Technical Report
ANALYSES OF ASSEMBLY LINE TESTING
Final Report
R. J. Mogavero and IV. R. Fairchild
Calspan No. NA-5194-D-II
"MISSIONS & EXHAUST GASES
Calspan Corporation
Buffalo, New York 14221
•jr
Formerly Cornell Aeronautical Laboratory, Inc.
-------
E PA-420-R-73-102
FEB 211974
c. J. domie
Calspan
On November 17, 1972 Cornell Aeronautical Laboratory (CAL) changed its name to Calspan Corporation and converted to
for-profit operations. Calspan is dedicated to carrying on CAL's long-standing tradition of advanced research and development
from an independent viewpoint. All of CAL's diverse scientific and engineering programs for government and industry are being
continued in the aerosciences, electronics and avionics, computer sciences, transportation and vehicle research, and the environ-
mental sciences. Calspan is composed of the same staff, management, and facilities as CAL, which operated since 1946 under
federal income tax exemption.
ANALYSES OF ASSEMBLY LINE TESTING
Final Report
R. J. Mogavero and W. R. Fairchild
Calspan No. NA-5194-D-II
Office of Mobile Source Air Pollution Control
Environmental Protection Agency
Ann Arbor, Michigan 48105
May 1973
Contract No. 68-01-0435
Calspan Corporation
Buffalo, New York 14221
-------
ACKNOWLEDGEMENTS
This study of assembly line test programs and activities related
thereto was performed for the Office of Mobile Source Air Pollution Control
of the Environmental Protection \gency.
This research effort was conducted by a project staff consisting
of, beyond the authors of the Report, Messrs. Arnold C. Keller, Paul F.
Przybylski, and William F. Gould. Their efforts and contributions were
significant.
During the course of the study, numerous meetings were held with
William R. Oliver, Joseph Somers, Gary Rossow, Ron Kruse and Ken Johnston
of the Environmental Protection Agency in Ann Arbor, Michigan. Their
collective guidance, comments and suggestions proved extremely helpful in all
phases of this study.
ii
-------
TABLE OF CONTENTS
Pape No.
ACKNOWLEDGEMENTS
LIST OF FIGURES
LIST OF TABLES
1 OVERVIEW AND SUMMARY OF RESULTS
1.1 Summary of Project Objectives
1.2 Summary of Results
1.3 Comments on Data Base
2 GENERAL METHODOLOGICAL APPROACH
2.1 Introduction and Overview of Emission Level Changes
2.2 Trends in Average Emissions Levels During a
Production Year
2.3 Short Inspection Tests and Their Impact Assessment
2.4 Audit Tests and Their Impact Assessment
3 FAILURE RATES AND REASONS FOR FAILURE OBSERVED FROM ASSEMBLY
LINE TEST PROGRAM DATA
3.0 Introduction
3.1 Seven Mode Inspection Tests
3.1.1 Overview of Failure Rates
3.1.2 Failure Rates: General Motors Corporation
a) Data Base
b) 7-mode Test Failure Rates
c) 7-node Test Failure Rates:
Adjusted For Test Malfunctions
Failure Rates: Ford Motor Company
3.1.3
3.1.4
a) Data Rase
b) 7-mode Test Failure Rates
Failure Pates: Chrysler Corporation
a) Data Base
b) 7~mode Test Failure Rates
n
vi
vii
1-1
1-1
i:l4
2-1
2-1
2-6
2-9
2-11
3-1
3-1
3-1
3-1
3-2
3-2
3-4
3-8
3-10
3-10
3-10
3-11
3-11
3-11
in
-------
TABLE OF CONTENTS (Cont'd)
3.2 Audit Tests
3.2.1 Overall Audit Test
3.2.2 Audit Test Failure
3.2.3 Audit Test Failure
3.2.4 Audit Test Failure
3.3 Audit Test Failure Reasons
Failure Rates
Rates : Heneral Motors
Rates: Ford
Rates: Chrysler
4 AN ASSESSMENT OF EMISSIONS REDUCTIONS ATTAINABLE FROM A
NATIONWIDE APPLICATION OF CURRENT ASSEMBLY LINE TEST
PROCEDURES
4.0 Introduction
4.1 F.missions Level Changes Associated With Trends
During a Production Year
4.1.1 Introduction
4.1.2 Methodology
4.1.3 Summary o f Results
4.1.4 Data and Computations: Trend Analysis
4.2 Emissions Level Changes Associated With Short
Inspection Tests
4.2.0 Introduction
4.2.1 Methodology
4.2.2 Data Base Available
4.2.3 Summary of Results
4.2.4 Data and Computations: Short Inspection
Tests
4.3 Emissions Level Changes Associated With Audit Tests
4.3.0 Introduction
4.3.1 Methodology
4.3.2 Data Base Employed
4.3.3 Summary of Results
4.3.4 Data and Computations: Audit Tests
4.4 An Analysis of Cross Emitters
Page No.
3-12
3-12
3-13
3-17
3-21
3-22
4-1
4-1
4-1
4-1
4-1
4-8
4-11
4-17
4-17
4-1 7
4-18
4-23
4-31
4-51
4-51
4-51
4-52
4-53
4-54
4-58
iv
-------
TABLE OF CONTENTS CCont'd)
Page No,
ALTERNATE ASSEMBLY LINE TEST PROCEDURES AND THEIR
IMPACT ON EMISSIONS 5-1
5.1 Introduction 5-1
5.2 Methodology and Analyses 5-1
5.2.1 Considerations of Modified Failure Rate
Standards 5-1
5.2.2 Consideration of Alternate 7 Mode and
Idle Test Rates 5-3
5.2.3 Consideration of Alternate Audit Sampling
Procedures 5-6
AN ASSESSMENT OF EMISSIONS REDUCTIONS ATTAINABLE WITH
CATALYTIC CONVERTER EMISSION CONTROL SYSTEMS PLANNED
FOR THF, 1976 MDDEL YEAR 6-1
6.1 Introduction 6-1
6.2 Methodology and Results 6-1
COST ANALYSES OF NATIONWIDE ASSEMBLY LINE TEST PROGRAMS 7-1
7.1 Introduction 7-1
7.2 Nationwide Costs of California Assembly Line
Test Procedures
7-2
7.3 Nationwide Costs of Alternate Assembly Line ^
Test Procedures
8 REFERENCES 8-1
APPENDIX A A-l
v
-------
LIST OF FIGURES
Figure No. Title Pape No.
2.1-1 Schematic of Typical Emissions Distribution 2-1
2.1-2 Temporal Patterns of Average Emissions Level 2-2
2.1-3 Schematic Flow of Emissions Distribution Changes 2-5
2.4-1 Skewed Emission Patterns 2-11
4.2-1 1972 General Motors Data Sample Sizes 4-19
7-node Tested Automobiles
4.2-2 1973 General Motors Data Sample Sizes 4-20
7-mOde and/or Idle Tested Automobiles
4.2-3 1973 Ford Data Sample Sizes 4-21
7-mode and/or Idle Tested Automobiles
4.2.4 1973 Chrysler Data Sample Sizes 4-22
7-mode and/or Idle Tested Automobiles
4.2-5a-c Short Test Patterns of Emissions Level Changes 4-23,26
1973 General Motors
4.2-6a-c Short Test Patterns of Emission Level Changes 4-27,28
1973 Ford
4.2-7a-c Short Test Patterns of Emission Level Chanpes 4-29,30
1973 Chrysler
VI
-------
LIST OF TABLES
Table No. Title Page No.
1.2-1 Overall 7-mode Failure Rates l-~
1.2-2 Quarterly 7-mode Failure Rates 1-2
1.2-3 Overall Audit Failure Rates 1-3
1.2-4 Summarized Emissions Level Changes - 1972 1-6
1.2-5 Summarized Emissions Level Changes - 1973 1-7
1.2-6 Summarized Emissions Level Changes - 1973 j.g
(tons per year)
1.2-7 Summarized Emissions Level Changes - 1973 i_i
(grains per mile)
1.2-8a Hydrocarbon Emission Level Changes
For Various Seven Mode/Audit Test Rates 1-11
1.2-8b Carbon Monoxide Emission Level Chanpes
For Various Seven Mode/Audit Test Rates
1-11
1.2-8c Oxides of N'itropen Emission Level Changes
For Various Seven Mode/Audit Test Rates 1-12
1.2-9 Audit Rate Related Emissions Changes with j ^
Catalytic Converter System
3.1-1 1972 Model Year Seven Mode Inspection Test 3-1
Failure Rates (by Manufacturer)
3.1^2 7-mode Test Failure Rates 3-4
General Motors: 1972 Model Year
(by Division and Production Quarter)
3.1-3 7-mode Test Failure Rates 3-6
General Motors: 1972 Model Year
(by Engine Size and Production Quarter)
3.1-4 7-mode Test Failure Rates 3-7
General Motors: 1972 Model Year
(by Engine/Division Category and Production Quarter)
3.1-5 7-mode Test Failure Rates: Adjusted for 3-8
Test Malfunctions
General Motors: 1972 Model Year
(by Division)
vii
-------
LIST OF TABLES (Cont'd)
Table No. Title Page No,
3.1-6 7-mode Test Failure Rates 3-8
Adjusted For Test Malfunctions
General Motors: 1972 Model Year
(by Engine Size and Production Year)
3.1-7 7-mode Test Failure Rates 3-9
Adjusted For Test Mai functions
General Motors: 1972 Model Year
(by Engine/Division Category and Production Ouarter)
3.1-8 7-mode Test Failure Rates 3-10
Ford: 1972 Model Year (by Engine Size)
3.1-9 7-mode Test Failure Rates 3-11
Chrysler Corporation: 1972 Model Year
(by Engine Size)
3.2-la Overall Audit Test Failure Rates 3-12
1972 Model Year
3.2-lb Overall Audit Test Failure Rates 3-12
1973 Model Year
3.2-2 Audit Failure Rates: 1972 General Motors 3-14
(by Division/Engine Category)
3.2-3 Audit Failure Rates: 1972 General Motors 3-15
(by Failure Type and Division)
3.2-4 Audit Failure Rates: 1973 General Motors 3-16
(by Division/Engine Category)
3.2-5 Audit Failure Rates: 1973 General Motors 3-17
(by Failure Type and Division)
3.2-6 Audit Failure Rates: 1972 Ford 3-18
(by Division/F.ngine Category)
3.2-7 Audit Failure Rates: 1973 Ford 3-19
(by Di vis ion/Engine Category)
3.2-8 Audit Failure Rates: 1972 Ford 3-20
(by Failure Type and Division)
3.2-9 Audit Failure Rates: 1973 Ford 3-20
(by Failure Type and Division)
3.2-10 Audit Failure Rates: 1973 Chrysler 3-21
(by Engine Category)
3.3-1 Audit Test Failure Reasons:1972 General Motors 3-23
(ranked by frequency of occurrence)
3.3-2 Audit Test Failure Reasons: 1973 General Motors 3-25
(ranked by frequency of occurrence)
viii
-------
Table No.
4.1-0
4.1-1
4.1-2
4.1-3
4.1-4
4.1-5
4.1-6
4.2-1
4.2-2
4.2-3
4.2-4
1.2-5
4.2-6
4.2-7
LIST OF TABLES (Cont'd)
Title Page No.
Trend Associated Emissions Changes 4-8
Average Hydrocarbon Fmissions From Audit Tests 4-9
(by Engine Category and Production Quarter)
1972 General Motors
Average Carbon Monoxide Fmissions From Audit Tests 4-10
(by Fngine Category and Production Quarter)
1972 General Motors
Average Hydrocarbon Fmissions From Audit Tests 4-13
(by F.ngine Category and Production Ouarter)
1972 Ford
Average Carbon Monixide Emissions From Audit Tests 4-13
(by Engine Category and Production Quarter)
1972 Ford
Average Hydrocarbon Emissions From Audit Tests 4-15
(by Engine Category and Production Ouarter)
1972 Chrysler
Average Carbon Monoxide Fmissions From Audit Tests 4-15
(by Fngine Category and Production Ouarter)
1972 Chrysler
F.missions Level Changes From Short Inspection Tests 4-23
Audit Test Emission Results 4-32
7-raode and Not 7-mode Tested Automobiles
1972 General Motors (Original Measurements)
Audit Test Emission Results 4-33
7-mode and Not 7-mode Tested Automobiles
1973 General Motors (Original Measurements)
Audit Test Emission Results 4-34
7-mode and Not 7-mode Tested Results
1973 Ford (Original Measurements)
Audit Test Emission Results 4-35
7-mode and Not 7-mode Tested Results
1973 Chrysler (Original Measurements)
Summary of 7-mode Test Effects on 4-36
Audit Results
Audit Test Emission Results 4-4\
Idle and Not Idle Tested Automobiles
1973 General Motors (Original Measurements)
ix
-------
LTST OF TABLES (Cont'd)
Table No. Title Page No.
4.2-8 Audit Test Emission Results 4-42
Idle and Not Idle Tested Automobiles
1973 Ford (Original Measurements)
4.2-9 Audit Test Emission Results 4-43
Idle and Not Idle Tested Automobiles
1973 Chrysler (Original Measurements)
4.2-10 Summary of Idle Test Effect on Audit Results 4-44
4.2-11 Audit Test Emission Results 4-48
7-mode/Idle Test Combinations
1973 General Motors (Original Measurements)
4.2-12 Audit Test Emission Results 4-49
7-mode/Idle Test Combinations
1973 Ford (Original Measurements)
4.2-13 Audit Test Emission Results 4-50
7-mode/Idle Test Combinations
1973 Chrysler (Original Measurements)
4.3-1 Estimated Emission Level Changes From 4-53
Nationwide Audit Testing at 2%
Sampling Level
4.4-1 Overall Cross Emitter Rates 4-59
1972 General Motors
4.4-2 Gross Emitter Rates - by Category 4-59
1972 General Motors
4.4-3 Averages and Standard Deviations of 4-60
Gross Emitters - by Category
1972 General Motors
5.2-1 7-mode Test Related Emissions Changes 5-5
5.2-2 Audit Rate Related Emission Changes 5-7
5.2-3 Extrapolated Audit Rate Related 5-8
Emission Changes
5.2-4 Audit Sample Allocations 5-12
1972 General Motors
5.2-5 Hydrocarbon <5> ij Values 5-13
1972 General Motors
5.2-6 Carbon Monoxide (5 ij Values 5-14
1972 General Motors
x
-------
LIST OF TABLES Cont'd)
Table No. Title Page No.
5.2-7 NOX c5ij Values 5-15
1972 General Motors
5.2-8 Emission Reduction Prom Optimal 5-16
2% Audit Sampling Allocation
1972 General Motors
5.2-° Comparative Audit Sampling Emission Levels 5-17
5.2-10 Emission Reductions From Optimal 5-17
2% Audit Sampling Allocation
1973 Model Year: Quarters 1 and 2
5.2-11 Extrapolated Emission Reduction Levels 5-18
6.2-1 Emissions Scaling Factors 6-3
6.2-2 1973 Emissions Standards Scaled
to 1976 Test Procedure 6-4
6.2-3 Reduction Factors For Catalytic Converters 6-4
6.2-4 Audit Rate Related Emission Changes
with Catalytic Converter Emission Systems 6-7
6.2-5 Audit Rate Related Emission Changes with
Catalytic Converter System (extrapolated
to all three manufacturers) 6-7
7.2-1 Average Assembly Line Test Costs Per Car
For Various 7-mode/Audit Test Rates 7-3
7.2-2 Nationwide Costs of Assembly Line Test
Programs For Various 7-mode/Audit 7-4
Test Rates
7.2-3a-c Cost of Emission Level Changes 7-5,6
XI
-------
1.
1.1
OVERVIEW AND SUMMARY OF RESULTS
Summary of Project Objectives
Stated very briefly, the focus of this nroject has been to organize
and summarize the results on automobile exhaust emissions levels as estimated
from the data base gathered because of the California Air Resources Roard
Assembly Line Test Program; to assess the nationwide reduction in emissions
which might be realized with currently existing emission control systems if
alternate assembly line test procedures were employed on a nationwide basis;
to extrapolate these results to proposed catalyst type emission control systems;
and to assess the nationwide costs of such programs.
The California Assembly line Test Pngrnm frr J972-7? consists of
the following:
1972 Seven mode test 25% of production
CVS audit test 2% of production
1973 Seven mode test 25% of production
Idle test 75% of production
CVS audit test 2% of production
The 75% idle test requirement for 1073 vns inplemented in stages over the
model year so that, at some stages, fractions of less than 75% were idle tested.
1.2 Summary of Results
The results of the many analyses undertaken will be briefly summar-
ized here. More detailed data underlying these results, as well as a
description of the methodology employed in approaching the numerous subtasks
is contained in subsequent chapters of this report. To the extent -feasible,
the results will be presented and discussed in the order mentioned in Section
1.1.
1-1
-------
Frequency and Patterns of Seven Mode Test Failures
The failure rates experienced on seven node tests applied during the
1972 irodel year varied considerably among manufacturers and at different
times of the production year within a manufacturer. The variation among
manufacturers is illustrated below.
Manufacturer
Overall Failure Rate
Seven Mode Tests
Chrysler
Ford
General Motors
25.5 %
24.8 %
5.3 %
Table 1.2-1
Overall Seven Mode Failure Rates
Throughout the production quarters of the 1972 model year, there
appeared to be a decrease and then increase in the seven mode test failure
rate, and this oattern seemed to hold for all three manufacturers. This
is illustrated below.
Manufacturer
Production Ouarte
;r
^veralI
1
2
3
4
5
General Motors
Ford
Chrysler
N. A.
N. A.
34. 0
6.19
21.0
16.3
4 .95
19.5
27.4
3.83
30.2
24.1
15.82
41 .1
36.6
5.3
24.8
25.5
Table 1.2-2
Quarterly Seven Mode Failure Rates
It is apparent that a gradual improvement in test results occurred
as the model year proceeded, "bottoming out" generally in one of the middle
quarters and then increasing dramatically in the fifth or buildout quarter.
While very noticeable increases in failure rates were observed throughout
this data base for the buildout quarter it. should be noted that this quarter
normally accounts for about 8% of yearly production.
1-:
-------
While these failure rates, within a manufacturer, were analyzed
and compiled on engine category, division, etc. bases, such detail will not
be presented here. These breakdowns can be reviewed in Chapter 3. For our
purposes, at present, it will suffice to say that large variations exist in
failure rates from one engine family to another, from one division to
another, etc. The time pattern in failure rates cited earlier appears,
however, to prevail in spite of these "category" differences.
Frequency, Patterns and Reasons For Audit Test Failures
The overall failure rates on audit tests are shown below for
the 1972 and 1973 model year data available.
Manufacturer
Failure Rate (%)
1972 Model Year
1973 Model Year
General Motors
Ford
Chrysler
7.0
15.7
N.A.
15.7
11 .5
17.37
Table 1.2-3
Overall Audit Failure Rates
These failure rates reflect the number of original test failures
as a percentage of the number of original audit tests performed by a
manufacturer. Within a given manufacturer there were large variations in
audit failure rates between divisions, engine families, engine sizes, etc.
This high degree of variability was true for all three manufacturers and is
elaborated on in Chapter 3. An interesting characteristic which prevailed
among all manufacturers represented in the data base was that on audit
test failures, the HC , CO and NO^ attributed failures tended to be disjoint;
that is, if a car failed on the HC standard it usually did not fail CO and/or
NO^. Consequently, the overall audit failure rates appear to be, by and large,
accounted for by the accumulated failure rates of cars which failed HC only, CO
only and NO^ only.
1-3
-------
One of the manufacturers (i.e. - General Motors) provided descriptors
for the reasons of an audit failure in their data base. Upon analysis of
these "reasons for failure", two major patterns presented themselves: First,
while there were many described reasons for failing an audit test a
comparatively small number of these "reasons for failure" accounted for 90%
of the number of audit failures recorded. From the 1972 data base 19 of the
52 listed reasons for failure accounted for 90.26% of the failures. From
the 1973 data base 8 of the 41 listed reasons for failure accounted for
90.4% of the failures. The second point observed is that the reason for
failure most frequently listed in both years covered was "rerun no repair"
accounting for 40.8% and 68.3% of the failures in 1972 and 1973 respectivelv.
This descriptor is applied when a car fails the audit test but no attributable
reason for failure could be isolated. This frequently occurring situation
suggests a high degree of variability among repeated tests on the same car.
Whether this variability is attributable to the so called "green engine
effect", to inherent variability of the test procedure, or possibly some
combination of both of these factors was not determinable from the data base
available for this study.
Nationwide Emissions Level Changes From Assembly Line Tests
Upon an analysis of the experiences with seven mode, idle and
audit testing during the brief history of the California
Assembly Line Testing Program it appears that an extension of a similar
program to a nationwide basis would increase emissions levels.
This analysis:
(]) determined changes in average CVS emissions from
the second to fifth quarters of 1972
(2) compared CVS emissions for cars that were seven mode
tested and those not. seven mode tested
(3) compared CVS emissions for cars that were idle tested
and those not idle tested
fl) compared initial and final CVS emissions for cars that
were audit tested.
1-4
-------
Tables 1.2-4 through 1.2-7 present, in summary fashion, the results of esti-
mating the changes in emissions levels, on a nationwide basis, if these
programs (i.e. - 25% seven mode testing, 75% idle testing and 2% audit testing)
were implemented. The estimated emission levels contained in these tables are
based on extrapolations which employed one years production volume and assumed
an average yearly mileage of 12000 miles per year for this automobile population.
Some brief interpretive comments are in order here. Regarding the
1972 data base (i.e. - Table 1.2-4) the only emissions changes estimated are
for hydrocarbon and carbon monoxide, since oxides of nitrogen were not
measured by C.V.S. techniques during that model year. The emission level
changes associated with a trend in the average emissions during the year
show a decrease in hydrocarbons and an increase in carbon monoxide. For
reasons discussed at some length in Chapter 4 it is felt that such changes
cannot, on the basis of the data in this study, be attributed to nor
disassociated from the Assembly Line Testing Program, and consequently shall
not be dwelt upon here. Seven mode test results, which were only estimable
ftom General Motors data, showed a decrease in. both hydrocarbon and carbon
monoxide ; however, these estimates were based on a relatively small sample
of automobiles seven mode tested and these were concentrated in a small number
of assembly plants and engine categories. Consequently, these are not to be
interpreted as typical of what results might be expected across all engine
types, manufacturers, etc. The audit changes are seen to be veiy small. In
summary then, while general patterns and magnitudes of emissions changes were
estimable from the 1972 data base, their overall utility is quite limited.
The 1973 data base yielded more useable results as related to
assembly line testing. Sample sizes of automobiles seven mode tested and/or
idle tested were large and comprehensive enough to lend more confidence in
the estimates of emission level changes associated with the tests applied on
a nationwide basis. The most interesting result from this analysis is that
1-5
-------
Emissions
Type
Source of
Emissions Level
Changes
Manufacturer
HC
cn
NOX
Trend
GM
-1555
+35750
NA
F ord
-1272
+82576
NA
Chrysler
-66
-108
NA
Total
-2893
+118218
NA
Seven Mode
GM
-2310
-29040
NA
Ford
NA
NA
NA
Chrysler
NA
NA
NA
Total
-2310
-29040
NA
Audit
CM
-122
-39
NA
F ord
NA
NA
NA
Chrysler
NA
NA
NA
Total
-122,
-39
NA
Total
-5324
+ 89H9
NA
Table 1.2-4
Summarized Emissions Level Changes - 1972
(tons per year)
- decrease in tons per year (nationwide)
+ increase in tons per year (nationwide)
1-6
-------
Source of
Emissions Type
Emissions Level
Change
Manufacturer
HC
CO
NOX
GM
-0.0236
+0.5417
NA
Trend
Ford
-0.0292
+1.8957
NA
Chrysler
-0.0338
-0.0558
NA
RM
-0.0350
-0.4400
NA
7-mode
Ford
NA
NA
NA
Chrysler
NA
NA
NA
CM
-0.00185
-0.02253
NA
Audit
Ford
NA
NA
NA
Chrysler
NA
NA
NA
Table 1.2-5
Summarized Emissions Level Chanpes - 197''
(grams per mile)
1-7
-------
Source of
Emissions
Emissions Type
Manufacturer
HC
CO
NOX
Level Changes
r.M
NA
NA
NA
Trend
Ford
NA
NA
NA
Chrysler
NA
NA
NA
Total
NA
NA
NA
Seven Mode
GM
0
-60500
+ 825
Ford
+ 872
0
+ 3815
Chrysler
-64
0
0
Total
+ 808
-60500
+ 4640
Idle
GM
+14355
+103950
+2970
Ford
-4185
0
-2943
C hrysler
0
+ 851
+ 45
Total
+10170
+104801
+ 72
Audit
GM
-121
-2317
-62
Ford
NA
NA
NA
Chrysler
-SO
-589
-34
Total
-151
-2906
-96
Total
+10827
+41395
+ 4616
Table 1.2-6
Summarized Emissions Level Changes - 1973
(tons per year)
1-8
-------
Source of
Emissions
Level Chanpes
Emissions Type
¦ fell L UI dv LU Iwl
IIC
CO
NOX
Trend
GM
Ford
Chrysler
NA
NA
NA
NA
NA
NA
NA
NA
NA
7-mode
CM
Ford
Chrysler
0
+0.02001
-0.0032
-0.9167
0
0
+0.0125
+0.0876
0
Idle
m
Ford
Chrysler
+0.2175
-0.0961
0
+1.575D
0
+0.0429
+ 0. (54 5
-0,0676
+ oron23
Audit
GM
Ford
Chrysler
-0.00183
NA
-0.00153
-0.0351
NA
-0.0297
-0.0009
NA
-0.0017
Table 1.2-7
Summarized Emissions Level Changes - 1973
(prams per mile)
1-9
-------
there appears to be an overall increase in emissions resulting from the
combined effects of seven mode, idle and audit tests. The idle tests, in
particular, dominate the overall impact and if one omits the idle test effect
it would appear that the nationwide emission change would he:
I) Hydrocarbon 657 ton per year increase
II) Carbon Monoxide 63406 ton per year decrease
III) Oxides of Nitrogen 4544 ton per year increase
This is a grossly different pattern than with any idle testing. The develop-
ment of these results and an extensive discussion of the data and methodology
on which they are based is contained in Chapter 4 of this report.
Nationwide Bmissions Changes From Alternate Assembly Line Tests
Tables 1.2-8 a,b and c summarize the estimated nationwide emission
level changes which would result from combinations of representative levels
of seven mode and audit test rates employed as alternates to the current
structure of the California assembly line test program. Note thnt idle tests
are not considered here, this being a result of the apparent increase in
emissions levels attributable to idle testing. It is further re-iterated at
this point that the changes in nationwide emissions levels estimated here
result from the "rectifying aspects" of the alternate tests considered;
that is to say, from the adjustments performed on failing cars and their
resultant impact on the overall distribution of emissions. Any changes in
emissions levels which may be attributable to trends (as discussed previously)
are not included here. The two most striking patterns apparent from these
data are that: (1) only carbon monoxide shows a decrease in emissions and,
(II) increases in audit test rates reflect nominal changes in emission levels
by comparison to the seven mode test effect.
1-10
-------
Seven Mode
Test Rate (%)
Audit Test Rate
:
1
2
3
5
10
25
+ 716
+ 624
+ 533
+ 34 9
-109
50
+ 1524
+ 1432
4 1341
+ 1157
+ 698.45
75
+ 2332
+ 2240
+ 2149
+ 1965
+1506.45
100
+ 3139
+ 3047
+ 2956
+ 2772
+2313 .45
Table 1.2-8a
Hydrocarbon Emission Level Changes
For Various Seven Mode/Audit Test Rates
(tons per year)
Seven Mode
Test Rate (%)
Audit Test Rate (%)
1
2
3
5
10
25
- 62701
-64902
- 671C3
-71505
-82510
50
- 123201
-125402
- 127603
- L32005
- 143010
75
- 183701
-185902
- 188103
- 192505
-203510
100
-244201
-246402
-248603
- 253005
-2'>4010
Table 1.2-8b
Carbon Monoxide Emission Level Chanpes
For Various Seven Mode/Audit Test Rates
{tons per year}
1-11
-------
Seven Mode
Test Rate (%)
Audit
Test Rate
'%)
1
2
3
5
10
25%
+ 4567
+ 4495
+ 4422
+ 4276
+ 3913
50%
+ 9207
+ 9135
+ W62
+ 8916
+ 8553
75%
+13847
+13775
+13702
+13556
+ 13193
100%
+184 87
+18415
+18 34 ?
+18196
+ 17833
Table 1 .2-8c
Oxides of Nitrogen Emission Level Chanpes
For Various Seven Mode/Audit Test Rates
(tons per year)
Nationwide Emissions Chanpes From Proposed Catalyst Type Systems
Viewing the catalytic converter systems to be employed on 1976
automobiles as essentially 1973 automobile engines feeding into a catalytic
converter, the nationwide emissions estimated from alternate assembly line
test programs are essentially fractional multiples of the levels estimated
under conventional designs and discussed earlier. Therefore, the emissions
levels and changes in emissions levels from 1973 assembly line testing are
reduced by a fixed percentage in order to assess emissions levels and
changes in emissions levels from assembly line testing of catalyst equipped
vehicles. One should be cautioned to note, however, that this extrapolation
technique assumes fully functioning catalysts in all cars and does not
account for intermittent failures of catalyst units which may occur.
Alternately, this technique does not include any beneficial effects which the
presence of assembly line testing may contribute to the quality control of
catalyst systems. With hot start tests considered ineffective in estimating
1-12
-------
emission reductions attributable to catalytic systems, the only alternate
tests considered were audit tests and the emissions changes anticipated are
summarized below:
Audit Rate
in %
Emission Level Change (tons per year)
HC
CO
NOX
1 %
- 16.79
-319.16
- 9.67
2 %
- 33.58
-638.31
- 19.34
3 %
- 50.37
-957.47
- 29.02
5
- 83.95
-1595.78
- 48.37
10 %
-167.91
-3191.55
- 96.75
Table 1.2-9
Audit Rate Related Emissions Chanpes
with Catalytic Converter System
It is apparent from these estimates that the anticipated magnitudes
of emission reductions with catalytic systems are quite small; however it should
be kept in mind that with the catalytic converters the magnitudes of both total
emission levels and incremental changes in emissions will be a small fraction
of the corresponding emissions components from a 1973 system. The ratio of
"change in emissions" to "total emissions level" will be the same for both the
1973 and 1976 (i.e. - catalytic systems) considered in this report.
Costs of Alternate Assembly Line Test Programs
Cost analyses performed on alternate assembly line test programs
are discussed at length in Chapter 7. Summarized very briefly here,
alternate configurations of assembly line test programs considered for use
with current engine/control system combinations covered average test costs
over a range of $5.20 to $28.00 per automobile and when extrapolated
to nationwide production yielded anticipated nationwide costs in the range
from 51 million to 274 million dollars per year. These cost estimates stem
1-13
-------
from two basic types of source data: First, estimates of the unit costs of
the component tests included in alternate test programs considered (i.e. -
seven mode test, idle test, and audit test) and secondly the sampling rates
associated with each type of test in a particular test program configuration
(e.g. - 25% seven mode, 75% idle and 2% audit testing). Cost effectiveness
of these test programs, in terms of Nationwide Testing Costs per Ton Emission
Change is discussed in Chapter 7.
Shown below are the estimated ranges of unit costs associated
with the three types of tests employed (at varying sampling rates) in the
alternate test programs considered.
Test Type
Unit Cost Range
(Dollars per Test)
Seven-mode
$8.00 - $20.00
Idle
$.32 - $1,20
Audit
$68.00 - $350.00
1.3 Comments on Data Base
The analyses and extrapolations performed in the course of this
study were based on data supplied by three automobile manufacturers, this
data being derived from assembly line testing programs required hy the
California Air Resources Board for automobiles sold in the State of
California. The data base employed did not include test information about
exhaust emissions from automobiles not assembly line tested. As in most
studies of this sort limitations on time and resources precluded the develop-
ment and/or acquisition of as comprehensive a data base as would ideally be
desired.
1-14
-------
2
2.1
GENERAL METHODOLOGICAL APPROACH
Introduction and Overview of Emission Level Changes
In presenting the methodological approach to assessing the
magnitudes of emissions reductions which would result if assembly line
testing procedures similar to those currently employed in California were
employed on a nationwide basis, it will be helpful to construct a frame-
work within which the plausible factors which operate combine to determine
general emission levels. Consider, for example, a situation in which a
single manufacturer produces a single automobile on which there is only
one engine/emission control system combination available.
If we were to measure the HC emission levels from a large
number of these automobiles (say by C.V.S. sampling) at a fixed point in time,
t, during the production year and plot a histogram of these emission
levels,we would obtain a "picture" or "histogram" of emissions which
would tend to be distributed in a manner similar to the schematic of
Figure 2,1-1.
Frequency of
Observations
Emissions Level "g"
Figure 2.1-1
Schematic of Typical Emissions Distribution
2-1
-------
Typically, such plots of emissions measurements will tend to be
skewed to the right. Let us refer to these emissions as being distributed
2
with some population averagef JJ , and population variance, c3~ ,
2
(i.e. C. a D( JLLt & t ))• Of fundamental importance in assessing changes
in emissions levels of new automobiles, over time during a production
year, is the ability to describe how the average emissions level
varies, with time, during that production year. Conceptually, one would
hope to be able to specify how the average emission level varies as a
function of the "time of production year" as depicted in Figure 2,1-2.
T
Average JU t
Emission
Level
(HC)
Time in The Production Year
Figure 2.1-2
Temporal Patterns of Average Emissions Level
If one were able to obtain sufficient data to describe the manner
in which JJL varies with time, a logical next question is to consider
what factors contribute to, and combine, to determine the dependence of
JLL^ on time. This project will, for the purpose of providing some
structure within which to approach this task, segment those factors into
two major categories:
I.) Specific Actions Taken in the Production/Inspection
Process which are clearly identifiable and whose
impact on emissions lend themselves to measurement
and isolation of effects. Included here would be
2-2
-------
factors such as the presence or absence of seven
mode testing, different levels of audit testing, etc.
II.) Long-Term Trends and Less Identifiable Factors: -
included here would be that nebulous "other"
category of the elusive and ill-characterized factors
which, individually are virtually impossible to
isolate, but collectively account for changes in the
emission levels patterns and distributions which are
otherwise unexplainable.
Reference to Figure 2.1-3 wi 11, perhaps, facilitate a clearer
understanding of this "segmenting" of factors accounting for emissions
level changes. At a fixed point in time, say t, the emissions levels
prevailing among all automobiles produced are thought of as being
distributed according to some distributional pattern with an average
yWot and variance (Dot, i.e. - D( t, <5 ot). Let us
think of this distribution as the one which would prevail if we were
able to measure the emissions levels of each automobile as it came off
the production line.
Now, over a "short period of time" some changes may occur to
these automobiles as a result of certain identifiable actions. Specif-
ically, a fraction of them (O^l) may be hot seven mode tested, a fraction
of them ( CX 2) may be idle tested, another fraction of them ( C. 3)
may receive both a hot seven mode and idle test, and finally a fraction
C OU) may receive no tests at all. Conceptually, we will describe
any changes which may result in the emissions distribution as a result of
the particular action (i.e. - testing type) taken by means of indicating
a modified set of parameters a-posteriori to the testing. For example,
at time t, the emissions were distributed D( O ot) whereas that
fraction, 0(l, of automobiles which were 7- mode tested only, at time
t will, after testing, be distributed D( Ml,t, c5\t) - possibly
different from prior to this testing type. In a manner similar to this the
impact of the other actions (i.e. - modes of inspection testing) are
2-3
-------
reflected in their changed distributions. After the respective tests are
performed on the portions of the incoming population of automobiles and
these tested subpopulations are pooled together again, we can conceive of
the resulting emissions distribution as being D(
-------
2-5
-------
2.2 Trends in Average Emissions Levels During a Production Year
Suppose that at the time t within a production year one can specify:
n(t) 5 the number of automobiles produced at time t of that
production year.
JXit) = average emissions level of the new automobile produced
at time t (in grams per mile).
pr = conversion factor which will convert grams per mile to
tons per year (assuming a fixed number of miles per
year driven by an "average" car).
Then
T
£ = r J n(-k)jxCi) dt (2.2.1)
o
represents the total average emissions (in tons per year) which will be generated
as a result of the production pattern and time pattern of average emissions.
This assumes that automobiles produced at time t will continue to generate
emissions at an average level yU.(t) throughout their driving life cycle.
Alternately, consider that instead of automobiles produced at time t
emitting at the average level ^-(t), one were to assume that the average emission
level were constant, say Then,
.T
<£"" « p jVWOdt
^o
^fQ n(od~fc (2.2.2)
would represent the total average emissions (in tons per year) if all new auto-
mobiles produced had the same average emissions level, i.e.,^Q^.
2-6
-------
Further, E^ - E, would represent the change in average emissions
which would result during the production year if average emissions follow the
pattern « t ^ T J rather than stay at the average level JjS^ for
all production during the year.
This admittedly simplistic approach to measuring change in average
emission levels stimulates a number of questions:
(1) Is sufficient data available to support an adequate charac-
terization of the functions XUt), fL(t), etc.? If not,
what modifications in this approach would be necessary to
enable a similar measurement to be developed which is in
harmony with available data?
(2) Even if the computations, or a reasonable approximation
thereof, can be attained, can one attribute the estimated
emission level changes to any causal factors?
Regarding (1) above, the data available for analysis in the report
were summary results covering various production quarters during 1972 and 1973
model years. This aggregated (in time) data precludes an analysis of emissions
level change in continuous time, however, modifications of the conceptual
approach to estimating changes in levels of emissions so as to e ploy aggre-
gated data will be presented in Chapter 4 of this report. The second point
(2) above is perhaps a more crucial one. The absence of any data on emissions
levels which were gathered on automobiles produced prior to the introduction
of assembly line testing procedures in California would appear to raise serious
questions about whether or not any trends (i.e., changes) in emissions levels
occurring over time during the production year can validly be attributed to
the presence of assembly line testing as the causal factor. Stated alternately,
the data base available for analysis only reflects the "after assembly line
testing" situation and there was not available any "before assembly line test-
ing" data against which to compare. Because of this, there has been no attempt
2-7
-------
in this report to either assess the statistical significance of changes in
emissions levels over time during a production year or to identify the
presence of assembly line testing as the cause of any such changes as may
appear significant. Rather, such changes in emissions levels as may be
estimated by the techniques briefly described earlier, are presented in the
hopes that such preliminary information on the magnitudes and patterns of
such changes during a production year may provide useful insight into a more
fundamental understanding of the nature of this type of testing.
2-8
-------
2.3 Short Inspection Tests and Their Impact Assessment
This section will focus on a discussion of the methodology employed
to assess the impact of those combinations of hot seven mode tests and die
ests which are applied to end of production line automobiles. In Figure 2.1-3
we can view this as that segment of the "flow" of automobiles through the
portion of the Assembly Line Test Procedure which terminates just prior
to the C.V.S. Testing Phase. The primary interest of this section, then,
is an analysis of the manner in which 7- mode testing, for example, at some
level will affect the average level of the emissions distributions.
Consider some end of line inspection test, T, say, the effect
of which one seeks to investigate. Specifically, we desire to estimate any
difference in average emissions levels which may exist due to the presence
of testing some fraction (j{ of automobiles produced by this test pro-
cedure .
Let be defined as the average (i.e. - 1st moment) of
the distribution of emissions of type j for the population of automobiles
which have been tested by procedure prior to audit testing. Con-
versely, let stffj be the mean of the distribution of emissions of
type j for the population of automobiles which have not been tested by
procedure ~T~ prior to audit testing. If it is true that the fraction
C( of the total population of automobiles have received test T , and
the fraction (l-O^ ) have not received test ~T , then the expected value
of the emissions resulting from audit testing from the total (and mixed)
population of automobiles is:
AAj. = AAT£-&-t 0
2-9
-------
If, in fact, the average level of emissions (as measured by
audit testing) in the presence of test procedure "T* (i.e. - is
significantly different from the average level in the absence of procedure
-J- (i.e. - Mt j), then the fraction, , of the population receiving
test T represents the rate at which the overall process mean will
deviate from j in "units" of ( j — vC/rj) grams per
mile. Consequently the term
(2.3.2)
represents the deviation in grams per mile of the mean level of emissions
when applying test T to 1000^ % of the automobiles produced from the
mean level when none of the automobiles receive procedure "[ ~
In order to express this averape emissions level in terms of
tons per year, one can convert as follows:
(2.3.3)
where (J. - overall mean emission level, as
a defined previously, of emission type j
j. 6
r = conversion factor (1.1 x 10~ tons/gram)
= average yearly mileage per automobile
= number of automobiles produced per year
2-10
-------
2.4
Audit Tests and Their Impact Assessment
This section focuses on an analysis of the impact on emissions
reduction occurring as a result of the screening or rectifying aspect of
the audit test procedures. Aside from the long term effect which may
occur simply due to the fact that some kind of emissions level monitoring
exists, an interesting question, in and of itself, is, to what extent
are overall emission levels changed if a 2% audit test is applied to all
cars produced in the nation. This section focuses on an analysis of just
that question.
Consider the following situation. Emissions measurements, via
C.V.S techniques as employed currently in California, are taken for a
single emittent (say, hydrocarbon) and the density function of the random
variable, C, , is determined. We shall think of this distribution of 0,
as characterizing the distribution of original measurements of C.V.S. tests
on this population of automobiles of a single engine category. Assume the
distribution to be of the form as shown in Figure 2.4-1.
Figure 2.4-1
Skewed Emission Patterns
The shape of this density is similar to that so commonly
observed in emissions measurements, i.e. - non-negative, skewed to the
right and approximately log-normally distributed. Shown in Figure 2.4-1
is a value of the standard,"^", which is to be interpreted as the
current standards in the California procedures. We shall denote:
2-11
-------
p- S
= ?{&} C2-4-1)
that is, p equals the probability that one observation on this population
(i.e. - emissions measurement on an automobile) exceeds the standard £ *
In a similar manner, we denote
'~P
o
= P{e* 2} (2.4.2)
that is, the probability that one observation on this population does not
exceed the standard.
Of basic concern is, what happens to average emissions levels
as a result of the audit sampling and any modifications which occur in the
emissions distribution because of it. We shall focus on three subsets of
the total population:
1. The set of automobiles not audit tested
II. The set of automobiles audit tested and passing the test
TIT. The set of automobiles audit tested and failing the test.
The intent is to determine the distributional form of emissions,
after the audit sampling, so as to relate this in some way to the audit
sampling rate, ^ , say. Let us consider how the distributional character-
istics of the three subsets described above will contribute to this -
"post audit sampling" emissions distribution.
2-12
-------
Case I - Automobiles Not Audit Tested
Given this population of automobiles not audit tested, one
one would anticipate that if they were C.V.S. tested their distribution
of emissions would be reasonably approximated by fQ(e)» the density in
Figure 2.4-1 which described the distribution of original measurements of
emissions. We thus assume:
f(e|not audit tested) = fQ(e)
It is further noted that (1-{U) is the fraction of the total population of
automobiles to which this density applies.
Case TI - Automobiles Audit Tested and Passing the Test
This subset of automobiles is restricted to those which,
upon C.V.S. testing, yielded emission values no greater than^. In
Figure 2.4-1 this is the set of emission values, e, less than or equal to
^ . Now, if sufficient retesting were performed on automobiles in this
subset one could characterize the distribution of emissions of such cars,
after the impact (if any) of initially being C.V.S tested and having
passed that test has been accounted for. Let us denote as t'ie
density function of the distribution of emissions from such automobiles.
It is important to note that this distribution is not, necessarily,
identical with that of cars not audit tested at all since this category
does not include failing vehicles. Further, this distribution is charac-
teristic of a fraction ^9 (1-p) of the total population of original
automobiles.
Case III - Automobiles Audit Tested and Failing the Test
This subset of automobiles is a fraction of the upper tail
of the density function of Figure2.4-1. Let us denote the density function
2-13
-------
characterizing this distribution of emissions as • As in the previous
case it is important to note here that f^Ce) is not, in general, identical
with f(e) for after automobiles are audit tested, have initially failed,
and are subsequently adjusted, repaired, etc. to a point where they can
pass a retest, the distribution of their emissions may be substantially
modified. The sum total of this modification is encompassed in f^(e).
Anainf as earlier, observe that this density will be characteristic of
a fraction the total population of original automobiles.
We reiterate, at this point, the central question of this
section. How does the average emission level change as a result of the
audit sampling procedure employed? Consider first the average emissions
which would prevail if no audit testing were employed. This is
e'°' 5 e. / e-F0ce)de
¦^o
Alternately, what average emissions level will prevail if audit
testing is executed at the rate Tt is readily seen that
= (i-p) £e-fo(e.)cle. -i- (3('-p) J~c4](q)cle ¦+ (3p
= (l-^Co ¦+• <2>i -*¦ pp (2.4.4)
CC <-*3
That is, the resulting average emissions level is a weighted
average of the average emissions levels of the three subsets of auto-
mobiles discussed earlier, the scaling weights being those fractions of the
total population falling into the respective subsets.
2-14
-------
The change in average emissions levels attributable to the
audit sampling rate, (3 , is the difference in these two averages, namely:
a — C° ¦> — C-fO
A 3 e - e>
A = eto>-|o-^e0-*^-p)e,-+-epez^
A - (3Cc5 - 0O-p)6»i - (3peTz C2.4.S)
We thus "have the change in average emissions (in grams per
mile) resulting from this sampling and rectifying procedure. The
discussion thus far has assumed a single automobile or engine category
and also a single emittant type. If these additional factors, as
well as different manufacturers, are incorporated a straightforward
modification of expression (2.4.5) above applies. Let us denote:
Jl - manufacturer's index; JL= 1,2,3 for
General Motors, Ford and Chrysler^
respectively
j = emission type index; j = 1,2,3 for
hydrocarbon, carbon monoxide and
oxides of nitrogen, respectively
- overall failure rate for manufacturer Jl,
in the audit tests
= average emissions level for emission
J" type j, manufacturer JL on original
measurements of automobiles not audit tested
— (»)
<3j; a average emissions level for emission type
a j, manufacturer Xon measurements of
automobiles which were audit tested and
passed the test originally
— (2)
Q^l » average final emissions levels for emission
J type j, manufacturer i. on measurements
of automobiles which were audit tested, failed
the test and were adjusted so as to pass a
subsequent C.V.S. test
2-15
-------
It is now possible to denote the change in emissions attributable
to a sampling rate, (3 , for a specific emission type and manufacturer.
This is a modification of expression^ .4.5)employing the above notation:
A /X —(°") \—0) —C2) ^
= -ftPzCVj. 1=1,2,3 (2.4.6)
These emissions changes are in grams per mile. To relate this
change to tons per year, it is necessary to account for the average
mileage (per year) an automobile is driven, and the number of automobiles
being manufactured. Defining
= average miles per year driven by a
typical automobile from manufacturer
= number of automobiles produced, per
year, by manufacturer
I"7 = conversion factor to relate grams per ^
mile to tons per mile (i.e. - f = 1.1 x 10~
tons/grams)
One obtains
V P-WNI*
the average change in emissions (in tons per year) for emissions type
j associated with manufacturer JL as a result of the audit testing program
as currently conducted in California.
2-16
-------
As outlined above, this methodology is employed for the
assessment of emissions levels changes which would be attained if audit
testing, as currently employed in California, were extended to a
nationwide basis. Specific results emanating from this type of analysis
applied to the currently available data base are contained in Chapter 4
of this report. Extension of this methodology in order to consider
alternate modes of audit sampling is contained in Chapter 5.
2-17
-------
3.
FAILURE RATES AND REASONS FOR FAILURE OBSERVED FROM
ASSEMBLY LINE TEST PROGRAM DATA
3.0 Introduction
An overview of the failure rates and descriptions of reasons for
failure is presented in this chapter. It will be helpful to have this
"baseline" understanding of current inspection and audit testing failure
rates being experienced by the various manufacturers whose data is em-
ployed in this study, prior to moving on to an analysis of emissions
reductions which may be attainable on a nationwide basis from these and
related testing programs.
3.1 Sfeven Mode Inspection Tests
3.1.1 Overview of Failure Rates
For the 1972 model year, the failure rates for the seven mode
inspection tests which were run are summarized in Table 3.1-1. This data
represents the overall results by manufacturer.
Manufacturer
Number of
Tests
Overall Failure Rates
(Seven Mode Inspection Tests)
Chrysler
Ford
General Motors
31669
81953
96519
25.5%
24.8%
5.29%
Table 3.1-1
1972 Model Year Seven Mode Inspection Test
Failure Rates (by Manufacturer)
3-1
-------
Within a manufacturer there are a variety of ways in which these
overall failure rates can be segmented for analyses. Further, variations
in the reporting format from one manufacturer to another can result in
differing interpretations of "failure rates". Because of these differences,
further discussion of the seven mode inspection test results will be
presented for each manufacturer.
3.1.2 Failure Rates: General Motors Corporation
a.) Data Base
Summary test data was available on seven mode tests performed for
the 2nd, 3rd, 4th and 4th plus buildout quarters of the 1972 model year only.
Reporting requirements for the 1973 model year tests are such that the 1973
seven mode test results are not available. For the 1972 data available the
total number of units initially tested, the number of units failing initial
tests and the number of units passing initial tests were reported (classified,
in each case, by Manufacturing Division/Engine Displacement/Carburetor barrels
categories).
An analysis of the failure rates on inspection tests was complicated
somewhat by variations in the reporting format of data from the different
production quarters for the 1972 model year. Specifically, for the second
and fourth plus buildout quarters, the results for units failing initial tests
tables contained (by category) data on the number tested and the number of
gross fails. The similar tables from the third and fourth quarters included
(by category) data on the number tested, number of gross fails and number
of good reruns.
Ambiguity in the definition of "gross fails" leads to some con-
fusion on how to handle a summary of failure rates. To illustrate, consider
third production quarter data for the Pontiac 400 C.I.D. (4 barrel) engines
as reported by GM:
Number of Initial Tests 884
Number of Failing initial Tests 33
Gross Fails among initial Tests 13
Good Reruns 10
3-2
-------
The gross fails may have resulted from a mixture of two sources.
On the one hand they could all be attributable to test equipment (and/or
procedure) malfunctions^ and as such, the corresponding test results would
represent invalid data points. Alternately, they could all be attributable
to unaccountably extreme variations in an engine's results even though there
was no test equipment (and/or procedure) malfunction. Here such data,
while undesirable, represents valid data and test results. If the former
case prevails we might adjust our interpretation of the source data as
follows:
Number of valid initial tests = Number of Initial Tests
Number of Gross Fails
= 884 - 13
= 871
Number of valid failing initial tests - Number of failing initial
tests - Number of
gross fails
= 33-13
= 20
and
% Failure Rate - o£ valid initial Jests31 ~ x 100
55 mx 100
¦ 2.3%
In the second situation cited above, we might consider the failure rate as:
Failure Rate (%) = (Number of Failing Initial Tests) *
(Number of Initial Tests) x 100
• m
= 3.74%
3-3
-------
These two alternatives represent the bounding cases, and if the
indicated gross fails consisted of a mixture of test equipment/procedure
related gross fails and engine related gross fails the failure rate for that
category would always be between these two extreme values. Consequently it
is illuminating to consider the discrepency in failure rates (for a given
category) which may occur,
b.) ^even Mode Test Failure Pates
Assuming that all gross failures are attributable to engine
variability and not test malfunctions, the failure rates for the 1972 model
year tests are summarized in Tables 3.1-2 thru 3.1-4.
Division
Production Quarter
(1972 Model Year)
Total
Year
2
3
4
5
Chevrolet
6.9
2.4
3.4
15.4
4.4
Pontiac
4.7
3.4
3.8
21.6
4.1
Oldsmobile
11.5
10.4
7.0
12.2
10.1
Buick
1.4
6.6
5.7
41.6
4.9
Cadillac
4.0
1.6
1.8
-
2.5
Overal1
6.19
4.95
3.83
15.82
5.29
Table 3.1-2
7 Mode Test Failure Rates (in %)
General Motors: 1972 Model Year
(By Division and Production Quarter)
3-4
-------
From Table 3.1-2 one observes a number of patterns:
I Some noticeable differences in the failure rates from one
division to another (e.g., Oldsmobile and Cadillac show strikingly
different failure rates).
II Within an operating division, there is the general trend of a
decreasing inspection test failure rate as the production year
proceeds, except for the buildout quarter in which there is a
consistent increase in the failure rate. This appears to indicate
that there is a gradual improvement in the ability of the pro-
duction process to satisfy the testing procedures as "experience"
in production of that model grows, with the exception the
buildout quarter.
Aggregating the data according to engine size (i.e., C.I.D.)
rather than operating division similar tone patterns appear, as seen from
Table 3.1-3.
3-5
-------
Engine
Production Quarter (1972 Model
Year)
Total
Size
2
3
4
5
Year
140
15.7
9.7
9.2
15.9
12.1
250
5.2
2.7
3.4
6.1
3.5
350
5.4
4.0
3.3
14.0
4.3
400
6.1
2.5
2.6
35.9
4.1
455
5.2
4.2
2.9
11.8
4.5
472
3.8
1.5
1.5
0.0
2.2
500
4.5
2.4
4.0
-
3.6
Overal1
6.19
4.95
3.83
15.82
5.29
i
• ->
Table 3.1-3
7 Mode Test Failure Rates (in %)
General Motors: 1972 Model Year
(By Engine Size and Production Quarter)
Tn Table 3.1-4, the seven mode test failure rates are nresentcd
when segmented into categories indexed by Di vision/Pi snl acement/Carbiiretnr
co'-Mnations for the production quarters available from the 1972
I'todel year.
3-6
-------
Engine
Category
Production Quarter
Total
2
3
4
5
Chev. 140-1
20.6
19.2
13.2
19.8
17.8
140-2
14.8
8.7
8.6
15.0
11.2
250-1
5.2
2.7
3.4
6.1
3.5
350-2
3.1
1.2
1.4
6.1
1.7
350-4
4.2
1.9
2.1
8.5
2.7
400-2
7.5
2.1
2.6
38.2
4.4
Pont. 350-2
5.4
3.6
5.5
10.7
4.7
400-2
5.0
2.6
2.8
23.5
3.5
400-4
2.9
3.7
2.5
20.0
3.2
455-2
4.9
6.5
4.9
40,0
5.8
455-4
5.4
2.6
3.2
43.5
4.2
Olds. 350-2
2.3
4.9
5.0
18.2
4.5
350-4
20.7
16.2
11.2
29.3
16.6
455-4
7.7
6.1
2.3
7.4
6.2
Buick 350-2
3.5
10.9
8.7
3.3
8.0
350-4
5.6
9.0
6.7
48.7
7.7
455-4
2.9
2.7
3.0
37.9
3.0
Cad. 472-4
3.8
1.5
1.5
0.0
2.2
500-4
4.5
2.4
4.0
-
3.6
Table 3.1-4
7 Mode Test Failure Rates (in %)
General Motors: 1972 Model Year
(By Engine/Division Category and Production Quarter)
3-7
-------
c. Seven Mode Test Failure Pates: Adjusted For Test "'al
-------
Hnpine
Catepory
Production Quarter C1 f72
*fodel Year)
Total
Year
2
3
4
5
Chevrolet 140-1
20.6
19.2
13.2
19. 8
17.8
140-2
14.8
8.7
8.6
15.0
11 .2
250-1
5.1
2.7
3.3
5.3
3.4
3S0-2
3.1
1.2
1 .4
5.3
1 .7
350- 4
3.6
1 .8
2.0
8.0
2.5
400-2
7.5
2.1
2.6
38.2
4.4
Pontiac 350-2
5.2
3.5
5.5
9.1
4.6
400-2
4.6
2.4
2.8
2 3.5
3.4
400-4
1.5
2.3
1 .0
20.0
1 .8
455-2
4.9
6.5
4.9
40.0
5.8
455-4
4 .8
2.3
3.2
43.5
3.9
Oldsmobile 350-2
0. 30
4.2
5.0
15.6
3.5
350-4
20.0
15.8
11 .2
28.4
16.2
455-4
7.5
6.0
2.3
7.2
6.0
Buick 350-2
| 3.2
10.8
8.7
3.3
7.9
350-4
5.6
8.9
6.7
48.7
7.6
455-4
2
1.5
3.0
37.^
2.6
Cadillac 472-4
3.5
1.2
1.1
0.0
1 .9
500-4
4.0
2.2
4.0
0.0
3.4
Table 3.1-7
7 Mode Test Failure Rates (in %)
Adiusted For Test Malfunctions
General Motors: 1972 Model Year
(Ry en pine/division category and nroduction quarter)
3-9
-------
3.1.3 Failure Rates: Ford Motor Company
a.) Data Base
Summary test data for 7_mode tests was available for the 2nd,
3rd, 4th and 4th plus buildout quarters of the 1972 model year. The Ford
data is broken down by engine size, across all divisions (Ford, Mercury and
Trucks), and does not include vehicles that failed because of test equipment
malfunctions.
b.) Seven MoHe Test Failure Hates
In Table 3.1-8 the seven mode test failure rates are nresenterl.
Here they are classified by en pine category and production quarter.
Engine
Production Quarter
Total
2
3
4
! 5
98-1V
40.6
39.2
34.8
50.0
38.8
122-IV
34.9
36.2
38.3
62.6
39.8
159-2V
-
34.9
61 .4
-
58.8
170-1V
0.0
26.7
24.5
0.0
22.9
200-1V
38.4
19.1
31.9
51.1
29.3
240-IV
18.0
12.1
14.3
5 .6
12.8
250-IV
38.0
36.5
37.5
55.8
37.5
302-2V
31.6
26.7
31.0
33.6
29.4
3S1-2V
15.7
6.2
18.7
28.9
12.8
351-4V
5.9
16.0
18.4
25.7
18.4
360-2V
31.6
32.3
40.5
49.4
37.6
390-2V
56.3
33.3
41.8
20.0
37.8
400-2V
9.5
3.6
8.7
14.1
7.7
429-2V
13.1
8.8
8.2
9.6
10.4
460-2V
20.8
14.0
14.0
19.8
16.3
Overall
21.0
19.5
30.2
41.1
24.8
Table 3.1-8
7 Mode Test Failure Rates (in %)
Ford: 1972 Model Year
(By Engine Size)
3-10
-------
3.1.4 Failure Pates: Chrysler Corporation
a.) Data Base
^umtnary test data for seven mode tests from the Chrysler fomora-
tion was available for the 1st through 5th (i.e. - buildout) quarter of the
1972 model year. This data was categorized by cnpine size and production
quarter, and reflected the number of initial tests and the number o^ initial
failures .
b.) Seven *'ode Test Failure Pates
In Table 3.1-9, the seven mode test failure rates are presented
for the 1972 model year Chrysler data. These failure rates are classified
by enpine category and production quarter.
i Engine
Production Quarter (If72
Hi del
Year
Total
1 Size
1
2
3
4
5
198
0.0
16.3
39.6
49.0
33.3
34.3
225
88.9
53.3
74.5
77.1
85.4
75.8
318
38.1
U.9
4.7
13.4
22.7
cn
00
340
6.0
3.2
3.5
0.6
2.5
3.2
360
44.3
32.8
23.6
9.4
10.7
20.7
400
14.1
12.3
4.8
4.1
6.8
8.7
440
10.9
8.3
3.5
3.8
8.9
7.1
Overall
34.0
16.3
27.4
24.1
36.6
25.5
Table 3.1-9
7 Mode Test Failure Rates (in %)
Chrysler Corporation: 1972 Model Year
(By Engine Size)
3-11
-------
2 Audit Tests
3.2.1 Overall Audit Test Failure Rates
Tables 3.2-1 fa) and (b) present the overall audit test failure
rates for the three manufacturers included in this study.
Manufacturers
Number of
Original
Tests
Overall Failure
Rate
Audit Test
General Motors
Ford
Chrysler *
9296
6S14
N. A.
7.0 %
15.7 %
N.A.
Table 3.2-1(a)
Overall Audit Test Failure Rates
1972 Model Year
* Audit test failure rates not confuted for 1972 model year Chrysler data.
Manufacturers
Number of
Original
Tests
Overall Failure
Rate
Audit Test
General Motors
Ford
Chrysler
4090
2901
1462
15.7 %
11.5 %
17.37 %
Table 3.2-1(b)
Overall Audit Test Failure Rates
1973 Model Year
3-12
-------
It should be noted that Table 3.2-1(a) is based on test data
covering production quarters 2 through buildout, within the 1972 model
year, and Table 3.2-l(b) is based on test data covering production
quarters 1 and 2 of the 1973 model year.
3.2.2 Audit Test Failure Rates: General Motors
Tables 3.2-2 through 3.2-5 summarize the audit failure rates as
determined from the 1972 and 1973 model year data available from General
Motors. In all cases here, the failure rate is the ratio of failures on
original tests divided by number of original tests. The columns in
Tables 3.2-3 and 3.2-5 refer to distinct failure type categories, that is,
entries in the column labelled HC refers to the failure rate for tests
which failed the HC standard only, the column HC and CO refers to the failure
fate for tests which failed both the HC and CO standards only, etc.
3- 13
-------
Division
F.n gine
Number
Number
Failure
Category
Tests
Failures
Rate %
Chevrolet
140-1,2
1041
96
9.2
250-1
289
19
6 .4
350-2
1661
116
7.0
350-4
827
53
6.4
400-2
862
44
5.1
Pontiac
350-2
419
33
7.8
400-2
133
5
3.8
400-4
154
31
20.1
455-2
52
4
7.7
455-4
152
7
4 .6
Oldsmobile
350-2
174
6
3.4
350-4
413
21
5.1
455-4
359
14
3.9
Buick
350-2
273
15
5.5
350-4
336
15
4.5
455-4
422
30
7 1
Cadi 11ac
472-4
476
52
10,9
500-4
116
7
6.0
Table 3.2-2
Audit Failure Rates (in %)
1972 General Motors
(by Division/Engine Category)
3-14
-------
Division
Failure Type or Combination
Total
HC
CO
HC and
CO
N0X
HC and
NOX
CO and
NOX
HC,
CO and
NOX
Chevrolet
2.1
3.6
0.3
0.9
0.1
0
0
7.0
Pontiac
3.1
3.3
1.5
0.4
0
0
0
8.3
Olds mobile
1.7
1 .0
0.0
1.7
0
0
0
4.4
Buick
1.3
1.3
0.4
2.7
0. 1
0.1
0
5.9
Cadi 1lac
0.2
8.1
0.0
1.7
0
0
0
10.0
Trucks
6.4
1.1
0.7
0.1
0
0
0
8.3
Table 3.2-3
Audit Failure Rates (in %)
1972 General Motors
(by Failure Type and Division)
3-IS
-------
Division
Engine
Category
Number
Tests
Number
Fails
Failure
Rate %
Chevrolet
140-1,2
338
53
16.0
250-1
69
7
10.0
350-2
723
65
10.0
350-4
268
27
10.0
400-2
206
37
18.0
Pontiac
250-1
8
1
13.0
350-2
186
43
23.0
400-2
96
21
22 .0
400-4
97
25
26.0
455-4
74
6
8.0
Oldsmobile
350-2
44
27
61 .0
350-4
161
4
2.0
455-4
196
9
5.0
Bui ck
350-2
185
50
27.0
350-4
124
15
12.0
455-4
266
90
34.0
Cadi 11 ac
472-4
254
16
5.0
500-4
64
4
6.0
Table 3,2-4
Audit Failure Rates (in %)
1973 General Motors
(by Division/Engine Category)
3-16
-------
Division
Failure Type or Combination
- Tiff-al
HC
CO
HC and
CO
NOX
HC and
NOX
CO and
NOX
HC, CO
and NOX
Chevrolet
2
8
1
3
0
1
0
13
Pontiac
2
2
0
18
0
0
0
22
Oldsmobile
2
4
2
2
0
0
0
10
Buick
10
9
5
1
1
1
0
26
Cadillac
1
3
0
1
0
0
0
6
Trucks
3
13
3
5
0
0
0
24
Table 3.2-5
Audit Failure Rates (in%)
1973 General Motors
(by Failure Type arid Division)
3.2.3 Audit Test Failure Rates: Ford Motor Company
Tables 3.2-6 through 3.2-9 summarize the audit failure rates as
determined from the 1972 and 1973 model year data available from Ford.
As in the previous section, the failure rate referred to here is the ratio
of original failures to number of original tests.
3-17
-------
Division
F.ngine
Number
Number
%
Category
Tests
Fails
Fail
Ford
98-1
210
9
4.0
122-2
1611
157
10.0
200-1
166
4
2.0
302-2
579
36
6
351-2
783
157
20
400-2
743
97
13
429-4
532
42
8
170-1
34
17
50
250-1
85
1
1
351-4
38
2
5
Me rcury
122-2
154
5
3
200-1
19
0
0
302-2
133
4
3
351-2
284
54
19
351-4
11
2
18
400-2
197
35
18
429-4
1
0
1
460-4
271
35
13
250-1
20
1
5
170-1
2
0
0
159-2
87
9
10
Trucks
240-1
77
9
12
302-2
184
68
37
351-2
44
5
11
360-2
197
71
36
351-4
18
2
11
390-2
34
1
3
Table 3.2-6
Audit Failure Rates (in %)
1972 Ford
(by Division/Engine Category)
3-18
-------
Division
Engine
Category
Number
Tests
Number
Fails
%
Fails
Ford
98-1
103
3
3
122-2
666
95
14
250-1
47
5
11
302-2
380
32
8
351-2
568
47
8
400-2
426
80
19
429-2
179
18
10
Mercury
302-2
108
11
10
351-2
9
1
11
460-4
271
15
5
Trucks
240-1
10
0
0
302-2
114
16
14
360-2
14
10
71
400-2
6
0
0
Table 3.2-7
Audit Failure Rates (in %)
1973 Ford
(by Division/Hngine Category)
3-19
-------
Division
Failure Type or Combination
Ford
HC
CO
HC, CO
NO
HC, NO
CO, NO
lie, CO, NO
TOTAL
1
7
-7
0
0
0
0
11
Mercury
2
7
1
2
0
0
0
12
Trucks
11
5
12
0
0
0
0
28
Table 3.2-8
Audit Failure Rates (in %)
1972 Ford
(by Failure Type and Division)
Division
Failure Type or Combination
Ford
V\C.
CO
W,, W
CI
w , co, w
TOTM,
3
3
1
4
0
0
0
12
Mercury
2
4
1
1
0
0
0
7
Trucks
10
2
3
2
0
0
0
18
Table 3.2-9
Audit Failure Kates (in %)
1973 Ford
(by Failure Type and Division)
3-20
-------
3.2.4 Audit Test Failure Rates: Chrysler Corporation
Table 3.2-10 summarizes the audit failure rate as determined from
the 1973 Chrysler data base .
Fjigine
Number
Number
%
Category
Tests
Fai Is
Fail
225
401
84
20.94
318
459
66
14.37
340
19
14
73.68
360
156
15
9.61
400
295
55
18.64
440
103
12
11.65
Spl. 8
29
6
20.68
Overall
1462
252
17.37
Table 3.2-10
Audit Failure Rates (in %)
1973 Chrysler
(by Engine Category)
3-21
-------
3.3
Audit Test Failure Reasons
The audit tests performed on the 1972 and 1973 model year pro-
duction for General Motors provided some diagnostic descriptions of reasons
for audit test failures. From the 1972 data base there were a total o* 716
audit test runs which failed the test; in 1973 there were 974 such failures.
These numbers include more than the number of original failures, for a
single car which failed originally and on one or more subsequent retests will
have generated more than one "failure" data point.
Tables 3.3-1 and 3.3-2 summarize the patterns of "reasons for
audit failures" as observed from the 1972 and 1973 General Motors data,
respectively. Each table presents the failure reasons, ranked, from most
frequent to least frequently occurring in their respective data sets.
3-22
-------
Descriptor of
Number of
% of
Cumulative %
railure Reason
Fai lures
Total
of Total
Rerun no Repair
292
40.78
40.78
Carb. Change
97
13.54
54.32
Electrical Conn Shy
37
5.16
59.48
Idle Low 75 RPM Slow
29
4.05
63.53
Rich Choke Action
26
3.63
67.16
Idle Stop Solenoid Bad
18
2.51
69.67
Miscellaneous Repair
18
2.51
72.18
Vacuum Conn Shy/Leaking
17
2.37
74.55
Improper Precondition
13
1 .81
76.36
Loose/Shy/Damaged Ipn.
13
1.81
78.17
Damaged/Stick Choke St.
12
1.67
79.84
Timing Was Advanced
11
1.53
81.37
Vacuum Hose Leak
10
1.39
82.76
Vacuum Hose Leak/Diset
10
1.39
84.15
Wrong/Broken Spark Pig.
10
1.39
85.54
Equipment Malfunction
9
1.25
86.79
Idle High 75 RPM fast
9
1.25
88.04
Brakes Dragging/Defe
8
1.11
89,15
Loose Mounting
8
1 .11
90.26
Wrong Plug Gap
8
1.11
91 .37
Engine Change
5
.69
92.06
Test Bench Malfunction
4
.55
92.61
Choke Rod Disconnected
3
.41
93.02
Incorrect Dwell
3
.41
93.43
^gs. Switch
3
.41
93.K4
Valves
3
.41
94.; 5
Diverter Value Defect
3
.41
94.(6
Driver Error
3
.41
95.07
Timing was Retarded
3
.41
95.48
Valve Lifter
3
.41
95.89
Accelerator Linkage
2
.27
96.16
Distributor Defect
2
.27
96.43
Idle Stop Solenoid Deft
2
.27
96.70
Incorrect/No Upshift
2
.27
96.97
Stove Pipe Disct/Shy
2
.27
97.24
Ran out of Gas
2
.27
97.51
Air Hose Disct/Shy
1
.13
97.64
Fluid Leak
1
.13
97.77
Probe Fell out
1
.13
97.90
C Table 3.3-1 continued on next page)
3-23
-------
(Cont'd)
Descriptor of
Number of
% of
Cumulative %
Failure Reason
Failures
Total
of Total
Air Cleaner Sw Sensor
1
.13
98.03
Air Pump Defect
1
.13
98,16
Data Instrument Failure
1
.13
98.29
Driver off Schedule
1
.13
98.42
Foreign Obj. in Carb.
1
.13
9ft. 55
Fu e 1 Pump De fe ct
1
.13
98.68
Tnop. Pre-Conditioninp
1
.13
98.81
Ignition Wires Crossed
1
.13
98.94
Throttle Blade Binds
1
.13
99.07
Computer Malfunction
1
.13
99,20
Overheater
1
.13
99.33
TCS Fuse
1
.13
99.46
TCS Temp, Override Swt
1
.13
99.59
Table 3.3-1
Audit Test Failure Reasons
(ranked by frequency of occurrence)
1972 Ceneral Motors
3-24
-------
Descriptor of
Number of
% of
Cumulative %
Failure Reason
Failures
Total
of Total
Rerun No Repair
665
.683
.683
Carb. Change
121
.124
.807
Idle Stop Solenoid Bad
25
,026
.833
Timing was Advanced
19
.020
.853
Miscellaneous Repair
18
.018
.871
Vacuum Hose Leak
13
.013
.884
Brakes Dragging/Defectivi
11
.011
.895
Equipment Mai function
9
.009
.904
Idle Low 75 RPM Slow
7
.007
.911
Incorrect Fast Idle
7
.007
.918
EGR Valve Defective
6
.006
.924
Engine Change
5
.005
.929
Loose Mounting
5
.005
.934
Distributor Defect
5
.005
.939
TCS/TVS Defect Disconn.
4
.004
.943
Vacuum Break Defect
4
.004
.947
Incorrect/No IJp Shift
4
.004
.951
Overheater
3
.003
.954
Wrong Plug Cap
3
.003
.957
Wrong/Broken Spark Pig.
3
.003
.960
Vacuum Hose Leak
3
.003
.963
Electrical Conn. Shy
5
.003
.966
Idle Higfi 75 RPM Fast
3
.003
.969
Incorrect Dwell
2
.002
.971
Loose/Shy/Damaged Ign.
2
.002
.973
TCS Temp Override Sect.
2
.002
.975
Veh. Not Driveable
2
.002
.977
Air Hose Disct/Shy
2
.002
970
Rich Choke Action
2
.002
981
Timing was Retarded
2
.002
.983
Vacuum Conn. Shy/Leaking
2
.002
.985
Driver Error
2
.002
.987
Delay Relay Defective
2
.002
.989
Incorrect Choke Index
1
.001
.990
Wrong/Damaged Ign. Wire
1
.001
.991
Idle Stop Solenoid Deft
1
.001
.992
Improper Calibration
1
.001
.993
Incorrect Choke Rod Set
1
.001
.994
Stove Pipe Disct/Shy
1
.001
.995
Egr. Valve Leaks
1
.001
.996
Fluid Leak
1
.001
.997
Table 3.3-2
Audit Test Failure Reasons
(ranked by frequency of occurrence)
1973 General Motors
3-25
-------
4. AN ASSESSMENT OF EMISSIONS REDUCTIONS ATTAINABLE FROM A
NATIONWIDE APPLICATION OF CURRENT ASSEMBLY LINE TEST PROCEDURES
Introduction
As discussed in Chapter 2, the analysis and assessment of emissions
reduction which may appear to be associated with the Assembly Line Testing
activities will focus on three contributing segments: (1) Trends, (2) Short
Tests, and (3) Audit Tests. This chapter presents the specific results of
analyses of the available data base and an attempt to assess the magnitude of
emissions level changes as well as identify the plausible contributors to
these changes.
4.1 Emission Level Changes Association with Trends During a Production
Year
4.1.1 Introduction
The discussion in Section 2.2 of Chapter 2 presented a conceptual
approach to estimating the magnitude of changes in emissions associated with
trends in average emission levels during a production year. In this section,
that approach is developed with respect to the data base available for this
project and the consequent constraints placed on the computational procedure.
Specifically, due to the availability of emission data and production data on
a quarterly basis, rather than continually throughout the year, some distinct
modifications in the estimation technique are necessary. These are developed
in the next section.
4.1.2 Methodology
Assume that in production quarters j there are 0j autos produced.
Consider a baseline quarter in which emissions are distributed D ( JJ o, c5"o).
Suppose in production quarter j the autos manufactured have emissions distributed
D(xUj,
-------
[1} If produced at average emissions level o, the emissions
in tons per year are:
Eo
= JJ o ' r7. M x Oj (4.1.1)
and
where /Uo ~ average emissions (gm/mile) level at baseline
P 2 conversion factor tons/gm
M x average yearly mileage
H j = number of autos produced in "quarter" 3:j=l,2,...,5
(2) If produced at average emissions level ytt j , the emissions
in tons per year are:
Ej » JJi'f ' M'^j (4.1.2)
where y(Aj = average emissions level (gm/mile) for j**1
production quarter
P" ^ M, and 0 j are as before.
The difference in emissions levels (in tons per year) as
a result of producing the j cars at level JJ j rather than level JJo is seen
to be:
- E0-E4 ^UovUj^Mr^ (4a-3)
Since this difference in emissions levels involves only one produc-
tion quarter, one can add these over all production quarters of a model year
subsequent to that quarter in which the baseline level JX o prevailed, obtaining
= Z P M (4.1.4)
as the change in emissions, relative to operating through the entire production
year at average emission level JJ(o. In particular, if the baseline emission
4-2
-------
level, Mo, were viewed at the level prevailing at the beginning of a model
year, then one would accumulate the changes over the 5 production quarters of
a typical model year (i.e., the four calendar quarters plus the buildout quarter);
thus:
c = ,
Schematically, it may be helpful to interpret expression 4.1.5 in the
context of the figure below. The horizontal dotted line represents the average
emissions level which is presumed to prevail at the beginning of a production
year (i.e. - at time to). The curved solid line represents a particular pattern
of how average emissions levels may vary, with time throughout the production
year. At times tj,...,tg the average emissions levels prevailing are 1,...,
^5 and let us assume that these can be viewed (as defined previously) as an
average level typical of all production in quarters 1,2,...,5. Application of
expressions 4.1.1 and 4.1.2, for each quarter, will yield the total emissions
resulting from that quarters production at the respective levels A o and
The "change" in emissions then (i.e. - expression 4.1.3) is then
seen to be attributable to the difference in emissions levels prevailing in the
respective quarters and as portrayed below.
4-3
-------
Three comments are appropriate at this point. First of all, the dis-
cussion thus far has not distinguished between types of emittants . Since HC,
CO and N0x are all of concern in this study, changes in emissions will ultimately
be estimated for all three types of emissions; however, for purposes of dis-
cussion and methodological development, the text will generally refer to a
"single emittant type." The second comment regards the "jth quarter production
level, Pj," employed in expression (4.1.5). Upon analysis of nationwide pro-
duction data for the three manufacturers considered in this study (i.e., General
Motors, Ford and Chrysler), it occurred that the fraction of yearly total pro-
duction which occurs in the various production quarters maintains a relatively
constant split among the 5 production quarters of a model year, as follows:
Production Quarter Fraction of National
of Model Year Production in Quarter
1 0.17
2 0.25
3 0.25
4 0.25
5 0.08
Hereafter we shall define:
(1) (Xj = fraction of national yearly production taking place
in production quarters j(j = l ,2,...,5)
(2) N = nationwide yearly production of the manufacturer
under consideration.
4-4
-------
and then
--No^ Cj-1 »2,... ,5)
til.
that is, the j quarters production will be specified as a fractional multiple
of the applicable nationwide production level.
The third comment concerns the average emission levels (i.e., Mo,
/^j's) employed in expression (4.1.5). Analysis of manufacturer's data
reveals that there is generally no single average emission level representative
of all engine categories, consequently, it is appropriate to make expression
(4,il.S) "engine category specific" by relating the average emissions levels to
a manufacturer's engine category as follows:
Define: (1) JJio = baseline average emission level for engine category L
(I ~ 1.~•,C)
(2) /Ulj = average emission level for quarter j and engine
category i
(3) ©; = fraction of a manufacturer's yearly nationwide pro-
a n> a
duction which is in engine category i (note 21 )
1=1'
Then, employing this additional notation, the previous computational procedure
is readily modified to account for "engine category specific" computations, as
follows:
4-5
-------
Note:
£ct=
= Z P M $Ld(6 isl
-------
E?1 = If
P M °*j) Z(4.1.7)
Alternately, if in production quarter j the average emission level were ij,
for engine category L , the corresponding emissions for that quarter would be:
and, for all quarters,
E = f r
.(•£) I.
1
L-,
- T M ) (4.1.8)
As previously, the change in emissions during the remainder of the
year, in lieu of staying at levels JyLLioj. , is:
^(X) __
- PMN r ol, E^u,^-r-MNli fjl.ftoL
j'l
-------
In summary then, the estimate of change in emissions, is a
"scaled" emissions change, conditional on measuring changes for the remainder
of the model year, starting from quarter Jt •
4,1.3 Summary of Results
Based on the data and computations discussed at length in Section 4.1.4
of this chapter, the estimated changes in emissions which would appear to occur
during the 1972 model year analyzed in this study, if the changes observed in
the California data base are extrapolated to nationwide production, are as
indicated in Table 4.1-0.
Emission Type
Manufacturer HC CO NQ^,
General Motors -1SSS* +35750** NA
Ford -1272 +S2576 NA
Chrysler -66 -108 NA
Table 4.1-0
Trend Associated Emissions Changes (Tons per Year)
*
A minus sign (-) indicates a decrease in tons per year, nationwide
A plus sign (+) indicates an increase in tons per year, nationwide.
As discussed at length in Chapter 2, these data must be interpreted
with extreme caution since they are from data sources operating only under the
existence of an operational California assembly line test program. Further, while
the General Motors and Ford values result from analysis of 4 production quarters'
data, the Chrysler data estimates a change over a 3-quarter period.
4-8
-------
Division
Engine Category
CID/bbls
Average Emissions Level (by quarter)
2
3
4
5
Chevrolet
140-1, 2
1.64
1.25
1.44
2.40
2S0-1
1.31
1.66
1.35
1.43
350-2
1.92
1.79
1.83
1.88
350-4
2.06
1.76
1.87
2.52
400-2
1.93
1.73
1.88
1.95
Pontiac
350-2
2.72
2.53
2.39
3.32
400-2
2.26
2.00
2.05
2.30
455-2
2.07
2.13
2.21
2.21
400-4
2.33
2.37
2.29
2.50
455-4
1.96
1.93
1.92
2.22
Oldsmobile
350-2
2.30
2.28
2.58
2.50
350-4
2.40
2.23
2.35
2.42
455-4
1.40
1.24
1.42
1.59
Buick
350-2
2.00
2.08
2.02
2.00
350-4
1.81
1.73
1.74
1.76
455-4
2.32
2.10
2.45
3.02
Cadillac
472-4
1.00
1.13
1.21
1.60
500-4
0.87
0.89
0.87
0.87
Table 4.1-1
Average Hydrocarbon Emissions from Audit Tests
(By Engine Category and Production Quarter)
1972 General Motors
(grams per mile)
4-9
-------
Division
Engine Category
CID/bbls
Average
Emissions
Level fbv o
uarterl
2
3
4
5
Chevrolet
140-1, 2
21-75
19.87
18.66
19.36
250-1
21.17
26.75
22 *04
22.39
350-2
27,95
28.18
31.71
32.13
350-4
18.02
18.52
23.35
24.38
400-2
24.91
22.42
27.37
27.35
Pontiac
350-2
28.76
24.26
22.34
26.60
400-2
27.81
18.08
19.45
22.00
455-2
28.00
25.46
26.57
26.57
400-4
26.84
25.55
26.11
32.56
455-4
17.50
18.50
17.50
22.50
Oldsmobile
350-2
24.86
24.18
26.00
40.20
350-4
19.23
16.84
18.75
17.00
455-4
15.36
16.80
24.88
19 .37
Buick
350-2
26.27
26.25
29.31
32 .00
350-4
17.94
18.70
20.87
22.00
455-4
21.31
21.29
26.61
29.44
Cadillac
472 -4
24.41
26.90
31.09
46.00 j
500-4
23.20
23.97
22.76
22.76 |
Table 4.1-2
Average Carbon Monoxide Emissions from Audit Tests
(By Engine Category and Production Quarter]
1972 General Motors
(grans per mile)
4-10
-------
4.1.4 Data and Computations: Trend Analyses
Tables 4.1-1 and 4.1-2 present the average hydrocarbon and carbon
monoxide levels of audit tests indexed by engine category (within division)
and production quarter for the 1972 General Motors data set. The similar data
for NO was not used since 7-mode tests were employed in 1972.for NOX
x
Measurements. Using these average emission levels, and the computational
procedures discussed in Section average emission levels, and the computational
procedures discussed in Section 2.2, the changes in emissions levels for the
remainder of the 1972 model year were determined. Specifically, the equation
(4.1.9) was used; that is:
£,(a>- w.i.io)
Observe that for the General Motors analysis the baseline emission
level used is the average emission level in the second production quarter and
consequently the emission level changes estimated here are in terms of changes
from quarter 2 of the 1972 model year.
It is further pointed out that, at the request of the manufacturers
supplying this data, the production figures regarding engine category produc-
tion as a fraction of total year production is being treated in a confidential
manner. Consequently, the actual values of | t =1,2,...,18^ will not be
recorded in this report but rather, the summary result from expression (4.1.10)
will be presented. Ford and Chrysler data will be treated in a similar manner.
For the GM data, one obtains:
(1) Hydrocarbons:
4-11
-------
= (1.1 x 10"6 tons/gm) (12000 —7^) (5 x 106 )
= 1555 tons per year decrease in HC.
0.02356 -Si-
mile
(2) Carbon Monoxide
, 1 f -\
C = P M Ni - 0.54167 gm/mile J
= 35750 tons per year increase in CO.
(3) Oxides of Nitrogen
Indeterminate since 1972 tests were by 7-mode methods.
Tables 4.1-3 and 4.1-4 present the average hydrocarbon and carbon
monoxide emission levels of audit tests, indexed by engine category and pro-
duction quartert for the 1972 Ford data set. As with all manufacturers, NO^
results were not analyzed. Proceeding as with General Motors, emissions changes
for that portion of the 1972 model year beyond quarter 2 were determined em-
ploying equation (4.1.9) and using baseline average emission levels equal to
second production quarter results.
4-12
-------
Engine Category
CID
Average Emissions Level (by quarter)
2
3
4
5
98
2.45
2.35
2.41
2.41
122
2.54
2.51
2.49
2.19
170
2.49
2.47
2.47
2.47
200
2.57
2.39
2.32
2.32
240
2.67
2.72
2.85
2.73
250
2.13
2.03
2.03
2.28
302
2.64
2.52
2.23
3.05
351
2.48
2.49
2.61
2.49
360
3.10
3.09
3.19
3.25
390
3.11
2.59
2.69
2.69
400
2.07
2.08
1.90
1.87
429
2.07
2.18
2.11
2.29
460
2.44
2.35
2.58
2.38
Table 4.1-3
Average Hydrocarbon Emissions from Audit Tests
(By Engine Category and Production Quarter)
1972 Ford
(grams per mile)
Engine Category
Average
Emissions
Level (by
quarter)
CID
2
3
4
5
98
122
22.93
28.59
29.18
26.56
170
16.88
17.58
17.68
17.68
200
23.10
23.46
24.21
24.21
240
24.84
30.67
33.06
32.17
250
23.14
21.12
21.12
23.88
302
19.35
23.40
18.98
35.34
351
22.46
28.65
34.58
30.88
360
25.61
29.36
30.44
33.04
390
22.40
25.77
20.10
20.10
400
34.28
32.77
25.91
21.13
429
24.28
26.18
25.64
30.08
460
28.19
28.30
42.58
33.02
Table 4.1-4
Average Carbon Monoxide Emissions from Audit Tests
(By Engine Category and Production Quarter)
1972 Ford
(grams per mile)
4-13
-------
From the Ford data, one obtains:
(1) Hydrocarbons
e= rMNjcf-.*, - fcv
L J"?- 5 l-l L-, d J
= (1.1 x 106 ) C12000 mil^S - ) (3.3 x 106 cars) .0292
gm car/yr JV J mile
= (43.56 x 103)(.0292) tons/yr
6 = 1272 tons per year decrease in HC.
(2) Carbon Monoxide
d = (43.56 x 103)(-1.8957) tons per year
= S2576 tons per year increase in CO.
(3) Oxides of Nitrogen
Indeterminate since 1972 tests were by 7-mode methods.
Tables 4.1-5 and 4.1-6 present the average hydrocarbon and carbon
monoxide emission levels of audit tests, indexed by engine category and pro-
duction quarter, foT the 1972 Chrysler data set. In contrast to General Motors
and Ford, the Chrysler data began with production quarter 3, hence, this was
used as a baseline.
4-14
-------
Average Emissions Level
Engine Family
(hy qiiav+
-------
Using the procedures, as earlier, with the Chrysler data one obtains:
(1) Hydrocarbons
£ = P M N3 (j? /^(3l) - j^0!)
= (1.1 x 10"6 ) (12000 mil^S )(1.5 x lo6 0.0338 -t
v gm car/yr'v yr ' mile
= 65.8 tons per year decrease in HC.
(2) Carbon Monoxide
(5 ¦ (1.98 x 103) { 0.05578 gm/mile ]
£ ~ 108 tons per year decrease in CO.
(3) Oxides of Nitrogen
Indeterminate since 1972 tests were by 7 Mode methods.
Trend estimates for the 1973 model year were not computed since,
during this project, data from only two production quarters of the 1973 model
year was available and this was judged insufficient to calculate lf73 trends.
4-16
-------
4.2 Emission Level Changes Associated with Short Inspection Tests
4.2.0 Introduction
The methodology for assessing changes in emission levels associated
with th« nresence or absence of hot 7-mode tests, idle tests, or combina-
tions thereof was described in Chapter 2, Section 2.3. Application of this
methodology to the particular data base available is presented in this
section.
4.2.1 Methodology
As employed in the 1973 California test procedures, two short
inspection test types are employed, namely, (i) the hot cycle 7-mode test and,
(ii) the idle test. Of interest here is an assessment of what (if any) change
in emissions levels (as measured by C.V.S. "audit" test procedures) results
from the presence of this short test procedure.
Considering the "hot cycle 7-mode test" as a "procedure T "
discussed in Section 2.3 of this report, we can investigate the manner in
wnicn the averajre level of emissions of type j depends on the fraction ( c( )
of automobiles receiving this test. Applying expressions 2.3.1 and 2.3.2
where T5 (i.e. - "presence of hot seven mode cycle test")
one obtains
ei.
—fM K) (4.2.1)
4-17
-------
In a similar fashion we can consider the "idle test" as a
"procedure T " as discussed in Section 2.3. Again applying expressions
2.3.1 and 2.3.2 as above, but in this case withT= I (i-e. T <=%> "presence
of idle test"), one obtains:
£, = /JjP h N
={S(/Uir 0-S')yUi,;1TPMM
~ ^ + W'N (4.2.2)
where 0( is the fraction of total production receiving die ests.
4.2.2 Data Base Available
For the analyses performed in this section, data from the 1972
production year of General Motors, and 1973 production year of General
Motors, Ford and Chrysler were employed. While a detailed description of
the data base available and/or employed on this project is contained in
Appendix A , the specific subsets of data employed, and the analysis
executed is described here.
Selected data was analyzed to assess the changes in emissions
levels which appear to have occurred due to combinations of the presence
or absence of hot seven mode testing and/or idle testing. The emissions
measurements analyzed were the hydrocarbon (H.C.), carbon monoxide (C.O.)
and oxides of nitrogens (NOX) levels, in grams per mile, as determined
from constant volume Sampling test procedures currently employed in the
California Assembly Line Test procedures. On 1972 model year data the
NOX measurements were not analyzed since they were developed by 7-mode
rather than C.V.S. testing. The data available, and the sample sizes
4-18
-------
of the various categories analyzed, are presented in Figures 4.2-1, 4.2-2,
4.2-3 and 4.2-4, and are grouped according to the manufacturer.
I. General Motors Corporation
There were test results from 9284 original C.V.S.
measurements on 1972 production year automobiles. Source data available did
not indicate whether or not these were idle tested, however, for each it was
stated whether or not a 7-mode test had been performed on the car.
Further, data from the third, fourth and fifth
(i.e. - buildout) quarter indicated, for those cars 7-mode tested, whether
or not the car had passed that test. Figure 4.2-1 illustrates the break-
down of the test/no test samples and their corresponding sizes.
Figure 4.2-1
1972 General Motors Data
Sample Sizes
7-roode Tested Automobiles
295 Cars for which
Pass/Fail 7*-mode
not indicated
4-19
-------
Figure 4.2-2
1973 General Motors Data
Sample Sizes
7-mode and/or Idle Tested Automobiles
4-20
-------
The 1973 General Motors data base available included C.V.S. test
results from the first and second production quarters of the model year.
Since Idle Testing was statutorily required on this model year, this data
base was capable of being segmented into a larger number of subcategories
than in the 1972 data. Figure 4.2-2 presents this breakdown.
II. Ford Motor Company
The 1973 Ford data base employed in this section
contained 2903 original C.V.S. measurements, each of these containing
an indication of whether or not 7-mode and/or idle tests had been per-
formed on the automobiles. This is presented in Figure 4.2-3.
Figure 4.2-3
1973 Ford Data
Sample Sizes
7-mode and/or Idle Tested Automobiles
4-21
-------
III. Chrysler Corporation
The 1973 Chrysler data base employed in this section
contained 1462 original C.V.S. measurements, each of these containing an
indication of whether or not 7-mode and/or idle tests had been performed
on the automobiles. This is presented in Figure 4.2-4.
1462
Original Measurements
385 Original Measurements on
Cars Only 7-mode Tested
55 Original Measurements on
Cars Only Idle Tested
254 Original Measurements on
Cars Both 7-mode and Tdle
Tested
968 Original Measurements
Cars Neither 7-mode
Nor Idle Tested
823 Original Measurements on
Cars Not 7-mode Tested
639 Original Measurements on
Cars 7-mode Tested
1153 Original Measurements on
Cars Not Idle Tested
309 Original Measurements on
Cars Tdle Tested
Figure 4.2-4
1973 Chrysler Data
Sample Sizes
7-mode and/or Idle Tested Automobiles
4-22
-------
4.2.3 Summary of Results
Table 4.2-1 summarizes the estimated emissions level changes
which would result from a nationwide application of hot 7-mode and idle
testing as is currently being performed for California production. To
TeiteTate, this would incorporate hot 7-mode testing of a 25% (randomly
selected) sample of production and idle testing of the remainder of
production
I Source of
Emissions Level
Mfr.
Emissions Type
Changes
HC
CO
NOx
7 Mode Test
(1972 Model Year)
GM
- 2310
- 29040
NA
F
NA
NA
NA
C
NA
NA
NA
Total
- 2310
- 29040
NA
7 Mode Test
(1973 Model Year)
GM
0
- 60500
+ 825
F
+ 872
0
+ 3815
C
64
0
0
Total
+ 808
- 60500
+4640
Idle Test
(1973 Model Year)
GM
+14355
+103950
+2970
F
- 4185
0
-2943
C
0
+ 851
+ 45
Total
+10170
+104801
00
+
Table 4.2-1
Emissions Level Changes From Short Inspection Tests (tons per year)
The specific data and computations on which these results are
based are presented in Section 4.2.4 of this chapter.
4-23
-------
In addition to the emissions changes attributed to 7-mode tests or
idle tests as presented in the previous table, it is informative to consider
the patterns of emissions level changes associated with automobiles receiving
only 7-node testing, only idle testing, neither 7-mode nor idle testing or,
finally, both 7-mode and idle testing. Such considerations are summarized,
schematically, in Figures 4.2-5 thru 4.2-7. Each of these charts portrays
the type of change in the audit test emission level as a result of a
particular combination of short tests. All changes portrayed start from the
•'reference point" of the average audit test emission level when no short test
has been executed. For example, consider Figure 4.2-5a, hydrocarbon emissions
on 1973 General Motors data. Relative to audit test levels on cars which
received no previous short tests, this chart reveals that cars which were
only 7 Mode tested do not differ in average HC emission levels, (i.e.,
No Significant Difference indicated on the broken line between the "Neither
Test" and "7-mode Test" cell). The average HC emission on cars idle tested
only reflected an increase relative to cars receiving Neither Test.
In general then, a broken line between two connected cells indicates
no difference in the average emissions level under the two associated test
conditions; a solid line indicates that the cell toward which the arrow points
reflects an increased (decreased) emission level as compared to the connected
cell according as an increase (decrease) note is along the connecting link.
4-24
-------
Figures 4.2-5 a,b
Short Test Patterns of Emission Level Changes
1973 General Motors
4-25
-------
Figure 4.2-5 c
Short Test Patterns of Emissions Level Changes
1973 General Motors
4-26
-------
b.) Carbon Monoxide Emission Chanpes
Figures 4.2-6 a,b
Short Test Patterns of Emissions Level Changes
1973 Ford
4-27"
-------
increase
c.) Oxides of Nitropen Emission Chan pes
Figure 4.2-6 c
Short Test Patterns of Emission Level Changes
1973 Ford
4-28
-------
( Neither^
Test
dec
rease
7 Mode
. Test
increase1
increase
b.) Carbon Monoxide Emission Changes
Figure 4.2-7 a,b
Short Test Patterns of Emissions Level Changes
1973 Chrysler
4-29
-------
c.) Oxides of Nitrogen Emissions Chanpes
Figure 4,2-7 c
Short Test Patterns of Emission Level Changes
1973 Chrysler
4-30
-------
4.2.4 Data and Computations: Short Inspection Tests
Tables 4.2-2 thru 4.2-5 present the summary results on the samples
of automobiles employed in the analyses of "hot 7-irode test" effects. The
sample statistics included here are based on the original C.V.S. measurements
taken on the tested automobiles (in contrast to rerun measurements on an
automobile which had failed its audit test). In this sense, the statistics
represent the average emissions to be anticipated upon the first C.V.S. test
to be run in audit test procedures.
Table 4.2-6 summarizes the results of significance tests performed
to investigate the possible existence of a significant change in average
emissions (on audit test results) as a result of an automobile having been
7-mode tested prior to the audit testing.
4-31
-------
Emission
Type
Sample Average
(gm/mile)
Sample Variance
2
C^m/mile )
Sample
Size
HC
CO
NOx
1.70
21.24
NA
0.42
95.31
NA
1117
1117
NA
7 Mode
Tested
Sample
HC
CO
NOx
1.84
23.0
NA
0.73
172.14
NA
8167
8167
NA
Not 7 Mode
Tested
Sample
Table 4.2-2
Audit Test Emissions Results
7-mode and Not 7-mode Tested Automobiles
1972 General Motors
(Original Measurements)
4-32
-------
Emission
Type
Sample Average
(gm/mile)
Sample Variance
(gm/mile2)
Sample
Size
HC
CO
NOx
2.20
25.43
2.32
0.63
244.17
0.40
2304
2304
2304
7 Mode
Tested
Sample
HC
CO
NOx
2.21
29.10
2.27
0.33
180.66
0.22
1781
1781
1781
Not 7 Mode
Tested
Sample
Table 4.2-3
Audit Test Emission Results
"7-mode" and "Not 7-irode1' Tested Automobiles
1973 General Motors
(original measurements)
4-33
-------
Emission
Type
Sample Average
(gm/mile)
Sample Variant
tgrn/mi 1 e^)
e Sample
Size
HC
CO
NOx
2.43
26.90
2.46
0.26
48.91
0.13
2505
2505
2505
7 Mode
Tested
Sample
HC
CO
NOx
2.36
26.95
2.11
0.16
66.6
0.19
398
398
398
Not 7 Mode
Tested
Sample
Table 4.2-4
Audit Test Emission Results
"7-mode" and "Not 7-mode" Tested Automobiles
1973 Ford
(original measurements)
4-34
-------
Emission
Type
Sample Average
(gin/mile)
Sample Variance
(gm/mile^)
Sample
Size
HC
CO
NOx
2,05
25.05
2.16
0.24
76.04
0.29
639
639
639
7 Mode
Tested
Sample
HC
CO
NOx
2.18
25.92
2.40
0.39
99.17
0.42
823
823
823
Not 7 Mode
Tested
Sample
Table 4.2-5
Audit Test Emission Results
7-mode and Not 7-mode Tested Automobiles
1973 Chrysler
(original measurements)
4-35
-------
Comparison of 7-mode Tested vs. Not 7-mode Tested Automobiles
Tested Statistic Employed:
-- *
21 ~ ( fi>?M SjjS
0 n, v>L
Summary of Results:
Model
Year
Manufacturer
Emission
Type
Significant
at 5% Level
Conclusion
1972
GM
HC
+8.092
Yes
¦LAfa ^
CO
+5.399
Yes
JJ 7M >
1973
GM
HC
+0.578
No
CO
+8.061
Yes
JJ-7H > Mm
NOx
-3.55
Yes
^ m
1973
Ford
HC
-3.139
Yes
M:Tm * A-Ant+
CO
0.116
No
MJih -
NOx
-17.50
Yes
yUw /U7M
1973
Chrysler
HC
-4.924
Yes
6J-7M H
CO
NOx
-1 .7778
-1.333
No
No
Xvi-7H ".X^n
Table 4.2-6
Summary of 7-mode Test Effects on Audit Results
* Introductory Engineering Statistics by T. Guttman and S.S. lVilks
J. Wiley and Sons
4-36
-------
For those model years and emission type combinations for which
there occured a significant change in emissions levels due to the presnece
of 7-mode testing the magnitude of emissions changes (in tons per year) were
computed. These computations are illustrated in the following material, and
employ the relationship
^ ^ (4.2.3)
as described earlier in this chapter (i.e., Section 4.2)
1972 General Motors:
Hydrocarbon:
(£u = (1.1 x 10 ^ tons/gram) (12,000 miles/year)
(5 x 106 autos/year)
[1.84 gm/mile + o((1.70 - 1.84) gm/mile]
= (66 x 103) (1.84 - 0.14o{)
= 121440 - 9240 tons per year HC
With o( = 0.25, one obtains a decrease of
/l(( ¦ 0(9240 = 2310 tons/year
II Carbon Monoxide:
Sa = (66 x 103) 123.0 gm/mile + 0( (21.24 - 23.0) gin/mi]
» 1,518,000 - 116,1600( tons per year CO
With ot 85 0.25, one obtains a decrease of
^ = 116,160o( ¦ 29040 tons per year CO
III Oxides of Nitrogen:
Not applicable since C.V.S, measurements not available
for this data base.
4-37
-------
1973 General Motors
I Hydrocarbon
Not applicable since there is no significant difference
in the average emission level under the presence or
absence of 7-node testing (see Table 4.2-6)
II Carbon Monoxide
Sl% = (66 x 10s) (29.1 gm/mile + o( (25.43 - 29.1) gm/mi}
= 1,920,600 - 242,0000( tons per year
With (X = 0.25, one obtains a decrease of
Az= 60,500 tons per year CO
HI Oxides of Nitrogen
C?i3 - (66 x 10J) (2.27 gm/mile + 0< (2.32 - 2.27} gm/mile)
= 149820 + 3300o( tons per year
With 0( = 0.25, one obtains an increase of
A13= 825 tons per year NOx
1973 Ford
Hydrocarbon
621 = f7 M NJa + o( (/Uin-\
= (1.1 x 10"6 tons/gm) (12,000 )
„ . _6 cars. car/yr
(3.3 x 10
{2.36 gm/mile + o((2.43 - 2.36)jgm/mile
= 102896 + 3488 o( tons per year
With = 0.25, one obtains an increase of
A2, = 872 tons per year HC
4-38
-------
II Carbon Monoxide
Not applicable since there is no significant difference
in the average emission level under the presence or
absence of 7-mode testing (see Table 4.2-6)
III Oxides of Nitrogen
&ZS = (43.6 x 103){2.11 gm/mi + o<(2.46 - 2.11)]gm/mi
= 91996 + 15260 tons per year
With
-------
Tables 4.2-7 thru 4.2-9 present the summary results on the samples
of automobiles employed in the analyses of the "idle test" effects. As in
the previous analyses of M7-mDde test" effects these statistics are based on
original C.V.S. measurements taken on the tested automobiles.
Table 4.2^o summarizes the results of significance tests performed
to investigate the possible existence of a significant change in average
emissions (on audit test results) resulting from an automobile having been
idle tested prior to the audit test.
4-40
-------
Emission
Type
Sample Average
(gm/mile)
Sample Variance
(gm/mile^)
Sample
Size
HC
CO
NOx
2.24
27.27
2.31
0.52
236.88
0.35
3623
3623
3623
Idle
Tested
Sample
IIC
CO
NOx
1.95
25.17
2.25
0.25 462
81.67 462
0.14 462
Not Idle
Tested
Sample
Table 4.2-7
Audit Test Emissions Results
"Idle" and "Not Idle" Tested Automobiles
1973 General Motors
(original measurements)
4-41
-------
Emission
Type
Sample Average
(gin/mile)
Sample Varian
(gm/Tnile2)
ce Sample
Size
HC
CO
NOx
2 .36
27.07
2.37
0.23
52.3
0.18
1587
1587
1587
Idle
Tested
Sample
HC
CO
NOx
2.48
26.72
2.46
0.25
50,30
0.12
1316
1316
1316
Not Idle
Tested
Sample
Table 4.2-8
Audit Test Emissions Results
"Idle" and "Not Idle" Tested Automobiles
1973 Ford
(original measurements]
4-42
-------
Emission
Type
Sample A.verap
p SampLe Variant
(gm/inile2)
J Sample
Size
HC
CO
NOx
2.13
26.69
2.37
0.26
ZQAZ
0.29
309
309
309
Idle
Tested
Sample
HC
CO
NOx
2.12
25.17
2.28
0.35
91.06
0.40
1153
1153
1153
Not Idle
Tested
Sample
Table 4.2-9
Audit Test Emissions Results
"Idle" and "Not Idle" Tested Automobiles
1973 Chrysler
(original measurements)
4-43
-------
Comparison of Idle Tested vs Not Idle Tested Automobiles
Test Statistic Employed:
^= (jx_ sir
J r>,
Summary of Results
Model
Manufacturer
Emission
Significant
Conclusion
Year
Type
at 5% Level
1973
GM
HC
-11.89
Yes
JJ,x <
CO
- 4.27
Yes
JJ.Y < jUt
NOx
- 3.47
Yes
J-i T<
1975
Ford
HC
8.511
Yes
JJiT >X'i
CO
- 1.313
No
yU i = Xiz
NOx
9.000
Yes
/Llf >X^r
1973
Chrysler
HC
- 0.302
No
JJ r -
CO
- 2.958
Yes
JJ.X < AJt
NOx
- 2,601
Yes
Table 4.2-10
Summary of Idle Test Effect on Audit Results
4-44
-------
For those emission types which revealed a significant change in
emissions levels due to the presence of idle testing the magnitude of
emissions changes (in tons per year) were computed. These computations are
illustrated in the material which follows and employ the relationship:
- P H ^£><3 ) J" (4.2.4)
1973 General Motors
Hydrocarbons
<£;, = (1.1 * 10'6 tons/gmi (12,000
(5 X 106 5H1)
yr ^
[1.95 gm/mi + o^(2.24 - 1.95) gm/mi]
= (66 x 103) [1.95 gm/mi + o((2.24 - 1.95) gm/mi]
=¦ 128700 + 191400C. tons per year HC
With o( = 0.75, one obtains an increase of
An = 14355 tons per year
II Carbon Monoxide
£\-L = (66 X 103) [25.17 gm/mi + c< (27.27 - 25.17) gm/mi]
» 1,661,220 ~ 138,600^ tons per year CO
/V
With
-------
1973 Ford
I Hydrocarbons
£i\ = (1.1 x 10"6 tons/gm) (12,000 )
_ _6 cars.
/7 7 «c&rsv
(3.3 x 10
[2.48 gm/mi + o( (2.36 - 2.48) gm/mi]
<-lt - 108,128 - 5580 o(, tons per year HC
With o( = 0.75, one obtains a decrease of
Asi = 4185 tons per year
II Carbon Monoxide
Not applicable since there is no significant difference
in the average emission level under the presence or
absence of idle testing (see Table 4.2-10)
III Oxides of Nitrogen
^23 = (43.6 x 103) {2.46 gm/mi + 5 (2.37 - 2.46)Jgm/mi
(Sz3i = 107256 - 3924o( tons per year NOx
With CX. = 0.75, one obtains a decrease of
A13 = 2943 tons per year
1973 Chrysler
I Hydrocarbons
Not applicable since there is no significant difference
in the average emission level under the presence or
absence of idle testing (see Table 4.2-10)
II Carbon Monoxide
= (1.1 x 10~6 tons/gm) (12,000 ml/S )
6 cars car/yr'
(1.5 x 10 £££)
[25.17 gm/mile +3 (26.89 - 25.17)]
(5^2 = 49836 + 3405c( tons per year CO
With ot = 0.75, one obtains an increase of
= 851.25 tons per year
4-46
-------
Ill Oxides of Nitrogen
C?3'3 = (1.98 x 103)|(2.28 gm/mi + 5 (2.37 - 2.28)^gm/mi)
= 4514 + 178.2ai tons per year NOx
C-'
With <*, = 0.75, one obtains an increase of
A3* = 44.5 tons per year
The current Assembly Line Test program calls for 100% of California
production to be "Inspection Tested", with 25% being 7-mode tested and 75%
being idle tested. In theory, then, there would be no overlap of these
testing procedures. In practice, however, it occurs that there is considerable
overlap (i.e., some automobiles receive both 7-node and idle tests), and
there are large numbers which are either 7 Mode tested only or Idle tested
only.* The occurrence of these four exhaustive and mutually exclusive com-
binations of testing permits an investigation of further, and more specific,
patterns which may occur in emission level changes as a function of testing
combinations.
Tables 4.2-11 thru 4.2-13 present the summary results, by manufacturer,
on the samples of automobiles employed in the analysis of "hot 7-mode and/or
idle test" effects on C.V.S. audit test emissions levels. It should be noted
that the data of these tables represents a further decomposition of the
samples, and consequently the statistics, contained in Tables 4.2-2 thru
4.2-5 as well as Tables 4.2-7 thru 4.2-9.
* Further, for the 1973 data base employed in this study (i.e. - the first
two production quarters), only 25% of production was required by California to
be idle tested; thus, some autos in the data base received no inspection
testing.
4-47
-------
Inspection
Test
Combination
Emissions
Type
Sample
Average
(gm/mile)
Sample
Variance
o
(gm/mile'6)
Sample
Size
7 Mode
HC
1 .92
0.23
217
Only-
CO
24.17
89.93
217
NOX
2.24
0.16
217
Idle
HC
2,24
0.33
1536
Test
Only
CO
29.58
196.18
1536
NOX
2.28
0.24
1536
Bo tfi
HC
2.23
0.66
2087
Types
CO
25.56
260.08
2087
NOX
2.33
0.42
2087
Neither
HC
1.97
0.26
245
Type
CO
26 .06
73.0
245
NOX
2.26
0.13
245
Table 4.2- 11
Audit Test Emission Results
7-mode/Idle Test Combinations
1973 General Motors
(original measurements)
-------
Inspection
Emission
Sample
Sample
Sample
Test
Type
Average
Variance
Size
Combination
(gm/mile)
2
(gm/mile )
7 Mode
HC
2.48
0.25
1242
Only
CO
26.62
48.74
1242
NOX
2.48
0.11
1242
Idle
HC
2.32
0.15
324
On ly
CO
26.63
64.42
324
NOX
2.08
0.17
324
Both
HC
2.38
0.51
1263
Types
CO
27.19
49.24
1263
NOX
2.45
0.15
1263
Neither
HC
2.51
0.17
74
Type
CO
28.36
74.63
74
NOX
2 .22
0.25
74
Table 4.2-12
Audit Test Emission Results
7-mode/Idle Test Combinations
1973 Ford
(original measurements)
4-49
-------
Inspection
Test
Combinations
Emission
Type
Sample
Average
(£m/mile)
Sample
Variance
2
(gm/mile )
Sample
Si ze
7 Mode
HC
1.96
0.20
385
Only
CO
23.39
66.32
385
NOX
1.97
0.21
385
Idle
HC
1 .83
0.14
55
Only
CO
23.85
67.92
55
NOX
1.97
0.19
55
Bo th
HC
2.20
0.26
254
Types
CO
27.55
80,59
254
NOX
2.46
0.27
254
Neither
HC
2.21
0.40
768
Type
00
26.07
101.17
768
NOX
2.43
0.43
768
Table 4.2- 13
Audit Test Emission Results
7-mode/Idle Test Combinations
1973 Chrysler
(original measurements)
4-50
-------
4.3
Emission Level Changes Associated With Audit Tests
4.3.0 Introduction
The methodology for assessing changes in emission levels associated
with the audit procedures was described in Chapter 2, Section 2.4. Application
of this methodology to the particular data base available is presented here.
For the 1972 model year such analysis was possible for the General Motors
Corporation only, since this was the only available data set which contained
final measurements of emission levels for those automobiles which failed the
audit tests originally. Further, within this model year, analyses were only
possible for Hydrocarbon and Carbon Monoxide emissions Csince NOx was measured
by 7 Mode rather than C.V.S. techniques).
For the 1973 model year analyses were possible for both General
Motors and Chrysler data, and further, since C.V.S. techniques were employed
on NOx measurements for this model year it was possible to consider all three
types of emissions (i.e., HC; CO; and NOx).
4.3.1 Methodology
The basic estimation procedure employed is the relation:
(BO-pa)] (4.3.D
as discussed in Section 2.4. From the data base available it is not possible
to estimate and consequently it was assumed that
— CO —\P),p
e* = ^
for each manufacturer anri emission type. Under this assumption, and coupled
with the fact that ~ ¦+('i-p) 3^ ^ » substitution
into expression 4.3.1 reduces to ^
(4.3.2)
The components of this expression were derived as follows:
4-51
-------
I p
(1.1 x 10 6 tons/gram)
II = 12,000 miles per year (assumed as a plausible yearly
average mileage for a typical automobile). This was
further assumed to be constant £ot all manufacturers
(i.e., i = 1, 2, 3).
III fsl^ = yearly nationwide production for manufacturer
IV
(3 = 0.02 (i.e., current procedures employed in California
call for 2% audit sampling)
V = For the ^-th tnanuf acturer this was taken as the ratio
of the number of failures on original audit tests
divided by the number of original audit tests.
VT = The average level of emissions of tvne > for cars
originally failing for thex-th manufacturer, from all
available original emissions measurements from the
audit tests.
, (0>,p
VII = The average level of emissions of type for cars
originally passing for the X-th manufacturer, from all
available original emissions measureiments from the
audit tests.
— ca>
VIII g, o = The average of the final emissions measurements from
the audit tests of manufacturer 1 (on emission type j.)
4.3.2 Data Base Employed
The data and information needed to estimate the parameters employed
in expression 4.3.2 were based on appropriately selected subsets of the 1972
and 1973 General Motors Audit Test data and the 1973 Chrysler Audit Test dat.
Estimation of the average emissions level for failing cars
_ Co) | |
- was conducted by using the set of original^ audit test results for the manu-
facturers involved. These sets included both passing as well as original measure-
4-52
-------
ments on failing cars and in that sense were representative of a population of
measurements taken on first tests of cars which failed off the production line.
For the data bases involved this included (7) 656 original measurements on 1972
General Motors cars, (II) 643 original measurements on 1973 General Motors cars,
and (III) 254 original measurements on 1973 Chrysler Corporation cars.
Estimation of the "final average emissions level of rectified cars
— f 21
which had originally failed the audit test" v ' was conducted by using
the set of final audit test results for the manufacturers involved. For the
data base involved this included: (1) 650 final measurements on 1972 General
Motors cars which originally failed, (IT) 640 final measurements on 1973
General Motors cars which originally failed and (TIT) 254 final measurements
on 1973 Chrysler cars which had originally failed.
4.3.3 Summary of Results
Table 4.3-1 presents a summary of the estimated emission level
changes which would result from a nationwide application of 2% Audit Sampling
as is currently being performed on California production.
Model Year
Manufacturer
Emission
Type
Emissions Level
Change
(tons per year)
1972
General Motors
Hydrocarbon
122
(decrease)
Carbon Monoxide
39
(decrease)
1973
General Motors
Hydrocarbon
121
(decrease)
Carbon Monoxide
2317
(decrease)
Oxides of Nitrogen
62
(decrease)
1973
Chrysler
Hydrocarbon
30
(de crease)
Carbon Monoxide
589
(de crease)
Oxides of Nitrogen
34
decrease)
Table 4.3-1
Estimated Emission Level Changes From Nationwide Audit Testing
at 2% Sampling Level
4-53
-------
4.3.4
The specific data and computations upon which the results of
Table 4.3-1 are based are presented here.
I Conversion Factor: [7 - 1.1 x 10 ^ tons per gram
II Average Yearly Mileage: 12,000 miles per year per car
assumed constant for all manufacturers
III Nationwide Productions
Manufacturer
I
Hi
(automobiles/year)
General Motors Corporation
Ford Motor Company
Chrysler Corporation
1
2
3
5.0 x 106
3.3 x 106
1.5 x 106
IV Sampling Rate: ^ = ,02 assumed constant for all manufacturers
V Failure Rates
Manufacturer
I
Pj> (*)
1972
1973
i
General Motors Corporation
1
7.06
15.74
jFord Motor Company
2
NA
NA
Chrysler Corporation
3
NA
17.37
4-54
-------
VI Average Emissions Levels: Original Measurements
Manufacturer
I
Emittaiit
J-
Average Emissions
Cgm/mile)
j
1972
1973
General Motors Corp.
1
HC
1
3.01
2.75
CO
2
38.59
38.78
NOx
3
NA
2.57
Ford Motor Co.
2
HC
1
NA
NA
CO
2
NA
NA
NOx
3
NA
NA
Chrysler Corp,
3
HC
1
NA
2.61
CO
2
NA
34.44
NOx
3
NA
2.-82
4 -55
-------
VII Average Emissions Levels: Final Measurements for Cars which
Originally Failed
Manufacturer
1
Emittant
1
J
Average Emissions
(gm/mile)
cT
1972
1973 ,
General Motors Corp.
1
HC
1
1.70
2.17
CO
2
22.64
27.63
NOx
3
NA
2.27
Ford Motor Co.
2
HC
1
NA
i
NA
CO
2
NA
NA
NOx
3
NA
NA
Chrysler Corp.
3
HC
1
NA
2.12
CO
2
NA
25.88
NOx
3
NA
2.33
yields.
Substitution of the data of I - VII above into expression 4.3.2
1972 General Motors
i £„ = nM,N,i3p.(et-ci*w)
= (1.1 x 10"6 gm/mile) • (12 x 103 miles/yr) •
(5 x 106 CaSS) • (0.02) (.0706) (3,01 - 1.70 gm/mi)
- 122.08 tons per year decrease in HC
n £n - r7 m, n, (3 p, (e?a- c.t1)
= (1.1 x 10~6) (12 x 103) (5 x 106) (.02) (.0706)
(38.59 - 22.64)
=1486.41 tons per year decrease in CO
4-56
-------
1973 General Motors
i <5„ ¦ pM,N,(5p. Ce^-e,',")
« (1.1 x 10"6) (12 x 103) (5 X 106) (.02) (.1574)
(2.75 - 2.17)
=120.51 tons per year decrease in HC
ii cri = pm.n, pp^St'-e"')
= (1.1 X 10"6) (12 x 103) (5 X 106) (.02) (.1574)
(38.78 - 27.63)
= 2316.61 tons per year decrease in CO
hi c,3 = P M,N. P- £,%)
= (1.1 x 1(T6) (12 x 103) (5 x 106) (.02) (.1574)
(2.57 - 2.27)
= 62.33 tons per year decrease in NOx
1973 Chrysler
i <531 ¦ HMjNs 31^ )
= (1.1 x 10"6) (12 x 103) (1.5 x 106) (.02) (.1737)
(2.61 - 2.17)
¦ 30.27 tons per year decrease in HC
n
-------
4.4
An Analysis of Gross Emitters
One facet of emissions patterns which is likely to be of some interest
and utility is that of outliers or gross emitters. While the statistical
pattern of emissions for a particular en pine may be at an acceptable average
level one may still be concerned about the "high tail" of the distribution,
namely what percentage of autos produced will have emissions beyond some
"tolerance limit" so to speak. In an attempt to develop some insight into the
extreme values resulting from audit test emissions measurements, the 1972
General Motors data base was analyzed. Because of time and funding limitations
this restricted subset of data was employed to analyze gross emitters, rather
than using the entire data set available during the project. At the request of
the Environmental Protection Agency a lower bound for defining a "Cross Fmitter"
was defined as that point which was both two standard deviations above the
average level of the emission type under analysis and higher than the applicable
emission standard. For the three emittants the development of the "Gross Emitter
Limits" is shown below:
Emission
Type
' Sample
Average (gm/mile)
Sample
Std. Dev.
Gross Emitter
Limit
California
Standard
HC
1.82
0.83
3. 48
3.2
CO
22.79
12 .78
58.35
39
NOX
1 .93
0.67
3.27
3.2
Since these limits exceed, for each emission tyt>e, the applicable
standard these limits will serve to define the gross emitters.
Working with the set of original measurements on 1972 General Motors
Audit Test results those values of HC, CO and NOX emissions measurements which
exceeded the limits were determined. (it should be recalled here that for the
data subset under consideration, the NOX data was obtained by 7-mode tests.)
The failures were classified according to whether only the HC limit was exceeded,
only the CO limit, etc. For the second thru fifth quarter, as well as yearly
4-58
-------
totals the overall failure rates are shown in Table 4.4-1, the failure rates
by Gross Bmitter Category are shown in Table 4.4-2 and Table 4,4-3 summarizes
the averages and standard deviations of Gross Emitters (by category) for the
1972 General Motors data.
Production
Number
Number
Percentage
Quarter
Original Measurements
Gross Emitters
Gross Emitters
2
2504
118
4.71
3
3656
112
3.06
4
2878
92
3.20
5
246
37
15.04
TOTAL
9284
359
3.87
Table 4.4-1
Overall Gross Emitter Rates
1972 General Motors
Gross Emitter
Type
Number of
Gross Emitters
Category
Rate (%)
% of Al!
Gross Emitters
HC
201
2 .17
56
CO
35
0.38
10
NOX
104
1.12
29
HC f, CO
13
0.14
4
HC § NOX
5
0.05
1.5
CO 5 NOX
0
0.0
0.0
HC, O) f, NOX
1
0.01
0.5
TOTAL
359
3.87
100.0
Table 4.4-2
Gross Emitter Rates - by Category
1972 General Motors
4-59
-------
Gross Emitter
Hydrocarbon
Carbon Mor
loxide
NOX
Category
X
s
X
s
X
s
nrr
HC
4.52
1 .22
26.88
9.78
1.73
0.44
201
CO
2.10
0.58
85.03
62.44
1.93
0.57
35
NOX
1.86
0.56
22.90
6.85
3.70
0.45
104
HC $ CO
5.38
2.65
90.54
16.93
1 .76
0.30
13
HC f, NOX
4.44
0.38
28.0
8.49
5.46
3.99
5
CO $ NOX
--
--
--
--
--
--
0
HC, CO $ NOX
3.91
--
76 .0
--
41.n
--
1
Table 4.4-3
Average and Standard Deviations of Gross Emitters
(by category]
1972 General Motors
(prams per mile)
4-60
-------
5 ALTERNATE ASSEMBLY LIME TEST PROCEDURES AND THEIR IMPACT
ON EMISSIONS
5.1 Introduction
Having observed the magnitudes of emissions changes which may be
related to segments of the assembly line testing as currently employed in
California and considering the assessment of changes in emissions which might
occur if such procedures were employed on a nationwide basis, the next logical
step is to consider how and to what extent variations in the structure of an
assembly line testing program may be devised and what emissions changes might
be obtained from alternate approaches. This chapter focuses on that issue.
A number of potential alternate assembly line testing structures
are discussed in Section 5.2. Essentially these focus on possible variations
in some of the controllable characterisitcs of the currently employed
California procedure. Specifically, variations in (I) the acceptable audit
test failure rate, (II) level of hot 7-mode testing, (III) level of idle
testing, and (III) level of audit sampling will be considered. Further, a
modification in the audit sampling procedures will be discussed in an
attempt to consider how alternate sampling strategies, within the confines of
a fixed sampling rate, may offer some promise of substantial improvements in
the rectifying capabilities of the audit testing phase of assembly line
testing programs.
5.2 Methodology and Analyses
5.2.1 Considerations of Modified Failure Rate Standards
Vfe shall first consider the issue of varying the acceptable audit
test failure rate associated with an assembly line testing program design.
Recall that the current California procedure has specified a 10% failure rate
limit. It is instructive to consider what might happen if this failure
criterion were modified, say to 5%Jor 15%, etc. Suppose an assembly line
5-1
-------
program specifies an audit failure rate limit of p<,(not necessarily equal to
10%), and further assume that audit sampling occurs at the rate (3. If, for
a given engine family of a manufacturer, the actual failure rate prevailing
is p , then subsequent to audit sampling (and the concomitant adjustment to
passing levels of automobiles which failed), the rectified fraction of failing
automobiles will be,
r ' (5.2.1)
This relationship merely reflects the fact that the audit screening, to
the extend that it only affects the environment by removing detected failing
cars, misses a fraction (l-(3) of those autos which are failures to begin v.ith.
It follows that if one seeks to assure that, on the average, outgoing failure
rates of "audited" engine families satisfy the legislated limit,p0 , then
necessarily
p0>, p - pO-fT) (^2)
or alternately,
(B > lr~ ) ~ (^O (5.2.3)
Expression 5.2.3 suggests that for a given engine family with
incoming audit failure rate p, a minimum audit sampling rate (30is necessary
if the audit process is to screen enough failures from the incoming process
to satisfy the maximum failure rate po . Given audit test failure rates, by
engine category within a manufacturer, as are summarized in Chapter 3, one
can assess how well or poorly the current audit rates (i.e., (3 = 2%) screen
the input failure levels and consequently contribute to meeting the 10%
desired failure rate limit. Perhaps a more central issue, regarding
variations in the magnitude of pa , is how changes in its level would be
reacted to by the manufacturers in so far as modifying process averages and
variabilities so as to satisfy the limit on failure rates.
Appropriate data was clearly not available during this project to
adequately characterize the manner in which emission distributions would
"adapt" to varying failure rate limits; thus this facet of alternate assembly
line programs is not pursued further.
5-2
-------
5.2.2 Consideration of Alternate 7-frode and Idle Test Rates
A second aspect of alternate assembly line program design is the
possible variation of 7-mode and/or idle test rates. Presumably, in an
attempt to decrease emission levels by manipulating testing rates one would
seek to increase test rates for those test types fox which the rate of change
in emissions is most negative. The analyses of 7-mode, idle, or 7-mode and idle
test effects on audit test results yielded very mixed results, and consequently,
it is not at all clear that increasing levels of 7-mode testing, or levels of
idle testing would yield a discernable improvement in emissions levels.
Since combinations of both 7-mode and idle tests on a single auto-
mobile appear to interact with one another and, more often than not (as
indicated, based on the preliminary analyses of Chapter 4), combine to in-
crease emissions levels, we shall consider only testing levels for which
7-rode testing and idle testing are mutually exclusive. Further, the results
from idle tests alone suggest that the presence of idle tests appear to in-
crease audit test emissions levels (i.e., 197J data for idle tested cars
suggest increased levels of HC, CO and NOx in contrast to no Idle testing).
These factors suggest that the most likely candidate for reducing emissions,
based on the analyses of data from this study, is to increase the testing
rates on hot 7-mode tests. It was observed in Section4.2.1 that, as a
function of 7-node testing rate, o( , the average emissions level can be
expressed as
£ = PM (5.2.4)
From the 1972 model year data base Ford and Chrysler analyses reflected no
appreciable changes in emissions levels (i.e., HC and CO) resulting from the
presence of 7-mode tests, and while the General Motors Data yielded a
significant trend, this was somewhat suspect in that the 7-mode tested cars
were from a restricted set of engine classes and assembly plants raising
some question of whether or not this trend is really attributable to the
7-mode test or to some other correlated factor. The 1973 data base yields
more promising results. While HC and NOx are seen to increase with 7-mode
5-3
-------
testing, CO decreases substantially. The question then is how to "balance"
increases in HC and NOx against decreases in CO. Clearly, some value
judgments will be needed to finalize this.
Employing expression (5.2.4) and denoting the total emissions
changes after summing over manufacturers, one obtains:
I Hydrocarbons
<£CHC>= JL F7 M -+ (5.2.5)
where the superscript denotes the manufacturer. This
reduces to:
= [121440 + O.C\]
+ [102,896 + 34880(1
~ [4316 - 257o(]
= 228652 + 3231 tons per year (5.2.6a)
II Carbon Monoxide
£co
= [1,920,600 - 242,000 ]
+ [1,165,020 + 0. o(]
+ [51,322 + O.ot]
£ - 3,136,942 - 242,000 (£ tons per year (5.2.6b)
III Oxides of Nitrogen
A**6*
£ = [149,820 + 3,30(&]
+ [91,996 + 15,260d]
+ [4752 + 0^]
= 246,568 + 18,560 0( tons per year (5.2.6c)
5-4
-------
In summary then, based upon available data, it would appear that
increasing 7-node test rates would modify HC, CO and NOx emissions levels as
indicated in expressions 5.2.6a-c. Emissions level changes at a representative
set of test rates (o() ^or 7-mode testing are presented in Table 5.2.1.
7 Mode Test
Rate
Nationwide
Emission Level Changes (tons per year) |
HC
CO
NOx '
.
25%
+ 808
- 60,500
t
+ 4,640 j
50%
+ 1616
-121,000
+ 9,280 i
75%
+ 2424
-181,500
+13,920 j
100%
+3231
-242,000 j
1
+18,560 |
1
Table 5,2-1
7-mode Test Related Emissions Changes
5-5
-------
5.2.3 Consideration of Alternate Audit Sampling Procedures
A third area of possible variation in the assembly line testing
procedure is centered on the audit testing segment. On the one hand, the
audit rate, , might be varied and on the other hand the audit rate might be
held fixed, but the manner in which the total sample of autos to be audited
may be altered.
Regarding the former approach, an assessment of the emissions level
changes is relatively straight forward. From expression (4.3.2) we have
tons per year decrease in NOx
5-6
-------
Emission level changes at a representative set o£ audit rates are
presented in Table 5.2-2.
Audit
Rate (0)
Emission
Level Changes (tons per year)
HC
CO
NOx
T
1%
- 60.56
- 1452.71
- 5ft. 01 f
2%
-121.12
- 2905.41
- 96.02 S
I
3%
-181.67
- 4358.12 1
-144.03 1
5%
-302.79
- 7263.56
|
-240.05
10%
. ... ... .... J
-605.58
-14527.05
-480,10
Table 5.2-2
Audit Rate Related Emission Changes
Recognizing that the data of Table 5.2-2 represents estimates based
only on General Motors and Chrysler data, it may be helpful to extrapolate to
what these levels might be if Ford data were available. Given that General
motors anH Oirvsler account for approximately 66% of the total production of
the "Bip Three" manufacturers and assuming that Ford experience on audit
tests would "approximate" the CM/Chrysler results, we can divide each entry
of Table 5.2-2 to estimate the emissions chanpes, on a nationwide basis,
when extended (under the limitations of the above assumptions) to all three
manufacturers. This result is summarized in Table 5.2-3.
5-7
-------
Audit
Rate
Fmission Level Changes
(tons per year)
HC
CO
NOX
1%
i :
'O !
H—»
- 2201.08
7 2.74
2\
-183.52
- 4402.14
-145.45
3%
-275.26
- 660 3.21
-218.23
5%
-4 5R.7"7
-11005.39
-363.71
10%
-917 .55
-22010.68
-727 .42
T ab 1 e 5.2-3
Extrapolated Audit Rate Related
Emission Chan pes
Thus far we have discussed anticipated emission level changes if
the rate of audi t sampling (p) is varied, but the manner of sampling is
retained as currently done. For example, at a 2% audit rate General Motors
would audit test approximately 100,000 automobiles on a nationwide basis and
these are tested on a pro-rated basis splitting them proportionally in each
production quarter of the year (i.e. 17%, 25%, 25%, 25% and 8%) and within
each quarter auditing 2% of each engine categories' production. The emissions
changes shown in Tables 5.2-2 and 5,2-3 are premised on this sampling
procedure.
We will now focus on an alternate approach to how one might
"allocate" the 100,000 automobiles to be auditedt thereby retaining the
equivalent of 2% sampling, yet hoping to improve upon the emissions
reduction attained. Consider a manufacturer with, say, different engine
categories and let flij denote the number of automobiles produced in
engine category i during production quarter j. Further, let Sij represent
the number of automobiles "audit tested" by the manufacturer from engine
category i in production quarter j.
5-8
-------
Define
= Suj.yn.tj j = 1,2 ,...5
- 1,23...c
(5.2.8)
It is seen that 0 ij represents the -fraction of production in
category i and quarter j which is audit tested. As assembly line testing
is currently applied it is true that 0ij « 0.02 - (3 for all i, jj that
is, 2% of production is audited on a category/quarter basis. Since
. S c
N - Z"Z 0* equals the yearly total production for this manufacturer and
since 3ij a H ij for all £, i ir the current scheme, It is clear that
S 2-
S £
--- 3 s: s; n;
J-• L-t
-------
(5.2.11)
and summing over all engine category and production quarter combinations
5 £ ^
«¦ f.lA*
-FMN|ER1nJ(et;-e?;)
(5.2.12)
What one would ideally seek is to define ^3 tj (or equivalently ^)L
-------
and let {<5^ : l,2j- 'W }
be defined so that
>6^
(5.2.15)
That is to say, ($_ jj) is the engine category cell which contains the
maximum <5ij value; cell (l^, ^2) contains the second largest value, etc.
The (5 ij term can be interpreted as an exptected or average change in
emissions rate (in prams per mile) resulting from detecting and correcting
— Cohf __ (Zi .
failing cars. The component (
-------
One can demonstrate that for the emissions type corresponding to
the given set i J for a particular production year allocating the
audit sampling as described in (5.2.16) maximizes the total emissions change
as in (5.2.13).
Let us illustrate with an example. Consider the 1972 General
Motors data. Tables 5.2-5 through 5.2-7 contain, for HC, CD and NOX
emissions respectively, the values of <5^- ~ ) indexed hy
engine category (i) and production quarter (_i). We shall consider an
"allocation" of audit inspected cars at a rate of (3= 2%. Estimating
1972 General Motors nationwide production at 5,000,000 automobiles this
yields 100,000 automobiles to be audit tested. We shall select engine
category/production quarter combinations on the basis of the hydrocarbon
values of
-------
Division
Fngine
Category
Class
Code
(i)
Production Quarter (j)
3
4
5
Chevrolet
140-1,2
1
.2979
.1102
.1001
.8250
250-1
2
.0275
.5494
.0567
0.0
350-2
3
.0529
.0469
.0789
.0695
350-4
4
.0852
.0432
.1353
.4675
400-2
5
.0292
.0473
.0442
0.0
Pontiac
350-2
6
.1193
.0094
.0426
.7200
400-2
7
.0076
.0100
.0S89
0.0
400-4
8
.0628
.2785
.1672
.1918
455-4
9
.1287
.1184
.000
.00
Oldsmobile
350-2
10
.0112
.0675
.4020
350-4
11
.0324
.0603
.0413
.1000
455-4
12
.0211
.0702
»1365
Buick
350-2
13
.0172
.095u
.0287
350-4
14
.0665
.0797
.0070
455-4
15
.0319
.0419
. 0489
.3097
Cadillac
472-2
16
.0355
.0153
.0611
.400
500-4
17
.0764
-.0560
.0020
Table 5.2-5
Hydrocarbon <5 ij Values
1972 Cteneral Motors
5-13
-------
Division
F.n pine
CI ass
Production Ou
jirtcr (i'
Category
Code
(i)
->
im
3
4
5
Chevrolet
140-1,2
1
1 .0651
.4204
.4421
.8250
250-1
2
1 .7500
13.216
1.2077
.0
350-2
3
.7614
.9411
1.9666
2. 123
350-4
4
1 .2837
.9204
1.6135
2.7325
400-2
5
.7779
.7175
1.1477
.0
Pontiac
350-2
6
1 .97
1 .339
.5472
4.0
400-2
7
.7600
. 3600
.6787
.0
400-4
8
3.3488
4.0460
1.6724
5.3778
455-4
9
.4620
.9065
.0
-.25
Oldsmobile
350-2
10
.084
.4590
1 3.002
350-4
11
,4082
.2385
I .2925
-.6000
455-4
12
. 7508
6.588
3.4125
Buick
350-2
13
.2054
1.2267
.0615
350-4
14
.1330
.6154
-.0528
455-4
15
.3440
1.4904
2.112
1.7749
Cadil1ac
472-2
16
1.3065
1.1590
3.44 37
20.0
500-4
17
.0
.2080
.120
Table 5.2-6
r*-
Carbon Monoxide O ij Values
1972 General Motors
5-14
-------
Division
Engine
Category
Class
Code
(i)
Production Quarter (j)
1
3
4
5
Chevrolet
140-1,2
1
.0605
.0551
.0835
.6171
250-1
2
- .0150
.4134
.0164
.0
350-2
3
.0033
.0051
.0218
.0521
350-4
4
.0039
.0027
.0108
+ .025
400-2
5
.0082
.0046
.0139
.2529
Pontiac
350-2
6
.0141
.0094
.0198
.0200
400-2
7
.0076
.0120
.0110
.0
400-4
8
.0157
.0643
-.0113
-0.0056
455-4
9
.0132
.0444
.0
.60
Oldsmobile
350-2
10
-.0028
.0027
-.0780
350-4
11
.0180
.0158
-.0055
.0600
455-4
12
.0277
.0275
.3528
Buick
350-2
13
.0080
.0270
.0390
350-4
14
.0350
.0561
.0458
455-4
15
.0794
.0064
.0530
.0966
Cadi 11ac
472-2
16
.0275
.0098
-. 01 34
.20
500-4
17
.1793
.0280
.0
Table 5.2-7
NOX
-------
75625 Chevrolet 250-J's to be produced in quarter 3 since we are constrained
to a total of 100,000 autos to be audited.
*
F.mploying expression (5.2.13) with the allocations i j of
Table 5.2-4 we obtain the estimated nationwide emissions reduction associated
with this audit nrogram, as shown below:
Emission Type
Reduction in
Tons per vear
HC
CO
NOX
874
I0.60S
535
Table 5.2-8
Emission Reduction From Optimal
2% Audit Sampling Allocation
1972 General Motors
Two points are noteworthy here: First, since 1972 test procedures
employed seven node testing for M0X, the emission reductions noted in Table
5.2-8 for NOX correspond to 7M results whereas, the HC and CD results are
from C.V.S. tests. The second point is more critical and centers on the
interpretation of the estimates of Table 5.2-8. As described earlier in
this section, these emissions changes represent upper bounds on the maximum
attainable emissions reduction from 2% audit testing during 1972. Mad one,
at the beginning of the 1972 model year, made the choice to allocate the
100,000 audit tests as show in Table 5.2-4, the reductions of Table
5-2-8 would have been attained. Most likely, such an allocation would not
be selected a priori. The utility, however of this computation is that it
demonstrates a rather dramatic potential for improvement from audit testing
5-16
-------
in contrast to the currently employed scheme where the 2% sample is pro-rated
across all engine categories and production quarters. This is seen more
clearly from a comparison of the estimated nationwide reductions from the
current procedure (See Table 4.5-1), and is summarized below:
Emissions Reductions
HC
CO
Current "2V
Audit Sampling
122 tons/yr.
39 tons/yr.
Optional 2%
Audit Sampling
874 tons/yr.
10608 tons/yr.
Table E .2-9
Comparative Audit Samplinp Emission Levels
The 1973 data base available facilitated a somewhat more
satisfactory application of this technique. Such data was available from
General Motors and Chrysler Corporation. Further, C.V.S. tests yielded
riGX as well as HC and CD test measurements, hence estimates for all three
emittent types were possible here. Without including all the tabular
data (i.e. -
-------
Since these estimated maximum attainable reductions were based on
data from the first 2 production quarters (i.e. - on approximately 42% of
the total years production), one can project this data to a full year's impact-
assuming that similar attainable improvements might be attainable in the
remainder of the year. Further, since Ford data was not available here,
these numbers were based on 66% of nationwide prodution (i.e. - H.M. and
Chrysler constitute roughly 66% of the "Big Three" total production); one
might extrapolate from this data assuming that the Ford experience, if
available, would be comparable to the G.M. and Chrysler results. These
steps, combined, would result in dividing the entries of Table 5.2-in bv
the constant (0.42) (0.66) and yield estimated yearly total changes for
G.M., Ford and Chrysler. These estimates are shown below:
Fmission
Extrapolated Nationwide
Fmission Reduction
(tons/yr.)
HC
461
CO
8084
NOX
711
Table 5.2-11
F: xtranol nted Emission Reduction Levels
5-18
-------
6. ASSESSMENT OF EMISSIONS REDUCTIONS ATTAINABLE WTTH CATALYST
CONVERTER EMISSION CONTROL SYSTEMS PLANNED FOR THE 1976
MODEL YEAR
6.1 Introduction
This chapter focuses on a consideration of what changes may result
from some form of assembly line testing program applied, on a nationwide basis,
to automobiles which will have the 1976 catalytic converter emission control
systems as they are currently anticipated.
6.2 Methodology and Results
The catalytic converter emission control systems to be considered
in this study will be limited to those planned for the 1976 model year
automobiles, and the corresponding standards (as of the recent E.P.A.
ruling which granted a one year delay of the 1975 emissions standards).
The 1976 (and later) Federal Standards for emissions which will be
employed in this analysis are shown below:
Emission Type
Hydrocarbon (H.C.)
Carbon Monoxide (C.O.)
Oxides of Nitrogen
(NOX)
Federal Standard
0.41 grams per mile
3,4 grams per mile
0.40 grams per mile
In attempting to draw upon the empirical evidence available from
the 1972, 1973 data base on emissions experience in California and infer from
this, and supplementary information, the potential emissions levels from a
"catalytic system", a number of key issues and assumptions must be considered.
6-1
-------
i.3 It is assumed for this analysis, that the 1976 engine/
catalytic converter systems will be typical of a 1973
type engine design feeding into a catalytic converter.
In light of this, it appears reasonable to utilize the
characterization of emissions levels obtained from the
1973 data base available to this project as descriptive
of the "inputs" to the catalytic convcrtors which will be
on 1976 automobiles. Consequently the average emission
levels, variabilities, patterns of distribution of HC,
CO and NOX emissions as derived from the 1973 California
data base will be employed.
ii-3 Considering the operational characteristics it appears
reasonable to argue that any form of "hot start test"
will be essentially useless in assessing emissions
reduction attributable to the presence of a catalytic
converter system. The very nature of a catalytic
converter is such that when hot it is extremely efficient
and removes the vast majority of emittants contained in
the input stream. The emissions which, it is anticipated,
will pass through the convertor will do so during the
"cold phase". Because of this characteristic, this
analysis will not consider assessing any changes in
emissions which may result under 'hot 7-mode tests"
or "idle tests" since these would really be hot start
tests. Consequently, of the alternate tests considered
in earlier segments of this report, there remains the
various audit tests applied. The remaining portions of
this chapter will concentrate on the audit tests.
iii.) It is assumed that the catalytic converters to be employed
on 1976 model year automobiles are "fixed percentage"
reducers of emissions. That is to say, a functiona
catalytic converter appears to be capable of removing
a constant percentage of emissions input to the
converter. Further this property will be assumed to
hold over all ranges of emissions levels input to the
converter.
iv.3 Fresh catalysts on cars are assumed fully operable and effective.
In light of assumptions ((i3 - ( iv J) discussed above, there are
essentially two steps which must be taken to assess the emissions reductions
associated with catalytic systems. The first is to estimate the "reduction
rate" at which anticipated catalytic converters will remove emissions from a
typical (i.e. - 1973 level) input stream; and secondly to apply these
-------
"reduction rates" to the 1973 levels estimated earlier in this report. Let
us consider these steps presently:
Catalytic Converter System Reduction Factors
Our task here is to determine a "reduction rate" for each emission
type (i.e. - HC, CO and W)X) which will describe the percentage reduction,
emr< an input stream, of emissions going through a catalytic converter.
Since test procedures have been modified for 1976 C.V.S. tests in
contrast to 1973 C.V.S. tests, a test procedure scaling factor, , is
necessary due to the different weighting of the "hot hag" and "cold bag"
emissions components in the 1973 versus 1976 C.V.S, procedures. F.mpirical
data and analyses supplied to F.P.A. by various manufacturers have facilitated
the determination of a multiplicative relationship between the emissions level
which would be obtained by testing the same auto with the "1973 procedure"
as compared to testing it with the "1976 procedure". We shall consider this
to be
" 1976 Procedure Emissions Level Measurement"
ip x
"1973 Procedure Emissions Level Measurement"
where (£) varies according as we are measuring HC, 00 and NOX. The
specific scaling factor values supplied by F.P.A. and to be employed in this
study are as follows:
Emission Type
Scaling Factor
M.C.
0.7
C.O
0.6
N.O.X.
1.0
Table 6.2-1
Emissions Scaling Factors
6-3
-------
Applying the scaling factor to the 1973 emission standard, one
obtains the "Comparable Standard" which should apply to a 1973 automobile
C.V.S. tested under the 1976 test procedure. These values are seen in
Table 6.2-2:
Emission Type
1973 California Standar
(1973 Test Procedure)
3 Scaling
Factor
1973 Standard
(19 76 Test Procedure
HC
GO
NOX
3.2 gm/mile
39 gm/mile
3.0 gm/mile
0.7
0.6
1.0
2.24 gm/mile
27.3 gm/mile
3.0 gm/mile
Table 6.2-2
1973 Emissions Standards:
Scaled to 1976 Test Procedure
In order to determine the "reduction factor" for a specific emission
type, under the "constant reduction assumption", we compute the ratio of
the "1976 standard for catalyst systems" to the "1973 standard adjusted to
the 1976 test procedure". This is shown in Table 6.2-3:
Emission Type
1976 Standard
(Catalyst System)
1973 Standard
(1976 test procedure)
Catalytic Con-
verters
Reduction
Factor ( P )
HC
CO
NOX
0.41 gm/mi.
3.40 gm/mi.
0.40 gm/mi.
2.24 gm/mile
27.3 gm/mile
3.0 gm/mile
0.183
0.145
0.133
Table 6.2-3
Reduction Factors For Catalytic Converters
6-4
-------
The rationale underlying the development of these "Reduction Factors"
contained in the rightmost column of Table 6.2-3 is that the 1973 and 1976
standards are feasibly attainable levels, and, piven that the 1976 systems
are essentially 1973 inputs to a catalytic converter, the degree to which
1976 standards are lower than 1973 standards is attributable to the %
reduction of the catalytic converter output stream relative to the input
stream. The results in Table 6.2-3 imply that HC emissions leaving the
converter are approximately 18.3% of the input levels; C.O. output is 14.5%
of CO input, and NOX output is 13.3% of input.
Emission Levels Leaving Catalytic Converters: Audit Tests
It will be recalled that according to the development of Chapter 4,
the emissions level change resulting from an audit test applied to 1973
systems at the rate ^ is:
£ =rMNPP(sf°- Sft>) C6.21)
Here, subscripts denoting manufacturer, engine category, emissions type, etc.
have been delated to facilitate discussion; however, the definitions of
terms contained in expression 6.2.1 are as before. Earlier discussion has
implied that if:
then
(O s emission level measured on a test of
a 1973 auto, or equivalently, at
input to a 1976 catalytic converter
£ = emission level measured at the output
of a 1976 catalytic converter
pe (6-2-2)
where Q = Catalytic Converter Reduction Factor, as
\ developed in Table 6.2-3
6-5
-------
Itlppears to be plausible that -
?= P{e*
= P{pe*
- Pi - ^ £ ¦.in* f
(6.2.3)
that is to say, under the "constant reduction rate assumption" one would
expect the failure rate to remain approximately constant. Further, since
Qg, it follows -readily that
*=> ~ (6.2.4)
That is, the average emissions upon first audit testing of catalyst systems
(i.e. - and averape emissions of failing cars which have been
(0 "\
rectified (i.e. - ) are related to their empirically estimable counter-
parts g (03'f and g (2) .
Substituting the results of 6.2.3 and 6.2.4 into 6,2.1, one obtains
E = PMN^Tr(e°-e
-------
prevail under a catalyst system
Audit
Rate ( ^ ]
Emission Level Chq
nges (tons per year)
HC
CO
NOX
1 %
2 %
3 *
5 %
10 %
- 11.08
- 22.16
- 33.25
- 55.41
-110.82
- 210.64
- 421.28
• 631.93
-1053.22
-2106.<12
- 6.39
-12.77
-19.16
-31.93
-63.85
Table 6.2-4
Audit Rate Related Fmission Chnnpes
with Catalytic Converter F.mission Systems
(Nationwide for GM ,md Chrysler)
Recalling thnt these levels were based on General Motors and
Chrysler data only, if we were to extrapolate to all three manufacturers
by pro-ratinp accordinp to nationwide production levels (i.e. - HM and
Chrysler constituting 66% of the "Bijt Three" production) we obtain the
estimates of Table 6.2-5.
Audit
Rate ( )
{•mission Level Changes (tons per year)
HC
CO
MOX
1 %
- 16.79
- 319.36
- 9.67
2 %
- 33.58
- 638.31
-19.34
3 %
- 50.37
- 957.47
-29.02
5 %
- 83.95
-1595.78
-48.37
10 %
-167.91
-3191.55
-96.75
Table 6.2-5
Audit Rate Related Emission Changes
with Catalytic Converter System
(extrapolated to all three manufacturers> nationwide)
6-7
-------
7. COST ANALYSES OF NATIONWIDE ASSEMBLY LINE TEST PROGRAMS
7.1 Introduction
Assessing the costs, on a nationwide basis, of alternate configura-
tions of assembly line test programs is, at best, a difficult task for at
least two reasons. First, there is asparcityof cost data available and that
which is available is quite variable in terms of estimates of performing a
given test. Secondly, there are many ways in which one can characterize and/
or interpret costs of a testing program.
Regarding the first issue, cost data was obtained from a variety of
sources, largely estimates of unit costs of 7-mode, idle and audit tests
by various personnel within E.P.A., various manufacturers, the California
Air Resources Board, etc. The reported costs were widely spread but the
range of costs ultimately developed were as follows:
1. 7-mode test Unit Costs: $8.00 - $20.00 per test
2. Idle tests: $.32 - $1.20 per test
3. Audit tests: $68.00 - $350.00 per test
For most of the analyses performed in this chapter average costs of $16.00,
$1.00 and $120.00 per test for 7-mode, idle and audit tests, respectively,
are employed in subsequent computation.
Regarding the second issue, two basic approaches were adopted in
assessing costs. On the one hand, for a given assembly line test configura-
tion one can determine an average test cost per automobile and extrapolate
this to nationwide production levels to develop a single total yearly cost.
Additionally, one can consider the cost/effectiveness of a test configura-
tion in the sense of assessing the estimated nationwide yearly cost per
unit (i.e. tons per year) change in emissions level. Both such approaches
were undertaken.
7-1
-------
7.2 Nationwide Cost of California Assembly Line Test Procedures
Given a unit cost for each 7-mode, idle and audit test, the
determination of the nationwide cost of applying the California procedures
is straight forward. Let us assume the following average unit costs for
testing.
i. 7 Mode Test - $16.00 per test
ii. Idle Test - $1.00 per test
iii. Audit Test - $120.00 per test
On a nationwide basis the total cost would be
Total Cost = (9,800,000 cars/yr) x
$16.00(0.25) + $1,00(0.75) + $120.00(0.02)
= (9,800,000 cars/year) ($7.15)
Total Cost = $70,070,000.00 per year
This amounts to an average of $7.15 per car produced for testing costs.
7.3 Nationwide Cost of Alternate Assembly Line Test Procedures
The alternate asseirbly line test procedures considered were, with
the exception of audit test variations on sampling strategies, of a nature
such that the nationwide costs are linear functions of the sampling or
testing rates. Let us define
i) (2s unit cost of the J.-th test type (^.= 1,2,3)
ii) cXa = sampling or test rate of l -th test type
* C / = 1,2,3)
iii) 7? = nationwide automobile production per year
The total cost of a specified testing program is readily specified as
Total Cost Per Year = T") x ( C(| Ct ¦+ 0^ Cz + ObC3 )
where (cXidi-t-cXtC*. + CtfsCj ) interpreted as the average test cost per car.
7-2
-------
If the testing rates or unit costs of tests are varied one will obtain a
range of total costs varying linearly in these parameters. The current
California procedures represent one special case of this situation, namely
OC, = 0.25, 04= 0.75 and (X3 =0.02 for 7 Mode, Idle and Audit tests
respecti vely.
In Chapter 5, a number of alternates were considered, these having
test rates of the form
[ Ofi > O j CX2. — CD y 0(3 > O 1
that is, some level of 7-mode testing (0(1 > O ) and audit testing (0(3 > O )
but no idle testing (i.e. - 0(z = O 3. For the representative ranges of
7-mode test rates and audit test rates incorporated in Tables 5.2-1 and 5.2-2,
the average testing costs per car using the representative costs of 316.00
per 7-mode test and $120 per audit test are as follows:
7 Mode
Test Rate (%
0
Aud^t Test Rfrte (%)
1
10
(T
25
50
75
100
T72TT
5.20
9.20
13.20
17.20
3.60
7.60
11.60
15.60
19.60
6.00
10.00
14.00
18.00
22.00
12.00
16.00
20.00
24.00
28.00
0.00
4.00
8.00
2.00
16.00
2.40
6.40
10.40
14.40
18.40
Table 7.2-1
Average Assembly Line Test Costs Per Car
For Various 7 Mode/Audit Test Rates (Dollars)
Assuming a nationwide auto production of 9,800,000 automobiles
"sr year, the corresponding total costs per year on a nationwide basis would
be as shown below, in Table 7.2-2.
7-3
-------
7 Mode
Test Rate (%}
Audit Test Rate (%)
0
1
~>
3
5
10
0
0
12
24
35
59
11 8
25
39
si
63
74
98
157
50
78
90
102
114
137
196
75
118
129
141
153
176
235
100
157
169
1 80
192
215
274
Table 7.2-2
Nationwide Costs of Assembly Line Test Programs
For Various 7 Mode/Audit Test Rates
(in millions of dollars per year)
If the nationwide testing cost for a particular combination of
7-jrode and audit test rates is divided by the corresponding magnitude of
emission level chanpes} one obtains some insight into the estimated cost
per ton of emission change for a particular test configuration. These
results are shown in Table 7.2-3 a through c. In reviewing these tables
one should be careful to recall that, based on the data analyses presented
earlier in this report, the combined effect of 7-mode and audit tests
together are estimated to increase HC and NOX levels while decre; sing
CO levels. Consequently, these tables reflect the ''cost per ton per year
change" in emissions levels where the change is an increase for Iff and
NOX but a decrease in CO. Tt is clear that in this situation a judgement
will need to be made to determine if the improvement in the CO situation
counters the degradation in the JIC and MOX situation for a given test level.
7-4
-------
7 Mode
Test Rate (%)
Audi
t Test R
ate (%)
1
2
3
5
10
25
50
75
100
71,229
59,055
55,317
53,830
100,962
71,229
62,046
59,074
138,837
85,011
71,196
¦64,953
280,802
118,410
89,567
77,561
-1,440,367
280,802
156,042
118,461
Table 7.2-3a
Cost of Hydrocarbon Increases
(Dollars per Ton per Year)
7 Mode
Test Rate (%)
Audit
Test Rate (%)
1
2
3
5
10
25
813
971
1103
1370
1903
50
731
813
893
1038
1371
75
702
758
813
914
1155
100
692
731
772
850
1038
Table 7.2-3b
Cost of Carbon Monoxide Decreases
(Dollars per Ton per Year)
7-5
-------
7 Mode
Test Rate (%)
Audit Test Rate (%)
1
2
3
5
10
25
11 ,167
14,016
16,735
22,919
4 0, 123
50
1,77?
11,166
I2.5RO
15,366
22,916
75
9,316
10,736
11 ,16*
12,983
I-?,??! 2
100
9,142
9,7 75
H ,8J6
15,365
Table 7.2-3c
Cost of NOX Increases
(Dollars per Ton per Year)
7-6
-------
8.
REFERENCES
1. CALIFORNIA ASSEMBLY-LINE TEST PROCEDURES, State of California
Air Resources Board, Adopted September 16, 1970, Amended February 17, 1971
2. CALIFORNIA ASSEMBLY LINE TEST PROCEDURES FOR 1973 AND SUBSEQUENT MODEL
LIGHT DUTY VEHICLES, State of California, Air Resources Board; December IS,
1971 Amended December 20, 1972
3. CALIFORNIA EXHAUST EMISSION STANDARDS AND TEST PROCEDURES FOR 1973 THROUGH
1976 MODELS GASOLINE POWERED MOTOR VEHICLES UNDER 6001 POUNDS GROSS
VEHICLE WEIGHT, State of California, Air Resources Board, Adopted
September 15, 1971 Amended December 18, 1972
4. PUBLIC HEARING ON CHANGES TO ASSEMBLY-LINE TEST PROCEDURES FOR 1973 MODEL
YEAR LIGHT DUTY VEHICLE, State of California, Air Resources Board,
October 17, 1972
5. REPORT TO LEGISLATURE: ASSEMBLY LINE TESTING, State of California, Air
Resources Board May 10, 1972
6. MOTOR VEHICLE ASSEMBLY LINE TESTING, Statistics Research Division
Research Triangle Institute; August, 1970
7. REPORT ON ASSEMBLY LINE EMISSION TESTING OF MOTOR VEHICLES, National
Air Pollution Control Administration, Bureau of Abatement and Control,
April, 1970
8. Final Report: Status of Industry Progress Towards Achievement of
The 1975 Federal Emission Standards For Light-Duty Vehicles; Urban Systems
Division, The Aerospace Corporation, El Segando, California July, 1972
9. AUTOMOBILE EMISSION CONTROL - THE STATE OF THE ART AS OF DECEMBER 1972;
Environmental Protection Agency, February» 1973
10. REPORT BY THE COMMITTEE ON MOTOR VEHICLE EMISSIONS, National Academy of
Sciences, February, 1973
8-1
-------
APPENDIX A
LISTING OF SOURCE DOCUMENTS FOR DATA USED
IN THIS STUDY
GENERAL MOTORS CORPORATION
Reports 1 through 17 were all supplied to the indicated agencies by:
Environmental Activities Staff
General Motors Corporation
General Motors Technical Center
Warren, Michigan 48090
1. Report of Assembly Line or Pre-Delivery Testing of Motor Vehicle
Exhaust Emissions. Period Covered October 1, 1971 Through
December 31, 1971.
To: California Air Resources Board
January 28, 1972
2. Report of Assembly Line or Pre-Delivery Testing of Light Duty
Motor Vehicle Exhaust Emissions. Period Covered January 1, 1972
Through March 31, 1972.
To: California Air Resources Board
May 3, 1972
3. Revisions to (2) above, dated July 7, 1972.
4. Report of Assembly Line or Pre-Delivery Testing of Light Duty
Motor Vehicle Exhaust Emissions. Third Quarter April 1,
1972 Through June 30, 1972.
To: California Air Resources Board
September 14, 1972
5. Report of Assembly Line or Pre-Delivery Testing of Light Duty
Motor Vehicle Exhaust Emissions. Fourth Quarter July 1, 1972
Through September 30, 1972.
To: California Air Resources Board
November 16, 1972
6. Report of Assembly Line or Pre-Delivery Testing of Light Duty
Motor Vehicle Exhaust Emissions. Period Covered Year to Date
September 1, 1971 Through March 31, 1972.
To: California Air Resources Board
July 10, 1972
A-l
-------
7. Report of 1973 Assembly Line or Pre-Delivery Testing of Light Duty
Motor Vehicle Exhaust Emissions: Start of Production Through
September 30, 1972.
To: California Air Resources Board
November 16, 1972
8. Report of 1973 Assembly Line or Pre-Delivery Testing of Light Duty
Motor Vehicle Exhaust Emissions for the State of California:
Fourth Calendar Quarter October 1 - December 31, 1972.
To: California Air Resources Board
February 1, 1973
9. Report of 1973 Assembly Line Pre-Delivery Testing of Light Duty
Motor Vehicle Exhaust Emissions for the State of California:
Year to Date September 1 - December 31, 1972.
To: California Air Resources Board
February 6, 1973
10. Report of 1972 Light Duty Vehicle Production: 1972 Model Year.
To: California Air Resources Board
November 16, 1972
11. Report of 1972 Model Production: Period Covered Start of 1972
Production Through September 30, 1971.
To: Environmental Protection Agency
October 27, 1971
12. Report of 1972 Model Production: Period Covered October 1, 1971
Through December 31, 1971.
To: Environmental Protection Agency
January 28, 1972
13. Report of 1972 Light Duty Model Production: Period Covered January
1, 1972 Through March 31, 1972.
To: Environmental Protection Agency
July 24, 1972
14. Report of Light Duty Model Production: Start of Production Through
June 30, 1972.
To: Environmental Protection Agency
August 31, 1972
15. Report of 1972 Light Duty Model Production: 1972 Model Year.
To: Environmental Protection Agency
November 16, 1972
16. Report of 1973 Light Duty Model Production: Start of Production
Through September 30, 1972.
To: Environmental Protection Agency
November 16, 1972
A-2
-------
17. Report of 1973 Light Duty Model National Production: Start of
Production Through December 31, 1972.
To: Environmental Protection Agency
February 1, 1973
FORD MOTOR COMPANY
Reports 18 through 29 were supplied by:
Automotive Emissions Office, Environmental and
Safety Engineering Staff
Ford Motor Company
The American Road
Dearborn, Michigan 48121
18. Second Quarterly 1972 California Assembly Line Testing Report;
January 25, 1972.
19. Third Quarterly 1972 California Assembly Line Testing Report;
April 25, 1972.
20. Fourth Quarterly 1972 California Assembly Line Testing Report;
Undated.
21. Final Quarterly 1972 California Assembly Line Testing Report;
October 24, 1972.
22. First Quarterly 1973 California Assembly Line Testing Report;
October 24, 1972.
23. Second Quarterly 1973 California Assembly Line Testing Report;
January 1973.
24. First 1972 Quarterly Report of the Number of Light Duty Vehicles
and Heavy Duty Engines Built for Sale in the United States;
November 30, 1971.
25. Second 1972 Quarterly Report of the Number of Light Duty Vehicles
and Heavy Duty Engines Built for Sale in the United States;
January 31, 1971.
26. Third 1972 Quarterly Report of the Number of Light Duty Vehicles
and Heavy Duty Engines Built for Sale in the United States;
April 28, 1972.
27. Fourth 1972 Quarterly Report of the Number of Light Duty Vehicles
and Heavy Duty Engines Built for Sale in the United States;
September 21, 1972.
A-3
-------
28. 1972 Model Year Nationwide Production Report; March 13, 1973,
29. Production Reports: 1973 Model Light Duty Vehicles Produced for
Sale in the United States; March 12, 1973.
CHRYSLER CORPORATION
Reports 30 through 38 were supplied by the:
Engineering and Research Office
Chrysler Corporation
Detroit, Michigan
30. Chrysler Corporation 1972 First Quarterly Report
California Assembly Line Test Report; October 29, 1971.
31. Chrysler Corporation 1972 Second Quarterly Report
California Assembly Line Test Report; January 26, 1972.
32. Chrysler Corporation 1972 Third Quarterly Report
California Assembly Line Test Report; April 26, 1972.
33. Chrysler Corporation 1972 Fourth Quarterly Report
California Assembly Line Test Report; August 9, 1972.
34. Chrysler Corporation 1972 Final Quarterly Report
California Assembly Line Test Report; September 29, 1972.
35. Chrysler Corporation 1973 First Quarter Report
California Assembly Line Test Report; November 30, 1972.
36. Chrysler Corporation 1973 Second Quarter Report
California Assembly Line Test Report; January 31, 1973.
37. 1973 Light Duty Vehicle and Heavy Duty Gasoline Engine Production
Report for Quarter Priding Septjejnber 30, 1972; November 1 , 1972.
38. 1973 Light Duty Veniele" aiidfflMSavyr-Du.tK /iqs.oline Engine Production
Report for Quarter Ending December :31 , 1972; February S, 1973.
A-4
------- |