Summary of ECTD Emission Laboratory
Correlation Programs
Fiscal Year 1974
by
Richard E. Lowery
October 3, 1974
Environmental Protection Agency
Office of Air and Waste Management
Office of Mobile Source Air Pollution Control
Emission Control Technology Division
Standards Development and Support Branch
Ann Arbor, Michigan
-------
ABSTRACT
This report summarizes the results of ECTD 1974 laboratory correlation
programs and infers conclusions relative to the state of current emission
laboratory correlation.
Data are presented and analyzed which show site-to-site correlation
demonstrated during ECTD programs. Regression analyses are presented
which detect the effect of barometric pressure and ambient humidity on
correlation results. Other factors influencing the degree of site-to-
site correlation, namely dynamometer characteristics, vehicle variability,
CVS accuracy, and gas analysis accuracy, are presented and discussed.
-------
Introduction;
During fiscal year 1974 the Emission Control Technology Division (ECTD)
conducted several inter-laboratory correlation programs with motor vehicle
manufacturers. Results of these studies were carefully analyzed by ECTD
personnel.
Purpose;
It is the intent of this report to summarize the results of ECTD 1974
correlation programs and to draw conclusions regarding the state of emission
laboratory correlation.
Summary of Programs;
ECTD correlation programs were conducted between the months of January
and June, 1974. Table 1 summarizes the various programs and the dates of
EPA involvement in them.
Table 1
Program Dates of EPA Participation
Japanese Automobile Manufacturers
Association (JAMA) February 4-13
Motor Vehicle Manufacturers
Association (MVMA) March 11 - May 17
Honda March 19 - April 9
Audi-NSU . May 31
The scope of each of the above programs was vastly different. The JAMA
program involved emission tests on two vehicles, dynamometer checks, and
.static gas exchanges. The MVMA program consisted of complete equipment
diagnostic checks, static gas exchanges, and emission tests on five vehicles.
The Honda program included dynamometer and CVS checks, static gas exchanges,
and'emission tests on four vehicles. The Audi-NSU study consisted of gas
exchanges and emission tests on one vehicle.
Analysis of Results;
The results of the above programs were analyzed to determine the degree
of site-to-site equivalency and the factors affecting the degree. The
following sections present those analyses.
Degree of Test Site Equivalency - Appendix A presents data relating to
vehicle emission correlation. Appendix A-l summarizes the specifica-
tions of the correlation vehicles. Appendix A-2 presents vehicle HC
-------
-2-
emissions normalized to the EPA mean value. Test-to-test variability
can be derived from the 95% confidence bands. These data show a large
vehicle-to-vehicle variation in the degree of HC correlation. Signifi-
cant differences are seen between measurements of EPA and the following
laboratories: Nissan, Honda, and CMC.
Appendix A-3 summarizes the degree of CO correlation. The normalized
values show significant discrepancies between EPA and (1) Nissan, and
(2) Honda, while other laboratories and EPA exhibit a high degree of
correlation.
Appendix A-4 presents the NOx emission correlation comparisons.
The only significant discrepancies are in the comparisons of Nissan and
Audi-NSU results to EPA.
Appendix A-5 shows the normalized C02 comparison. Significant dis-
crepancies are shown between EPA and results from (1) Honda, (2) CMC,
(3) AMC, and (4) Ford.
Effect of Barometric Pressure on Test Results - Appendix B-l pre-
sents the results of regression analyses of barometric pressure versus
emissions. It should immediately be noticed that barometric effects are
highly vehicle dependent. CO emissions most consistently correlate with
the barometer, only the Toyota vehicle shows no correlation. All other
vehicles exhibit the tendency of increasing CO emissions for decreasing
barometer. HC emissions show correlation on several vehicles, in general,
HC'emissions are inversely proportional to pressure. NOx correlates with
the barometer in several cases, with a general increase in NOx for in-
creasing pressure. C02 values correlate well with Honda and MVMA data,
however the trend is inconsistent.
Effect of Humidity on Test Results - Appendix C presents regression
analyses data relating ambient humidity to emission test results. In
general, no correlation was found between humidity and any emission, in-
cluding corrected NOx. Consequently, the current method of correcting
NOx emissions for humidity was not shown to be in error.
Discussion;
Previous discussion has related barometric pressure and ambient
humidity to test results. Other factors were believed to influence
site-to-site correlation, and those factors will now be discussed.
1. Dynamometers - Perhaps the most pronounced variable in all
correlation programs was the dynamometer. Highly different configura-
tions of roll size, roll spacing, inertia drive system, power absorber
-------
-3-
capacity, and vehicle restraint system were employed in various correla-
tion programs. In addition to these physical differences, the problems
of accurately calibrating the dynamometer power absorber and speed meter
are significant. Consequently, the effect of various dynamometers on
test results could possibly be important, however, available data cannot
quantify such effects. Future correlation efforts should be centered on
learning more about the influence of the dynamometer characteristics on
emission test results.
2. Vehicle Variability - Changes in the test vehicles are often a
significant problem in judging the results of a correlation program.
Test-to-test variability is usually acceptable, except that driver in-
fluences can sometime make HC and CO repeatability poor. But site-to-
site variability is a major problem, especially since many of the vehicles
in question were transported thousands of miles by air to the EPA labora-
tory for testing. The effect of vehicle changes is unmeasurable, and
thus an unknown variable in any program. Extensive preconditioning of
correlation vehicles at EPA prior to testing should help minimize this
problem.
3. Constant Volume Sampler (CVS) - The CVS accuracy is also a
possible source of variations in test results. The small amount of
correlation work investigating this area (MVMA study) showed poor
results of propane injection checks on several CVS systems. Perhaps
more effort needs to be expended to quantify the influence of the CVS
on test results.
4. Gas Analysis System - The phase of the correlation tests which
probably contributed least to correlation problems was the gas analysis
phase. Exchanges of static gases with all involved laboratories produced
very comparable results. However, as laboratory-to-laboratory correlation
continues to improve, the area of gas analyzer calibration and maintenance
cannot be ignored.
Conclusions and Recommendations;
An analysis of the results of ECTD 1974 correlation programs shows
that good inter-laboratory correlation can be obtained. There are,
however, several factors which may decrease the degree of laboratory-
to-laboratory emission correlation. These factors include barometric
pressure, dynamometer characteristics, vehicle variability, and CVS
inaccuracies.
The influence of dynamometer variables on test results could be
most readily determined because such influences are usually seen in NOx
and CC-2 emissions, which are most repeatable. Therefore, it is recommended
that future correlation studies be designed to quantify the effects of
dynamometer variables on test results.
-------
-4-
The effect of vehicle variability on correlation results is extremely
difficult to quantify. Therefore, it is recommended that all correlation
vehicles be carefully preconditioned in an attempt to minimize such vari-
ability.
Barometric pressure effects, like vehicle effects, are very difficult
to isolate. The best method of studying these effects would be an
environmental chamber, which is currently unavailable. Another way of
determining barometer influences is to collect all available data on
vehicles undergoing replicate tests. It is therefore recommended that
EPA collect barometric pressure data on all correlation vehicles and
carefully examine such data for trends in barometric effects. Such
analysis must take care to eliminate the confounding effects of dyna-
mometer differences and vehicle variability.
-------
-5-
References
1. "Emission Laboratory Correlation Study Between EPA and the Japan
Automobile Manufacturers Association, Inc.," Richard E. Lowery;
EPA Report; April, 1974.
2. "Emission Laboratory Correlation Study Between EPA and Honda Motor
Company, Inc.," Richard E. Lowery; EPA Report; April, 1974.
3. "Emission Laboratory Correlation Study Between EPA and Audi-NSU
Auto Union," Richard E. Lowery; EPA Report; June, 1974.
4. "Emission Laboratory Correlation Study Between EPA and the Motor
Vehicle Manufacturers Association of the United States, Inc.,"
Richard E. Lowery; EPA Report; September, 1974.
-------
Appendix A
Vehicle Emission Correlation Data
-------
1975 Model Year Correlation Programs
Correlation Vehicle Specifications
Vehicle
Designation
1
2
3
4
5
6
7
8
9.
10
11
12
13
*
Manufacturer Inertia
and Model Weight
Toyota Carina 2500
Toyota Carina 2500
Datsun 610 2750
Datsun 610 2750
Honda Civic CVCC 2000
Honda Civic CVCC 2000
Honda Civic 2000
Honda Civic CVCC 2000
CMC Caprice 4500
AMC Hornet 3000
Chrysler Coronet 4500
Ford Maverik 4000
Audi Fox . 2500
CMC Repca I 4500
No. of
Cylinders
4
4
4
4
4
4
4
4
8
6
8 .
8
4
8
C.1.D. Control System
96.9 Eng. Mod., Air Inj.
96.9 Catalyst, EM, AI
119.1 EGR, Air Inj.
119.1 Catalyst, EGR, AI
90.8 Eng. Mod.
90.8 Eng. Mod.
75.5 Air Injection
90.8 Eng. Mod.
• 350 , Catalyst, EGR, AI
232 EGR
- 318 Catalyst, EM, EGR
302 Catalyst
97 Eng. Mod.
350 Eng. Mod.
>
ts
•o
tt>
Cu
H-
X
* Round-robin vehicle used in MVMA program.
-------
1975 Model Year Correlation Programs
Vehicle HC Emissions Comparisons
Laboratory Symbols
EPA
Toyota Motor Co.
Nissan Motor Co,
Honda Motor Co .
General Motors Corp.
American Motors Corp.
Chrysler Corp.
Ford Motor Co,
Audi-NSU Auto Union
5^ A f i : -i -I • \_ i Q
f *' o r t / r i o
Vehicle Designations
12 -*-* 13
JAMA Correlation
Feb. 4-13, 1974
EPA Site i"5
Honda Correlation
Mar. .19-Apr. 9, 1974
?;PA Site y'f
MVMA Correlation
May 6-17, 1974
T^A Site "5
Audi-NSU -._•
May 31 *?
rite *5 "•-•
-------
1975 Model Year Correlation Programs
Vehicle CO Emissions Comparisons
CO
T3
cfl
pa
QJ
o
0)
10 C
C O
O O
•H
co s^s
M U">
W 'O
O tfl
O
CO
0> 0)
•H 3
0 t-l
0)
N
o
3
'•^tr-iS^EFrHtr:-™ :vfSrn
Laboratory Symbols
EPA
Toyota Motor Co.
Nissan Motor Co.
Honda Motor Co.
^General Motors Corp.
+ American Motors
Chrysler Corp.
Ford Motor Co.
Auto
. -:~.--.-—:|-tr:47-..-'vT :r.:'..:rq
JAMA Correlation
Feb. 4-13, 1974
EPA Site *5
Honda Correlation
Mar. 19-Apr. 9, 1974
EPA Site #fi
I MVMA Correlation
~T*-May 6-17, 1974
Rite #5
Audi-NSU
' May 31
sit*. #5
-------
1975 Model Year Correlation Programs
Vehicle NOX Emissions Comparisons
CO
•a
rt
pa
0)
y
c
0)
•a
•H
14-1
co c
c o
o c_>
•H
00 tr-S
co m
2 co
0)
0) 3
i-l iH
O cd
01
T)
cu
N
Laboratory Symbols • ^4~- :iilr."il^^iJi^^B^iEp?^
""~~~ sTn-ii^^t^ru: -ciifPntirJ
""
'frrh:-
^^; ^Toyota Motor Co.
BNissan Motor Co.
Motor Co.
"ri "^^
^Chrysler Corp.
Motor Co.
HAudi-NSU Auto Union
..v ...---i- > i
5 -i~f 6 4—t- 7 --+ 8
Vehicle Designations
3 -»-4- 4
m
Jfir:-1"-:
JA^lA Correlation
Feb. 4-13, 1974
EPA Site :"5
Honda Correlation
Mar. 19-Apr. 9, 1974
F.PA Site *fi
• I
-H^-
MVMA Correlation
May 6-17, 1974
EPA RltP.' #.S
Audi-NSU
May 31
Site.- *-5.
-------
CD •
TJ
CO
pq
1975 Model Year Correlation Programs
Vehicle C0£ Emissions Comparisons
~'
'4 '-i f. i:rr
Uii
Laboratory Symbols
EPA
Toyota Motor Co .
Nissan Motor Co.
Honda Motor Co.
1.°. -! 11 ^"f -
General Motors Corp. .....i^ij^if
rtfer-rU":!^7
-ur -r;u.
. Tt.71-- i rzrr"_,r| t^
'American Motors Corp, . ..itfiJt-^p :-jS:j:.:::.::-I[-;:; -.j^L
I Chrysler Corp. '.-..
Ford Motor Co. ;
* Audi-NSU Auto Union
IM
••*-••• ' — -T1 .—*..-
^•:i .^XF
f—. 3 ^rrf 4 ^4 5 -H4- 6 -+^f 7
Vehicle Designations
h
JAMA Correlation
1 Feb. 4-13, 1974
I Honda Correlation I
-*f«-Mar. 19-Apr. 9, 1974 -**+
* •>-•«« i ft t . JL s- *
MVMA "Cor r elation
May 6-17, 1974
EPA Site #5
"Audi-NSU
•\ May 31 .
-*-'.2 Site #5 i
-------
Appendix B
Regression Analyses of
Test Results vs. Barometric Pressure
-------
Appendix B-l
Linear Regression Analyses
Test Results vs. Barometric Pressure
Regression Correlation Coefficients
Vehicle
Designation*
1
2
3
4
5
6 .
7
8
9
10
HC
.3826
-.847
-.937
-.524
-.850
.339
-.583
-.933
-.781
.416
CO
.0711
-.782
-.909
-.960
-.812
-.761
-.965
-.959
-.937
-.422
NOX
-.6223
.475
.944
.936
.901
-.670
.914
-.682
.654
.887
C02
-.281
-.314
-.226
.377
.888
-.982
.920
.987
.850
.809
Fuel
Economy
.402
.554
.735
.280
-.928
+ .988
-.810
-.982
-.854
-.723
* See Appendix A-l for vehicle specifications.
-------
Appendix B-2
Effect of +1.0 Inch of Mercury Barometric Change
Based on Least - Squares Linear Curve
Vehicle Gram/Mile Change
Designation HC CO NOV
1 .13
2 -.25
3 -.63
4 -.07
5 -.08
6 .01
7 -.27
8 -.13
9 -.16
10 .04
HC
1 10.0
2 -32.7
3 -45.7
4 -13.0
5 -21.5
6 1.8
7 -17.4
8 -22.7
9 -15.9
10 0.8
.23
-1.48
-12.02
-2.79
-0.79
-0.48
-2.33
-1.54
-1.14
-0.85
Percent of
CO
1.9
-15.2
-78.1
-61.2
-24.1
-8.7
-21.7
-29.0
-12.1
-49.1
*»
-.28
.11
.40
.30
;i .12
-.05
.16
-.16
.26
.18
Mean Change
NOX
-14.1
5.5
24.7
20.2
8.6
-3.4
10.9
-8.4
9.2
12.5
CO?
-12
-6
-6
5
5
-28
13
15
132
62
CC-2
-2.9
-1.4
1.6
1.2
1.4
-8.9
4.1
4.2
26.4
7.6
Fuel
Economy*
0.8
0.5
1.5
0.2
-0.3
2.4
-0.7
-0.9
-4.2
-0.6
Fuel
Economy
4.0
0.0
7.5
0.1
-1.2
8.5
-2.7
-3.8
-24.3
-6.1
* Fuel economy changes are in miles/gallon.
-------
Appendix C
Regression Analyses of
Test Results vs. Ambient Humidity
-------
Appendix C-l
Linear Regression Analyses
Test Results vs. Ambient Humidity
Regression Correlation Coefficients
Vehicle
Designation*
1
2
3
4
9
10
HC
.5132
.4427
.4404
-.1718
.5000
-.7760
CO
.0494
.7079
.2500
.6145
.6162
.2960
NQV
.7198
-.2025
-.0382
-.5023
-.2854
-.7376
CO?
.4186
.6614
.2272
.1651
-.5178
-.9297
Fuel
Economy
-.5101
-.1510
.1506
-.4996
.5042
.8493
Note: NOX results are corrected for ambient humidity per Federal
Register specifications
* See Appendix A-l for vehicle specifications,
-------
Appendix C-2
Effect of a 10 Grain H20 per Lb. Dry Air
Ambient Humidity Increase
Based on Least-squares Linear Curve
Gram/Mile Change
Vehicle
Designation
1
2
3
4
9
10
1
2
3
4
9
10 -
HC
.010
.029
.079
-.005
.063
-.016
HC
0.8
3/8
5.8
-1.0
6.2
-8.9
CO
.010
.313
.891
.402
.889
.112
Percent
CO
0.1
6/;
5.8
8.8
9.5
6.4
\
N0y
.021
-.011
-.004
-.036
-.063
-.027
of Mean Change
NOV
1.1
-0.6
-0.2
^2.4
-1.0
-1.3
CO?
1.2
5.0
28.1
0.5
-49.1
-13.2
CO?
0.3
1,2
7.0
0.1
-9.8
-1.6
Fuel
Economy*
-.067
-.035
.083
-.076
1.487
.141
Fuel
Economy
-0.3
-0.2
0.4
-0.4
8.9
1.3
* Fuel economy changes are in miles/gallon.
------- |