
1.30363636
1 .28113434
0.17300303
6. 22 330000
1.64122833
0.19313375
98.272
NOX
44
3.15353132
3.4 7 939.321
2.36'30300
4 .17 'TOT00
0.22952743
0.07222557
15.182
SNOKE
44
2.72272727
1 .19 j 6 15 l ?
0.23.5 30003
5.5£"333?'"03
1 . 43668/776
0. 1 836981 9
44.023
l.S> IDLfc
PTHRUST
44
8.97313636
0.22195320
8.57730307
9.2254:7003
£.04926983
0.03346297
2.474
P 3
4 1
63.39704 5 4 i
4.14^43954
55.09333237
70. 923F0i'33
17. 19296C«4
0.62=09923
6.540
T3
44
4 37.41590G82
" 7 . 47 ":3321 9
352.39393303
496.35??r£?r3
2254.291732 V 2
7 .15/78613
10.855
FA
44
.0.014 834 5 5
3. 00.75 3 993
3.01350333
3.316*0333
3.03300335
0.00.7.09345
4.053
CO
44
12.94Q50B91
3. & 3 4.9271 4
9.110.300.3?
21. 83333333
9.235 3235"
0. 4573968.3
23.435
HC
44
0. 1 5.i3S354
<3.17 2~ ii37n
0.02.700.035
0.81333333
3.32933S7;
3.02S36321
110.565
NOX
44
4 . 1 2.3*09/79
0.72 ".9164;
2.7433J330
5 .7£".330333
0.5240S457
3.10913451
17.567
SMOKE
44
£.61i6363i
3.9i J9455 J
0. 2.3 '''33033
4.73333033
0.84446089
0.13853625
35.160
PTHRUST
44
30.37'00305
3. 00.*J..7/7
33.03.3.30.33 7
33.03^33333
3.00333333
0.33 ".30303
0.000
P 3
44
1 3S - 1 3 1 3
j. 12 90927';
122. 7333000.3
154.041.3303 :r
33 . 3433338';
1.37526251
6.568
T3
44
654.]S132727
59 .05 ',:5993
5*7.00733333
719. 2 .'-''333333
3487.75335^43
0.90321692
9.028
FA
44
0.31783864
3.03.7353 13
0.01623333
0.01-:T.3333
0 .003.3337
3.00.<'23684
0.00101832
100.00000000
100.00450000
0.03300104
0.00016519
0.001
P3
30
171.56657695
10.42613719
152.51300000
191.07300000
108.70433663
1.69134273
6.077
T3
38
958.38158158
68.66313081
840.50300000
1049.00300000
4714.62553322
11.13862055
7. 164
FA
38
0.02714474
0.00164213
0.02410000
0.02940000
0.00000273
0.000265 39
6.050
CO
33
3.06032632
1.08758037
1.44000000
5.4 3300000
1.18283215
3.17542891
35.536
HC
33
0-/3133303'
3.33 ",;3J0.;
3.31 •'333 "J
3.313: ;;3.~:
3.0 J.-33333
3.30333333
3.333
NOX
33
15.23530033
2.8771624!
1 1.973J0300
21.91333303
8 . 27fT3635 1
0.46573737
18.635
SMOKE
33
4.1157894/
1 .27*72.71.3
1 . 7U3B'lB3:i
8. 303333,35
1.63301422
0.257J2 135
31 .020
-------
Table 13
CFM56 COMBUSTOR RIG - HC REGRESSION SUMMARY
GENERAL LINEAR MODELS PROCEDURE
*£>
I
in
DEPENDENT VARIABLE:
SOURCE
MOOEL
ERROR
CORRECTED TOTAL
PARAMETER
INTERCEPT
T3
L*HC NA"U?AL LOG(HC)
9* '¦UK O- SQUARES
'.55.02231252
85 22.32196483
87 1?7.84447711
ES~IM*TE
8.23545584
-0.02J485S2
T f3R HO:
PA ITI
0.0001
0.0001
r VALUE
584.17
STD ERR03 OF
ESTIMATE
C.39535633
C'. 00C 97171
PR > F
0.0001
SYD DEV
0.51514239
R-SOUARE
0.871675
C.V.
12.1495
LNHC MEAN
1.22217993
-------
Table 14
CFM56 COMBUSTOR RIG - CO REGRESSION SUMMARY
"ENERAL LINEAR MODELS PROCEDURE
<£>
DEPENDENT VARIABLE: LNCO
SOURCE PF
MODEL 1
ERROR 86
CORRECTED TOTAL S7
PARAMETER
INTERCEPT
T2
NATURAL LQC(CO)
SUM OF SQUARES
]1.66575453
2.97533233
14.6416R6P3
ESTIMATE
S.A7 590815
~0.00 644263
7 FOR J0:
PA*AMETtK=0
32.30
- 12.35
MEAN SQUARE
11 .66575450
0 03463336
PR > ITI
0.0001
0.0001
F VALUE
337.12
STD ERROR OF
ESTIMATE
0.142C8231
C.00035C89
PR > F
0.000!
STD DEV
0. 18602111
R-SQUARE
0.73c7t9
C.V.
6.4658
LNCO MEAN
2.87699378
-------
Table 15
CFM56 COMBUSTOR BIG - NOX REGRESSION SUMMARY
GENERAL LINEAR MODELS PROCEDURE
to
I
DEPENDENT VARIABLE: LNNOX
SOURCE DF
MODEL 2
ERROR 211
CORRECTED TOTAL 213
PARAMETER
INTERCEPT
T3
HAMB
NATURAL LOG(NOX)
SUM OF SQUARES
35.55307798
1.17695298
86.73003096
ESTIMATE
0.46742286
0.00250325
-20.70200748
T FOR HO:
PARAMETERS
30.28
120.40
-32.91
MEAN SQUARE
42.77653899
0.00557798
PR > 1TI
0.0001
0.0001
0.0001
F VALUE
7668.83
STD ERROR OF
ESTIMATE
0.01543587
0.00002075
0.62901159
PR > F
0.0001
STD DEV
0.07468585
R-SQUARE
0.986430
C.V.
3.8478
LNNOX MEAN
1.94101868
-------
Table 16
CFM56 COMBUSTOR RIG - SMOKE REGRESSION SUMMARY
GENERAL LINEAR MODELS PROCEDURE
<£>
t—1
00
DEPENDENT VARIABLE:
SOURCE
HODEL
ERROR
CORRECTED TOTAL
PARAMETER
INTERCEPT
LNP3
LNSMK NATURAL LOG(SMOKE)
DF SUM OF SOUARES
1 4.. 64120782
212 88.26363650
213 92.904844 32
ESTIMATE
-0.31230670
0.27750009
T FOR HO:
PARAMETERS
-0.81
3.3*
MEAN SQUARE
4.64120782
0.41633791
PR > ITI
0.4207
0 .0010
F VALUE
11.15
STO ERROR OF
ESTIMATE
0.38714494
0.08311332
PR > F
0.0010
STD DEV
0.64524252
R-SOUARE
0.049957
c.v.
66.3910
LNSMK MEAN
0.97188252
-------
Table 17
NOX CORRECTION FACTOR COEFFICIENTS - TABULAR SUMMARY
NOX CORRECTION COEFFICIENT SUMMARY
S
C
0
H
A
H
N
E
A
N
T
M
A
T
P
0
N
N
C
M
S
3
T
B
M
3
3
X
C
0
L
P
T
C
3
C
B
M
M
M
I
D
A
L
A
0
S
0
S
R
E
E
E
N
E
S
E
N
E
T
E
T
S
A
A
A
P
E
S
S
S
T
F
D
F
D
Q
N
N
N
R
GTCP85-98CK APU
BLEED * SHAFT
APU
9
1.0507
0.001677
0.000331
-22.84
2.360
0.7859
357
56.4
5.20
4.00
T56 STANDARD JETA
PR"2
PZ
16
0.0120
0.001785
0.000292
.
.
0.7278
190
30.0
1.43
2.00
TS6 LEAN JETA
PR-2
P2
16
0.3258
0.002018
0.000181
.
.
0.8993
188
30.0
2.04
2.00
T56 STANDARD OETA
PR-3
P2
IS
0.0567
0.002853
0.000220
.
.
0.9232
283
45.0
2.42
3.00
T56 LEAN JETA
PR»3
P2
16
0.5212
0.001469
0.000193
.
0.8054
282
45.0
2.56
3.00
T56 STANDARD JETA
PR-4
P2
16
0.3925
0.001945
0.000199
0.8722
355
60.0
2.99
4.00
T56 LEAN JETA
PR«4
P2
16
0.3400
0.002173
0.000208
.
0.8867
356
60.0
3.09
4.00
T56 STANDARD JETA
PR-5
PZ
16
0.6065
0.001627
0.000210
.
0.8107
417
75.0
3.65
5.00
TS6 LEAN JETA
PR-5
PZ
16
0.1884
0.002699
0.000166
.
0.9499
417
75.0
3.81
5.00
T63-A-5A
WALL FILMt FA IN
EO)
PZ
9
-9.9541
0.003030
0.001026
.
.
0.9478
622
60.7
5.33
6.20
T63-A-5A
PRES ATOM! FA IN
EQ)
P2
9
-5.2671
0.001999
0.000622
.
0.8229
600
60.2
5.83
6.20
T63-A-5A
AIR BLAST< FA IN
EO)
PZ
9
-5.3716
0.001925
0.000487
.
0.8827
600
60.5
5.51
6.20
PT6A-50
IDLE TO TAKEOFF
P2
15
0.7037
0.002668
0.000072
.
.
0.9906
637
126.0
11.17
8.70
ALLISON 501K
LOU RPM
PZ
7
0.3612
0.003046
0.000369
.
0.9317
403
66.3
5.02
9.40
ALLISON 501K
HIGH RPM GT 90X
RATE
P2
19
0.5855
0.002882
0.000292
.
0.8514
599
135.9
20.23
9.40
TPE331-5-251
APPROACH TO TAKEOFF
PZ
19Z
0.5517
0.002864
0.000301
-14!13
0.094
0.7584
655
142.0
10.56
10.37
TYNE
IDLE TO CRUISE
PZ
28
0.4647
0.002602
0.000486
-17.83
1 .580
0.8419
700
180.0
7.70
13.50
ALFSOZ COMBUSTOR RIG
IDLE TO APPROACH
T1
66
0.5328
0.002725
0.000097
-20.09
1.220
0.9316
456
75.0
4.99
5.11
JT150-4 P&W CANADA
IDLE TO TAKEOFF
Tl
1Z
0.3697
0.003205
0.000109
-24.35
5.040
0.9898
624
152.0
9.44
9.65
ALF502 ENGINE
IDLE TO TAKEOFF
T1
56
0.7147
0.002400
0.000051
-15.01
3.100
0.9773
698
154.0
8.19
10.70
TFE731-2
IDLE TO TAKEOFF
Tl
28
0.4411
0.002936
0.000097
.
0.9722
754
186.0
15.88
13.00
TFE731-3
IDLE TO TAKEOFF
Tl
14
0.5427
0.002881
0.000125
.
.
0.9780
792
197.0
18.49
14.60
CFM56 COMBUSTOR RIG
IDLE
T2
44
0.5638
0.001994
0.000189
-17.96
0.970
0.8954
370
48.4
3.16
CFMS6 COMBUSTOR RIC
2.5MDLE
T2
44
0.5989
0.002267
0.000171
-21.55
0.940
0.9289
437
63.4
4.12
.
CFM56 COMBUSTOR RIG
APPROACH
T2
44
1.0522
0.001744
0.000155
-21.38
1.080
0.9062
654
139.0
7.48
.
CFM56 COMBUSTOR RIG
CLIMB
T2
44
0.5770
0.002352
0.000114
-19.31
0.910
0.9356
918
166.0
13.19
.
CFM56 COMBUSTOR RIG
TAKEOFF
TZ
38
0.4949
0.002425
0.000121
-16.95
1 .350
0.9292
958
171.6
15.24
.
CFM56 COMBUSTOR RIG
IDLE TO TAKEOFF
T2
214
0.4674
0.002503
0.000021
-20.70
0.630
0.9864
958
17Z.0
I5.Z4
11.70
TF30
IDLE TO TAKEOFF
T2
178
0.2605
0.003324
0.000036
-12.01
1.340
0.9801
740
210.0
13.44
14.20
SPEY 511
IDLE TO TAKEOFF
T2
1Z0
-0.8503
0.004674
0.000142
0.9016
826
265.0
8.0Z
18.90
0T9D-7 VORBIX
PILOT FA LT .009
T2
61
-0.5154
0.003514
0.000125
.
0.9301
887
314.0
13.99
20.40
0T9D-7A
IDLE TO TAKEOFF
T2
181
-0.1987
0.004156
0.000048
-19.94
2.370
0.9771
845
258.7
24.68
20.40
JT9D-7F
IDLE TO TAKEOFF
T2
94
-0.3103
0.004422
0.000081
-16.12
4.240
0.9714
820
268.9
27.45
21.20
JT9D-7 OVERHAUL
IDLE TO TAKEOFF
T2
1 14
-0.4312
0.004307
0.000064
0.9760
899
283.0
34.19
22.30
RB211-22B COMBUST #1
IDLE TO TAKEOFF
T2
51
-0.6017
0.004290
0.000093
.
0.9888
907
351.0
27.98
25.00
RB211-22B COMBUST #2
IDLE TO TAKEOFF
TZ
16
-0.7541
0.004412
0.000133
.
0.9874
901
341 .0
27.13
25.00
RB211-22B ALTITUDE
IDLE TO TAKEOFF
T2
Z1
-0.7016
0.004357
0.000119
.
.
0.9857
849
260.0
20.8?
25.00
JTBD-9
IDLE TO TAKEOFF
T4
134
0.3033
0.003204
0.000082
0.9209
719
198.4
20.94
15.90
JTBD-17
IDLE TO TAKEOFF
T4
85
0.4502
0.002989
0.000109
•
.
0.9013
740
207.8
15.27
17.00
-------
Table 18
NOX CORRECTION FACTOR COEFFICIENTS - SUMMARY STATISTICS
NOX CORRECTION COEFFICIENT SUMMARY
VARIABLE
LABEL
N
MEAN
STANDARD
MIHIMUM
MAXIMUM
RANGE
DEVIATION
VALUE
VALUE
SAMPLES
SAMPLE SIZE
34
55.00000000
61.56888234
7.00000000
214.00000000
207.00000000
CONSTANT
REGRESSION CONSTANT
34
-0.43273824
2.18601102
-9.95410000
1 .050700 '"0
11.004800jO
T3C0EF
•NOX TEMPERATURE COEFFICIENT
<4
0.00232826
0.00090634
0.00!46300
0.004674-0
C.003205UU
T3STD
T3COEF STANDARD ERROR
:4
0.00021138
0.00020130
0.00002100
0.001025O0
0.00100500
HAMBCOEF
NOX HUMIDITY COEFFICIENT
i'j
-18.30200000
3.96678095
-24.35000000
-12.010000 00
12.340000 JO
HAMBSTD
HAMBCOEF STANDARD E"ROR
10
2.19740000
1.56617611
0.09400000
5.04000000
4.94500000
RSQ
COEFFICIENT OF DETERMINATION'-SQUARED)
34
0.91305471
0.07437623
0.727300J0
0.93060jOu
u.26280000
T3MEAN
MEAN CCMS'JSTOR INLET TEH°tRATURE t D F <
14
519.8E235294
225 . 39094 2 (31
188.00000000
958.OOOOOOuO
770.OOOOOOuO
P3MEAN
MEAN COMBbSTOR INLET PRESSURE( ? S IA :
34
152.994 i1765
97 . 1 1157888
30.00000000
351.00000000
321.0jOOOOuO
NOXMEAN
MEAN NOX EMISSION r«5EX
34
1 1 .90441 176
9.05046330
1 .43000000
34. 130000i'C
32.76000000
PR
RATED EKGINE PRESSURE RATIO
34
11 .53029412
7.32116131
Z.OOOOOOOu
25.000000
23.00000CD1
I
^ NOTE: ONLY CFM56 IDLE TO TAKEOFF COEFFICIENT FROM TABLE 17 INCLUDED IN ABOVE SUMMARY
-------
Table 19
HC CORRECTION FACTOR COEFFICIENTS - TABULAR SUMMARY
HC CORRECTION COEFFICIENT SUMMARY
ENGINE
MODES
CLASS
SAMPLES
CONSTANT
T3COEF
T3STD
RSQ
HCMEAN
PR
IPR
GTCPB5-98CK APU
BLEED & SHAFT
APU
9
9.5625
-0.029100
0.004059
0.8801
0.49
4.00
4.00
T63-A-5A
WALL FILM INJECTION
P2
9
34.4800
-0.010745
0.003386
0.8232
4.68
6.20
T63-A-5A
PRESSURE ATOMIZER
P2
7
16.3600
-0.008126
0.003307
0.8034
0.69
6.20
.
ALLISON 501K
HIGH RPM LT 30XRATED
P2
21
6.0081
-0.009915
0.001105
0.8091
2.32
9.40
.
PT6A-50
IDLE TO APPROACH
P2
8
6.2118
-0.019231
0.000619
0.9938
12.04
8.70
2.20
TPE331-8
IDLE
P2
4
6.6488
-0.017719
0.002916
0.2064
1.54
10.30
3.20
ALF502 COMBUSTOR RIG
IDLE TO APPROACH
T1
66
3.8194
-0.008304
0.000486
0.8201
5.22
5.11
1.73
JT150-4 PAW CANADA
IDLE TO APPROACH
T1
6
6.9240
-0.016619
0.002715
0.9035
2.84
9.65
1.85
TFE731-2
IDLE TO APPROACH
T1
21
4.4835
-0.006693
0.000618
0.9214
20.58
13.00
1.87
TFE731-3
IDLE TO APPROACH
T1
16
4.4715
-0.008575
0.002653
0.7231
8.96
14.60
2.00
ALFS02 ENGINE
IDLE TO APPROACH
T1
28
4.4205
-0.011038
0.000387
0.9690
5.16
10.70
2.30
SPEV 511
T3 LE 600 DEG.F
T2
77
6.7523
-0.009061
0.000343
0.9029
121.50
18.90
2.20
TF30
IDLE & APPROACH
T2
82
5.6067
-0.012370
0.000378
0.9362
12.45
14.20
2.70
CFH56 COMBUSTOR RIG
IDLE & 1.5 IDLE
T2
88
8.2555
-0.023486
0.000972
0.8717
1.30
11 .70
3.40
RB211-22B COMBUST #2
IDLE
T2
8
7.3528
-0.010160
0.001919
0.8236
50.09
25.00
3.50
RB211-22B ALTITUDE
IDLE i. APPROACH
T2
6
9.8035
-0.014348
0.000607
0.9929
66.96
25.00
3.50
JT9D-7 VORBIX
PILOT FA GE 0.009
T2
23
6.4740
-0.018116
0.004094
0.4825
1.46
20.40
3.95
JT9D-7 OVERHAUL
IDLE & APPROACH
T2
62
8.7491
-0.015041
0.000397
0.9599
27.00
22.30
3.95
0T9D-7A
IDLE
T2
49
7.6726
-0.011815
0.000839
0.8083
23.48
20.40
3.95
JT9D-7F
IDLE
T2
28
8.3483
-0.014810
0.000748
0.9378
17.87
21 .20
3.95
0T80-9
IDLE
T4
66
5.3951
-0.011670
0.001033
0.6659
16.56
15.90
2.24
JT8D-17
IDLE
T4
36
5.9325
-0.014036
0.001049
0.8404
10.37
17.00
2.47
-------
Table 20
HC CORRECTION FACTOR COEFFICIENTS SUMMARY STATISTICS
HC CORRECTION COEFFICIENT SUMMARY 3
VARIABLE
LABEL
N
MEAN
STANDARD
MINIMUM
MAXIMUM
RANGE
DEVIATION
VALUE
VALUE
SAMPLES
SAMPLE SIZE
22
32.72727273
28.24077063
4.00000000
88.00000000
84.00000000
CONSTANT
REGRESSION CONSTANT
22
8.35147727
6.40408561
3.81940000
34.48000000
30.66060000
T3COEF
HC TEMPERATURE COEFFICIENT
22
-0.01368032
0.00543549
-0.02910000
-0.00669300
0.02240700
T3STD
T3C0EF STANDARD ERROR
22
0.00157409
0.00129571
0.00034300
0.00409400
0.00375100
RSQ
COEFFICIENT OF DETERMINATION!R-SQUARED)
22
0.82160000
0.18049226
0.20640000
0.99380000
0.78740000
NCMEAN
MEAN HC EMISSION INDEX
22
18.79818182
28.33646257
0.49000000
121.50000000
121.01000000
PR
RATED ENGINE PRESSURE RATIO
22
14.08454545
6.47804845
4.00000000
25.00000000
21.00000000
IPR
IDLE PRESSURE RATIO
19
2.89263158
0.85054731
1 .73000000
4.00000000
2.27000000
I
fsj
-------
Table 21
CO CORRECTION FACTOR COEFFICIENTS TABULAR SUMMARY
CO CORRECTION COEFFICIENT ANALYSIS
ENGINE
NODES
CLASS
SAMPLES
CONSTANT
T3COEF
T3STD
RSO
COMEAN
PR
IP*
GTCP85-98CK API)
BLEED * SHAFT
APU
9
5.8608
-0.008326
0.000881
0.9274
18.15
4.00
4.00
T63-A-5A
WALL FILM(FA IN EQ)
P2
9
21.7700
-0.006354
0.001523
0.8734
16.17
6 ¦ 20'
T63-A-5A
PRES ATOM(FA IN EQ)
P2
9
16.1700
-0.004859
0.001742
0.7019
9.90
6.20
T63-A-5A
AIR BLAST(FA IN EO)
P2
9
16.5900
-0.003449
0.000738
0.9116
10.79
6.20
ALLISON 501K
HIGH RPM LE 30X RATE
P2
22
5.3696
-0.005739
0.000829
0.7056
10.42
9.40
PT6A-50
IDLE TO APPROACH
P2
8
5.1577
-0.008222
0.000186
0.9969
35.08
8.70
2.20
TPE331-5-251
IDLE TO 2.0 IDLE
P2
144
4.9222
-0.006378
0 .001201
0.1657
22.00
10.37
3.45
TYNE
IDLE TO CRUISE
P2
28
7,1201
-0.009577
0.001243
0.6953
26.84
13.50
4.00
ALF502 COMBUSTOR RIG
IDLE TO APPROACH
T1
66
4.7618
-0.004177
0.000267
0.7923
36. 10
5.11
1.73
JT15D-4 R&V CANADA
IDLE TO APPROACH
T1
6
5.5153
-0.005613
0.000213
0.9943
26.83
9.65
1.85
TFE731-2
IDLE TO 50% RATED
T1
16
5.4579
-0.U05752
0.000458
0.9184
58.94
13.00
1.87
TFE731-3
IDLE TO 50* RATED
T1
8
5.5458
-0.006097
0.000798
0.9068
47.07
14.60
2.00
ALF502 ENGINE
IDLE TO APPROACH
T1
28
5.3600
-0.006459
0.000171
0.9822
40.89
10.70
2.30
SPEY 511
IDLE TO TAKEOFF
T2
150
6.7451
-0.007242
0.000157
0.9349
132.30
18.90
2.20
TF30
IDLE TO TAKEOFF
T2
180
5.8252
-0.007146
0.000073
0.9820
56.31
14.20
2.70
CFM56 COMBUSTOR RIG
IDLE & 1.5 IDLE
T2
88
5.4769
-0.006443
0.000351
0.7967
25.69
11 .70
3.40
RB211-22B COMBUST #1
IDLE TO TAKEOFF
T2
49
7.2056
-0.007024
0.000220
0.9559
99.55
25.00
3.50
RB2!1-22B COMBUST *2
IDLE TO TAKEOFF
T2
15
7.2690
-0.007149
0.000474
0.9459
107.33
25.00
3.50
RB211-226 ALTITUDE
IDLE TO TAKEOFF
T2
21
8.1424
-0.007864
0.000287
0.9753
163.42
25.00
3.50
0T9D-7 VORBIX
PILOT FA GE 0.009
T2
23
7.7941
-0.014862
0.001809
0.7627
14.83
20.40
3.95
0T9D-7 OVERHAUL
IDLE & APPROACH
T2
62
7.9885
-0.010337
0.000388
0.9222
66.13
22.30
3.95
JT9D-7A
IDLE TO APPROACH
T2
63
7.8671
-0.009779
0.000354
0.9203
60.42
20.40
3.95
JT9D-7F
IDLE TO APPROACH
T2
31
7.5716
-0.010320
0.000659
0.8942
45.83
21.20
3.95
JT8D-9
IDLE TO APPROACH
T4
79
4.9610
-0.CO5638
0.000182
0.9257
40.03
15.90
2.24
0T8D-17
IDLE TO APPROACH
T4
43
4.8839
-0.005335
0.000155
0.9667
33.05
17.00
2.47
-------
Table 22
CO CORRECTION FACTOR COEFFICIENTS SUMMARY STATISTICS
CO CORRECTION COEFFICIENT ANALYSIS
VARIABLE
LABEL
N
MEAN
STANDARD
MINIMUM
MAXIMUM
RANCE
DEVIATION
VALUE
VALUE
SAMPLES
SAMPLE SIZE
25
46.84000000
48.64726782
6.00000000
180.00000000
174.00000000
CONSTANT
REGRESSION CONSTANT
25
7.65326600
4.21369507
4.76180000
21.77000000
17.00820000
T3C0EF
CO TEMPERATURE COEFFICIENT
25
-0.00720564
0.00240191
-0.01486200
-0.00344900
0.01141300
T3STD
T3C0EF STANDARD ERROR
25
0.00061436
0.00051998
0.00007300
0.00180900
0.00173600
RSQ
COEFFICIENT OF DETERMINATIONR-SQUARED )
25
0.86217200
0.17264962
0.16570000
0.99690000
0.83120000
COMEAN
MEAN CO EMISSION INDEX
25
48.16280000
39.39699402
9.90000000
163.42000000
153.52000000
PR
RATED ENGINE PRESSURE RATIO
25
14.18520000
6.65251840
4.00000000
25.00000000
21.00000000
I PR
IDLE PRESSURE RATIO
21
2.98619048
0.85584155
1.73000000
4.00000000
2.27000000
i
ro
-------
Table 23
SMOKE NUMBER CORRECTION FACTOR COEFFICIENTS TABULAR SUMMARY
EPA SMOICE NUMBER CORRECTION COEFFICIENT SUMMARY
ENGINE
MOOES
CLASS
SAMPLES
CONSTANT
P3COEF
P3STD
RSQ
SMKMEAN
PR
GTCP85-98CK APU
BLEED & SHAFT
APU
15
1.2375
0.6394
0.1964
0.4492
41.49
4.00
TPE331-5-Z51
APPROACH TO TAKEOFF
P2
19Z
-0.5323
0.6471
0.1011
0.1774
14.06
10.37
ALF502 ENGINE
IDLE TO TAKEOFF
T1
55
-6.984B
2.0754
0.0940
0.9020
23.50
10.70
CFMS6 COMBUSTOR RIG
IDLE TO TAKEOFF
T2
214
-0.3123
0.2775
0.0831
0.0500
4.12
11 .70
SPEV 511
IDLE TO TAKEOFF
T2
120
-0.3525
0.7771
0.0445
0.7780
17.15
18.90
JT9D-7 VCRBIX
PILOT FA LT 0.009
T2
24
-3.4921
1.1907
0.0793
0.9112
29.84
20.40
JT9D-7A
IDLE TO TAKEOFF
T2
69
-3.6831
0.9507
0.1439
0.3945
6.04
20.40
0T9D-7 OVERHAUL
IDLE TO TAKEOFF
T2
36
-3.5236
0.9207
0.1937
0.3993
7.80
22.30
RBZ11-22B COMBUST #1
IDLE TO TAKEOFF
T2
42
-0.8485
0.4992
0.0729
0.5357
8.67
25.00
RB211-22B COMBUST *2
IDLE TO TAKEOFF
TZ
13
-3 .2239
0.9097
0.1803
0.6984
10.60
25.00
JTED-9
IDLE TO TAKEOFF
T4
94
-3.8398
1.3038
0.0285
0.9579
21 .26
15.90
JT8D-17
IDLE TO TAKEOFF
T 4
61
-4.3727
1.4063
0.0640
0.8929
22.95
17.00
to
I
N)
cn
-------
Table 24
SMOKE NUMBER CORRECTION FACTOR COEFFICIENTS SUMMARY STATISTICS
EPA SMOKE NUMBER CORRECTION COEFFICIENT SUMMARY
VARIABLE
LABEL
N
MEAN
STANDARD
MINIMUM
MAXIMUM
RANGE
DEVIATION
VALUE
VALUE
SAMPLES
SAMPLE SIZE
12
77.91656667
66.45635387
13.00000000
214.00000000
201.00000000
CONSTANT
REGRESSION CONSTANT
12
-2.49400833
2.322S5414
-6.98480000
1.23750000
8.22230000
P3COEF
SMOKE PRESSURE COEFFICIENT
12
0.96646667
0.47964638
0.27750000
2.07540000
1.79790000
P3STD
P3COEF STANDARD ERROR
12
0.10680833
0.05789169
0.02850000
0.19640000
0.16790000
RSQ
COEFFICIENT OF DETERMINATION R-SQUARED )
12
0.59554167
0.30652553
0.05000000
0.95790000
0.90790000
SMKMEAN
MEAN EPA SMOKE NUMBER
12
17.29000000
11.07448993
4.12000000
41.49000000
37.37000000
PR
RATED ENGINE PRESSURE RATIO
12
16.80583333
6.49695377
4.00000000
25.00000000
21.00000000
<£>
I
N)
ON
-------
10. FIGURES
10-1
-------
NO* i
e.z -
N I
0
X 1 a TAMB = 85° F
i
E >
r-f 6.o *
1 i
5 !
Si
I
0 !
N b.e +
1
N I
:i
E !
5.0 ~ a TAMB = 59°F
I !
L I
0
1
5.:
5. C
4.8 ~
I
TAMB = 39° F
A TAMB = 19 F
AMBIENT HUMIDITY = 0.0037 lb H20/lb AIR
AMBIENT PRESSURE = 29.92 IN Hg
COMBUSTOR INLET PRESSURE (P3) = 75 PSIA
TAMB = AMBIENT TEMPERATURE DEG. F
430
T3
470 440 4S0
COM3 JSTOR INLET TEMPERATURE'DEG. F)
43 3
55'.'
Figure 1 VARIATION OF OXIDES OF NITROGEN (NOX) EMISSION INDEX WITH
AMBIENT TEMPERATURE — ALF 502 COMBUSTOR RIG, APPROACH MODE
-------
FULL SCALE
ENGINE TESTS
COMBUSTOR
RIG TESTS
CONTRIBUTED
TEST DATA
FROM INDUSTRY
CONTROLLED
AMBIENT
CONDITIONS
UNCONTROLLED
AMBIENT
CONDITIONS
CONTROLLED
AMBIENT
CONDITIONS
AIRESEARCH
GTCP85 98CK
APU
TPE331-5-251M
P2
AVCO
AVCO
GE
ALF502
T1
ALF502
T1
CFM56
T2
DATA ANALYSIS
Figure 2 DATA OVERVIEW - EPA AIRCRAFT EMISSION AMBIENT EFFECTS PROGRAM
10-4
-------
TPE 331-5-251
0.3000E+0 3
0
1
in
B 0.7<>00e*03
U
R
N
E
R 0.6300E+03
I
L
E 0.6200E*03
T
M 0.5600E*G3
P
E
R
A
T 0.5000E+03
U
R
E
0.4&OOE+03
D
E
G
TAKEOFF
i CLIMB
1 CRUISE
APPROACH
o.3aooe+03
0.3200E»03
0.2600£»03
1 1
, 2.0 IDLE
l 1.5 IDLE
IDLE
0.2000E+33
0.0
0.60 00c-02
0.1200E-01
0.1800E-01
0.2400fc-01
0.3000E-01
AMBIENT HUMIDITY
L * WATER/L9 £) * Y A I K
Figure 3 BURNER INLET TEMPERATURE VERSUS AMBIENT HUMIDITY - TPE 331-5-251
-------
AMBIENT TEMPERATURE ( f)
0 40 90 >20
I S
STANDARD DAY
— CONDITION
i i
9
7
20
It
5
JIO J20
AMBIENT TEMPERATURE (°K)
360
240
Ambient temperature correciion factors (o -0.9 ami o - 0.6).
3.0
3 0
M -10
0-0 9
Pt'-7
10
0
03
0&
AMBIENT HUMiniTY/GM "»°
Igm DRY AIR
Ambient humidify cormiion factors l« - U.V und o = 0.6).
(b)
Figure 4 EFFECT OF ENGINE PRESSURE RATIO AND EQUIVALENCE RATIO ON NOX
TEMPERATURE AND HUMIDITY CORRECTION FACTORS - FROM BLAZOWSKl
REFERENCE 5
10-6
-------
0 1 1 1 1 L
-30 —20 -10 0 10 20
_L_
30
_L
40 50
AMBIENT TEMPERATURE (°C)
Figure 5 ANALYTICALLY PREDICTED AMBIENT TEMPERATURE CORRECTION
FACTORS FOR HYDROCARBONS. FROM BLAZOWSKI AND MARZEWSKI
REFERENCE 9.
10-7
-------
PLOT OF CO*HC
LEGEND: A « 1 OBS , B - 2 OBS , ETC
HC
H
C
E
H
I
S
S
I
0
N
1
N
D
E
X
(
L
B
H
C
/
1
0
0
0
L
B
F
U
E
L
)
A A
A
A
AAA
AA
A E AA C
ACBCBACCBA
A B AA
AAA
A A A A A AA AAA A
AAA A AA A8B AA
AAA AA
I ~-
0
10
20 30
CO = CO EMISSION INDEX (LB CO/IOOO LB FUEL)
40
50
Figure 6 HC El VERSUS CO El - CFM56 COMBUSTOR RIG @ IDLE AND 1.5 IDLE
-------
PLOT. OF LNCO*LNHC LEGEND: A - 1 OBS , B - 2 OBS .ETC
2 ~
O
-1
-2
-3
-4 ~
LNHC l«—
2.0
A AA A A
A A
AA A
B A
A AA
A A
AA A
A A
A A A A A
A A A
A A A
A A AA A
A A AA
A
AAA
A A
BB
AA
A A A
A C
A A A
AAA
B A
2.5
3.0
LNCO - NATURAL LOG(CO EMISSION INDEX)
3.5
4.0
Figure 7 NATURAL LOG OF HC El VERSUS NATURAL LOG OF CO El - CFM56
COMBUSTOR RIG @ IDLE AND 1.5 IDLE
-------
GENERAL
CORRECTION
FACTOR
APPROACH
TEST TO
TEST
VARIABILITY
LITERATURE
SURVEY
ENGINE
TO
ENGINE
VARIABILITY
INDUSTRY
CONTRIBUTED
DATA
EPA
SPONSORED
DATA
CORRECTION
FACTOR
COEFFICIENT
SENSITIVITY
ANALYSIS
REGRESSION
MODEL
ASSESSMENT
AND
REFINEMENT
ENGINE
PARAMETERS
AND CLASS
T3, P3, HAMB
COEFFICIENT
TRENDS
REGRESSION
MODEL
FORMULATION
LNHC - f(T3)
LNCO - f(T3l
LNNOX - f (T3, HAMB)
LNSMK - f(LNP3)
Figure 8 GENERAL CORRECTION FACTOR DEVELOPMENT OVERVIEW
10-10
-------
PAMB-Z9.31 IN HC , HAHB-0.0018 LB H20/LB AIR , P3-49 PSIA
PLOT OF TAMB*T3 LEGEND: A « 1 OBS , B - 2 OBS , ETC
T3
460
440
420
A
A
I
N 400
L
E
T
A
A
©
I
380
36C
340
320
A
A
A
A
300
280 *
I * ~ ~ * *
0 20 40 6C 80 100 120
TAMB • AMBIENT TEMPERATURE*DEG. F)
Figure 9
COMBUSTOR INLET TEMPERATURE VERSUS AMBIENT TEMPERATURE -
CFM56 COMBUSTOR RIG AT IDLE
-------
LNHC
2.0
PAMB-29.31 IN HG , HAMB-0.0018 LB H20/LB AIR , P3-49 PSIA
PLOT OF T3*LNHC LEGEND: A - 1 OBS , E = 2 OBS . ETC
0
1
N
A
T
U
R
A
L
L
0
€
(
H
C
E
M
1
S
S
I
0
N
1
N
0
E
X
)
1.5
1.0
C.5
0.0
-0.5
-1.0
-1 .5 «¦
280
300
320
340 360 380
T3 = COMBUSTOR INLET TEMPERATUREfDEG. F)
400
420
440
Figure 10 NATURAL LOG OF HC El VERSUS COMBUSTOR INLET TEMPERATURE -
CFM56 COMBUSTOR RIG AT IDLE (PRESSURE AND HUMIDITY FIXED)
-------
PAMB-29.31 IN HS , HAMB-0.0018 LB H20/LB AIR . P3*49 PS1A
PLOT OF T3*LHCO LEGEND: A - 1 OBS . B - 2 CCS , ETC
LNCO
3.6
3.5
3.4
3.3
3.2
3.1
3.0
2.9
I+--
280
300
320
340 3S0 380 400
T3 » COMBUSTOR INLET TEMPERATURE'DEG. F>
420
440
Figure 11 NATURAL LOG OF CO El VERSUS COMBUSTOR INLET TEMPERATURE -
CFM56 COMBUSTOR RIG AT IDLE (PRESSURE AND HUMIDITY FIXED)
-------
50 300 850 ^00 9s0 :00"
T3 = C0W3UST0R INLET TEMPERATUF.E( DEG. F>
Figure 12 NATURAL LOG OF NOX El VERSUS COMBUSTOR INLET TEMPERATURE -
CFM 56 COMBUSTOR RIG AT CLIMB {PRESSURE AND HUMIDITY FIXED)
-------
LNNOX
TAMB-59 DEC F , PAMB-29.31 IN HS , P3-166 PSIA , T3-890 BEG F
PLOT OF HAMB*LNNOX LEGEND: A - 1 OBS , B - 2 OBS , ETC
2.65
0
1
II
A
T
U
R
A
L
L
0
G
(
N
0
X
E
N
1
S
S
1
0
N
1
N
D
E
X
)
2.60
2.55
2.50
2.45
2.40 ~
1+
0.000
0.002
HANB
0.004 0.006
AMBIENT HUM1DITV(LB H20/LB DRY AIR)
0.008
0.010
Figure 13 NATURAL LOG OF NOX El VERSUS AMBIENT HUMIDITY - CFM56 COMBUSTOR
RIG AT CLIMB (PRESSURE AND TEMPERATURE FIXED)
-------
LNHC =
f(T3)
LNCO = f(T3)
LNNOX = f(T3, HAMB)
LNSMK - f(LNP3)
WHERE:
T3
=
COMBUSTOR INLET TEMPERATURE
HAMB
s
AMBIENT HUMIDITY
LNP3
¦
NATURAL LOG COMBUSTOR INLET PRESSURE
LNHC
=
NATURAL LOG HC El
LNCO
=
NATURAL LOG CO El
LNNOX
=
NATURAL LOG NOX El
LNSMK
=
NATURAL LOG SMOKE NUMBER
Figure 14 AMBIENT EFFECTS PROGRAM REGRESSION MODEL SUMMARY
10-16
-------
o
I—'
-J
0.5
0.4
>
o
2
O
H
<
E
<
>
u.
O
H
2
uj
g
iE
u.
LU
8
-------
EMISSIONS AT STANDARD CONDITIONS
MEASURED EMISSIONS
EXAMPLE - CFM 56 COMBUSTOR RIG TEST
.002503 T3jccT -20.7 HAMBtest
NOX ElTESJ = 1.5956 e test e test
NOX EItest = 1.5956 e 002503 T3STD e -20.7 HAMBSTD
NOX EISTD
CF =-
NOX EItest
CF = e-002503 'T3STD " t3TEST> e-20.7 (HAMBTEST - HAMB
Figure 16 CORRECTION FACTOR STRUCTURE
10-18
-------
PLOT OF T3-m_ LEGEND: A - 1 OBS , B - 2 OBS , ETC
HC
7
B
A A
6
AAA
e
A
AA AB
AAA
BA AB ACA
A ABADBB
AA DBC BBA
A A C AIEA
A
E
AC
BN
DAA BGA EFC DIB AEOC
CDC
200
400
.600 800
T3 « COMBUSTOR INLET TEMPERATURE(DEG. F)
1000
1200
Figure 17 HC El VERSUS COMBUSTOR INLET TEMPERATURE - CFM56 COMBUSTOR RIG,
IDLE TO TAKEOFF
-------
PLOT OF T3*LNHC LEGEND: A - 1 OBS . B = 2 OBS , ETC
LNHC
0
1
Ki
o
-3
-5 +
I +—
200
A
AA
1
B
A
1
1
1
A A
BA
AA
N
1
B
A
0 ~
A
T
1
AA
AB
U
1
A
R
1
A
A
AA
A
1
A
B
AA
L
1
-1 ~
AA
A
AAA
VA A
L
1
A
A
BB
0
1
A
A
G
1
A
A
(
1
AAA
H
1
A
A
C
)
-2 ~
1
A
A
AA
A A
BBA
A AA
AA
AB
A
AA
A
C
A
C
B
AB
AN
DAA BGA EFC
DLB AEOC
CDC
400
T3
600 BOO
COMBUSTOR INLET TEMPERATURE'DEG. F)
1000
1200
Figure 18 NATURAL LOG HC El VERSUS COMBUSTOR INLET TEMPERATURE - CFM56
COMBUSTOR RIG, IDLE TO TAKEOFF
-------
PLOT OF T3*CO LEGEND: A - 1 OBS , B - Z OBS , ETC
CO
so
A
L
B 40
C
0
/
1
0
0
0
30
L
B
F
U
E
L 20
10
BA
A
AA
B
A
A
B AAA
A AAA A
B AB A
A BBA
AAA
A A A
AA B
AA AA A
B A
B B
A
B AA
CAAACA
A EBA
A
AB
A
A
G AL
AB
A
AA A A
B A AC AE
A CA DAB
A AA
ABA ABJA
BIA BDB
A
A
B B
ACA
I+--
200
400
T3
600 800
COMBUSTOR INLET TEMPERATURE(0EG. F>
1000
1200
Figure 19 CO El VERSUS COMBUSTOR INLET TEMPERATURE - CFM56 COMBUSTOR RIG,
IDLE TO TAKEOFF
-------
PLOT OF T3*LNCO LEGEND: A - 1 OBS , B » 2 OBS , ETC
0
1
ro
4.0 ~
3.5
3.0
H
A 2.5
T
U
R
A
L
2.0
A
AA
BA
AA A
0.0 ~
LNCO I~—
200
A
AB
AA
C
A
B
A
AB ABA
BC
CBA
AAAA
AAA
AA AA
A A
A B
B A
A
A
B AAB
BAA B
A CAA
CB
1
1
A
1
A
1.5 +
A
A
1
B
A
A
1
A
A
A
A
1
B
B
A
AA
A
1
A
B
A
AA
1
A
B
C
A#
A
A
1 .0 ~
A
B
A
C
A
B
A
D
1
A
B
D
B
1
0
A A
B
1
AB
A
1
A
BA
1
A
AA
CAB
0.5 +
B
A
1
A
A
1
A
A
A
BA
BA
A C
B
AB
BA
A
A A
B
A B
A
AA
AA
A
400
T3
BOO 800
COMBUSTOR INLET TEMPERATURE!DEG. F>
1000
1200
Figure 20 NATURAL LOG CO EJ VERSUS COMBUSTOR INLET TEMPERATURE - CFM56
COMBUSTOR RIG, IDLE TO TAKEOFF
-------
PLOT OF T3*NOX LEGEND: A * 1 OBS . B = 2 OBS , ETC
NOX
25
0
1
ro
W
20
IS
AA
AB
AB
CA
E
10 +
B
L
1
AC
1
B
A
1
B
B
B
A
1
B
AB
B
A
A
1
|
B
A
F
A
B
1
1
CA
A
AD
5 +
AAA AB
A
BA
AD
CB BA AS
A A
AA AD
BB FCC
BB AB
AA ADA
A A
AA B
A AE
A
A
A
AA
A
A
B
AB
AB
A
A A
B
BB
B
A
A
A
CA
A
BA
C
A A
B
B
BC
AA
B
B
A
AA
A
AA
B
AA
l+ —
200
400
T3
600 800
COMBUSTOR INLET TEMPERATURE< DEG. F)
1000
1200
Figure 21 NOX El VERSUS COMBUSTOR INLET TEMPERATURE - CFM56 COMBUSTOR RIG,
IDLE TO TAKEOFF
-------
3.S +
PLOT OF T3*LNN0X
LEGEND: A = 1 OBS , B = 2 OBS , ETC
0
1
NJ
3.0
N
A
T
u
R
A
L
L
0
G
(
N
0
X
)
2.5
2.0
1.5
1.0
B
BA
BA
B
A
B A
B
A
AB
B
AA
C
B
A
B
B
A
B
A
D
B
B
A
AA
B
AC A
AA
A
BB
AC
B
A
BA
BB
AD
A
BC
BA
AB
C
CA
CB
A
AB
A
A
AA
B
AA
A
BA
AA B
A A
C
AA
AB
BA
AA B
A
A
AA CBA
A B
C
AA BAB
A
A
A
A
A
B
AA A
r
AA
B
AB
A
AB
AA
A
C
AA A
A
A
A A
A
0.5 +
LNNOX I+--
200
400
600 800
COMBUSTOR INLET TEMPERATURE( DEG. F)
1000
1200
Figure 22 NATURAL LOG OF NOX EI VERSUS COMBUSTOR INLET TEMPERATURE -
CFM56 COMBUSTOR RIG, IDLE TO TAKEOFF
-------
PLOT OF HAMB*NOX LEGEND: A » 1 OBS , B = 2 OBS , ETC
NOX
25
L
B 20
IS
10
A
A
A A
A
B
BA
BA
A
BB
A
C
A
AB
BB
A A
B
A A
B A
A AA
AA A A
A AA A
AA
AAAA
A AB
CBBAA
CABSff A
ABB B
A
AA
AA
A A
A
A
BA
A A
AB
B
A A
AA D
A
AAA
AA
CA
B
C
BCA
A BCA A
AA
AA
A B
A
A
A
AA
AA
A A
AAB
BAA
A A
AA
B
AAAA
B C
A
A A CAA
A A BCA A
I
0.000
0.005
0.010 0.015 0.020
HAMB = AMBIENT HUMIDITV
-------
3.5 ~
PLOT OF HAMB'LNNOX LEGEND: A ¦ ! OBS , B = 2 OBS , ETC
0
1
K>
O
3.0
N
-A
T
U
R
A
L
L
O
G
<
N
0
X
»
2.5
Z.O
1.5
1.0
AB
B
& A
BC
A
BB
A
C
BA
AC
A A
B
A A
A
A
=A
BA A
BA
A
A AA
A AB
CAAAA
A B
BAAB A
B
A
B 1
A
AA
A
0.5 ~
LNNOX 1+
O.OGO
AA
B A
AA
D
A C
BA A
AA
AA
CA
B
B
A
BB
A
A
A BA
A
AA
AA
AA
c.oos
AA A
AA
A B
A
B A
A
A AA
B
A A A A
A A
A C
A
A
B A
AA
AA
B A
0.010 0.015 0.020
HAMB ¦ AMBIENT HUMIDITY!LB H20/LB DRV AIR)
0.025
0.030
Figure 24 NATURAL LOG OF NOX El VERSUS AMBIENT HUMIDITY - CFM56 COMBUSTOR
RIG, IDLE TO TAKEOFF
-------
12 ~
PLOT OF P3*SM0KE LEGEND: A « 1 OBS , B = 2 OBS
ETC
O
to
s
M
O
K
E
10
N
6
A
A
A
U
A
M
A
A
B
A
A
E
B
C
R
A
AA
A
A
AA
AA
A
A
A
AA
AA
A
A
4
B
A' B
B
A
CB
AA
A
A
B
A A
A
A
B
AA
A
A A
AB
A
A
A
A
C
D A
B
A
A
A
A
B
AB
AD AB A
BB
A AA
B
AA
A
B
A
AA
B A
A
A
A
A
A
AA
BC
A
BC
A
AA
. AB
A
A
A A B
AA
AA
A
A
2
A
BA
A
BA
A
AB
A
0B
B
A
A
AC
A
A
C
A
A
A
A
A
A
A
A
A
A
A
A
A
B A
A
BA
A
SMOKE I~"
10
60
80
100 120 140
P3 - COMBUSTOR INLET PRESSURE(PSIA}
ISO
180
Figure 25 SMOKE NUMBER VERSUS COMBUSTOR INLET PRESSURE - CFM56 COMBUSTOR
RIG, IDLE TO TAKEOFF
200
-------
PLOT or P3*LNSMK LEGEND: A - 1 OBS , B - 2 OBS , ETC
LNSMK I
3
0
1
ro
oo
N
A
A
2 ~
A
T
A
AA
BA
U
A
AA
A
A
A
R
C
D
A
A
B
A
A
BB
AAA
A
A A
L
B
B
B
B
A
B
CC
BA
A A
A
E
A
ABA
AA
A
B
AA
AA
EB
A
L
AE
AB B
BC
A
AA
AB
AB
AA
B
BA
A
0
1 *«
BA
B
A
C
B
BA
A
AA
A
A
c
A
AC
BA
AC
B
A
B
A
(
A
A
AB
A
AA
s
CB
A
AA
A
M
B
B
A
CB
A
B
A
0
AA
A
C
A
K
A
A
E
A
)
0 ~
A
A
A
BA
A A
-2 +
I ~ ~ «• + + + + + + -
40 60 80 100 120 140 160 180 200
P3 - COMBUSTOR INLET PRESSURE(PSIA>
Figure 26 NATURAL LOG OF SMOKE NUMBER VERSUS COMBUSTOR INLET PRESSURE-
CFM56 COMBUSTOR RIG, IDLE TO TAKEOFF
-------
HC El = 3848.5658 * e" 023486 # T3
CO El = 239.1063 * e ' 0®6443 * T3
NOX El = 1.5956 « e -002503 * T3 » e"20-7 * HAMB
SMOKE = .7318* P3 2775
WHERE
P3 = COMBUSTOR INLET PRESSURE - PSIA
T3 = COMBUSTOR INLET TEMPERATURE - DEG F
HAMB = AMBIENT HUMIDITY- LB H20/LB DRY AIR
Figure 27 REGRESSION COEFFICIENT SUMMARY CFM56 COMBUSTOR RIG
10-29
-------
REGRESSION
ANALYSIS
SUMMARIZE COEFFICIENTS FOR
EACH ENGINE AND CORRELATE TO
ENGINE OPERATING PARAMETERS
NOX
T3COEF
PR
PERFORM REGRESSION
ANALYSIS FOR
SINGLE ENGINE
USING BASIC
MODELS
LNHC = f(T3)
LNCO = f (T3)
LNNOX = f(T3, HAMB)
LNSMK - f (LNP3)
SUMMARIZE
REGRESSION
COERFICIENTS
FOR SINGLE ENGINE
HC
T3 CO EF
CO
T3 CO EF
NOX
73 CO EF
HAMB CO EF
SMOKE •
P3COEF
SUMMARIZE ENGINE
OPERATING
PARAMETERS FOR
EACH ENGINE
PRESSURE RATIO
IDLE PRESSURE RATIO
CONTINUE FOR EACH
ENGINE IN DATA BASE
Figure 28 CORRECTION FACTOR COEFFICIENT VERSUS ENGINE OPERATING
PARAMETER - DEVELOPMENT APPROACH
-------
ENGINE = Engine or Rig used to develop regression equation.
MODES = Power modes included in analysis.
SAMPLES = Number of test points in analysis.
CONSTANT = Constant term in regression equation (based on T3 data in Deg. F
of P3 data in PSIA)
T3C0EF = Combustor Inlet Temperature Regression Coefficient
(input T3 data in Deg. F)
T3STD = Standard Error of T3C0EF
HAMBCOEF = Ambient Specific Humidity Regression Coefficient
(input HAMB data in lb H^O/lb dry air)
HAMBSTD = Standard Error of HAMBCOEF
P3C0EF = Combustor Inlet Pressure Regression Coefficient
(input P3 data in PSIA)
P3STD = Standard Error of P3C0EF
2
RSQ = Coefficient of Determination for Regression, 0 < R ^1.
T3MEAN = Mean Combustor Inlet Temperature (T3) in Deg. F of test data
and defined as follows:
HC: Mean Idle T3
CO: Mean Idle T3
NOX: Mean Takeoff T3
Note: Where NOX takeoff data was not available, T3MEAN
corresponds to mean T3 at highest thrust level
(e.g., approach, climb, or cruise) for which data
was available.
P3MF.AN = Mean Combustor Inlet Pressure (P3) in PSIA of test d
and defined as follows:
HC: Mean Idle P3
CO: Mean Idle P3
NOX: Mean Takeoff P3
SMOKE: Mean Takeoff P3
Note: Where Takeoff data was not available, P3MF.AN
corresponds to mean P3 at highest thrust level
(e.g. approach, climb, or cruise) for which data
was available.
HCMEAN = Mean Idle HC EI.
COMEAN = Mean Idle CO EI
NOXMEAN = Mean Takeoff NOX EI -- Note: If Takeoff data unavailable
NOXMEAN corresponds to mean NOX EI at highest power
setting measured.
SMKMEAN = Mean Takeoff EPA Smoke Number -- Note: If Takeoff data unavailable
SMKMEAN corresponds to mean EPA Smoke Number at highest
power setting measured.
PR = Rated Pressure Ratio -- Note: If Takeoff data unavailable,
PR corresponds to P3MEAN/14.7 PSIA
IPR = idle Pressure Ratio (computed from engine test data)
CLASS = EPA Engine Class Designation
Figure 29 CORRECTION FACTOR NOMENCLATURE
10-SI
-------
PLOT OF PR*T1MEAN LEGEND: A = 1 OSS , B = 2 OBS , ETC
T3MEAN I
1200 +
0
1
OJ
l\J
M
E
A
N
C
0
M
B
U
S
T
0
R
1
N
L
E
T
T
E
M
P
E
R
A
T
U
R
E
(
0
E
G
F
>
1CC0
A
A
COO
A A
A
500
A A
A
:oo
A B
?00
10 15
PR - RATED ENGINE PRESSURE RATIO
20
25
Figure 30 T3MEAIM VERSUS RATED PRESSURE RATIO PR, AMBIEIMT EFFECTS
DATA BASE
-------
PLOT OF MNEAK«T3COSF
LEGEND: A " 1 OBS ,# - 2 OBS , ETC
T3COEF
O.OOSO
M
0.0045
N
0
X 0.0040
T
c
it
p
E 0.0035
R
A
T
U
R
E 0.0030
C
0
fc!
sr F 0.CO25
Z. I
H C
CO I
Z E
UJ N
t 0.0920
M
o
z
g 0.0015
0.0010 '~
I ~-
0
T-56
A S P E V 511
ARE2H-223 COMB J ST #2
AJT9D-7F
RB21I-22B ALTITUDEA A
JTSD-7 OVERHAUL ARE211-?ib COMBUST #1
A 0T9D-7A
LIPFERT
A JT9C-7 VORBIX
ALL Tr-CN 7-T1K LOW RPM
A TF30
A 0T J fcL---4 AJT6C-9
A Tb3-A-5AHI(-H rphL GtA90S RATE TFE7o 1-2 ^JTSLl-17
A1S6
kALFi.02 RIG
iT56 A PT6A-50
'A""*""" A A TFE721-3
A TPE :1 -5-251
ATYNE
A CFM56 RIG IDLE TO TAKEOFF
A Ac*Xt*
.tff. V" ALFbi;2 EN-SIHEA A AcFMSO TAKEOFF
4t(l ,, CFM5E CLIMB
A^\&-
T56
T63-A-BA
4T56 GTCPC5-98CK APU
A ATS6
AT56
ACFKES APPROACH
50
100
150 200 250 300
P3MEAN - MEAN COMBUSTOR INLET PRESSURE(PSIA>
350
400
Figure 31
NOX COMBUSTOR INLET TEMPERATURE COEFFICIENT T3COEF VERSUS
COMBUSTOR INLET PRESSURE P3MEAN
-------
100
90
80
70
60
50
40
Lipfert
30
20
r—4
-------
1X0CP
u
U N U
uu
M
L
ut \m l
* t
* ML U
M
U MM
M
L
L L
U
M
U M L
U M
M
L L
U U
\m u
L M
M L
t
UMU
M \J
UMLL *
*1 L
L
M = TEMPERATURE COEFFICIENT
U - UPPER CONFIDENCE BOUND
L - LOWER CONP~
a - nc
0*000 ~
100
190
*00
F3HCAN
MEAN COMBUSTOR INLET PRESSURE (PSIA)
i* o
300
350
*60
Figure 33 NOX TEMPERATURE COEFFICIENT VERSUS COMBUSTOR INLET PRESSURE P3MEAN
CONFIDENCE BOUNDS
-------
T3C0CF I
0.0050 +
0
1
O-l
ON
>¦
H
>
0.0045
N
0
i( 0.0040
0.0035
U
P.
E G.0030
C.0C25
0.0020
CO
z
Ui
CO
o
z
u
<
LU
cc
u
S 0.0015
o.ooto
T3COEF = .001735 + .000107 * PR
10 15
PR - RATED ENGINE PRESSURE RATIO
20
25
Figure 34 IMOX COMBUSTOR INLET TEMPERATURE COEFFICIENT T3COEF VERSUS
RATED ENGINE PRESSURE RATIO PR
-------
T3C0EF I
0.C.3S0 ~
A
0.0045
N
0
X 0.0C40
T
E
H
P
E 0.0035
R
A
T
U
R
E 0.0030
>
0
o
H
E
1
u>
>
F
F
0.0025
M
H
I
H
c
Z
T
Ul
CO
E
O
¦ Z
N
T
0.0020
s
111
C
o
z
0.0015
0.0010 ~
l+-
0
±SFEY 511
RB2 1.. -22B ALTITUDE
A
RB2H-22B COMS'JST 4*2
At JT9D-7F
A
A 0T9D-7 OVERHAUL
A RE2II-228 COMBUST #1
JT9D-7A
ALLISON 5:-lK
LOW^RPM
AT6--A-5A
A JTSD-7 VCJRBIX
A TF3C
A JT15D-4
Aote:-9
Aalfe"2 Rie
Atyn
A ALFS02 ENSI-NE
AJTBt-'.7 TFEVsl-3
A A A ALLISON 501K
A TFE?31-3-«-3l VPC";!-2 HIQH RPM GT 90X RATE
A PT6A-50
A
' CFME6 RIG
* ^ ^T63-A-SA
A TE5
.A
A GTCPfi j-<3K.":vr APil
10
15 20 25
NOXMEAN - MEAN NOX EMISSION INOEX
30
35
Figure 35 NOX COMBUSTOR INLET TEMPERATURE COEFFICIENT T3COEF VERSUS
MEAN NOX EMISSION INDEX NOXMEAN
-------
PLOT OF NOXMEAN«T3MEAN LEGEND: A - 1 OBS , # « Z OBS , ETC
T3MEAN I
1200 ~
0
1
w
00
M
E
A
N
C
0
H
B
U
S
T
0
R
1
N
L
E
T
T
E
M
P
E
R
A
T
U
R
E
<
D
E
e
F
)
1000
BOO
GCO
400
200
cV£> NEWER TECHNOLOGY
S P E V 5!lA
' «¦- -
-------
COMBUSTOR INLET TEMPERATURE (T3)
Figure 37 NOX EMISSION INDEX VERSUS COMBUSTOR INLET TEMPERATURE
UNCONTROLLED AND CONTROLLED ENGINES
10-39
-------
0
1
•pfc
o
T3C0EP
0.000
-0.005
£ C
CO
z
111
CO
o
CO
<
Ul
a.
o
z
-0.010
-0.015
-0.020
-0.025
-0.030 ~
l+—
1.5
HC CORRECTION COEFFICIENT SUMMARY
PLOT OF IPR*T3C0EF LEGEND: A ' 1 OBS , • - 2 OBS , ETC
4 TFE731-2
a\
ALF50Z RIG \ ~ A SPEY 511
TFE731-3
^AJTBD-g
ALF502 ENGINE ¦
RB211-22B COMBUST #2
A TF30
~
0T9D-7A
AJTBD-17
RB2I1-22B ALTITUDE
A JTibD-4
A TPE331-8
A PT6A-50
JT9D-7 OVERHAUL
JT9D-7F
~
0T9D-7 VORBIX
A CFH56 RIG
GTCP85-98CK APU
A
IPR
3.0
IDLE PRESSURE RATIO
3.5
Figure 38
HC COMBUSTOR INLET TEMPERATURE COEFFICIENT T3COEF VERSUS
ENGINE IDLE PRESSURE RATIO IPR
-------
CO CORRECTION COEFFICIENT ANALYSIS
PLOT OF IPR*T3COEF LEGEND: A - 1 OBS , • - 2 OBS , ETC
T3COEF I
0.000 ~
-0.002
0
1
-U
>-
I-
>
P
M
z
ui
C/9
CD
Z
C/3
<
III
AC
a
z
O -0.004
T
E
M
P
E
R
A
T
u
R
E
-0.006
-0.008
-0.010
-0.012
A
ALF502 RIC
-\>
A0T8D-17
—^ A _
ALF502 ENGINE
ASPEV 511
APT6A-S0
, TF3D
TPE331-S-2S1
RIG
RB211-22B COMBUST *1
RB211-22B COMBUST #2
A
RB211-22B ALTITUDE
—- CTCP85-98CIC APU
JT9D-7A A A TYNE
0T90-7 OVERHAUL
0T9D-7F
-0.014
-0.016
A
JT9D-7 VORBIX
.5
2.0
2.5 3.0
IPR • IDLE PRESSURE RATIO
3.5
4.0
Figure 39
CO COMBUSTOR INLET TEMPERATURE COEFFICIENT T3COEF VERSUS
ENGINE IDLE PRESSURE RATIO IPR
-------
P3C0E* I
2.0 +
I
I
I
I
I
0
1
-c*
NJ
A
>
P
V)
z
LU
CO
o
Z
«n
<
uj
oc
o
s
M
0
K
E
F
R
E
S
s
u
R
E
C
0
E
F
F
1
C
T
E
N
T
L.C
. 0 +
I
r.5
~ AL if2 "NGINE
J i L U ' i '
A vo-g
Av":D-T VDSBIX
A GTCP85-98CK APU
A TF?331---251
A A A
0T9D-7A Olsn-7 f,L. i .-220
O.'ERHAUL CC'-'SUST #2
A SFtY 51'
A
re:ii-2":b
COMBUST *1
C' :r: COMBUSTOR RIG
0.0
10 IS
PR - RATED ENGINE PRESSURE RATIO
20
25
Figure 40 SMOKE NUMBER COMBUSTOR INLET PRESSURE COEFFICIENT P3COEF
VERSUS RATED ENGINE PRESSURE RATIO PR
-------
A ALF502 ENGINE
A JTM-17
AuT8D-9
AJT9D-7 VORBIX
JT9D-7A
A A ARB211-22B COMBUST #2
JT90-7
OVERHAUL
ASPEV 511
A TPE331-5-251- A CTCP8B-98CIC APU
ARB211-22B COMBUST #1
ACFM56 COMBUSTOR RIG
11 20 30 .40
SVXMEAN = MEAN EPA SMOKE NUMBER
Figure 41 SMOKE NUMBER COMBUSTOR INLET PRESSURE COEFFICIENT P3COEF
VERSUS MEAN SMOKE NUMBER SMKMEAN
-------
I
I
I
5.0 «•
I
7 c +
g2- ,
" I
< 1
u- i
Z
2 2.P +
&
01
- =eT3COEF* # s * • »
O < I 3
is
s
s
#
0
#
C 0
T
#
#•
#
#
n
1#
0
Figure 42 NOX TEMPERATURE CORRECTION FACTORS
-------
t
I
2.S ~
I
! CHAMR = eHAMBCOEFMHAMBREF HAMB,^)
HAMB CO EF "A^NOX
I - "-10.
i e
i •
Z.O ~ 9
HAMBCOEF—14. • 9
I * w 9
I # #
HAMB CO EF--18. * * 9 9
I S * * 9
g 1 S S S # 3 #
J- HAMBCOEF S s * # 9
g 1.5 ~ * -22. S S * * 9 9
5 i «•» s s s # # o
2 HAMBCOEF s s **99
O I .. *»*« 3 ! 3 « » M
P | * * S S # # 9 »
0 I «"*»SS#9»
K I S 3 * 9 9
OC ! ..**#»»
8 1.0 ~ . * • 9 9 • .
" , « » # e * •
f i e @ ® #
S J e
1 >
3 I
X |
£ 1
§ o.s *
i
I
I
I
I
I
i
0.0 ~
I* * +
-0.03 -0.02 -0.01 -0.00 0.01
AMBIENT HUMIDITY (REFERENCE) - AMBIENT HUMIDITY (MEASURED)
Figure 43 NOX HUMIDITY CORRECTION FACTORS
-------
0
1
-c*
ON
O 6
S
<
u.
Z
o
p
u
Ui
EC
GC
8 4
t-
<
K
UJ
&
2
UJ
T3 CO EF - -.020
T3 CO EF - -.016
#
T3 CO EF = -.012
T3 CO EF ¦ -.008
T3 CO EF - .004
# 0
T3,
HC
= •T3 C0 EF (T3REF " ^MEAS1
(# e
#
0
-100
-BO
50
T3(REFERENCE)-T3(MEASURED)
Figure 44 HC TEMPERATURE CORRECTION FACTORS
100
-------
0
1
¦£>
vl
K
I,
IL
1
u
ua
AC
tc
8 2
HI
cc
i
S
h :
8
T30CEF --XJ180
T3COEF- .0128
P 9
T3COEF"jOOW e
•T3COEF --0064
T30OEF --J0O32
00
0
#
0
#
#
s
#
s
#
t
#.
-SO 0 50
T3
-------
I
2.5 *
I
I
I
I
I
Co, -teB§E-\
^SMOKE ^\,EAsJ
P3 CO EF
P3 CO EF = 3.6
# P3C0EF =235
0
1
4^
00
s
0 I
H
3
u.
z 1.5 -
2 i
h- i
a
1 :
8 i
uI |
5 i.o r
M !
M ,
uj !
AC
0.
UJ
*
O
s
u
« # *
® #
* • - * s P3C0EF «2.10
# $ S
9
0 ? V
p * 4' j
? 0 *> :: j; « ~
0 0 ^ C - . .
0 # •*
• s s
*- J 3 s
# , . = & §
* * * S # -3 -J
S 3 •* 0 9
1
.
* *
*
*
i*
3
3 5 # #
1
*
ft
* *
*
j
3
S
# .i O 5
;
* « f. *
s
£
S 3
#
e
0 ?
c
s
w4 ,>
0
I
s c s s
-*
*
I
«¦ Q
o
-------
"TRUE" CORRECTION FACTOR - •-002,T3REF ' T3MEAs'
"ACTUAL" CORRECTION FACTOR - •¦004(T3REF * T3MEAS*
I ~ +
100 -50 0 SO 1
T3-T3
-------
JO
30
"TRUE" CORRECTION FACTOR =e-22 -HAM3< MEASURED)
Figure 48 CORRECTION FACTOR SENSITIVITY ANALYSIS NOX HUMIDITY CORRECTION
PERCENT ERROR = ((TRUE CORRECTION-ACTUAL CORRECTION)/TRUE
CORRECTION)*100
-------
60
"TRUE" CORRECTION FACTOR - t" 017
"ACTUAL" CORRECTION FACTOR - •--012-T3
-------
'TRUE" CORRECTION FACTOR = e' ^'^REF " T3MEAS>
"ACTUAL" CORRECTION FACTOR = e' °°7(T3REF ' "^MEAS1
10
0
1
tn
K>
-10
-20
-30 ~
I + ~ ~ + ~ -
-100 -50 0 50 100
T3(REFERENCE>-T3(MEASURED)
Figure 50 CORRECTION FACTOR SENSITIVITY ANALYSIS CO TEMPERATURE CORRECTION
PERCENT ERROR = ((TRUE CORRECTION-ACTUAL CORRECTION)/TRUE
CORRECTION)*100
-------
ERROR
20
"TRUE CORRECTION FACTOR ¦ C,3ref/p3meas)1*5
ACTUAL CORRECTION FACTOR - *75
0
1
V)
(/¦
-10
-20
-30 ~
|4 ~ -
0.7 0.8 0.9 1.0 1.1 1.2 1.3
P3(REFERENCE>/P3
Figure 51
CORRECTION FACTOR SENSITIVITY ANALYSIS SMOKE PRESSURE CORRECTION
PERCENT ERROR = ((TRUE COR RECTI ON-ACTUAL CORRECTION)/TRUE
CORRECTION)* 100
-------
ENGINE
MODE
CFM56 RIG
IDLE
1.5 IDLE
APPROACH
CLIMB
TAKEOFF
UNCORRECTED
CV (%)
15
18
16
18
19
CORRECTED
CV (%)
Specific Engine General
Coefficient Coefficient
6
5
7
5
5
Reduction
60
72
56
72
74
Reduction
6 60
7
10
6
6
61
35
67
68
ALF502 RIG
IDLE
1.5 IDLE
APPROACH
13
13
11
15
4
3
5
69
77
64
68
3
3
_3
6
77
77
73
65
CV = Coefficient of Variation
where CJ- = standard deviation
M * mean value
o~
^6
Figure 52 NOX AMBIENT EFFECTS CORRECTION SUMMARY
10-54
-------
ENGINE
MODE
UNCORRECTED
CV (%)
CORRECTED
CV (%)
Specific Engine General
Coefficient Coefficient
CFM56 RIG
IDLE
98
34
Reduction Reduction
65 31 68
ALF502 RIG
IDLE
1.5 IDLE
48
36
18
23
62
36
20
24
58
33
61
25
54
25
53
CT
CV = Coefficient of Variation =
where cr = standard deviation
= mean value
!Figure 53 HC AMBIENT EFFECTS CORRECTION SUMMARY
10-55
-------
ENGINE
MODE
UNCORRECTED
CV (%)
CORRECTED
CV (%)
Specific Engine General
Coefficient Coefficient
CFM56 RIG
IDLE
1.5 IDLE
25
23
12
15
Reduction
52
35
Reduction
19 24
24 -4
ALF502 RIG
IDLE
1.5 IDLE
APPROACH
23
15
20
21
6
6
10
10
74
60
5£
54
7
8
n_
1.4
70
47
45
36
CV = Coefficient of Variation = —
uc
where
-------
ENGINE
MODE
UNCORRECTED
CV (%)
CORRECTED
CV (%)
Specific Engine General
Coefficient Coefficient
Reduction
Reduction
ALF502
IDLE
1.5 IDLE
APPROACH
79
90
31
67
76
83
26
62
16
78
87
28
64
5
0~"
CV = Coefficient of Variation =
u
-------
HYPOTHETICAL
EMISSION y
INDEX
(7«v ~ «Vi ' Y)
(Vj - yl ~ CTy
UNCORRECTED
HYPOTHETICAL
EMISSION y
INDEX
REFERENCE
CORRECTION FACTOR CORRELATING PARAMETER
CORRECTED
REFERENCE
CORRECTION FACTOR CORRELATING PARAMETER
Figure 56 HYPOTHETICAL EMISSION INDEX CORRECTION
10-58
-------
APPENDICES
A-l
-------
APPENDIX A
DATA BASE SUMMARY
ENGINE SUMMARY BY CLASS
ENGINE
CLASS
TESTS
SOURCE
SMOKE
MODES
COMMENTS
ALF60Z
T1
56
AVCO (EPA SPONSORED DATA)
Y
IDLE
TO
TAKEOFF
LIMITED HUMIDITY VARIATION
ALFS02 RIG
T1
66
AVCO (EPA SPONSORED DATA)
Y
IDLE
TO
APPROACH
NONSTANDARD AIR FLOW AT IDLE ft l.S IDLE
JT15D-4
T1
12
PRATT (r WHITNEY CANADA
N
IDLE
TO
TAKEOFF
.
TFE731-2
T1
S3
CARRETT
N
IDLE
TO
TAKEOFF
TFE731-3
T1
28
CARRETT
N
IDLE
TO
TAKEOFF
.
CFM56 RIC
T2
214
CE (EPA SPONSORED DATA)
Y
IDLE
TO
TAKEOFF
REDUCED P3 AT TAKEOFF
JT9D-7A
T2
181
PRATT ft WHITNEY
Y
IDLE
TO
TAKEOFF
PILOT LOT DATA
JT9D-7F
T2
94
PRATT ft WHITNEY
Y
IDLE
TO
TAKEOFF
PILOT LOT DATA
JT90-7 OVERHAUL
T2
114
BRITISH MINISTRY OF DEFENCE
V
IDLE
TO
TAKEOFF
BRITISH ENGINE OVERHAUL LTD.
0T9D-7 VORBIX
T2
86
NASA CLEAN COMBUSTOR PROGRAM
Y
IDLE
TO
TAKEOFF
MODEL 27E. PILOT FA ABOVE/BELOW 0.009
RB211-22B COMBUST #1
T2
52
ROLLS-ROYCE (SEA LEVEL DATA)
Y
IDLE
TO
TAKEOFF
LIMITEO HUMIDITY RANGE
RB211-22B COMBUST #2
T2
16
ROLLS-ROYCE (SEA LEVEL DATA)
Y
IDLE
TO
TAKEOFF
LIMITED HUMIDITY RANGE
RB211-22B
T2
21
ROLLS-ROYCE (ALTITUDE DATA)
N
IDLE
TO
TAKEOFF
10700 METER SIMULATED ALTITUDE
SPEV 511
T2
152
ROLLS-ROYCE
Y
IDLE
TO
TAKEOFF
.
TF30
T2
194
ALLEN-SLUSHER FAA DATA
N
I0LE
TO
TAKEOFF
FIXED AREA EXHAUST REPLACES AFTERBURNER
JT8D-9
T4
137
PRATT ft WHITNEY
Y
IDLE
TO
TAKEOFF
PILOT LOT DATA
JTBD-17
T4
85
PRATT ft WHITNEY
Y
IDLE
TO
TAKEOFF
PILOT LOT DATA
ALLISON SOIK
P2
81
DETROIT DIESEL ALLISON
Y
10500 RPM,13800 RPM
STATIONARY GAS TURBINE
PT6A-S0
P2
15
PRATT ft WHITNEY CANADA
N
IDLE
TO
TAKEOFF
.
TPE331-S-251
P2
336
CARRETT (EPA SPONSORED DATA)
Y
IDLE
TO
TAKEOFF
WATER!HUMIDITY) NOT FULLY VAPORIZED
TPE331-8
P2
31
CARRETT
N
IDLE
TO
TAKEOFF
.
TVNE
P2
28
ROLLS ROYCE
Y
IDLE
TO
CLIMB
.
TS6 RIC STANDARD
P2
no
USAF AFAPL (JP4 .TEMPERATURE)
N
I0LE
PARAMETRIC RIG TEST
TS6 RIC LEAN
P2
64
USAF AFAPL < JP4 .TEMPERATURE)
N
IDLE
PARAMETRIC RIG TEST
T56 RIC RICH
P2
16
USAF AFAPL (0P4 .TEMPERATURE)
N
IDLE
PARAMETRIC RIG TEST
TS6 RIC STANDARD
P2
64
JUSAF AFAPL (JETA,TEMPERATURE>
N
IDLE
PARAMETRIC RIG TEST
TS6 RIC LEAN
P2
64
USAF AFAPL (JETA.TEMPERATURE>
N
IDLE-
PARAMETRIC RIG TEST
TS6 RIG STANDARD
P2
84
USAF AFAPL (JP4 .PRESSURE)
N
IDLE
PARAMETRIC RIG TEST
TS6 RIC RICH
P2
24
USAF AFAPL (0P4 .PRESSURE)
N
IDLE
PARAMETRIC RIG TEST
TS6 RIC STANDARD
P2
60
USAF AFAPL (JETA,PRESSURE>
N
IDLE
PARAMETRIC RIG TEST
T63-A-5A WALL FILM
P2
9
DETROIT DIESEL ALLISON
Y
N/A
PARAMETRIC RIC TEST
T63-A-5A PRES ATOM
P2
9
DETROIT DIESEL ALLISON
Y
N/A
PARAMETRIC RIG TEST
T63-A-SA AIR BLAST
P2
9
DETROIT DIESEL ALLISON
Y
N/A
PARAMETRIC RIG TEST
CTCP85-98CK
APU
240
GARRETT (EPA SPONSORED DATA)
Y
BLEED+SHAFT.OTHER
WATER(HUMIDITY) NOT FULLY VAPORIZED
GTCP85-98CK
APU
10
CARRETT
N
BLEED+SHAFT
MAXIMUM POWER
GTCPBS-98D
APU
8
GARRETT
N
BLEED+SHAFT
MAXIMUM POWER
CTCPBS-129
APU
6
CARRETT
N
BLEED+SHAFT
MAXIMUM POWER
CTCP660-4
APU
6
CARRETT
N
BLEED+SHAFT
MAXIMUM POWER
TSCP700-4
APU
6
GARRETT
N
BLEED+SHAFT
MAXIMUM POWER
-------
APPENDIX B
COMBUSTOR INLET CONDITIONS RELATED TO AMBIENT CONDITIONS
The compressor discharge temperature (or the combustor inlet tempera-
ture) of a jet engine that corresponds to a particular discharge pressure
or pressure ratio depends also upon overall compressor efficiency and inlet
temperature to the compressor. Its computation is based upon the definition
of overall compressor efficiency, which is the ratio of ideal isentropic
change of enthalpy (energy) of the air as it passes through the compressor
to its actual change.
charge enthalpy and the related discharge temperature. Their use for this
purpose can be explained best by example using typical values for compressor
characteristics:
Gas tables* have been prepared that aid in the computation of dis-
Given: Pressure ratio = 20
T2 inlet temperature = 580°R
/^c efficiency = 0.85
Procedure: Look up 580°R in the table for air and note the
corresponding value for h, enthalpy, and P , relative
pressure, a computation parameter:
h
2
138.66 Btu/lb and P =1.78
r„
Compute P^
= P x pressure ratio = 35.6
3 r2
Look up T3 and corresponding to P
T3 = 1333°R h3 = 325.47 Btu/lb.
B-l
-------
These are the ideal values for state 3. To find the
actual values, use the definition of ,
^ ^actual
Ah
ideal
n
325.47 - 138.66 _ 219.78 Btu/lb
85
Then h3 = 219.78 + 138.66 = 358.44 Btu/lb
actual
Look up T3 . that corresponds to the value of
setuax
h3
actual
T3 , = 1459 °R.
actual
The computation of accurate values for actual T3, such as is done
for performance analysis, requires this kind of procedure. However, a simpl
method exists, which may suffice for some purposes. In fact, evidence of it
2
use can be found in the literature of the gas turbine industry. This is
manifested by a simple formula for T3, which is based upon the definition
for , and the assumption of a constant specific heat for air. Its
derivation follows:
Ah. , , C (T3. , . - T2)
ideal p ideal
r\ c = a = ¦ en
V L± h „ . C (T3 „ . - T2)
actual p actual
k-1
a"d "ideal " T2 (S) " (2)
where P3/P2 is simply the compressor pressure ratio, k is the ratio of
specific heats for air, and is the specific heat at constant pressure
B-2
-------
Substituting Eq. (2) into Eq. (1) and solving for T3 we get:
aCtUaly
T3
actual
T2
/fc
(B)
k-1
- 1
+ T2
The error incurred by using this formula increases with temperature,
i.e., pressure ratio. In order to evaluate this effect, discharge temperature
was computed using both methods over a range
of
pressure ratios
and the results
are presented in a table below.
ERROR DUE TO FORMULA FOR
DISCHARGE
TEMPERATURE T2 = 580oR,(
L
= 0.85
T3
Pressure Ratio actual
T3 _ ,
actual
Percent Error
by formula, °R
by table, 0
R
10 1215.0
1196.7
1.51
15 1376.8
1345.5
2.27
20 1503.6
1459
2.97
Note that the purposes of this report, T2 is equivalent to TAMB,
the ambient temperature of the testing chamber.
*Keenan, J.H. and Kaye, J., Gas Tables, J. Wiley 5 Sons, Inc.-, 1963.
^For example, see Shaw, H., "The Effects of Water, Pressure, and Equivalence
Ratio on Nitric Oxide Production in Gas Turbines," Trans, of the ASME.
J. of Eng. for Power. July 1974. ~ "
B-3
-------
APPENDIX C
MULTICOLLINEARITY AND AIRCRAFT EMISSIONS DATA
One problem which can arise in the development of regression models
to predict the effect of changing combustor inlet conditions on emission
levels is that of multicollinearity. Multicollinearity arises whenever,
either in the population or in the sample, various of the explanatory
variables stand in an exact or almost-exact linear relation to each other.
When multicollinearity occurs, it is as if members of a subset of explanatory
variables always act in unison. As a result, the data lack sufficient
independent variation to allow one to sort out the separate effect of each
independent variable. The greater the degree of multicollinearity that
exists, the more arbitrarily and unreliably does least squares regression
allocate the sum of the unexplained variation among the individual
explanatory variables. This phenomenon therefore can result in coefficient
estimates which are particularly sensitive to changes both in the model
specification and data used to develop the model.
Table CI has been constructed to analyze the impact of multi-
collinearity on a subset of the ambient effects program test data.
Presented in this table are the correlation coefficients for LN(NOX), T3,
LN(P3) and HAMB (ambient humidity), and the regression coefficients and
their standard errors for a variety of NOX model formulations. The data
used to develop this table is the CFM56 combustor rig, all thrust levels.
The first point of note in this table is the high degree of correla-
tion (.91507) between combustor inlet temperature T3 and the logarithm of
combustor inlet pressure LNP3. This situation suggests that a model using
both these terms may be subject to multicollinearity problems. In order to
analyze the impact of this potential problem, five regression models were
developed and are summarized in Table CI. In the first model (LNN0X=f(T3)),
a T3 coefficient of .002481 was determined. Similarly, the model
LNN0X=f(LNP3) yielded a LNP3 coefficient of 1.1177. When both T3 and LNP3
C-l
-------
are used together to predict LNNOX as in regression #3, a substantial change
in the estimated coefficients occurs. The T3 goes from .002481 to .001670,
a change of -32.7%, and the LNP3 coefficient changes from 1.1177 to 0.4101,
a change of -63.3%. This volatility in the estimated T3, LNP3 coefficients
and the high correlation between these two variables suggests that multi-
collinearity is present to a considerable degree.
On the other hand, the addition of the HAMB to the equation represents
a substantial "real" improvement in the estimation process. As shown in
Table CI the correlation between T3 and HAMB is only 0.03248 and the
correlation between LNNOX and HAMB is -0.23270. Ambient humidity therefore
represents a reasonable predictor of NOX emission levels that is not as
highly related to T3 as is P3. The inclusion of the HAMB term results in a
substantial reduction in the residual sum of squares (SS Qr) and only
changes the T3 coefficient from .002481 to .002503 (0.9%). In this case,
multicollinearity does not pose a serious problem.
The trends illustrated by the above example were also demonstrated
by the contributed data analyzed during the program. In fact, since much
of the contributed data had relatively small sample sizes and was collected
for the most part in undesigned experiments, the degradation in the
coefficient standard errors was more pronounced than in the CFM56 data.
Table C2 presents a summary of NOX regression coefficients and standard
errors for the Pratt 5 Whitney JT9D-7A Pilot Lot data. As shown in this
table, the addition of combustor inlet pressure adds little to the predictive
2
ability of the equations (same R ) and the LNP3 standard error (.1075)
indicates that the LNP3 coefficient (.0109) is not significantly different
from zero.
While the above examples illustrate the instability which can be
introduced in the estimation of T3 and LNP3 regression coefficientsf they do
not necessarily imply that independent combustor inlet temperature and pressure
correction factors can never exist. Had sufficient test data existed in the
data base for a variety of engines whose engine control strategies result in
independent T3 and P3 variation, separate correction coefficients for both T3
and P3 could have been developed and P3 coefficient trends established,
C-2
-------
However, only a limited number of engine tests were available in which both
temperature and humidity were held constant while pressure was allowed to vary
independently.
In order to establish representative values for potential HC, CO
and NOX combustor inlet pressure correction coefficients, small subsets of the
CFM56 combustor rig data were analyzed in which both ambient temperature and
humidity were held constant and ambient pressure permitted to vary from 26.0 in
Hg to 32.3 in Hg. Each of these data sets contained six observations.
Regressions were performed using the functional models:
LNHC = f(LNP3)
LNCO = f(LNP3)
LNNOX = f(LNP3)
with TAMB, HAMB constant.
These models imply a P3 correction factor of the form
(p3 \ P3 COEF
meas l
"ref )
where P3C0EF is the LNP3 regression coefficient determined above.
For HC, a statistically significant P3 effect could not be established.
For CO, on the other hand, a combustor inlet pressure coefficient of -1.46 was
found. For ambient pressures in the range 26.0 to 32.3 in Hg, the magnitude of
this coefficient implies a CO pressure correction factor ranging from 1.24
to 0.89. For the range of ambient pressures typically experienced during
emission testing (e.g., 29.92 ±1 in Hg), this CO pressure correction coeffi-
cient provides correction factors from 1.05 to 0.95. or a maximum correction
of 5 percent. Similar analyses were performed on the NOX data and a combustor
inlet pressure coefficient equal to 0.34 was determined, A correction
coefficient of this magnitude will provide NOX pressure correction factors
ranging from 0.95 at 26.0 in Hg to 1.03 at 32.3 in Hg or a 5 percent maximum
correction.
C-3
-------
While the above analysis of small subsets of the CFM56 combustor rig
data suggests that independent T3 and P3 correction factors for CO and NOX may
be applicable to this particular engine, an independent P3 correction factor
could not be established for other engines in the data base which for the most
part were subjected to less extensive testing than the CFM56 rig. Since the
magnitude of the CFM56 pressure correction factors for CO and NOX cited above
were less than 5 percent for normal pressure excursions from reference day
conditions, only a combustor inlet temperature coefficient was employed in the
development of correction factors presented in this report.
Still another effect of multicollinearity problems can be demon-
strated by examining variance maps for the data in question. A brief discussion
of variance mapping is given below.
In general, the linear model which gives the true emission level, y,
of a particular engine is given by:
$1 fl * ^ f2 * ^ f3
CD
where ^ i = 1,2,3.. are constants, ... are basis functions of
combustor inlet temperature (T3), combustor inlet pressure (P3), ambient
humidity (HAMB), and e is the random error.
Since e is a random variable, the responses observed at each
(T3, P3, HAMB...) point also constitute a random variable. As a result
it is only possible to obtain from the observations an equation of the form:
' s 'iViVVj'- (2)
A (2
where y is an estimate of y and bj, i = 1,2,3 ... is an estimate of
C-4
-------
The variance function is controlled by three considerations:
1) The type of basis functions employed in the
regression model.
2) The positions in T3, P3, HAMB space, called
design points, at which emission measurements are
taken, and
2
3) The magnitude of the error variance (J- at each
design point.
Var $) varies at different coordinates in T3, P3, HAMB space.
At some points the response can be estimated with relatively little error;
at other positions the error can be quite large. The variance in the esti-
mated response at a point P in T3, P3, HAMB space is given by
Var (y) = x (X'X)"1 x» (J~2 (3)
where x is a vector obtained by evaluating each of the basis functions
at the particular point P and X'X is matrix of the least squares normal
equations. Therefore, for every point P, Var is actually a variance
function. By dividing both sides by
-------
One should observe that the normalized variance (equation 4) is a
function of the inverse of the X'X matrix. The effect of multicollinearity
presents itself by noting that, as the dependence between explanatory
variables increases, the diagonal terms in matrix corresponding to
these variables tend toward infinity. In more practical terms, this means
that the variance function itself increases with the degree of multi-
collinearity. In addition, the variance (standard errors) of the regression
coefficients are directly related to the (X'X)-1 matrix. As terms in this
matrix increase, due to multicollinearity problems, the magnitude of the
coefficient standard error also increases.
The increase in the variance function due to multicollinearity in
the explanatory variables T3 and P3 is demonstrated as follows.
Figures CI and C2 are variance contour maps for the CFM56 data as a
function of (1) T3 and (2) T3 and P3. (Note that in Figure CI the
contours are constant with respect to P3 because it was not included in
the basis function for this map. The variance contours are plotted vs T3
and P3 in this figure for consistency only.) The contour lines trace a
^ z
constant value (as indicated on each line) of Var(y)/
-------
TABLE G1
NOX REGRESSION COEFFICIENTS VS. MODEL FORMULATION
GE CFM56 COMBUSTOR RIG - IDLE TO TAKEOFF
(214 Observations)
MODEL
COEFFICIENT/STD ERROR
CONSTANT
LNP3
T3
HAMB
SSERROR
R2
CI)
LNNOX=f(T3)
.3049
(.0361)
-
.002481
(.000051)
-
7.22
.9168
(2)
LNNOX=f(LNP3)
-3.2315
(0.1393)
1.1177
(0.0299)
-
-
11.43
.8682
(3)
LNNOX=f(T3,LNP3)
-1.0584
(0.1754)
.4101
(.0519)
.001670
(.000112)
-
5.57
.9358
(4)
LNNOX=f(T3,LNP3,HAMB)
-.3971
(.0578)
.2567
(.0169)
.001994
(.000036)
-19.28
( 0.44)
0.56
.9936
(5)
LNNOX=f(T3, HAMB)
.4674
(.0154)
-
.002503
(.000021)
-20.70
(0.63)
1.18
.9864
LNNOX
T3
LNP3
HAMB
CORRELATION COEFFICIENTS
LNNOX T3 LNP3
1.00000
0.95748
1.00000
0.93176
0.91507
1.00000
HAMB
-0.23270*
0.03248
-0.05482
1.00000
When effect of varying thrust levels is removed by fitting LNNOX=f(T3), the HAMB-LNNOX correlation
is -0.91437. ^ ) = Coefficient Standard Error
-------
TABLE C2 NOX REGRESSION COEFFICIENTS vs MODEL FORMULATION
PRATT § WHITNEY JT9D-7A PILOT LOT DATA, IDLE TO TAKEOFF
(181 Observations)
MODEL COEFFICIENT/STD ERROR R2
CONSTANT LNP3 T3 HAMB
(1) LNNOX=£(T3,LNP3,HAMB) -.2319 .0109 .004122 -19.73 .9771
(.3289) (.1075) (.000340) (3.16)
(2) LNNOX=f(T3.HAMB) -.1987 - .004156 -19.94 .9771
(.0362) (.000048) (2.37)
CORRELATION COEFFICIENTS
LNNOX
T3
LNP3
HAMB
LNNOX 1.00000
.98386
.97840
.06671
T3
1.00000
.98231
.16356
LNP3
1.00000
.03893
HAMB
1.00000
( ) = Coefficient Standard Error
C-8
-------
EQUATION: LNNOX = F(T3)
8.
B-
a:
•—. o
CO £¦
a_
Ll)
cc
ZD
CO
?\
LU
tr ~
o_
LU
cc.
LU
2
tr
=3
CD
8-
K3-
(s
02
.01
_T_
35
.005
T-
50
.005
T
65
.01
-T"
80
~r
95
.02
^
110
(X101 )
.03
20
BURNER INLET TEMPERATURE(DEG. F)
125
Figure CI CFM56 VARIANCE MAP
C-9
-------
EQUATION: LNNOX = F(T3, LNP3)
T
35 50 65 80 95
BURNER INLET TEMPERATURE(DEG. FI
(X101 )
Figure C2 CFM56 VARIANCE MAP
C-10
-------
APPENDIX D
COMBUSTOR RIG - FULL-SCALE ENGINE CORRELATION
This appendix summarizes investigations undertaken to analyze the
correlation between combustor rig and full scale engine emissions. This
analysis attempts to partially answer two interrelated questions:
1) Can emissions from combustor rigs operating at lower
combustion inlet pressures than corresponding full scale
engine be related to emissions from the full scale
engine itself?
2) Is it reasonable to include correction factor
coefficients developed using data from combustor
rigs in the development of general full scale engine
correction factors?
In order to investigate these questions, EPA-sponsored data
on the T1 class ALF502 combustor rig and full scale engine were
analyzed to determine the nature and extent of the rig-engine correlation.
The ALF502 data was selected for this analysis since it represents the
only data source for which sufficient ambient effects tests were performed
on both the full scale engine and corresponding combustor rig.
The approach taken to assess the rig-engine correlation is suggested
by several commonly used correlation methods summarized below which
emphasize the role of combustor inlet pressure P3 (or its equivalent T3)
on the emissions process.
D-l
-------
SELECTED RIG-ENGINE CORRELATION TECHNIQUES
NASA
P3 .
HC = HC . rig
GE
engine rig P3
& 6 engine
CO = CO
P3 . \ N
ng \
'engine I P3engine)
P3 . < -°"37
NOX . = NOX . 1 —
engine rig \ P3
5 & » engine
P3 .
SMOKE . I rig
SMOKE . rig P3 I
engine 6 \ engine J
y.
In the analysis taken, individual equations of the form
LN(Pollutant) = f(l, LNP3)
were developed for HC, CO and NOX for both the engine and rig data. The
natural logarithm of combustor inlet pressure LNP3 was selected for this
analysis rather than combustor inlet temperature T3 so that direct compari-
sons with NASA and GE correlations could be made. Table D1 summarizes
the results of these regressions. This regression analysis approach was used
in the analysis since test data with the rig and engine operating under the
same ambient conditions was not available.
D-2
-------
The approach used to assess the degree of rig engine correlation
is graphically illustrated in Figure D1 which presents the LNHC vs LNP3
rig and engine regression lines. The combustor rig and full scale engine
can be correlated if the rig data operating at a combustor inlet pressure
P3 . can be extrapolated to determine the emission levels of the full
rig
scale engine operating at P3 where P3 > P3 . . In mathe-
6 r " engine engine' rig
matical terms, the ability to correlate rig and engine emissions can be
established if both the slope (P3C0EFF) and intercept (CONSTANT) of the
rig and engine regression equations are equal. Under these conditions,
the low combustor inlet pressure rig data can be treated as an extension
of the full scale engine data. Since the slopes of the rig and engine
equations are equal, the same emissions variation in response to changing
ambient conditions can be expected even though the absolute emission levels
may differ. Under the above circumstances, it is reasonable to include the
rig correction factor coefficients in the development of a general correc-
tion factor scheme.
In order to assess the equality of rig and engine intercepts and
slopes, Table D2 has been constructed. This table presents the 95%
confidence bounds for the CONSTANT and P3C0EF regression terms presented
in Table D1. A comparison of the degree of overlap between the confidence
bands suggests that a meaningful difference cannot be found between the
rig and engine coefficients. For example, the confidence limits on the
rig and engine CO pressure coefficients could be illustrated as follows:
RIG
ENGINE
LOWER CONFIDENCE
BOUND
-1.3537
P3C0EF
-1.1741
UPPER CONFIDENCE
BOUND
-.9945
-1.4711
•1.1987
,9263
ALF502 CO PRESSURE COEFFICIENT CONFIDENCE LIMITS
D-3
-------
As seen in this illustration, a considerable degree of overlap exists
between the two CO coefficient confidence limits with the result that a
meaningful difference between rig and engine does not exist. Similar
conclusions can be drawn about the other confidence limits presented.
It should be noted that the confidence band analysis presented above
is not a rigorous test for the equality of slopes and intercepts of two
regression equations. Instead, it is presented as a satisfactory approxi-
mation relying to a great extent upon an engineering assessment of the
practical significance of any small differences found in the model coefficients.
A more rigorous analysis using the data from Table D1 and the method for
comparing two linear bivariate regression lines presented in Appendix F
was performed and also confirms the hypothesis that ALF502 rig and full scale
engine emissions can be correlated.
To determine a suitable pressure correlating factor for HC, CO, and
NOX, the weighted averages (based on sample size) of the rig and engine
pressure coefficients were determined and are summarized below:
P3C0EF (Weighted Average)
HC -2.1(3
CO -1.18
NOX 0.76
ALF502 RIG/ENGINE WEIGHTED AVERAGE PRESSURE COEFFICIENTS
The method used to compute engine emissions from rig emissions is
illustrated in Figure D1 for the representative case of HC.
LNHC = LNHC . - 2.36 (LNP3o -LNP3 . ,
eng rig eng rig)
P3
LNHC . - 2.36 LN —
ri® P3 .
rig
D-4
-------
Taking inverse logarithm
HC
eng
-2.36
or
HC
eng
2.36
The corresponding CO and NOX rig engine correlations are summarized
below:
CONCLUSIONS
Using the limited data available on the ALF502, the ability to
correlate combustor rig and full scale engine emissions has been demonstrated.
Of particular importance in this analysis has been the ability to demon-
strate that the response to changes in ambient conditions (as seen in
P3C0EF) is of equal magnitude between the rig and full scale engine
even though the absolute emission levels between rig and engine may differ.
CO
engine
1.18
engine
D-5
-------
TABLE D1
ALF502 RIG-ENGINE CORRELATION ANALYSIS
REGRESSION SUMMARY
LN(POLLUTANT) = CONSTANT + P3COEF*LNP3
HC
EI
Observations
Constant
Std Error
P3COEF
Std Error
Mean Square Error
Variance (LNHC)
R2
Rig
66
9.8189
.6809
-2.2855
.1773
.2271
.8041
.7218
Engine
21
10.7168
1.2174
-2.5894
.3404
.0199
.0766
.7528
CO
EI
Observations
Constant
Std Error
P3COEF
Std Error
Mean Square Error
Variance (LNCO)
R2
66
7.8733
.3448
-1.1741
.0898
.0582
.2105
.7276
21
7.9157
.4869
-1.1987
.1362
.0032
.0154
.8031
NOX
EI
Observations
Constant
Std Error
P3COEF
Std Error
Mean Square Error
Variance (LNNOX)
R2
66
-1.5512
.1752
0.7289
.0456
.0150
.0738
.7994
28
-1.8299
.0791
0.8196
.0210
.0012
.0673
.9832
D-6
-------
TABLE D2
ALF502 RIG-ENGINE CORRELATION ANALYSIS
COEFFICIENT CONFIDENCE BOUND SUMMARY
Oc. = 0 . 05
CONSTANT P3COEF
HC RIG 8.4571 to 11.1807 -2.6401 to -1.9309
EI ENGINE 8.2820 to 13.1516 -3.2702 to -1.9086
CO RIG 7.1837 to 8.5629 -1.3537 to -.9945
EI ENGINE 6.9419 to 8.8895 -1.4711 to -.9263
NOX RIG -1.9016 to -1.2008 .6377 to .8201
EI ENGINE -1.9881 to -1.6717 .7776 to .8616
D-7
-------
X
LU
a
V)
M
(ENGINE) 10-72
(RIG) 9.82
LNHC
RIG
U LNHC
O
O
<
ce
3
lnhceng
LNP3
RIG
RIG EQUATION
SLOPE = -2.29
WEIGHT AVERAGE SLOPE « -2.36
ENGINE EQUATION
SLOPE » -2.59
LNP3
ENG
LNP3
NATURAL LOG OF COMBUSTOR INLET PRESSURE
Figure D1 GRAPHICAL INTERPRETATION OF ALF502 RIG ENGINE CORRELATION
D-8
-------
APPENDIX E
COMPARISON OF TWO LINEAR BIVARIATE REGRESSION LINES
Given two regressions:
= number of points for Regression 1
Y2 = b2 x2 + a2 n2 = number of points for Regression 2
Test the hypothesis that: b^ = b2
2 2
given that (0" j5 ( 0" 2 are un'cnown an^ unequal
yx 1 ^ yx
The test statistic t is given as:
yb2
calc Sn
b
and
is compared to a t distribution with V D.F
2 2
(S ) (Sz )
2 y* 1 y* ,
s d = 1 + "
b S S
X1 X2
where
Sx = VarfXj) • (n2 - 1)
Sx = Var(x2) • (n2 - 1)
-------
2 2
(S ) = estimate of (Q~ ) = Residual Mean Square 1
yx j yx ^
2 2
(S ) = estimate of (Q" ) = Residual Mean Square 2
yx 2 yx 2
^t 2 2 ^1 ~ °1 ~ 2
r r/Vj + (i - Kr/v2
v2 = "2 " 2
ts^A
(s2 ) S + (s2 )_ s
yx J x2 yx 2 x1
2
S is the pooled variance estimate of b^, b2
b
and the degrees of freedom are weighted by
their respective variances.
DECISION RULE:
IF tcaic^tv Accept Hypothesis eg bj = b2
t
IF tcalc > \ Reject Hypothesis eg bj f b2
REFERENCES:
1. Documenta Geigy, Scientific Tables, ed. by K. Diem § C. Lenther,
7th edition, Geigy Pharmaceuticals, Division of CIBA-Geigy Corp.,
Ardsley, New York 10502.
2. Intro. Engineering Statistics, Guttman § Wilks.
E-2
-------
APPENDIX F
SAMPLE EMISSION INDEX CORRECTION
This appendix illustrates the steps required to correct the emission
indices for a particular engine. For the purposes of this illustration, the
HC, CO, and NOX CFM56 combustor rig emission indices will be corrected using
the coefficients developed specifically for this engine. The correction
coefficients used in this example are:
HC:
CO:
NOX:
T3COEF
T3COEF
T3C0EF
HAMBCOEF
-0.023486
-0.006443
0.002503
¦20.70
The reference engine conditions to which the emission levels were
corrected are:
Reference Engine Conditions
Mode
% Rated
T3 (Deg. F)
P3 fPSIA)
Idle
6
353
48.6
1.5 Idle
9
417
63.7
Approach
30
628
139.0
Climb
85
890
148.4
Takeoff
100
941
168.6
These reference conditions were obtained from the manufacturers'cycle deck
for the rig operating under the following reference conditions:
Ambient Temperature = 59 DEG. F
Ambient Pressure = 29.31 IN HG
Ambient Humidity = 0.0075 lb ^O/lb Air
It should be noted that the correction to "standard day" reference conditions
is identical to the process outlined in this appendix except that values of
-------
T3 £ and P3re£ corresponding to an ambient, pressure of 29,92 IN HG and an
ambient humidity of 0.00634 lb F^O/lb Air are used.
The corrected emissions indices are generated on a mode-by-mode
basis using the following expressions:
^ -.023486 * (T3 . - T3,. )
HCmDD = HCmcac * e Ref Meas
CORR MEAS
CO = CO e".006443 * (T3 - T3 )
CORR MEAS
noxcorr
* .002503 * (T3_ _ T3 ) * - 20.7(HAMBD p-HAMB.. )
NOXj-p.,, * e Ref- Meas e Ref Meas'
where T3 _ is different for each mode.
Ker
Tables F1 through F6 present plots of both the measured (uncorrected)
and the corrected HC, CO and NOX CFM56 rig emission indices. The vertical
lines in these figures correspond to the reference combustor inlet temperatures
to which the data were corrected. Ideally, the corrected emissions data
should be a horizontal line with a minimum vertical spread.
F-2
-------
HC t
e ~
GE CFM56 COHBUSTOR RIG * CORRECTION FACTOR ANALYSIS
PLOT OF T3*HC LEGEND: A - 1 OBS , B - 2 OBS , ETC
I
W
A A
A A A
A
AAA
A
A A A A A
AA AB A A A A
2B0
300
320
340
360
400
420
440
T3 » COHBUSTOR INLET TEMPERATURE(DEC. F)
Figure F1 UNCORRECTED HC EMISSIONS INDEX versus COMBUSTOR INLET
TEMPERATURE - CFM56 COMBUSTOR RIG, IDLE
-------
GE CFM56 COMBUSTOR RIG - CORRECTION FACTOR ANALYSIS
PLOT OF T3*HCCORR LEGEND: A - 1 OBS , B - 2 OBS . ETC
HCCORR I
8 *
•n
i
4^
A A
A A
I ~—
280
300
AA
A
320
A A
A A
A A A
A A A A BAA
A A A
A A
3*0 360 380 400
T3 - COMBUSTOR INLET TEMPERATURE(DEG. F>
420
440
Figure F2 CORRECTED HC EMISSIONS INDEX versus COMBUSTOR INLET TEMPERATURE -
CFM56 COMBUSTOR RIG, IDLE
-------
GE CFM56 COMBUSTOR RIG - CORRECTION FACTOR ANALYSIS
PLOT OF T3*CO LEGEND: A - 1 OBS , B - 2 OBS , ETC
CO I
60 ~
50
tn
I 20
N
0
E
X
40
30
10
BA
A A
A A
A
A
A A
AAA
AB
A A
A A
A AA
A
AA
A A
AA
A
A A
250
300
T3
350 400
COMBUSTOR INLET TEMPERATURE( DEG. F)
A A
BA A A
A
A A
ABA A
D AAA A
A
450
500
Figure F3 UNCORRECTED CO EMISSIONS INDEX versus COMBUSTOR INLET
TEMPERATURE - CFM COMBUSTOR RIG, IDLE AND 1.5 IDLE
-------
GE CFM56 COMBUSTOR RIG - CORRECTION FACTOR ANALYSIS
PLOT OF T3*COCORR LEGEND: A « I OBS , B - 2 OBS , ETC
COCORR I
60 ~
50
I
0\
40
3C
20
I
N
0 10
E
X
BA
AAA
A AA
A A
AA
AA
A
B
A
A
AA A
AA
B
BA
A
AA
A A A
BA A
BAA A
C A AA
l«~-
250
300
350 400
T3 - COMBUSTOR INLET TEMPERATURE< OEG. F)
450
500
Figure F4 CORRECTED CO EMISSIONS INDEX versus COMBUSTOR INLET TEMPERATURE -
CFM56 COMBUSTOR RIG, IDLE AND 1.5 IDLE
-------
TJ
CE CFN56 COMBUSTOR RIG - CORRECTION FACTOR ANALYSIS
PLOT OF T3*NOX LEGEND: A - 1 OBS ,8=2 OBS , ETC
NOX
30
25
20
15
10
AB
AB
AA
B
B
AB
A
CA
AAA AB
BA AA AO A A
AO BB FCC AA B
CB BA AG BB AB A AE
A A AA ADA A
B
AC
A
A
A
B
AD
A
AB
AB
A
A A
B
BB
B
A
A
.A
CA
A
BA
C
A.A
B
B
BC
AA
B
B.
A
AA
AA
AA
400
600 800
T3 - COMBUSTOR INLET TEMPERATURE(DEG. F)
1000
1200
Figure F5
UNCORRECTED NOX EMISSIONS INDEX versus COMBUSTOR INLET
TEMPERATURE - CFM56 COMBUSTOR RIG, IDLE TO TAKEOFF
-------
T1
I
00
GE CFMS6 C0HBUST0R RIG - CORRECTION FACTOR ANALYSIS
PLOT OF T3*NOXCORR LEGEND: A » 1 OBS , B - Z OBS . ETC
NOXCORR I
30 ~
C
0
R
R
E
C
T
E
0
N
0
X
E
M
1
S
S
I
0
N
1
N
D
E
X
25
20
15
10
A
BC CA
AC
I
A
BA BB BAA BB CB
0 A ACAB C AAFCA
AH ACC FED
B
C
AI
A
A
B
A.
A
A
A
B.
A
BC
A BB
B
BC
A
AA
B A.
0
AB
A
A A
A C.
AB
BB
AA
B
BA ADA
DB
AE
AAA
I *—
200
400
T3
eoo soo
COMBUSTOR INLET TEMPERATURE
-------
APPENDIX G
COMPUTATION OF EPA EMISSION PARAMETERS (EPAP) FROM
EMISSION INDEX AND SPECIFIC FUEL CONSUMPTION
The EPA Emission Parameter, or EPAP^, for each gaseous emission
component is computed from the Emission index for each component
and operating mode and from the specific fuel consumption for the
corresponding mode after it has been corrected to standard conditions.
Thrust specific fuel consumption (TSFC) is used in the case of a jet or
fanjet engine, and brake specific fuel consumption in the case of a
propjet engine (BSFC). The EPAP is, of course, a cumulative result
obtained by summation over the complete EPA operating cycle of taxi/idle,
takeoff, climbout, approach and taxi/idle.
EPAP is specified in units of pounds of pollutant per 1000 pounds of
thrust-hr. EI is defined in units of pounds of pollutant per 1000 pounds
of fuel burned. SFC is defined as fuel flow rate, pounds per hour
per pound of thrust, or pounds of fuel per brake horsepower-hour. The
commonality of terms among these parameters suggests that they are related
and an expression for this is written by incorporating the time-in-mode
of operation (TIM), as follows:
j
ETIM.
(EI..) (SFC.) (—i)
J 60
The subscript i indicates the specific pollutant and j denotes the
prescribed mode of operation or power level.
SFC based upon measured values for actual fuel flow rate and thrust
or power is corrected to SFC at standard conditions.by the relation:
G-l
-------
i/yyf&
SFCcorr. = Thrustraeas/7
FF SFC
meas/ 7V " meas
y[B~
Ambient Temperature °R , Ambient Pressure PSIA
where ^ — and 7 =
519 14.7
because the standard ambient temperature is 519 °R and standard pressure
14.7 psia.
Figures G1 and G2 illustrate a typical application of the above SFC
correction formula for the case of the CFM56 combustor rig. In Figure G1
the measured fuel flow in lb/hr for each of five modes is plotted versus
ambient pressure. As shown in this figure, uncorrected fuel flow rises
significantly with increasing pressure. Figure G2 presents comparable data
after the fuel flow has been corrected using the ambient temperature and
pressure method outlined above. The corrected fuel flow as desired is
independent of the ambient pressure. A similar plot versus ambient tempera-
ture also demonstrates the effectiveness of this correction method.
G-2
-------
B.S236Z+B4
O
W
K a.*53S=*34
F
L
0
V
B.iBBBt+B*
B.35ff5E+J4
B.3B33E*Bi
B.ZS0eS+B4
S>.2BB2£*B*
B.ISBBE+B4
B.IBB3E+B*
B.5BB3i+Z3
0^0 +•*•»*•*•« +**ft***ft*ft4.*ft***««**4.**ftftft***«4.«ft*ft«ft*ftft+*»*ftftftftttft4.**ft****ftft4.*«««***ft*4.•«*••***« 4**ftftftftftft ft4
B.ZSBBE+BZ B.2663E+32 B.ZZ2Bt*B2 B.293BE+B2 B.314BZ+32 B.32331*32
AMBIENT PRESSURE
IN H6
Figure G1 MEASURED FUEL FLOW VERSUS AMBIENT PRESSURE - CFM56
COMBUSTOR RIG, IDLE TO TAKEOFF
-------
0. 5C0BH +04
ft «**+******•*•+******«•*+*«*****••+** •****•*+*««***
c
0
R
R
E
0.4500EfJBT4
0.4000E+04
0.3S00E+04
CI
I
¦b.
L
0
u
0.3000E+04
0.Z500E+04
0.2000E+04
0.1500E+04
0.1000E+04
0.5000E +03
g _& +w* ***•***+**«•*•***+*«
0.2500E+02 0.2660E+32 0.Z820E+0Z 0.Z980E+02 0.3140E+02 0.3300E+02
AMBIENT PRESSURE
IN H G
Figure G2 CORRECTED FUEL FLOW VERSUS AMBIENT PRESSURE - CFM56 COMBUSTOR
RIG, IDLE TO TAKEOFF
-------
|