EPA/AA/CTAB/TA/83-2
A STUDY OF THE RELATIONSHIP
BETWEEN EXHAUST EMISSIONS
AND FUEL ECONOMY
by
Jensen P. Cheng
Larry C. Landman
Robert D. Wagner
May, 1983
NOTICE
Technical Reports do not necessarily represent final EPA decisions or
positions. They are intended to present technical analysis of issues using
data which are currently available. The purpose in the release of such
reports is to facilitate the exchange of technical information and to inform
the public of technical developments which may form the basis for a final
EPA decision, position or regulatory action.
prepared by
U.S. Environmental Protection Agency
Office of Air, Noise and Radiation
Office of Mobile Sources
Emission Control Technology Division
Control Technology Assessment and Characterization Branch
2565 Plymouth Road
Ann Arbor, Michigan 48105
-------
EPA/AA/CTAB/TA/83-2
A STUDY OF THE RELATIONSHIP
BETWEEN EXHAUST EMISSIONS
AND FUEL ECONOMY
by
Jensen P. Cheng
Larry C. Landman
Robert D. Wagner
May, 1983
NOTICE
Technical Reports do not necessarily represent final EPA decisions or
positions. They are intended to present technical analysis of issues using
data which are currently available. The purpose in the release of such
reports is to facilitate the exchange of technical information and to inform
the public of technical developments which may form the basis for a final
EPA decision, position or regulatory action.
prepared by
U.S. Environmental Protection Agency
Office of Air, Noise and Radiation
Office of Mobile Sources
Emission Control Technology Division
Control Technology Assessment and Characterization Branch
2565 Plymouth Road
Ann Arbor, Michigan 48105
-------
CONTENTS
Page
I. Summary 1-1
II. Background II-l
III. Fleet Level Effects
A. Fuel Economy of 1974 and Newer Vehicles
Compared to Uncontrolled Vehicles III-l
B. Fuel Economy of 1975 and Newer Vehicles
Compared to Previous Year Vehicles III-7
C. Fuel Economy of Federal versus California
Vehicles . III-8
D. The Relationship Between Emission Standards
or Levels and In-use Fuel Economy 111-20
IV. Data Analysis IV-1
A. Two Variable Linear Regressions IV-6
B. Two Variable Linear Regressions with
Stratification by Model Year IV-11
C. Stepwise Backward Regression Analysis IV-29
D. Residual Analysis IV-34
-------
Page
E. Subconfiguration-Matched Data Analysis IV-39
F. Special Engine Over Time Trends IV-43
G. Sensitivity Coefficient Study IV-53
H. Data Stratification by Emission Control System. . IV-55
I. Is There a "Knee" in the Relationship between
Tailpipe Emissions and Fuel Economy? ...... IV-69
V. Individual Manufacturer Discussions
A. General Motors V-l
B. Ford V-36
C. Chrysler V-53
D. Toyota V-65
E. Nissan V-73
F. Honda V-87
G. Volkswagen V-96
H. Toyo Kogyo V-104
I. American Motors V-112
VI. Appendixes
1. Prediction of Federal Test Procedure
Results from Hot Emissions Data 1-1
2. Calculation Methodology for Fuel
Economy Change Allocation 2-1
3. Variability Estimates for Emissions
and Economy over the FTP 3-1
-------
Page
4. Two Variable Linear Regression Plots
of Emissions versus Economy 4-1
5. Results from Stepwise Backward Regression of
Vehicle Parameters and Emissions versus Fuel
Economy • 5-1
6. Residual and Ratio Regressions 6-1
7, Plots of Emissions and Fuel Economy
from Special Engines over Time 7-1
8. Multicollinearity Discussion 8-1
9. Plots of Ton-Miles per Gallon versus
Emissions by Emission Control System 9-1
10. Plots of Ton-Miles per Gallon versus
Emissions by Emission Control System
and Transmission Type 10-1
11. Adjusted Fuel Economy versus Emissions
by Emission Control System and
Transmission Type 11-1
-------
SECTION I
I. Summary
This report contains analyses of the fuel economy/emissions relationships
for 1981 model year and earlier vehicles. The 1981 model year Federal
emission standards are the most stringent Federal standards to date. "A
Study of the Relationship Between Exhaust Emissions and Fuel Economy" is the
title for this report. However, this does not mean that these are the only
two important variables. The relationship between the fuel economy and
exhaust emissions does not exist in isolation, but rather is intertwined
with many other vehicle characteristics. These associated characteristics
include, but are not limited to; driveability, performance, costs, octane
requirement, production lead time, and fuel economy/exhaust emissions
control technology. Ideally, an examination of fuel economy/exhaust
emissions relationships would include quantification of the concurrent
interactions with these other vehicle characteristics. Unfortunately,
neither this report nor any other reports available to EPA have identified
data allowing such quantificaton.
This report is briefly summarized as follows: First, the background for the
report was reviewed, including past EPA reports on fuel economy and
emissions. Second, a literature search identified other reports which'
addressed fuel economy and emissions. Third, the 1981 car fleet was
examined and its fuel economy/emissions performance assessed.
The historical fuel economy and emission standards trends show an overall
trend of increasing fuel economy and decreasing emissions over time. This
is not meant to imply that the reduction in emissions is necessarily a
causal factor in the fuel economy rise. Many other vehicle characteristics
including weight and engine displacement were also changing over this same
time period.
An important observation to make and the major conclusion of this study is
that the twin goals mandated by Congress for the early 1980's have
essentially been met. Fuel economy has been improved substantially and
emissions have been greatly reduced.
1-1
-------
SECTION II
II. Background
This report has been prepared to examine the relationship between exhaust
emissions and fuel economy of automobiles. This report treats the
relationship between emissions and fuel economy in more detail than any
previous EPA work. This subject is of considerable interest to EPA due to the
importance of any changes in fuel economy on cost to the consumer of various
emission standards. The Congress of the United States is also interested in
the relationship between exhaust emissions and fuel economy. The interest of
the Congress is expressed in Section 203(a) of the Provisions of Public Law
95-95 which do not amend the Clean Air Act. The Congressional interest in the
subject was an important consideration in the decision to conduct this study.
"Emission control reduces fuel economy." Is it true or false? This cliche is
typical of many statements which are made in reference to an interaction
between vehicle emissions and fuel economy. Three very noticeable
shortcomings are evident in this cliche. First, the vehicle(s) for which the
emission/fuel economy relationship is being cited is not defined. Many
automobile design parameters can affect a vehicle's emissions and/or fuel
economy. Secondly, the implied comparison made in the example statement does
not identify the basis for comparison. Gain, loss, reduction, improvement,
better and worse mean nothing without a clearly stated basis of comparison.
Finally, the type of emission control system is not stated. During the time
period through which new vehicles have been subject to emission standards, the
type of pollutants, the pollutant levels and the test procedures have all
changed. In addition, the State of California has its own set of emission
standards, which differ from the Federal standards.
II-l
-------
EPA has studied and reported extensively on both vehicle fuel economy and
exhaust emissions. Several of these reports have addressed the relationship
between fuel economy and emission standards. Prior to a review of some
conclusions from these reports, a brief discussion of fuel economy will be
presented.
Fuel economy of automobiles is often expressed in terms of miles per gallon
(MPG). Many methods are used to generate mpg values. EPA has developed test
procedures which measure a vehicle's mileage over an "urban" driving cycle
and a "non-urban" or "highway" driving cycle. Exhaust gas is monitored as a
vehicle is operated over these cycles and the fuel economy is calculated
using the carbon balance method discussed below.
To calculate fuel economy, in miles per gallon (MPG), from an emission test,
the following equation applies:
Miles _ gins carbon/gal of fuel
Gallon gms carbon in exhaust/mile
The carbon in the fuel is:
grams Cfuel = grams fuel x
gallon
molecular wt. C
molecular wt. fuel
= (2798) x (.866)
= 2423
where:
2798 is a density typical of test gasoline, in grams/gallon; and
.866 is typical of the weight fraction of carbon in the fuel.
II-2
-------
The carbon in the exhaust is contained in the unburned fuel hydrocarbons
(HC), carbon monoxide (CO), and carbon dioxide (CO™), as follows:
grams CHf, = gm HC x molecular wt. C
molecular wt. HC
gm HC x (.866)
grams GCO = gm CO x molecular wt. C
molecular wt. CO
gm CO x (.429)
grams C = gm C02 x molecular wt. C
2 molecular wt. C02
gm C02 x (.273)
So we have:
Miles = 2423
Gallon (.866gmHC + .429gmCO + .273gmC02)/miles
or
MPG = 2423 x miles traveled
.866gmHC + .429gmCO + .273gmC02
following amounts of carbon compounds: HC, 9 grams; CO, 124 grams; CO-,
Example: In a 10-mile test, a car's exhaust emission measurements show the
following ai
3641 grams.
Using the above equation, the fuel economy is:
MPG = 2423 x 10
.866(9) + .429(124) + .273(3641)
24,230 = 23.0 MPG
7.8 + 53.1 + 993.7
II-3
-------
The measured fuel economies observed over these two driving cycles are the
urban mileage (MPG ) and the highway mileage (MPG ). The MPG and
MPG, can be used to calculate a composite mileage (MPG ) by means of a
weighted harmonic averaging technique.
Suppose a motorist takes the following three trips:
200 miles, using 15.0 gallons;
100 miles, using 9.4 gallons;
140 miles, using 11.8 gallons.
The fuel economies of these three trips are:
200 miles
15.0 gal.
100 miles
9.4 gal.
140 miles
13.3 MPG;
10.6 MPG;
11.9 MPG.
11.8 gal.
If he merely averages the individual trip MPG's he gets:
(13.3 + 10.6 + 11.9)/3 = 11.9 MPG
But this is incorrect. the motorist traveled 440 miles and used 36.2
gallons, so his overall fuel economy was:
440/36.2 = 12.2 MPG
To get the correct fuel economy for multiple trips, the following equation
must be used:
Miles total miles traveled
Gallon total gallons used
II-4
-------
If the individual trip lengths and fuel economy values are known, but the
gallons used are not known> the proper equation is:
miles.. + miles? + ... +
MPG
miles.. miles^ miles,,
4-4-
T . . . i
where miles = length of trip "x"; MPG = gas mileage to trip "x"; and
X. X
N = number of trips.
For a number of test trips of the same length, the above equation is
equivalent to:
miles x N
MPG = - -
where miles = the standard test length and N = the number of tests.
The above equation simplifies to:
N/SumN
which is the harmonic average of the MPGs from the tests.
To calculate the composite MPG from known city and highway MPG's, the
apportionment of total mileage between city and highway driving must be
used. If a motorist drives 55% of his mileage in the city and 45% on the
highway, his composite fuel economy is:
II-5
-------
MPG = t0tal
.55(total miles) .45(total miles)
City MPG Highway MPG
1
T55~ ~ T?5~
MPGC MPGR
In addition to the carbon balance method, fuel economy can also be
calculated from vehicle speed and fuel consumption rate, as follows.
Miles = Miles/Hour
Gallon Gallons/Hour
But fuel consumption rate is related to the engine power output (not the
power rating) by the expression:
Gallons Hp Ibs fuel Gals
Hours X HP-Hr x Ib
So fuel economy is:
Miles _ (Mi/Hr) (Ibs/gal)
Gallon ~ (HP) x (Ib/HP-Hr)
or
MPH x Df
MPG
HP x SFC
Where Df is fuel density, pounds per gallon (approximately 6.2 for
gasoline); and
SFC is specific fuel consumption, pounds per hour per horsepower output.
SFC is a commonly-used engineering term directly related to engine
efficiency. The more efficient an engine is, the less fuel it needs to
deliver a given power output. For a typical gasoline fuel, the relationship
between SFC and engine efficiency is approximately:
SFC = 13'5
Efficiency
(An efficiency of 13.5% corresponds to an SFC of 1.0 Ib/HP-Hr)
II-6
-------
Substituting,
MPH x Df MPH x 6.2 x Eff.
MPG =
HP x 13.5/Eff. HP x 13.5
So we see that fuel economy is a function of speed (MPH), engine load (HP),
and engine efficiency according to:
MPG = .46 - x Efficiency
for a typical gasoline fuel.
Example: An intermediate size car requires an engine output of 26 HP to
cruise at 50 MPH. The engine efficiency for this condition is 22.0% (SFC =
0.614). Using the above equation, the fuel economy is:
MPG = .46 x 50/26 x 22 = 19.5 MPG
To cruise at 70 MPH, the same car requires 51 HP, and the engine efficiency
is 25.4% (SFC = 0.532). The fuel economy is:
MPG = .46 x 70/51 x 25.4 = 16.0 MPG
Also, note that the instantaneous fuel economy of any automobile can range
from zero to infinity. Examples are a vehicle stopped with an idling engine
(zero MPG) and a vehicle coasting with the engine stopped (infinite MPG).
The remainder of this section lists some conclusions which have been
presented in previous EPA reports. The reader will notice that some fuel
economy estimates for the same subject group of automobiles are different
between reports. This is a function of the dynamic process of estimation.
As time has passed, the data available to EPA have changed, and analysis
technqiues have been improved.
II-7
-------
An early EPA report* contained the following conclusions:
Vehicle weight is the single most important parameter affecting
urban fuel economy; a 5000 pound vehicle demonstrates 50% lower
fuel economy than a 2500 pound vehicle.
The fuel economy loss for 1973 vehicles, compared to uncon-
trolled (pre '68) vehicles, is less than 7%. The average fuel
economy loss due to emission control for all controlled (68-73)
vehicles is 7.7%.
Data on 172 1970 and 1971 GM cars did not demonstrate any effect
on fuel economy of reduced compression ratio.
The Diesel and stratified charge engines show better fuel economy
than the conventional engine with the Diesel showing a fuel
economy improvement of more than 70%.
A year later, in October 1973, another EPA report** stated the following
concerning emissions/fuel economy relationships:
The sales weighted average fuel economy loss due to emission
controls (including reduction in compression ratio) for 1973
vehicles, compared to uncontrolled (pre-1968) vehicles, is 10.1%.
However, vehicles less than 3,500 pounds show an average 3% gain
(attributable to carburetor changes made to control emissions)
while vehicles heavier than 3,500 pounds show losses up to 18%.
The size of these losses, however, is highly dependent on the type
of control systems the manufacturer has chosen to use.
The reduction in compression ratio employed by most manufacturers
to enable their vehicles to operate on 91 octane gasoline has
resulted in a 3.5% fuel economy loss.
* Fuel Economy and Emission Control, U. S. Environmental Protection
Agency, Office of Air and Water Programs, Mobile Source Pollution
Control Program, November 1972.
** A Report on Automobile Fuel Economy, U.S. Environmental Protection
Agency, Office of Air and Water Programs, Office of Mobile Source
Air Pollution Control, October 1973.
H-8
-------
By the 1975 model year, catalytic converter equipped vehicles were
being produced in sizeable numbers. In the Fall of that year, the EPA
again reported on automotive fuel economy. This report* presented the
following conclusions:
There is no simple or inherent relationship between fuel economy
and the emissions standards that new cars are required to meet;
especially misleading is the contention that fuel economy always
becomes poorer as emissions standards are made more stringent.
With the use of catalyst technology, the average fuel economy of
1975 cars is nearly 24% better than the 1974 models, although
their emissions are lower than the 1974's. In fact, fuel economy
of the 1975 fs is as good as cars built before emission controls
were introduced.
There is no guarantee of superior fuel economy through the use of
catalytic converters. 1975 cars using catalysts can give excel-
lent or poor fuel economy, depending on the manufacturer's overall
design. Cars which do not use catalysts can also give excellent
or poor economy, again depending on the overall design.
At the beginning of 1975, EPA published a report which was an overview
of the technical status of automobile emission control.** The chapter
of this report which covered fuel economy contained the following
discussion on the effects of emission standards:
The net effect on fuel economy of a given emission standard
depends on the combination of control techniques used to achieve
compliance. Analysis of EPA certification data has clearly shown
that the fuel economy performance of nominally identical cars
(e.g. same weight, engine size, axle ratio, etc.) can be
significantly different while the emissions are nearly the same.
* Factors Affecting Automotive Fuel Economy, U.S. Environmental
Protection Agency, Office of Air and Waste Management, Mobile Source
Air Pollution Control, Emission Control Technology
Division, September 1975.
** Automobile Emission Control - The Technical Status and Outlook as of
December 1974, A Report to the Administrator, U.S. Environmental
Protection Agency, Mobile Source Pollution Control Program, Emission
Control Technology Division, January 1975.
II-9
-------
The difference in fuel economy is the result of the difference in
the usage of fuel efficient control technology. At a fixed
emission level fuel economy is a function of the usage of fuel
efficient control technology.
Erroneous conclusions about emission control and fuel economy can
result from ignoring differences in the control technology
available and looking at the effect of different levels of
emission control using a particular control system. With a fixed
emission control system fuel economy is a function of the degree
of emission control required. When alternative control approaches
are not considered, changes in emission level can only be achieved
by altering basic engine calibrations such as spark timing.
Since minimum emissions and maximum fuel economy are usually not
simultaneously achieved, lower emission levels with a given system
result in degraded fuel economy.
Besides looking at emission control system/emission level/fuel
economy relationships with the control system held constant or the
emission levels held constant, it is also possible to consider the
fuel economy level held constant. With a fixed level of fuel
economy the degree of emission control achievable depends on the
type of control technology used. For example, the change in
emission level from uncontrolled to the 1975 Federal Interim
levels (1.5, 15, 3.1) has been accomplished at a fixed fuel
economy level by selection of lean engine calibrations and
catalytic exhaust treatment systems.
The three underlined statements above are three different ways of
looking at control system/emission level/fuel economy relation-
ships. Each statement is a two-dimensional analysis of a three
dimensional problem, however. Unless one fully understands the
three dimensional aspects of the tradeoffs involved, it is pos-
sible to be misled about the expected impact of a particular
emission standard. Since there is no fixed relationship between
fuel economy and emission standards, it is impossible to guarantee
a change in fuel economy by a change in.emission standards.
Every year since 1975 EPA authors have written Society of Automotive
Engineers (SAE) papers that discuss the current status and trends in fuel
economy and, more importantly for this report, discuss estimates of the
various reasons why the fuel economy of one group of vehicles might be
different from the fuel economy of another group. Some of the differences
have been attributed to emissions or to emission standards in the past.
11-10
-------
A summary of these reports is shown below.
Previous Reports That Contain Fuel Economy and Emissions Discussions
Year
SAE Paper
Title
1978
780036
Light Duty Automotive Fuel Economy..
Trends Through 1978
1979
790225
Light Duty Automotive Fuel Economy..
Trends Through 1979
1980
800853
Passenger Car and Light Truck Fuel
Economy Trends through 1980
1981
810386
Light Duty Automotive Fuel Economy.
Trends through 1981
1982
820300
Light Duty Automotive Fuel Economy.
Trends through 1982
1983
830544
Light Duty Automotive Fuel Economy.
Trends thru 1983
This section has only touched upon some highlights from previous EPA technical
reports which addressed the emissions versus fuel economy issue. The
following bibliography is provided to assist those readers who wish to obtain
the reports which were reviewed in preparation of this section.
11-11
-------
Bibliography
1. Automobile Emission Control - A Technology Assessment As Of December,
1971, U.S. Environmental Protection Agency, Mobile Source Pollution
Control Program, Office of Air Programs, January 1, 1972.
2. Fuel Economy and Emission Control, U.S. Environmental Protection Agency,
Office of Air and Water Programs, Mobile Source Pollution Control
Program, November, 1972.
3. A Report on Automobile Fuel Economy, U.S. Environmental Protection
Agency, Office of Air and Water Programs, Office of Mobile Source Air
Pollution Control, October, 1973.
4. Automobile Emission Control - The Development Status As of April 1974, A
Report to the Administrator, U.S. Environmental Protection Agency,
Mobile Source Pollution Control Program, Emission Control Technology
Division, April, 1974.
5. Automobile Emission Control - The Technical Status and Outlook as of
December 1974, A Report to the Administrator, U.S. Environmental Pro-
tection Agency, Mobile Source Pollution Control Program, Emission
Control Technology Division, January, 1975.
6. Tradeoffs Associated With Possible Auto Emission Standards, A Report to
the Administrator, U.S. Environmental Protection Agency, Mobile Source
Pollution Control Program, Emission Control Technology Division,
February, 1975.
7. Factors Affecting Automobile Fuel Economy, U.S. Environmental Protection
Agency, Office of Air and Waste Management, Mobile Source Air Pollution
Control, Emission Control Technology Division, September, 1975.
11-12
-------
8. Automobile Emission Control - The Current Status and Development Trends
as of March 1976, A Report to the Administrator, U.S. Environmental
Protection Agency, Office of Mobile Source Air Pollution Control, Tech-
nology Assessment and Evaluation Branch, Emission Control Technology
Division, April, 1976.
9. Automobile Emission Control - The Development Status, Trends, and Out-
look as of December 1976, A Report to the Administrator, U.S. Environ-
mental Protection Agency, Office of Air and Waste Management, Mobile
Source Air Pollution Control, Emission Control Technology Division,
April, 1977.
11-13
-------
SECTION III
III. Fleet Level Effects
III A. Fuel Economy of 1974 and Newer Vehicles Compared to Uncontrolled
.Vehicles
The summary of most fuel economy/emission relationships studies is repre-
sented in a figure like that shown in figure III.A-1. In general, those
that claim fuel economy penalties for controlling emissions have estimates
in the region labeled A. In region A, for each and every percentage
reduction there is a penalty and the percentage value of the penalty
increases as the stringency increases. EPA can fairly be said to have made
estimates like those shown in B in the past - a small effect - it could go
either way. Almost nobody can be found that espouses the relationship
labeled as C in the figure.
In order to investigate this issue with some data, SAE paper 810386* was
studied. The nominal emission standards in figure 4 of that reference were
transformed to percent reductions from an uncontrolled base of 8.7 HC, 87.0
CO, 3.5 NOx. The base, nominal emissions standards and percent reductions
are shown in table III.A-1, along with the constant-weight-mix fuel economy
ratio. The term "nominal emission standard" is used to denote the most
typical emission requirement, and is used to account for those cases where
waivers (1981 Federal) or optional standards (California) make a single
triplet of values not precisely correct.
The analysis is based on each model year's total fuel economy change from
the pre-emission control level and total emission standards change, also
from the pre-control level. The data in table III.A-1 are presented
graphically in figure III.A-2 through figure III.A-4. It should be pointed
out that these figures are at a constant weight mix, so that the im-
provement in fuel economy due to reduced weight cars takin3 a larger
J.A. Foster, J.D. Murrell, and S.L. Loos, SAE Paper 810386, "Light Duty
Automotive Fuel Economy....Trends through 1981", U.S. EPA.
III-l
-------
Figure III. A-l
Concepts of Changes in Fuel Economy with
Changes in Stringency of Emission Standards
o
(_>
UJ
a
UJ
o
oc
o
o
CD
a
UJ
a
o
o
o
CO
o
o
UJ
UJ
20.00
40.00 60.00
STRINGENCY
80.00
1QO.QQ
EXPRESSED flS PERCENT REDUCTION FROM UNCONTROLLED
III-2
-------
Table III. A-l
Percent Emission Reductions
and
Fuel Economy Ratios Compared to Uncontrolled Emissions
and
Fuel Economy
(Nominal Emissions Standards)
Year
pre
1974
1974
1975
1975
1976
1976
1977
1977
1978
1978
1979
1979
1980
1980
1981
1981
Fed or Gal
- control
Fed***
Gal***
Fed
Gal
Fed
Gal
Fed
Gal
Fed
Gal
Fed
Gal
Fed
Gal
Fed
Gal
HC Standard
(% Red)
8.7 (0%)
3.0 (66%)
2.9 (67%)
1.5 (83%)
0.9 (90%)
1.5 (83%)
0.9 (90%)
1.5 (83%)
0.41 (95%)
1.5 (83%)
0.41 (95%)
1.5 (83%)
0.41 (95%)
0.41 (95%)
0.41 (95%)
0.41 (95%)
0.39* (95%)
CO Standard
(% Red)
87 (0%)
28. (68%)
28 (68%)
15 (83%)
9.0 (90%)
15 (83%)
9.0 (90%)
15 (83%)
9.0 (90%)
15 (83%)
9.0 (90%)
15 (83%)
9.0 (90%)
7.0 (92%)
9.0 (90%)
3.4 (96%)
7.0 (92%)
NOx Standard
(% Red)
3.5 (0%)
3.1 (11%)
2.1 (40%)
3.1 (11%)
2.0 (43%)
3.1 (11%)
2.0 (43%)
2.0 (43%)
1.5 (57%)
2.0 (43%)
1.5 (57%)
2.0 (43%)
1.5 (57%)
2.0 (43%)
1.0 (71%)
1.0 (71%)
0.7 (80%)
MPG Ratio**
1.00
1.00
0.94
1.14
1.07
1.26
1.12
1.30
1.14
1.30
1.14
1.28
1.17
1.32
1.26
1.43
1.39
* 0.39 NMHC treated equal to 0.41 THC for this study
** From SAE paper 810386, page 13, figure 4
*** Approximate 1975 FTP equivalent
III-3
-------
Figure III. A-2
Fuel Economy of the Federal and California Fleets
Divided by the Fuel Economy of Uncontrolled Vehicles
Versus Stringency of the HC Standard
FIXED WEIGHT MIX
POLLUTflNT: HC
<=>
C3
CM*
o
z
a
UJ
a
UJ
o
cc
03
a
UJ
a
C3
CD
OJ ..
C3
CO
o
o
UJ
_
UJ
u_
a
o
: 77F, 78F
\ 79F
. 76F
75F
74F
. 76C
• 75C
8 IF
80C
80F
80C
79C
77C,
8C
74C
68.00
76.00 84.GO
STRINGENCY
92.00
100.00
EXPRESSED RS PERCENT REDUCTION FROM UNCONTROLLED
III-4
-------
Figure III. A-3
Fuel Economy of the Federal and- California
Fleets Divided by the Fuel Economy of Uncontrolled
Vehicles Versus Stringency of the CO Standard
FIXED WEIGHT MIX
POLLUTflNT: CO
CM
z:
o
o
0
LU
LU
O
LU
O
tn
o
2:
>-
CQ
Q
LU
a
C3
CD
CN . .
a
CD
o
o
LU
8 IF
,77F, 78F • 81C
/79F • 80F
: 76F -80C
.79C
!77C, 78C
. 75C
74F
74C
S3.QQ
76.00 84.QD
STRINGENCY
32.CQ
100.00
EXPRESSED flS PERCENT REDUCTION FROM UNCONTROLLED
III-5
-------
• " Figure III. A-4-
Fuel Economy of the Federal and California Fleets
Divided by the Fuel Economy of Uncontrolled Vehicle;
Versus Stringency of the NOx Standard
FIXED WEIGHT MIX
POLLUTflNT: NOX
o
o
LU
_J
LU
ZD
U.
a
LU
_l
O
oc
O
CJ
CO
a
U-l
a
C3
CO
C3
OJ
CO
CD
o
CJ
LU
LU
C3
C3
-76F
.75F
74F
: ffl, 78F
' 79F
.79C
. 76C :77C, 78C
. 75C
81F
80C
81C
74C
00
0.CG
40.00 60.00
STRINGENCY
80.00
100.00
EXPRESSED flS PERCENT REDUCTION FROM UNCONTROLLED
III-6
-------
fraction of the fleet is not shown. Something else other than weight is
responsible for the trends. The figures show that since 1975 there have
been no combinations of emissions standards that have caused fuel economy to
fall below the line. There is not much of a trend for HC and CO, but that
could be because they both fall between about 66% and 95% in percent
reduction space. For NOx there is a wider spread, from 11% reduction to 80%
reduction, but there does not seem to be a consistant trend here either.
One thing is clear, however, if there is a trend such as A (figure III A-l)
in the data it is not showing up at the constant weight level of
aggregation. If anything, the data seem to reflect a trend more like the
"C" example.
III. B. Fuel Economy of 1975 and Newer Vehicles Compared to Previous Year
Vehicles
Another way of associating fuel economy changes with emission standards is
to compare each model year's fuel economy and emission standards with those
of the preceding year. This has been done using two methods.
The first method considers the Federal and California vehicle fleets
separately and considers fuel economy of these fleets on a constant weight
basis.. The constant weight basis is the 1978 model year weight mix of the
Federal fleet. Referring to the emission standards given earlier in table
III.A-l for the various model years, this approach assigns (for example) a
35% reduction in the 49-States NOx standard for 1977 (2.0 grams/mile) com-
pared to 1976 (3.1 grams/mile), and a 50% reduction in the 49-States NOx
standard for 1981 (1.0 grams/mile) compared to 1980 (2.0 grams/mile). The
weight-normalized fuel economy changes for the 49-States and California
fleets can then be linked to the emission standards changes of the same two
subfleets.
The second method treats year-to-year changes in emission standards as a
value equal to the 49-States change weighted 90% and the California change
weighted 10%. This yields an emission change representing a pooled
50-States value. The corresponding 50-States year-to-year changes in fuel
economy are determined using a different analytical technique than the
III-7
-------
weight-normalization used above. This technique isolates the fuel economy
change due to factors other than changes in weight mix, changes in engine
mix within the weight classes, and changes in transmission mix shifts within
discrete weight/engine combinations. The method is described more fully in
appendix 2 of this report. Suffice it to say here that the resulting fuel
economy change, dubbed "system optimization", provides a better estimate
(than weight-normalized) of the fuel economy change due only to emission
standards and the emission control technology employed to meet those
standards.
Table III.B-1 gives the results of these two methods of year-to-year
analysis. Using either method, this table shows that along with
year-to-year changes in emission standards which have always been downward,
fuel economy changes have always been upward (with the single exception of
1979, when fuel economy dropped in response to factors other than emissions
standards, which were unchanged from 1978). Figures III.B-1, III.B-2, and
III.B-3 illustrate the data from table III.B-1.
In general it should be noted that the fuel economy went up in some years
when the year-to-year change in a given emissions standard was zero. These
are plotted on the figures as zero percent reductions. Figure III.B-1 shows
that fuel economy tended to go up when no change was made in the HC emission
requirements. As in the case where the comparison was made to uncontrolled
vehicles as a base, figures III.B-1, III.B-2, and III.B-3 indicate no
penalty for changes made from one year to the next even when these changes
are substantial reductions in emissions. In fact the resulting fuel economy
effects are shown to be positive, not negative.
III. C. Fuel Economy of Federal versus California Vehicles
There is yet another method of analysis that can be applied to the certifi-
cation data. Within a given model year, the system optimization fuel
economy difference between the 49-States and California fleets can be com-
pared, along with the differences in 49-States and California emission
standards. This would indicate, at a fixed point in time, the fuel economy
effects of two levels of emission standards. If one assumes that at a given
III-8
-------
Table III. B-l
Changes in Emission Standards and Fleet
Fuel Economy Compared to the Previous Year Values
Method 1 Method 2
(Federal and California Fleets Separately) (Federal and California Fleets Combined)
Year
Change in Weight-Normal
Emission Std. FE Change**
Change in 50-State
Emission Std.
50-States
System Optim.
FE Change***
74-75
75-76
76-77
77-78
78-79
79-80
80-81
Fed
Cal
No
Fed
Cal
No
No
Fed
Cal
Fed
Cal
HC,
CO,
HC,
CO,
NOx,
Change
NOx,
HC,
NOx,
Change
Change
HC,
CO,
NOx,
CO,
NOx,
CO,
NOx,
-50%
-46%
-69%
-68%
-5%
-35%
-54%
-25%
Fed,
Cal,
Fed,
Cal,
Fed,
Cal,
+14%
+14%
+11%
+5%
+3%
+2%
(1.
(1-
(1.
(1-
(1.
(1-
14)
14)
11)
05)
03)
02)
No Change
-73%
-53%
-33%
-51%
-50%
-22%
-30%
Fed,
Cal,
Fed,
Cal,
Fed,
Cal,
-2%
+3%
+3%
+8%
+8%
+10%
(0.
(!•
(1.
(1.
(1.
(1.
98)
03)
03)
08)
08)
10)
HC,
CO,
-52%
-48%
NOx, -0.
No
HC,
NOx
No
No
HC,
CO,
NOx
CO,
NOx
Change
-5.
5%
4%
, -34%
Change
Change
+11.5% (1.12)
+ 8.8% (1.09)
+ 2.8% (1.03)
+ 0.5% (1.01)
- 2.3%* (0.98)
-66%
-48%
, -3.
3%
-48%
+ 2.6% (1.03)
+ 2.7% (1.03)
, -48%
* More than half of this change in FE was due to test procedure changes
(dynamometer loading) made to restore the rigor of the 1975 procedure - See SAE
790225, Appendix C. The net system optimization change independent of this test
procedure effect is estimated at -1.0%.
** Calculated values from figure 4 of SAE Paper 810386, page 13.
*** From SAE Paper 800853, page 36.
III-9
-------
Figure III. B-l
% Change in Fuel Economy Versus
% Change in HC Standard
from Previous Year
POLLUTRNT: HC
CD
CD
CD'
oo
CM
OJ
cc
oc
LU
ID
ZD
O
UJ
cc
Q_
O
u_
o
o
UJ
UJ
u_
UJ
CD
CD
CD
CD
CD
CO
CD
0
oi - -
I
CD
CD
X= System Optimization
• = Weight - Normal
75C
75F
X '75
76
80
(No change)
• 80C, 81F
.76C
* 77F, 79C
78X ?3F,~ 73C
79v 79F
•+•
-80.00
-60.00
-40.00
-20.00
0.00
20.00
X CHflNGE IN EMISSION STflNDflHD FROM PREVIOUS YEflR
111-10
-------
Figure III. B-2
% Change in Fuel Economy Versus
% Change in CO Standard
from Previous Year
POLLUTflNT: CO
CD
CO
CM
OJ
cc
-------
Figure III B-3
Change in Fuel Economy Versus
% Change in NOx Standard
From Previous Year
POLLUTflNT: NOX
cc
a:
LU
cn
13
o
LU
or
a_
a
cc
o
o
LU
_J
LU
U_
•z.
LU
o:
O
o
OJ -.
CN
C3
C3
o
a
cd
C3
I
C3
O
81F
•81C
. 80C
X 77FX
(No change)
75C- • 75F
X 75
' 76F
X 76
. 76C
80 x . 79C, 80F
X 78
N78F, 78C
79* 79F
-80.00
-SO.OO
-40.00
-20.00
0.00
20.00
V. CHflNGE IN EMISSION STRNDflRD FROM PREVIOUS YEflR
111-12
-------
point in time, roughly the same emission control and fuel economy technology
is available to the Federal (49-States) and California fleets, then the
technological capability could be said to be the same. This is probably
roughly the case, but it must be pointed out that the emission control
technology used in California, in a given year in some cases is more
sophisticated, due to the use of California as a proving ground for the
newer technology. The results of using this method are given in table
III.C-1. Figures III.C-1, III.C-2 and III.C-3 illustrate the data of table
III.C-1.
Looking at table III.C-1 and figures III.C-1 through III.C-3 it can be seen
that the nature of the relationship has changed. Comparing 49-States to
California for the same year generally shows California to be more stringent
in emissions and lower in fuel economy.
From table III. C-l the Federal vs. California data appear at first to be
five equations in four unknowns:
-73h - 40c -25n + k = -11.8f
-73h - 40c -25n + k = -11.If
-73h - 40c -25n + k = -10.If
-Oh + 29c -50n + k = -4.6f
and -Oh +106c -30n + k = -1.7f
where: h, c, and n = % change in HC, CO, and NOx coefficients respectively
f= % change fuel economy and k = constant.
At second glance, however, the first three equations have no possible unique
solutions for h, c, and n, since the right sides of the three equations are
unequal. Also, the constant terms (k) have no basis in reality, since the
FE change due to no emissions changes must be zero (treating emission
differences as the sole source of FE change). Interpreting the -11.8,
-11.1, and -10.1 figures as random variances around the true A FE that comes
from a combined -73%AHC, -40% A CO, and -25% A NOx, we can average them
and construct 3 equations. Setting k = 0, there are now 3 equations in 3
unknowns.
111-13
-------
Table III. C-l
Federal vs. California Emissions
and
Fuel Economy (System Optimization Method)
Difference in Emission System Optimization
Year Standard FE Difference*
*From SAE paper 800853, page 50,
and SAE paper 810386, page 12
1977 HC, -73%
CO, -40% -11.8% (0.88)
NOx, -25%
1978 HC, -73%
CO, -40% -11.1% (0.89)
NOx, -25%
1979 HC, -73%
CO, -40% -10.1% (0.90)
NOx, -25%
1980 HC, "same"
CO, +29% -4.6% (0.95)
NOx, -50%
1981 HC, "same"
CO, +106% -1.7% (0.98)
NOx, -30%
111-14
-------
Figure III. C-l
% Difference in Fuel Economy Versus % Difference
in HC Li .idard for the California Fleet Compared
to th3 Fo-eral Fleet in the Same Model Year
POLLUTflNT: HC
o
'cr
o_
o
t—
en
en
o
o
LU
UJ
o
LU
CJ
cc
UJ
Q_
CD
CD
I
oo
I
CN
I
CO
I
(No Difference)
CD
81
80
O 79
-------
Figure III. C-2
% Difference in Fuel Economy Versus % Difference
in CO Standard for the California Fleet Compared
to the Federal Fleet in the Same Model Year
POLLUTflNT: CO
CD
CC
fvl
O-
o
UJ
en
en
>-
o
z
a
UJ
UJ
UJ
CD
UJ
C_J
oc
UJ
Q_
CD
C3
C3
i
00
I
C3
O
OJ
I
(No Difference)
O 81
O 80
O 79
O 78
O 77
-60.00
-20.00
20.00
60.00
100.00
140.00
7. CHflNGE IN EM IS. STRNDflRD ( CflLIF VS FED ) IN THE SflME YEflR
111-16
-------
Figure III. C-3
% Difference in Fuel Economy Versus % Difference
in NOx Standard for the California Fleet Compared
to the Federal Fleet in the Same Model Year
POLLUTflNT: NOX
o
•—I
I—
-
o
o
C_>
UJ
_l
UJ
UJ
-------
-73h -40c -25n = -11.Of
+29c -50n = -4.6f
+106c -30n = -1.7f
The solution to this system is:
h = 0.110246, c = 0.011964, n = 0.098939
(all decimals are necessary)
11.0 = 11.8 + 11.1 + 10.1
So the dependence of % A FE on % A X, where X = HC, CO, NOx is:
% A FE = .110246 (% A HC) + .011964 (% A CO) + .098939 (% A NOx)
Note that all coefficients are positive, meaning that if any emission
standard goes down (tighter control), FE also goes down (suffers a
penalty).
In the ballpark sense, FE responds at 1/10 sensitivity to HC and NOx changes
(lowering either standard by 50% would penalize fuel economy by 5%), but is
less sensitive to CO (lowering the CO standard by 50% would penalize FE by
.012 x 50 = 0.6%).
The previous equation can be used to calculate a fuel economy difference for
sets of standards which may be of particular interest:
1. Base 0.41 HC, 3.4 CO, 1.0 NOx
2. "80" 0.41 HC, 7.0 CO, 2.0 NOx
3. "Hybrid" 0.41 HC, 7.0 CO, 1.5 NOx
The HC change for all cases is zero. For Base versus "80" or Hybrid the CO
111-18
-------
change is (7.0 - 3.4)/3.4 = 1.058824 (105.8824%). The NOx changes are
1.0000(100%) and 0.5(50%).
"80" vs. Base = 0.110246(0) + 0.011964(105.8824) + 0.098939(100)
"80" vs. Base = 0 + 1.266770 + 9.8939 = +11.2%
Hybrid vs. Base = 0 + 0.922320 + 4.946950 = +6.2%
Note that this is the only fuel economy relationship quantified. The other
positive trends could also have been quantified.
As a check on the accuracy of the equation relating Federal and California
fuel economy in the same model year for more general application, the change
in fuel economy from the 1980 to the 1981 Federal standards was compared.
In.the previous calculations, this would be the negative of "80" vs. Base or
a predicted loss of 11.2% fuel economy for this change in emission
standards. This compares poorly with the 2.7% increase in fuel economy from
1980 to 1981 due to system optimization (from table III.B-1), both in sign
and in magnitude, and leads to the conclusion that while the
California/Federal relationship for prior years may be quantifiable, that
relationship has little utility when one attempts to predict changes due to
differences in Federal or 49-State emission standards.
In summary, California vehicles have achieved less fuel economy than their
Federal counterparts in recent years. The difference has been declining and
is now 1.7% less.
II.I-19
-------
III. D. The Relationship Between Emission Standards or Levels and In-use
Fuel Economy
III. D.I Background
For the past four years EPA and DOE have analyzed differences between EPA
test fuel economy and "on-road" fuel economy of in-use cars. The purpose of
that work was to measure fuel economy shortfall, that is, the offset between
"on-road" fuel economy and fuel economy as measured on EPA tests. The fleet
shortfalls reported in the 404 Report* are typical of the results of these
early analyses;
Model Year Road Shortfall (vs. EPA Composite MPG)
1974
1975
1976
1977
1978
1979
- 6.9%
-12.4%
-19.2%
-19.6%
-19.2%
-16.1%
Note that the shortfall increases with time (model year) through 1977, after
which it decreases. Before attempting to relate this trend to possible
emissions influences, however, we must consider the matter of vehicle tech-
nology.
Since model year 1978, manufacturers have introduced changes both in engine
technologies and in drivetrains that result in a much less homogeneous mix of
technologies in the U.S. fleet. Recent analysis disaggregates the large
sample of in-use vehicles in the current data base (model years 1978 to 1980)
into technology-specific subgroups to consider the shortfall for each tech-
nology subgroup.
EPA, "Passenger Car Fuel Economy: EPA and Road", Report EPA-460/3-80-010,
September, 1980.
111-20
-------
For the purposes of this discussion, the following technologies are defined
as "conventional" technologies.
0 Front engine, rear wheel drive vehicles
0 Spark ignition (S.I.) gasoline engines
° Carburetor for fuel metering
0 Automatic transmissions
The 1974-78 vehicles studied in early analyses employed predominantly
conventional technologies. "Alternative" technologies are defined as follows:
0 Gasoline fuel-injection for fuel metering
0 Front engine, front-wheel drive vehicles
0 Diesel engines
0 Manual transmissions
A contractor working for both EPA and DOE has recently performed a
statistical analysis of the current in—use data base to study shortfall for
technology-specific subgroups of vehicles. The results of this analysis sug-
gest that vehicles utilizing some of these alternative technologies had lower
shortfall than conventional technology vehicles.
The objective of this discussion is to examine the engineering principles that
govern the MPG performance of alternative technologies on the EPA test pro-
cedure and under in-use conditions. The results can be used to interpret the
observed shortfalls of alternative technologies. The discussion examines each
of the four technologies in comparison to their conventional technology
counterpart. Thus, manual transmissions are compared to automatics, fuel in-
111-21
-------
jected S.I. engines to carburetted S.I. engines, front-wheel drive vehicles to
rear-wheel drive vehicles and Diesel engines to carburetted S.I. engines. The
synergy between multiple alternative technologies is not examined in this
analysis, although it may be that vehicles with combinations of these tech-
nologies will exhibit greater reductions in shortfall than will any single
application of technology.
Differences in shortfall of the four technologies in comparison to conven-
tional technologies are associated with responses to the following performance
factors:
0 The response of the technology to the EPA test procedure (FTP and
HFET). The EPA fuel economy is measured over fixed driving cycles on
a dynamometer and the technology's response to the EPA procedure
will affect the shortfall.
° The in-use maintenance characteristics of the technology. The state
of tune of the vehicle is an important contributor to shortfall.
0 The response of the technology to non-FTP driving conditions. The
effects of cold ambient temperatures, extended idling and different
driving cycles are also important contributors to shortfall.
Manual Transmissions
Earlier studies on shortfall showed that between model years 1974 and 1977,
shortfall for manuals started at a level on par with automatic transmissions
in 1974 and increased to approximately 10 percent above the shortfall for
automatics by 1977. EPA* explained that manual transmission shift points were
variable in the hands of the user, whereas the automatic was calibrated to
shift at preprogrammed points. It was also suggested that vehicle owners
drove, on the average, more aggressively than on the EPA cycle, i.e., the
shifts are generally at higher engine speed/load points thus leading to higher
Ibid.
111-22
-------
shortfall for manual transmission equipped vehicles in comparison to those
equipped with automatic transmissions.
New analyses by Ford, GM, and DOE contractors, with much larger and more
recent data bases show the reverse to be true, i.e., manuals have less short-
fall than automatic transmissions. The data bases included vehicles from
model years 1978 through 1980. While the earlier results may be suspect
because of the small sample of vehicles with manual transmissions in the data
base, there may be two technological reasons behind this change: (i) EPA
changed the manual transmission shift points during the FTP for model year
1979 and succeeding years. In previous years, manufacturers were allowed to
specify shift points and patterns which were often unrealistic, but which
tended to maximize fuel economy on the FTP. EPA changed the test requirements
to restore the rigor of the 1975 procedure, and appears to have succeeded in
reducing shortfall for vehicles equipped with manual transmissions. However,
automatic transmissions were unaffected by this ruling, leading to one
explanation for the observed change in relative shortfall. (ii) As fuel
economy became more important, manufacturers optimized shift points for
automatic transmissions so that upshifts occured as early as possible and
downshifts were delayed during the FTP cycle. Torque converter lock-up was
also calibrated for maximum effectiveness on the FTP, i.e., the lock-up is
engaged more frequently on the test than in "real-world" driving. This may
have led to an increase in shortfall for newer automatics in comparison to
automatic transmission equipped vehicles of earlier years. This, in turn,
provides a second explanation for the observed changes in relative shortfall
between manual and automatic transmissions.
Fuel Injection
EPA's Emission Factors program annually tests a large sample of vehicles on
the FTP and checks these vehicles for maladjustments and mistuning. EPA has
noted that carburetor idle mixture and choke maladjustments are the most
common malperformances in these vehicles and that the adjustments were pre-
dominantly richer than the manufacturers' recommendations, leading to fuel
economy reductions as large as 10 percent depending upon the degree of
maladjustment.
111-23
-------
Fuel injection systems could be expected to have lower shortfall than car-
buretted vehicles primarily because most users are less likely to tamper with
fuel injection systems than carburetted vehicles, for two reasons. One reason
for tampering and maladjustment is customer dissatisfaction with drive-
ability, and it is generally conceded that fuel injection vehicles offer
superior driveability in comparison to carburetted vehicles. The second
reason for a possible relatively low level of maladjustments in fuel injected
vehicles is user unfamiliarity with the system, leading to reduced tampering.
However, these differences could diminish for MY 1981, because the choke and
idle mixture will be much less easy to adjust in use.
Fuel injected vehicles also should have lower shortfall than carburetted
vehicles because, at low ambient temperatures and during transients, fuel
injected vehicles need less enrichment than carburetted vehicles. A Canadian
study* found that fuel injected vehicles showed smaller decreases in fuel
economy at cold ambients in comparison to carburetted vehicles. This is the
result of fuel being delivered to the intake ports, rather than to the intake
manifold as in carburetted vehicles where excess fuel is required to ensure
adequate evaporation and transport of fuel to the cylinders. The precise
delivery of fuel possible in fuel injected vehicles also reduces the need for
enrichment during transient operation. Operation at ambient temperatures
below those encoutered on the FTP, and rapid accelerations and decelerations
not encountered on the FTP, are important causes of shortfall.
Front-Wheel Drive
On the chassis dynamometer, FWD vehicles are penalized relative to rear wheel
drive (RWD) vehicles, regardless of whether the power absorption unit (PAU)
setting on the dyamometer is determined by coastdown techniques or by EPA's
"cookbook" method. To a large extent it is believed that this is because the
FWD vehicle places a greater load on the driven wheels than a RWD vehicle.
For example, a FWD vehicle with a 60/40 weight distribution will have 33
percent more load on the driving wheels than
N. Ostrouchov, "Effects of. Cold Weather on Motor Vehicle Emissions and
Fuel Consumption - II", SAE Paper 790229, 1979.
Ul-24
-------
a RWD vehicle with a 55/45 weight distribution. These distributions are
nominal values for each type. The additional load on the driving wheels
increases the tire rolling resistance, and hence, the actual horsepower
required to drive the vehicle on the chassis dynamometer.
If the EPA "cookbook" technique is used to determine the PAU setting, these
losses are uncompensated as the present formula does not differentiate between
RWD and FWD vehicles. Although coastdown techniques correct this situation
somewhat, the FWD vehicle is still penalized. This is because the PAU setting
is determined for tractive effort at 50 mph during the coastdown procedure.
At speeds below 50 mph the FWD vehicle must still overcome a greater load on
the dynamometer than an equivalent RWD vehicle. This is because rolling
resistance is the dominant force at low speeds, while aerodynamic drag is more
important at high speeds. Since about 75 percent of the mileage in the EPA
test is accumulated below 50 mph, the FWD vehicle suffers a net fuel economy
penalty in the FTP compared to the RWD vehicle; this would translate into
decreased shortfall for FWD vehicles in comparison to RWD vehicles.
A second, possibly minor, effect which could reduce the shortfall of FWD
vehicles relative to RWD vehicles is the influence of the cooling fan during
the FTP test. Since the vehicle is stationary during the test, cooling air is
provided by an external fan. This fan operates at constant speed and does
not provide adequate cooling air at higher vehicle speeds. Most FWD vehicles
have electric fans which are controlled thermostatically; these fans remain
on during much of the FTP, because of the inadequate external cooling, unlike
the real world situation. The electrical power absorbed by the cooling fan is
supplied by the engine through the alternator. This power requirement lowers
the fuel economy on the FTP relative to "on-road" conditions, resulting in
decreased shortfall for FWD vehicles. RWD vehicles, a majority of which
employ engine driven fans, do not have this difference in operating mode
between FTP and "on-road" conditions. While the effect of the cooling fan is
believed not as significant as the effect of tire loading, the cooling fan
does make a contribution to shortfall reduction of FWD vehicles relative to
RWD vehicles.
111-25
-------
Diesel
Diesel engines differ markedly from spark ignition (gasoline) engines in their
fuel delivery systems and the type of combustion employed. These principal
technological differences could account for decreased shortfall in comparison
to spark ignition engines. The reasons are as follows:
(1) The fuel injection system in the Diesel engine is factory calibrated
to tight tolerances and sealed at the factory. Spark ignition engines
on the other hand, have carburetors where the user can adjust the
idle air/fuel ratio and the choke. As detailed earlier, EPA has
found that a large majority of spark ignition engines are maladjusted
to richer air/fuel ratios, leading to increased shortfall. Diesels
on the other hand, are rarely maladjusted.
(2) In Diesel engines, the fuel is injected directly into the cylinder
head, unlike carburetted S.I. engines where fuel is generally
distributed via an intake manifold to the individual cylinders. At
low ambient temperatures, Diesel engines need very little enrichment
in comparison co an S.I. engine, where extra fuel is needed to ensure
adequate vaporization and transport of fuel to the cylinders.
(3) Diesel engines are less sensitive to transient and non-FTP driving
modes than carburetted S.I. vehicles in that the fuel consumption
changes are less dramatic. This is due to two reasons: (a) the
Diesel consumes very little fuel at idle in comparison to an S.I.
engine (up to 70 percent less). Thus, extended idling, common in
inner city driving, has a much smaller impact on the fuel consumption
of Diesels than of S.I. engines; (b) transient operation such as
sharp accelerations and decelerations not encountered on the .FTP
increase fuel consumption dramatically for an S.I. engine but not as
much for a Diesel engine. A carburetted S.I. engine has very poor
transient air/fuel ratio control and requires special devices such as
acceleration enrichment pumps and deceleration fuel shutoffs to
I11-26
-------
control the transient air/fuel ratio. Diesels, on the other hand, use an
accurate system of fuel metering where transient operation does not
disturb the combustion process as much.
The combined effect of reduced tampering, lower enrichment at cold ambients,
and decreased sensitivity to variations in driving cycle should reduce the
shortfall for Diesel engines in comparison to S.I. engines.
III. D. 2. Technology-specific In-use Data
Against the background above, which theorizes why some advanced technologies
should have lower shortfall behavior, we can now examine what actual data
reveal.
Figure III.D-1 shows the shortfall characteristics of a number of vehicle
technologies expressed as the ratio of road gallons per mile to EPA 55/45
gallons per mile, plotted against the EPA 55/45 MPG (this ratio is the same as
EPA MPG/Road MPG).
The slanted lines are regression lines for three technology subclasses of
vehicles: all are rear wheel drive, carbureted (gasoline) cars. The topmost
line, labeled "Automatic" consists almost completely of standard (non-lockup
or overdrive) automatics, while the "Manual" line represents just that. The
band around the Manual line is the 95% confidence interval. The 95%
confidence interval for Automatics is within the width of the line.
The line labeled "Auto-lockup and/or 0/D" has been added to the chart on the
basis of a recent finding* that automatic overdrive gives a shortfall between
that of manuals and standard automatics. The assumptions that this line has
a slope parallel to the other two lines, and that it also represents lockup
automatics, whether or not overdriven, are our own assumptions.
* South, "1978 to 1980 Ford On-Road Fuel Economy", SAE Paper
810383, February 1981.
111-27
-------
Figure III. D-l
i
to
CO
Impact of
Technology
on Shortfall
1S78-1980 CARS
I D/M/RWD
®C/A/FWD2
15
20
I I
30 35
EPA Composite MPG
-------
The spots and vertical bands on the figure require additional explanation.
Each spot represents a particular technology, and the vertical bands span the
95% confidence intervals. For some technologies, these intervals are large
because the sample sizes are small. The technology descriptions are:
Fuel System:
FI = Fuel injected gasoline
C = Carbureted gasoline
D = Diesel
Transmission:
A
M
Automatic
Manual
Drive:
FWD = Front
RWD = Rear
The subscripts for some groups designate two inherently different subsets of a
given technology. This becomes obvious when looking at table III.D-1, which
gives the number of cars, and model types, included in each spot.
111-29
-------
TABLE III. D-l
EXPLANATION OF FIGURE III. D-l
Technology Group
(1) FI/A/FWD1
(2) FI/A/RWD
(3) D/A/FWD
(4) D/A/RWD
(5) D/M/RWD
(6) C/M/FWD2
(7) C/A/FWD1
(8) FI/A/FWD2
(9) C/A/FWD2
(10) FI/M/FWD
(11) C/M/FWD1
(12) D/M/FWD
(13) C/M/ FWD1
No. Cars
8
14
2
191
40
90
26
5
83
73
184
72
411
Models Included
All Eldorados
13 Sevilles + 1 BMW
Toronados
Half GM, rest Mercedes & Peugeots
39 Peugeots, 1 Mercedes
Fiestas & Japanese, all 1978 models
All GM: Rlvieras, etc.
Rabbits & Foxes
78-79 Japanese, Omnis, 1980 X-cars
Rabbits, Foxes, Sciroccos
Same as (6) plus Omnis, all 1979's
Rabbits & Dashers
1980 model Fiestas
111-30
-------
Interpretation
The following can be gleaned from figure III.D-1 and table III.D-1:
the behavior of the eight Eldorados in set (1) is anomalous.
0 Fuel injection makes no significant difference in shortfall, compared
to carbureted versions of similar vehicles.
0 The high shortfall for the 90 cars in set (6) can be attributed to
the 1978 test procedure, not to the cars themselves; nearly 600 of
the same types of cars (but 1979-80 models) included in sets (11)
and (13) show no such high average shortfall. For all three sets,
(6), (11) and (13), average road MPG is consistently about 30;
they were rated by the 1978 test at about 36, and by the 1979-80
test at about 31.
0 Rear wheel drive Diesels have essentially the same shortfall as
RWD manual gasoline cars.
0 Transmission type makes no significant difference in shortfall for
front wheel drive cars.
0 The location of set (12), the VW front wheel drive Diesels, on the map
would appear to be due to their front wheel drive rather than
their "Dieselity".
0 The bottom line for all of this is that these data suggest four
distinguishable technology groupings, each with its own shortfall
algorithm:
Front wheel drives, 1/R = 1.02/E
RWD Manuals and Diesels, 1/R = 0.70/E + 0.016
111-31
-------
RWD Automatics, lockup or 0/D, 1/R = 0.77/E + 0.016
(gasoline)
RWD Std. automatic (gasoline) 1/R = 0.85/E + 0.016
where R = Road MPG and
E = EPA 55/45 MPG
Implication
When unique shortfall behaviors of specific technology strata are accounted
for, time trends in aggregated fleet shortfalls can result from time changes
in the mix of the various technologies. As an example, this type of analysis
was done for Chrysler Corporation, who has changed from almost entirely
"conventional" vehicle technology in 1975 to almost entirely "advanced"
vehicle technology in 1981. Table III.D-2 shows the results. In the earlier
years, we would expect increasing shortfall for the conventional part of
Chrysler's fleet simply because the shortfall algorithm for these vehicles is
MPG dependent, and overall fleet MPG is increasing. In fact, the calculated
shortfall for conventional vehicles continues to worsen through 1981. As
advanced technologies increase to significant market penetrations in the later
years, however, their lower shortfall characteristic influences overall fleet
behavior more and more, and aggregate fleet shortfall decreases.
Understanding now the significant role of vehicle technology changes in the
shaping of time trends in fuel economy shortfall, let us compare technology
specific calculated shortfalls for the entire fleet of manufacturers with the
latest observed shortfalls, in table III.D-3.
111-32
-------
Table III. D-2
Model
Year
1975
1976
1977
1978
1979
1980
1981
EPA
MPG
14.7
15.9
15.7
17.8
19.8
20.0
25.6
Example of Variation in Calculated,
In-Use FE Shortfall (Chrysler Corp
Technology-Specific
. Domestic Fleet)
Shortfall for
RWD. Automatics % Use of Advanced
(Conventional Technology) Technology *
- 7.7%
- 9.2%
- 9.0%
-11.0%
-12.4%
-12.3%
-14.3%
3%
6%
8%
12%
24%
88%
96%
Fleet
Shortfall
- 7.6%
- 9.0%
- 8.9%
-10.3%
-10.7%
- 5.5%
- 5.1%
(*) Manual or lockup transmissions, overdrive, front wheel drive.
111-33
-------
Table III. D-3
Similarity Between Observed fleet Shortfall and
Calculated Technology-Specific Shortfall (All Manufacturers)
Model Observed Calculated Technology Federal
Year Fleet Shortfall1 Specific Fleet Shortfall Emission Stds.'
1975 -8.9%
1976 -9.4%
1977 -15.4%
1978 -15.0%
1979 -10.2%
1980 - 3.1%^4^
1981
-7.2%
-9.6%
-10.6%
-11.7%
-11.5%
-10.4%
- 9.7%
1.5/15/3.1
1.5/15/3.1
1.5/15/2.0
1.5/15/2.0
1.5/15/2.0
.41/7/2.0
.41/3.4/1.0
(1) From Contractor analysis (EEA), January 1981 (unpublished)
(2) Calculated using only technology-specific shortfall behavior,
no model year dependence assumed.
(3) HC/CO/NOx grams per mile.
(4) Sample size much smaller than prior model years.
111-34
-------
Note the parallel increasing/decreasing trends for the observations and the
calculations. Three "step changes" in observed shortfall are not fully ex-
plained satisfactorily by the technology-specific theory: the six percent in-
crease in shortfall from 1976 to 1977, the five percent decrease from 1978 to
1979, and the seven percent decrease from 1979 to 1980.
Taking the second of these first, there was no change in emission standards
between 1978 and 1979, so the five percent shortfall decrease cannot be
related to emissions effects. As mentioned earlier, loopholes in the EPA test
procedures.were closed in 1979, and we attribute this improvement in shortfall
to the test procedure.
The 1976 to 1977 shortfall increase came simultaneously with a tightening of
the NOx standard from 3.1 to 2.0 grams per mile, and could be concluded to be
an adverse by-product of the more stringent NOx standard.
In the case of the 1979 to 1980 improvement in observed shortfall, there was
at the same time a tightening of the HC and CO (but not NOx) standards, so it
could be concluded in this case that reduced shortfall is a beneficial
by-product of the more stringent HC and/or CO standards.
The HC/CO conclusion relationship is less strong than the NOx conclusion,
because 1980 data used for this study represented a much smaller sample size.
A test case for the premise that tighter NOx control adversely affects
shortfall will be the in-use data from 1981 model cars, as 1981 NOx is cut in
half from the 1980 level. If a relationship exists between shortfall and the
NOx standard change, it should certainly appear in 1981 cars. Unfortunately,
there are no 1981 in-use fuel economy data currently available;
111-35
-------
SECTION IV
IV. Data Analysis
The computer data bases used to support the analyses for this document were
derived from two EPA sources; the Emission Factors data base, and the
Certification Division data base which includes tests performed by EPA and by
the individual manufacturers. From these sources, three data bases were
generated for use in this study. The data bases in this study will be
referenced as the EF, CERT/EPA and CERT/MFR data bases respectively. They
only include data for vehicles tested under the 1975 Federal Test Procedure
(FTP) and the EPA Highway Fuel Economy Test Procedure (HFET).
Certification records used to create the CERT/EPA and CERT/MFR data bases
were gathered from five certification data bases. These include VEHSUM,
CERTESTSUM, and DFRSLTS, which are part of the MICRO Information System, and
files 1200D and 1202D-MFR (as they appeared on February 7, 1981). The VEHSUM
file contains vehicle descriptions taken from the vehicle identification
sheet. CERTESTSUM includes emission test results- and other pertinent
information from the test data sheet. Engine family deterioration factors
came from DFRSLTS. 1200D included bag data and 1202D-MFR included
information on whether tests were performed at high altitude or not.
The data in CERTESTSUM comes from two sources; a) tests run at the EPA
emissions laboratory and b) tests performed at the manufacturers'
facilities. The CERT/EPA data base, which was created specifically to
support this report, is a compilation of selected information from the five
Certification Division data bases discussed above, but only for vehicles
which were tested at EPA. Likewise, the CERT/MFR data base was limited to
vehicles tested at the manufacturers' facilities. The information selected
for inclusion in the CERT/EPA and CERT/MFR files is discussed below.
IV-1
-------
From the file VEHSUM, all vehicles with a Car Line Code* of less than 50,000
were selected, except those whose Sales Class* was Both Trucks, California
Trucks, or Federal Trucks. The object was to select only light duty vehicles.
The test results used from CERTESTUM were restricted to those gathered under
the FTP and the HFET. Tests with Certification Test Dispositions* of VOID,
NEWTEST and ZEROMILE were not included in the CERT/EPA and CERT/MFR data
bases. For the FTP, the Test Types* included were emission data tests,
durability tests, and fuel economy tests. For the highway fuel economy test
procedure, only emission data tests and fuel economy tests were used.
The resulting data set from CERTESTSUM was then joined to the resulting data
set from the VEHSUM files. They were matched by Internal Vehicle Number* and
by Version Number*. In order to eliminate duplicate test results in the new
data bases, carryover data were assigned to a single model year which
corresponds to the last model year the data were used.
Using information from the certification data files 1200D and 1202D-MFR,
high altitude tests were removed, and bag data for the EPA tests were added
to the new data base. Bag data were not available for manufacturers' tests.
This data base was then segregated into the CERT/EPA and EF files. Table
IV-1 lists the field names which were included in the CERT/EPA and EF files.
The CERT/MFR file was identical to the CERT/EPA file except that the CERT/MFR
file did not contain bag emissions data.
Combined fuel economy values were calculated for each vehicle which had both
FTP and HFET data. The objective was to calculate a combined fuel economy
value for each vehicle configuration for which both urban and highway
miles/gallon (mPg) data were available. A method was selected to achieve
These are field names for the data base. Explanations of these terms
can be found in the August 28, 1980 letter from Robert E. Maxwell,
Director, Certification Division of EPA, to light-duty motor vehicle
manufacturers. This letter transmitted the Supplement to the Application
Format for Certification of Light-Duty Motor Vehicles, 1981 Model Year.
IVT2
-------
Table IV-1
Information Contained in the Data Bases Used for This Studv
DATA BASE VARIABLES (FIELDS) SUMMARY
DATA BASF CERT-EPA
YEAR
Bl.HC (BAG 1 HO
Bl.CO (BAG 1 CO)
Bl.NOX (BAG 1 NOX)
B2.HC (BAG 2 HO
B2.CO (BAG 2 CO)
B2.NOX (BAG 2 NOX)
B3.CO (BAG 3 CO)
B3.NOX (BAG 3 NOX)
FTPHC (FTP HO
FTPCO (FTP CO)
FTPNOX (FTP NOX)
HHPG (HWY MPG)
CMPG (COMR MPG)
FEHC (HWY HO
FECO (HWY CO)
FENOX (HWY NOX)
SITE (TEST SITE)
VEH. (VEH. NO.)
TEST (TEST SE(J.)
FTPBP (FTP riARO. P.)
OOOMtMILE
FEBP (HWY BARO. P.)
MAKE (MAKE OF AUTO)
MFR«MFG(MANUF . )
MDYR»MYR(MODEL YEAR)
MODL'MOL (MODEL)
VTRN.TRAN(TRANSM)
»BBL'CARB<* OF B8L.)
XCYL (* OF CYLIN.)
VNRT'INER( INERTIA WT.)
AOHP'HP (ACTUAL DYNO HP)
AIR.1NJIAIK INJECTED?)
CATA (CATALYST)
THER (THERMAL REACTOR)
ENKM'ENFY (ENG FAMILY)
EGR
75-
7V
X
X
X
X
x •
X
A
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
80
X
X
X
X
X
X
X
X
X
X
X
A
X
X
X
X
X
X
X
X
X
11
A
X
X
X
X
X
HS
81 79 IV
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
V
X
A
A
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
A
A
X
A
A
'x
X
X
X
X
X
X
X
X
A
X
A
A
X
X
X
X
X
X
X
X
X
A
X
X
A
X
A
X
X
X
X
A
X
X
X
X
X
X
X
X
A
A
A
A
A
X
( "x" MEANS THE VARIABLE
FMISSION FACTORS
-------
Table IV-1 (con't)
DATA SASE VARIARLESIFIELDSI SUMMARY < "x" MEANS THE VAPIAHLE is AVAILABLE )
DATA BASE
<-
«SITF
YEAR
MAL.DISPIMAL ADJ UISP)
AXLft»Axr>A(AXLE PATIO)
MODI (ENG MODIFICATION)
PCV
TIMING
INT« ( INTERNAL NO. )
VRSN ( VERSION)
VID (VEH. IOENT.)
CLIN (CAR LINE)
OVOR (OVERDRIVE)
BOKE
STRK (STROKE)
RTHP (RATED HP)
ETYP (ENG. TYPE)
**CRB (MO. OF CAr<8S)
FIN (FUEL INJ.)
CMPR (COMP. RATIO)
N/VR ( N/V RATIO)
ECS1 (EM CONT SYS 1 )
ECS? (EM CONT SYS ?)
ECS3 (EM CONT SYS 3)
ECS4. (EM CONT SYS u)
ECSS (EM CONT SYS 5>
FTYP ( FUEL TYPE )
SACL (SALES CLASS)
VSS (SHIFT SPEED CODE)
ENCD (ENG COUE)
UFV1 (DURA FAC VEH)
TRBO (TURBOCHARGED)
VTYP (VEH TYPE)
RCNO (RUNNING CHANGE »)
CRCD (CRITICAL CODE)
TNUM (TEST NUMHER)
TTYP (TEST TYPE)
WCHG (RUNNING CHANGF)
TPRO (TEST PROCEDURE)
RTST (RETEST CODE)
AOHP (ACTUAL OYNO HP)
CTD (CERT TEST DISP.)
FEU (F.E. TEST OISM
TAYR (TEST ACTIVE YP)
ETW (EOUIV TEST WT)
CdKT-EPA EMISSION FACTORS (EF)
75-
79
A
X
X
X
X
X
A
X
X
X
X
X
X
X
A
X
X
X
X
X
X
A
X
A
A
X
X
A
X
X
X
X
X
X
X
X
X
A
80
X
X
X
X
X
A
X
X
A
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
A
X
A
X
X
A
X
X
X
HS . LA'JiV KMSF
81 79 79 78 n fb fS f* /J 12 /9 11
XXXXXX XX
X X X X X x.
X
X
X
X
X
X
A
X
X
X
X
x •
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
IV-4
-------
this, while simultaneously attempting to avoid combining city and highway
data for different configurations of the same vehicle. The last four digits
of the city and highway test numbers were compared. Those test numbers which
were within — 3, for vehicles having the same internal vehicle number and
version number, were used in calculating the combined fuel economy value.
This method of "pairing" city and highway tests allows more than one combined
fuel economy value to be generated for vehicles which had more than one set
of paired tests.
EPA's emission factors data base included data on vehicles from 1972 thru
1981. The EF data base, which was also created specifically to support this
report, included information selected from EPA's emission factors data base.
Table IV-1 lists the field names which were included in the EF file.
Again, only FTPs and HFETs were used. Other tests such as inspection and
maintenance tests and steady state tests were not used.
IV-5
-------
IV. A. Two Variable Linear Regressions
This was the initial step in a series of plotting and regression exercises.
Three large data bases (CERT/EPA, CERT/MFR and EF) were used in this work.
First, two dimensional scatter plots of fuel economy and emissions for each
of the following were generated.
1) UMPG* versus FTPHC**
2) UMPG versus FTPCO**
3) UMPG versus FTPNOx**
4) UMPG versus FEHC***
5) UMPG versus FECO***
6) UMPG versus FENOx***
7) HMPG* versus FTPHC
8) HMPG versus FTPCO
9) HMPG versus FTPNOx
10) HMPG versus FEHC
11) HMPG versus FECO
12) HMPG versus FENOx
13) CMPG* versus FTPHC
14) CMPG versus FTPCO
15) CMPG versus FTPNOx
16) CMPG versus FEHC
17) CMPG versus FECO
18) CMPG versus FENOx
*UMPG, HMPG, and CMPG respectively mean urban fuel economy as measured on
the 1975 Federal Test Procedure (FTP), highway fuel economy as measured on
the EPA highway test and composite fuel economy as calculated by EPA from
urban and highway economy respectively.
**FTPHC, FTPCO, and FTPNOx mean HC, CO, and NOx emissions, respectively, as
measured on the 1975 FTP.
***FEHC, FECO, and FENOx mean HC, CO, and NOx emissions, respectively, as
measured on the EPA highway test.
IV-6
-------
For the sake of brevity, these are called "the 18" plots. These 18 plots
were generated for each of the three data bases for all the data in each
data base. A total of 54 plots were generated.
There was no apparent linear or other relationship between any of the two
variables. Some of the plots are included as appendix 4 to this report.
The various plots that were generated were not totally adequate for
visualizing lines or curve shapes because multiple points in one spot appear
as a number (or x which means 10 or more points at the same spot) which is
difficult to visually weight compared to single points. Linear regressions
were performed that corresponded to each of the plots. A summary of the
regressions is presented in table IV. A-l through table IV. A-3.
Summary
There are several items of interest that can be deduced from these tables.
First, there is no correlation for any of the regressions as evidenced by
values of r-squared which always have a zero as the first digit to the right
of the decimal point. Second, the slopes of the regression lines (COEFF in
the tables) are consistently negative except for occasional, positive values
for HC and CO as measured on the highway fuel economy test. Negative slopes
mean that, as the emissions increase, the fuel economy decreases. Positive
slopes mean, as emissions increase, fuel economy increases. The final item
of interest is that highway HC and CO emissions are not regulated.
No readily apparent linear or other relationship between exhaust emissions
and fuel economy is seen at this level of analysis. Clearly, no negative
impact on fuel economy due to emissions can be identified in these data.
There may be several reasons for this. These reasons include a lack of
stratification of the data and the effects of other variables on fuel
economy. Other approaches used later will be improved in some areas, but by
necessity the number of data points available for analysis will be reduced
with increasing stratification.
IV-7
-------
Table IV. A-l
Two Variable Linear Regressions .lI
'"" l^'tjiJ?1701'''- the EPA Laboratory
CERTIFICATION MICRO DATABASE DATA REGRESSION RUNS SUMMARY
I
00
CASE TAYR XXXX XXXX XXXX
1
2
3
4
5
6
7
8
9
10
1 1
12
13
14
15
16
17
18
FILE: CERT/EPA
XXXX XXXX XXXX XXXX DEP(Y)
UMPG
UMPG
UMPG
UMPG
UMPG
UMPG
HMPG
HMPG
HMPG
HMPG
HMPG
HMPG
CMPG
CMPG
CMPG
CMPG
CMPG
CMPG
FEB. 27, 1981
APTl 1 A I "^ mTAi
AV^ 1 UAL
1 IND(X) SIZE
FTPHC
FTPCO
FTPNOX
FEHC
FECO
FENOX
FTPHC
FTPCO
FTPNOX
FEHC
FECO
FENOX
FTPHC
FTPCO
FTPNOX
FEHC
FECO
FENOX
4605
46O5
4605
2975
2975
2975
2977
2977
2977
4123
4123
4123
2977
2977
2977
2975
2975
2975
1 !_/ 1 M L.
SIZE
5753
5753
5753
5753
5753
5753
5753
5753
5753
5753
5753
5753
5753
5753
5753
5753
5753
5753
PAGE NO.
COEFF
-o
-0
-2
3
0
- 1
-0
-0.
-3
5
0
- 1
-o
-0
-2
3
0
_ i
.5468
. 1968
. 2268
.0094
. 1799
.0281
.2919
.2683
. 1631
.6186
. 1999
.3075
.3914
. 2300
.5551
.8268
. 1933
1262
1
CONST
18.
19
21
18.
18
20.
26
27
30
26.
26.
28
21 .
22
24.
20
21
23,
,814
,627
608
204
.395
, 191
,222
,505
.389
,001
,448
,623
,464
,498
,757
,835
109
O73
R-SO STD.ERR
O
0
0
0.
0
0,
O,
0
0
0
0
O
0,
0
0
0
0.
0.
00092
. 019O7
0481 1
00803
OO280
,02363
. 000 1 2
.O1474
O4854
.01325
OO179
01942
,00033
01606
04694
.O1O32
00257
02252
5
5
5
5,
5,
5
7
7
7
7
7
7
6
6
6
6
6.
6,
.7657
.7131
.6279
,6870
,7019
.6421
.7974
.7401
. 6O62
.8347
,8800
.8102
.4044
.3538
.2533
.3733
3982
3339
-REMARKS
AcLual size is less than total size in this and subsequent computer outputs
due to the method used in construction of the data base. In general the
difference between actual and total size is an indicator that not all urban
tests have paired highway tests and that not all highway tests have paired
iu-l>;-m tests. This difference is not an indicator that data were ignored or
delete:' '< or any reason .
-------
IV. B. Two Variable Linear Regressions with Stratification by Model Year
Various stratifications of the data analyzed in section IV. A of this report
could be done. Intuitively,the stratification of choice appears to be by
emission control system; however, data base limitations would not permit
this stratification without greatly reducing the available data or without
modifying the available data. The stratification chosen was by vehicle
model year (or for certification data by test active year). This
stratification was accomplished with minor losses of data. Stratification
by emission control system was accomplished later by data modification (see
section IV. H).
Plots were generated for each of the three data bases for each of the 18
emission versus fuel economy relationships for each model year. Again, no
linear or other relationships between fuel economy and emissions were seen.
Those plots were not reproduced for inclusion in this report.
Linear regressions were again run using the 18 emission and fuel economy
variables for each of the 3 data bases stratified by model year. The
results of these regressions are presented here in table IV. B-l through
table IV. B-3. In the following discussion of these results, only cases
where sample size ("actual size" in the tables) was greater than or equal to
25 are considered.
Correlation
Of the 126 cases (144 minus 18 cases with less than 25 data points) gen-
erated for the CERT/EPA data base, r-squared did not exceed 0.100 in any
case.
A total of 135 cases were generated for the Emission Factors data base.
r-squared exceeded 0.100 in 9 of these cases. The range of r-squared was
from 0.12 to 0.24 for those 9 cases. The emission variable was CO as
measured on the 1975 FTP in each case. The fuel economy variables were
urban fuel economy for model years 1966 through 1971 and model year 1978,
IV-11
-------
highway fuel economy for model year 1978, and composite fuel economy for
1978. These values of r-squared are considered to represent very weak, if
any, correlation. The six cases between 1966 and 1971 are consistent and
could be further investigated; however, they are for vehicles that are so
old that they are no longer of much importance.
R-squared was over 0.100 (.175 was the actual value, see case number 8) once
for the 93 cases in the CERT/MFR data base. This case was urban fuel
economy as the dependent variable and HC emissions from the FTP as the
independent variable for the 1982 model year. This can be found on table
IV. B-3 line 8. TAYR corresponds to model year.
Slopes of the Regression Lines
The slopes of the regression lines for all ten cases for which r-squared
exceeded 0.100 were all negative.
Even though there was little or no correlation for the regression lines, the
slopes of the other regression lines were investigated. The purpose of
looking at the slopes was to see if there was any consistency in the signs
of the slopes between the 3 data bases. The results are shown in table IV.
B-4. There are two things of interest in this table. First, the in-use
data base almost always shows negative slopes for the emissions/fuel economy
relationships. Malfunctions and maladjustments could possibly explain the
negative slopes for HC and CO. This speculation is not based on the results
of an analysis, however. An explanation for a negative NOx slope could be
the impact of vehicle weight on NOx and fuel economy. This was not
confirmed and is also speculative.
Second, there was some consistency between the CERT/EPA and CERT/MFR data
bases in terms of the numbers of cases with positive slopes.
IV-21
-------
Slopes of the Regression Lines for 1981 and 1982 Model Year Vehicles
The 1981 and 1982 model years are of particular interest in this study
because 3-way catalyst technology is extensively used in these model years
and because 3-way catalyst technology at this point in time appears to be
the technology most vehicle manufacturers will use to meet .emission
requirements in the near future.
«r
It is sufficiently important to repeat that the correlation for tue
regression lines was very weak or non-existent; however, as can be seen from
Table IV. B-5, the slopes of the regression lines for the 1981 model year
are surprisingly consistent between the CERT/EPA and CERT/MFR data bases.
The slopes for the regressions of 1982 model year data are provided for the
few cases for which slopes are available.
Summary
Stratification of the data bases by model year did not provide any real
improvements in correlation between emissions and fuel economy. Consistent
directional trends for the slopes of the regression lines between the 1981
EPA and manufacturer data bases are evident.
IV-22
-------
Table IV. B-4
Direction of the Regression Lines for the Three Data Bases
Dependent
Variable
UMPG
UMPG
UMPG
< UMPG
I
M
U)
UMPG
UMPG
HMPG
HMPG
Stratified by Model Year
Independent Data Base Number of
Variable CERT/ CERT/ Cases with
EPA EF MFR + Slope
FTPHC x 5
x 0
x 3
FTPCO x 3
x 0
x 2
FTPNOx x 2
x 1
x 4
FEHC x 5
x 1
x 4
FECO x 5
x 0
x 3
FENOx x 4
x 1
x 4
FTPHC x 6
x 0
x 3
FTPCO x 3
x 0
x 2
Total Number
of Cases
7
15
6
7
15
6
7
15
6
7
6
5
7
6
.5
7
6
5
7
6
5
7
6
5
-------
Table IV. B-4 (con't)
Direction of the Regression Lines for the Three Data Bases
Dependent
Variable
HMPG
HMPG
HMPG
M
1
£ HMPG
CMPG
CMPG
CMPG
CMPG
Stratified by Model Year
Independent Data Base Number of
Variable CERT/ CERT/ Cases with
EPA EF MFR + Slope
FTPNOx x 1
x 0
x 3
FEHC x 6
x 1
x 4
FECO x 4
x 0
x 2
FENOx x 4
x 0
x 4
FTPHC x 5
x 0
x 3
FTPCO x 3
x 0
x 2
FTPNOx x 1
x 0
x 3
FEHC x 5
x 1
x 4
Total Number
of Cases
7
6
5
7
6
5
7
6
5
7
6
5
7
6
5
7
6
5
7
6
5
7
6
5
-------
Table IV. B-4 (con't)
i
(0
Ul
Direction of the Regression Lines for the Three Data Bases
Dependent
Variable
CMPG
CMPG
Stratified
Independent Data
Variable CERT/
EPA
FECO x
FENOx x
by Model Year
Base
CERT/
EF MFR
x
X
X
X
Number of
Cases with
+ Slope
5
0
3
4
1
4
Total Number
of Cases
7
6
5
7
6
5
-------
Table IV. B-5
Slopes of the Regression Lines for the 1981 & 1982 Model Year
Dependent
Variable
UMPG
UMPG
UMPG
UMPG
UMPG
UMPG
HMPG
Independent Data Base~ Slope
Variable CERT/ CERT/ for '81 MY for '82 MY
EPA EF MFR
FTPHC x - 9.43
x -13.45 -23.25
FTPCO x - 1.06
x
x - 0.45 - 0.33
FTPNOx x 2.30
x
x 0.30 5.72
FEHC x - 8.39
x
x - 0.59
FECO x - 0.86
x
x - 0.52
FENOx x 1.89
x
x 2.84
FTPHC x -16.18
x
x -19.26
-------
I
to
Table IV. B-5 (con't)
Slopes of the Regression Lines for the 1981 & 1982 Model Year
Dependent
Variable
HMPG
HMPG
HMPG
HMPG
HMPG
CMPG
Independent Data Base ~ — Slope
Variable CERT/ CERT/ for '81 MY for '82 MY
EPA EF MFR
FTPCO x - 1.16
x
x - 1.30
FTPNOx x 1.31
x
x 1.32
FEHC x - 9.36
x
x 0.59
FECO x - 1.03
x
x - 0.81
FENOx x 2.19
x
x 3.40
FTPHC x -18.05
x
x -19.61
-------
-------
IV. C. Stepwise Backward Regression Analysis
Stepwise backward regression analysis was investigated as another approach to
see if there was any fuel economy/emissions relationship and if so, what its
nature might be.
The process basically involves starting out with a multiple linear regression
of MPG as a function of several variables. Then the least important variable
as determined by the partial correlation coefficient is dropped and the
analysis repeated and the next least important variable dropped, and so on
until all of the coefficients are at least 0.1. The process can be started
with a large number of variables, the relatively more important variables can
be identified and the less important ones discarded for further analysis.
The independent variables that were chosen to start out the process were:
N/V ratio, road load horsepower, displacement, weight, rated horsepower, and
the six emission values. The dependent variables were the three fuel economy
values.
The 1981 CERT/EPA data base was analyzed as a unit, then stratified by manu-
facturer and transmission type. In contrast to the low values of the
correlation coefficient found in cases using the single independent variables
discussed earlier, the R-squared values obtained with this approach were
substantially higher, typically in the 0.7 to 0.9 range.
We must be aware that this procedure can indicate that a variable is
statistically insignificant for any of several reasons which include:
1. The variable may actually be virtually unrelated to fuel economy.
2. The effects of the variable may be partially masked by one or
more other variables. For instance, the effects of differences
in rated horsepower may be obscured by related changes in com-
IV-2 9
-------
pression ratio or displacement. Similarly, variations in N/V
ratio could be hidden by axle ratio differences. Appendix 8
includes a discussion of collinearity (i.e., to determine whether
the variables are truly independent of one another). Emissions
did not appear to be collinear to any other variable.
3. The variable may strongly influence fuel economy; however, that
influence may be hidden if the range (i.e. variations) of that
variable is very small. For example, the following summary
indicates that the least significant variable in determining the
urban fuel economy for the 1981 Ford vehicles with manual trans-
missions was the reciprocal of the test weight (i.e. ETWM1).
However, an examination of those test vehicles indicated that
there were only two weight ranges. The 3000 to 3125 pound
vehicles were offered with 140 and 200 C.I.D. engines, while the
2375 to 2500 pound cars were offered only with a 98 C.I.D.
engine. Similarly, ETWM1 is also the least significant variable
for Nissan's urban fuel economy, and, again, the maximum
variation of the test weight is only 125 pounds within each basic
engine. Thus, in both of those cases, a variable (i.e. ETWM1)
was indicated to be the least significant when, in actuality,
ETWM1 is one of the most important variables in determining urban
fuel economy.
In reviewing the results of this analysis, it was determined that the HC and
CO highway emissions (FEHC and FECO respectively) were consistently,
statistically insignificant variables, and thus could be omitted when we
perform a multiple linear regression of each of MPG , MPG,, and MPG as
a function of the remaining variables. Results of those regressions (except
those which were stratified by manufacturer) are in appendix 5. Compression
ratio and axle ratio do not appear in these results because they were
eliminated on the basis of partial correlation in earlier results.
Summary tables that describe the results of the process are shown below.
IV-30
-------
Table IV. C-l
SUMMARY OF STEPWISE BACKWARD ANALYSIS
FOR URBAN MPG FOR MODEL YEAR 1981
Level of
Stratification:
Order in which the variables
were dropped:
<3
I
OJ
All Data
Conventional Automatic FEHC
Trans. (CA)
Conventional Manual
Trans (CM)
Lockup Auto.
Trans. (LA)
All Chrysler
All Ford
All GM
All Honda
All Nissan
All Toyota
Ford/CA
GM/CA
Ford/CM
GM/ LA
FEHC
FEHC
FEHC
FEHC
FENOX
FTPCO
FTPCO
FEHC
ETWM1
FEHC
FEHC
VDHP
ETWM1
FEHC
FECO
FTPCO
FECO
FTPCO
DISP
FTPHC
FTPHC
DISP
FECO
FTPNOX
RTHP
FECO
FEHC
RTHP
DISP
FTPHC
FENOX
FECO
FTPHC
RTHP
FECO
FTPHC
FTPNOX
DISP
FTPCO
FEHC
VDHP
FTPHC
FTPNOX
DISP
RTHP
FTPNOX
FEHC
FENOX
FTPNOX
DISP
RTHP
VDHP
NSVR
FENOX
FENOX
VDHP
VDHP
VDHP
FEHC
FECO
VDHP
ETWM1
FEHC
VDHP
NSVR
FECO
FECO FENOX
NSVR
VDHP FECO
FTPNOX NSVR
VDHP FECO
FTPHC VDHP
FTPCO FTPHC
FTPCO FTPHC
FTPCO
FENOX
-------
Table IV. C-2
SUMMARY OF STEPWISE BACKWARD ANALYSIS
FOR HIGHWAY MPG FOR MODEL YEAR 1981
Level of
Stratification:
Order in which the variables
were dropped:
M
U>
ro
All Data
Coventional Automatic
Trans. (CA)
Convent! onan Manual
Trans (CM)
Lockup Auto.
Trans. (LA)
All Chrysler
All Ford
All GM
All Honda
All Nissan
All Toyota
Ford/CA
GM/CA
Ford /CM
GM/LA
FECO
FENOX
FENOX
RTHP
FTP CO
VDHP
FECO
FEHC
FEHC
FTPNOX
FTPHC
NSVR
FEHC
FEHC
FEHC
FEHC
FECO
FEHC
DISP
FEHC
FENOX
ETWM1
ETWM1
FEHC
FTPCO
FTPNOX
NSVR
RTHP
FENOX
FECO
FEHC
FTPCO
FENOX
FTPHC
FTPCO
FTPNOX
FTPNOX
• ETWM1
FEHC
FENOX
DISP
FTPCO
FTPHC
FECO
FTPNOX
FTPCO
FEHC
FTPHC
FTPHC
FTPHC
FECO
DISP
FTPCO
FECO
FTPNOX FTPCO
FTPHC
FEHC VDHP FTPHC
FECO
FTPHC
VDHP FECO FTPCO DISP
FECO DISP VDHP
FENOX
FEHC FECO FTPCO
FECO FENOX
FTPHC
-------
Table IV. C-3
SUMMARY OF STEPWISE BACKWARD ANALYSIS
FOR COMBINED MPG FOR MODEL YEAR 1981
U>
Level of
Stratification:
Order in which the variables
were dropped:
All Data
Conventional Automatic
Trans. (CA)
Conventional Manual
Trans. (CM)
Lockup Auto.
Trans. (LA)
All Chrysler
All Ford
All GM
All Honda
All Nissan
All Toyota
Ford/CA
GM/CA
Ford/ CM
GM/LA
FECO
FEHC
FECO
FEHC
FTPNOX
FTPHC
FTPCO
FEHC
FTPNOX
RTHP
FEHC
FECO
FEHC
RTHP
VDHP
FENOX
FEHC
FECO
FENOX
FTPCO
FECO
DISP
ETWMI
FEHC
FTPHC
NSVR
ETWMI
FEHC
FEHC
FTPHC
DISP
FTPCO
DISP
RTHP
VDHP
ETWMI
FECO
FTPNOX
VDHP
DISP
DISP
FECO
FENOX
FTPCO
VDHP
RTHP
FTPCO
FEHC
FTPHC
FTPNOX
DISP
DISP
FTPCO
FENOX
FTPCO
FECO
FENOX
VDHP
FEHC
FECO
FENOX
VDHP
FEHC
FTPCO
FECO
FEHC
FTPHC
FTPNOX VDHP
VDHP FTPHC
FTPNOX
FTPHC
FTPHC VDHP
FTPNOX-- •
FENOX
-------
IV. D. Residual Analysis
Another analytical approach that was used was a residual analysis based on
multiple independent variable regression equations. Briefly, the approach is
as follows. Get a regression equation for MPG as a function of several
different variables, but not emissions. Compute the residuals by using the
equation to predict the actual data. The residuals are the differences
between, or the ratios of, the predicted and actual MPG data. Plot and
regress .hese residuals versus emissions to see if any relationships can be
seen. Ii this way most of the fuel economy important variables can be
"accounted for" and what is left can be explored to see if there is a residual
fuel economy/emissions relationship. The data base utilized here was the
CERT/EPA base for the 1981 model year (less Diesels).
The form of the basic multiple regression equation to use was considered
next. Rather than specify just one form it was decided to use 3 forms since
it was not known if the form of the basic MPG equation would influence the
result. We looked for equations that were published and that had reasonable
2 2
R values (R values of about 0.9 were considered desirable). Three
equation forms were selected; they are called the Murrell 1975, Bascunana
1979, and Cheng 1981 equations.
The Murrell 1975 equation came from SAE paper 750958, entitled "Factors
Affecting Fuel Economy" by J. Dillard Murrell. The equation contains 5
variables: (CID x N/V~'8), ETW, (RTHP/ETW), (RTHP/CID), and (CMPR°'4
-1)/(CMPR°'4).
The Bascunana 1979 equation came from SAE paper 790654 by Jose L. Bascunana,
entitled "Derivation and Discussion of a Regression Model for Estimating the
Fuel Economy of Automobiles". The 1979 Bascunana equation contains as
independent variables: ETW, displacement, and N/V ratio. In contrast to many
other equations, the 1979 Bascunana equation involves a product of variables
instead of a sum.
IV-34
-------
The 1981 Cheng equation was developed in this study as an adjunct to the
sttpwise backward regression analysis study discussed in section IV. C of this
report.
The general functional forms of the equations are shown below.
Functional Forms of the MPG Equations
Equation Form
1975 Murrell* MPG = A(CID x N/V)~°'8 + B(ETW)~°'67
+ C(RTHP/ETW) + D(RTHP/CID)
+ E(CMPR°'4 -1)/CMPR°*4 + F
1979 Bascunana* MPG = A[(ETW)3 (DISP)b (N/V)°]
1981 Cheng MPG = ACETW"1) + B(VDHP) + C(DISP) + D(RTHP)
+ E(N/V) + F
For the data bases that they were derived from with the specific constants
listed in the references an estimate of the "goodness of fit" is shown below:
2
R as an Indication of "Goodness of Fit"
Equation
1975 Murrell
1975 Murrell
1979 Bascunana
1981 Cheng
1981 Cheng
1981 Cheng
(Type of MPG)
(urban)
(highway)
(composite)
(urban)
(highway)
(composite)
R2 (original ref)
0.89
0.87
0.95
0.90
0.86
0.90
* Equivalent test weight (ETW) was substituted for inertia weight which was
the original variable in the Murrell and Bascunana equations.
IV-35
-------
For the 1975 Murrell and the 1979 Bascunana equations it was thought
inappropriate to use them with their originally determined coefficient
values. Instead, the functional forms were kept the same and new
coefficients were derived using the MY 1981 data base used in this report.
The results are shown in the table below.
Updated Coefficients Using MY 1981 Data Base
1975 Murrell Equation
Coefficients
Coeff .
A
B
C
D
E
F
Original
urban highway composite
7,909 19,930 N.A.
2,824 2,876 N.A.
-66.82 5.839 N.A.
0.1224 -2.259 N.A.
34.76 39.37 N.A.
-19.74 -25.24 N.A.
R2
Updated
urban
7,501
5,764
-47.36
-5.025
16.20
-17.03
0.90
highway
29,878
3,956
227.2
-16.80
-1.843
-10.69
0.88
composite
14,989
5,602
36.28
-10.09
-1.700
-8.915
0.91
1979 Bascunana Equation
Coefficients
Coeff.
A
a
b
c
Original
urban highway composite
N.A. N.A. 37,246
N.A. N.A. -0.4653
N.A. N.A. -0.4031
N.A. N.A. -0.4150
R2
urban
141,482
-0.8659
-0.1889
-0.2403
0.90
Updated
highway
959,401
-0.7840
-0.3301
-0.6450
0.88
composite
436,918
-0.8831
-0.2323
-0.4086
0.90
IV-36
-------
1981 Cheng Equation
Coefficients
A.
B.
C.
D.
E.
F.
urban
70,077
0.2908
-0.0044
-0.0291
-0.1047
4.0439
R2 = 0.90
highway
80,599
-0.4058
-0.0301
-0.0271
-0.4356
34.546
0.86
composite
77,444
0.0458
-0.0084
-0.0366
-0.2187
13.560
0.90
Using the updated coefficients, and the original functional forms for all 3
equations, a residual analysis was carried out. Two forms of residuals were
studied, actual MPG minus predicted MPG, (called the delta-residual), and
actual MPG divided by predicted MPG, (called the ratio-residual). The
residuals were scatter plotted against the six emission variables. The
initial stratification was based .on all the 1981 data implying 108 scatter
plots (3 fuel economy dependent variables x 3 equations x 2 residuals per
dependent variable x 6 variables). The plots did not suggest any linear or
other relationship between the residuals of fuel economy and emissions.
Upon examining the results of the 108 plots, linear regressions were performed
2
for all the plots, and all the r terms were so small as to be
insignificant. The regression results are presented in appendix 6. The
2
largest of the r values, 0.042, was for the ratio-residual of combined fuel
economy (MPG ) regressed against urban NOx (FTPNOX). This result implies
that only 4.2% of the variability in the percentage by which the actual MPG
differs from the MPG predicted by the 1979 Bascunana equation could be
explained as a linear function of the urban NOx; the remaining 95.8% was not
2
related to FTPNOX. Since most (58 out of 108) of the r values were less
2
than 1% (the average r was only 1.1%), only a small proportion of the total
variation of any of the 18 residuals can be explained by a linear function of
the 6 emission variables.
IV-37
-------
Another way to consider this type of information is to forget residual
2
analysis and use improvement in R and standard error as an indicators that
emissions do or do not relate to fuel economy-. The following data illustrate
multiple regressions using the Cheng equation with and without the six
emissions as independent variables.
Dependent
Variable
UMPG
HMPG
CMPG
—Without Emissions—
Standard
R2 Error(SE)
.898
.864
.897
1.760
2.776
2.064
—With Emissions— -Change in:—
Standard
R2 Error(SE) R2 SE
+.005 +.059
+.003 +.103
+.005 -.038
.903
.867
.902
1.819
2.879
2.026
These data indicate that inclusion or exclusion of the urban and highway
emissions in the Cheng equation makes virtually no difference. It can be
concluded that for the 1981 EPA certification data base, exhaust emissions
have essentially no identifiable relationship to fuel economy after accounting
for variability in fuel economy due to vehicle design parameters. This
conclusion is restricted to the 1981 EPA certification data base due to
somewhat limited ranges for emissions (these are expanded in section IV. H)
and collinearity of variables in the Cheng equation (see appendix 8).
IV-38
-------
Section IV. E. Subconfiguration - Matched Data Analysis
Another analysis method was to restrict the study to the 1981 CERT/EPA data
base and then stratify by the following ten (10) parameters:
1. Manufacturer
2. Engine Family
3. Displacement
4. Vehicle I.D.
5. Transmission
6. ETW
7. Dynamometer Horsepower
8. Axle Ratio
9. N/V
10. Compression Ratio
The purpose of such a stratification was to identify, as closely as
possible, the effect of engine calibration on emissions and fuel economy for
each vehicle that was tested at the EPA laboratory in more than one
configuration. Also this tended to eliminate test variability which was due
to either differences among the test vehicles or to differences among the
vehicle testing laboratories. The disadvantage of such a stratification was
that only 27 such vehicles were identified, each of which was tested at EPA
in at least two (2) versions.
In examining the results the following relationships between the urban fuel
economy (MPG ) and the regulated emissions were seen.
1. For FTPHC:
a. MPG increased as FTPHC decreased, in 18 of the 27 cases,
IV-39
-------
b. MPG decreased as FTPHC decreased, in 8 of the 27 cases, and
u '
c. MPG remained unchanged in one (1) of the 27 cases.
2. For FTPCO:
a. MPG increased as FTPCO decreased, in 23 of the 27 cases, and
b. MPG decreased as FTPCO decreased, in four (4) of the 27
cases.
3. For FTPNOX:
a. MPG increased as FTPNOX decreased, in nine (9) if the 27
cases, and
b. MPG decreased as FTPNOX decreased, in 18 of the 27 cases.
The following chart illustrates the seven (7) combinations of outcomes that
we found in examining those 27 groups of data.
As MPG increases, the level of the corresponding emissions were
observed to:
Outcome 1
Outcome 2
Outcome 3
Outcome A
Outcome 5
Outcome 6
Outcome 7
Number of
Occurrences
10
7
5
2
1
1
1
FTPHC
Decreases
Decreases
Increases
Increases
Increases
Decreases
Unchanged
FTPCO
Decreases
Decreases
Decreases
Increases
Increases
Increases
Decreases
FTPNOX
Increases
Decreases
Increases
Increases
Decreases
Increases
Decreases
IV-40
-------
Thus, from this rather limited set of data (shown in table IV. E-l), we can
infer that decreasing the level of the FTPHC or FTPCO emissions is not
usually detrimental to MPG . But decreasing the level of FTPNOX emissions
usually accompanies a decrease in MPG .
IV-41
-------
Table IV. E-l
t
UEfi
FORD
FORO
FORD
FORO
FORO
FORD
FORO
FORO
FOHD
FORO
FORO
FORD
FORD
FORO
FURD
FORO
FORO
FORO
FORD
FORD
FORD
88
vin
1E2-1.6-F-441
1E2-1.6-F-441
1K2-2.3-C-2B5
1K2-2.3-C-285
1K2-2.3-C-2BS
1K2-2.3-C-2B5
1K2-2.3-C-28S
1K2-2.3-G-286
1K2-2.3-G-286
1K2-2.3-G-286
1K2-2.3-G-2B6
1K2-2.3-G-286
1Z2-2.3-C-290
122-2. 3-C-290
1Z2-2.3-C-290
1S1-4.2-F-345
1S1-4.2-F-345
1A1-5.8W-G-259
1A1-5.8W-G-259
1A1-5.8H-G-259
1AI-5.8W-G-2S9
COMQ79
COM079
V
E
B
1
3
4
5
S
0
0
1
1
1
0
1
1
1
2
0
1
1
2
i
UQUE
6.4
6.4
10.4
10.4
10.4
10.4
10.4
10.4
10.4
10.4
10.4
10.4
8.9
8.9
B.9
11.0
11.0
12.6
12.6
12.6
12.6
18:*
CLUE
98.0
98.0
140.0
140.0
140.0
140.0
140.0
140.0
140.0
140.0
140.0
140.0
140.0
140.0
140.0
255.0
255.0
351.0
351.0
351.0
351.0
131:8
$
L
4
*
4
4
4
4
4
4
4
4
4
4
4
4
4
8
B
8
8
8
B
8
C B F
R B I
a L 2
1 2 N
1 2 N
1 2 N
1 2 N
1 2 N
1 2 N
1 2 N
1 2 N
1 2 N
1 2 N
1 2 N
1 2 N
1 2 N
1 2 N
1 2 N
1 2 N
1 2 N
1 2 N
1 2 N
1 2 N
1 2 N
1 §N
COMP.
BallQ
B.8
8.8
9.0
9.0
9.0
9.0
9.0
9.0
9.0
9.0
9.0
9.0
9.0
9.0
9.0
8.2
8.2
8.3
6.3
8.3
8.3
gig
AXLE
T
R
BAIIfl UlY H.
3.59
3.59
3.08
3. 08
3.08
3.08
3.08
3.08
3.08
3.08
3. OB
3.08
3.45
3. 45
3.45
3.08
3. OB
2.73
2.73
2.73
2.73
§:iU
43.0
43.0
43.0
43.0
43.0
43.0
43.0
44.0
44.0
44.0
44.0
44.0
37.0
37.0
37.0
28. 0
28. 0
23.0
23.0
23.0
23.0
3H
J£ • O
M4
M4
M4
M4
M4
M4
M4
M4
M4
M4
M4
M4
MS
MS
MS
L4
L4
L4
L4
L4
L4
b3
EMISSION
CONTROL
SISIEU
EGR.PMP.OXD.RED
EGR.PMP.OXD.RED
EGR.PMP.OXD.3CL
EGR.PMP.OX0.3CL
EGR.PMP.OXD.3CL
EGR.PMP.OX0.3CL
EGR.PMP.OXD. 3CL
EGR.PMP.OXD.3CL
EGR.PMP.OXD.3CL
EGR.PMP.OXD. 3CL
EGR.PMP.OXD.3CL
EGR.PMP.OXD.3CL
EGR.PMP.OXO
EGR.PMP.OXD
EGR.PMP.OXO
EGR.PMP.OXD.RED
EGR.PMP.OXO. RED
EGR.PMP.OX0.3CL
EGR.PMP.OXD.3CL
EGR.PMP.OX0.3CL
EGR.PMP.OXD.3CL
i§B:WP:85B:3a
irsTg
80534B
806864
806678
806574
807680
807366
807453
805808
805901
806116
606183
80637S
805831
B06191
80630S
B05B84
806720
805919
806204
806299
806572
mm
E.m_
2375
2375
3125
3125
3125
3125
3125
31 ^5
3125
3125
3125
3125
3125
3125
3125
3750
3750
4500
4500
4500
4500
388
UtiEfi
27.9
27.3
21.3
21.4
21.4
22.1
22.2
22.9
23.0
23.3
23.2
23.0
21.2
20.0
20.6
18.1
17.0
13.8
13.9
13.6
13.7
18:*
GM
GM
GM
GM
GM
GM
GM
JRT
JRT
SAAB
SAAB
UEHC CIPCQ EJEH04
P0288
P0288
C09023
COB023
COB023
CA7403-A
CA74Q3-A
XJ52/16
XJS2/16
JKT XJ52/1S
JHT XJ52/1S
JRT XJ52/15
JRT XJ52/15
668
668
TOYOTA 81-FE-4
TOYOTA 81-FE-4
TOYOTA 81-FE-17
TOYOTA 81-FE-17
TOYOTA fll-FE-13
TOYOTA 81-FE-13
1
1
1
2
0
2
0
1
0
0
0
1
2
0
1
2
3
0
1
11:1
10.3
10.3
10.3
14.5
14.5
11.3
11.3
11.3
11.3
11.3
11.3
9.2
9.2
7.8
7.8
10.2
10.2
11.5
11.5
265.0
265.0
268. S
268.5
268.5
367.0
367.0
258.0
256.0
258.0
258.0
258.0
258.0
111:1
88.6
88.6
108.0
108.0
144.4
144.4
i
8
B
8
8
8
6
6
6
6
6
6
4
4
4
4
4
4
4
4
1
1
1
1
0
0
0
0
0
0
0
0
0
0
1
1
1
1
1
2
2
2
2
0
0
0
0
0
0
0
0
0
0
2
2
2
2
2
2
fi
N
N
N
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
N
N
N
N
N
N
8:3
8.3
8.3
8.3
8.2
8.2
8.
B.
8.
B.
8.
8.
?:i
9.0
9.0
9.0
9.0
9.0
9.0
§:J1 31:8 t3 i8B:Mp;8!B:3Eh
2.41 30.2 L3 EGR.PMP.OX0.3CL
2.41 30.2 L3 EGR.PMP.-OX0.3CL
2.41 30.2 L3 EGR.PMP.OXD.3CL
3.08 38.2 A3 EGR.PMP.OXD.3CL
3.08 38.2 A3 EGR.PMP.OXD.3CL
3.07 40.6 A3 3CL
3.07 40.6 A3 3CL
3.07 40.6 A3 PMP.3CL
3.07 40.6 A3 PMP.3CL
3.07 40.6 A3 PMP.3CL
3.07 40.6 A3 PMP.3CL
3.89 43.7 MS EGR.3CL
3.89 43.7 MS EGR.3CL
3.31 51.7 K4 EGR.PLS.OXD
3.31 si.7 M4 EGR.PLS.OXO
3.73 55.6 A3 EGR.PMP.3CL.OTR
3.73 55.6 A3 EGR.PMP.3CL.OTR
3.58 34.5 A4 EGR.PMP.3WY.OTR
3.58 34.5 A4 EGR.PMP.3WY.OTR
R03041
804167
803482
803S69
805865
606869
807294
806388
806429
806478
806524
80666
-------
Section IV. F. Special Engine Over Time Trends
In an attempt to examine the question of fuel economy versus emission stan-
dards, a review of five major engine displacement groupings was undertaken
to see the effects on fuel economy with increasingly stringent Federal
emission standards. The model years selected (1979, 1980, and 1981) were
chosen because these three years have three different sets of emission
•»
standards and represent the most recent automotive and emission control
technology. The engine groupings chosen represented high production volume
designs that in turn represent the trend for future automotive applications.
The objective was to match up, within the limits of the test data which were
used to generate the EPA fuel economy label values,* similar vehicles of
differing model year, and see how fuel economy varied with differing
emission levels. The matching was to incorporate as many possible vehicle
and powerplant design parameters as the data base was capable of providing,
in order to eliminate to the maximum extent possible the influence of
vehicle and power plant design differences. This was not always possible,
due to the variability of product offerings from model year to model year.
In every case, however, the closest possible matching was arrived at
utilizing the EPA Test Car List data base. The power plant and vehicle
design parameters considered for matching are noted below:
Powerplant Design Parameters
Swept Volume (displacement)
Induction System (carburetor design i.e., 2 barrel, 4 barrel, etc.)
Compression Ratio
Manufacturer's Rated Power Output
Emission Control System
* EPA Test Car List, Second Edition 1979, Second Edition 1980, and Second
Edition 1981.
IV-43
-------
Vehicle Design Parameters
Weight
Transmission Type
Axle Ratio
N/V Ratio
Road Load Horsepower (RLHP)
By "walking through" the noted model years and matching the above noted
parameters to the maximum extent possible within a given engine displacement
grouping, data were compiled to address the fuel economy versus emissions
question. The following five tables are the results of this "walk through"
procedure. After carefully reviewing the tables and their associated
graphs, it becomes apparent that no readily discernible trend is evident
with respect to fuel economy versus emissions since there are instances when
fuel economy increases accompany both increases and decreases in emissions.
These data are all plotted and included in this document as appendix 7.
Looking at the data for various engine families, some tentative conclusions
can be arrived at.
Ford 2.3 L Powerplant (Table IV. F-l)
For automatic transmission equipped vehicles, the slight increase in fuel
economy is of the magnitude expected with a decrease in RLHP with
progressing model years. For manual transmission vehicles, an increase in
fuel economy was noted with decreasing emissions, which coincided with a
major change in emission control technology.
Chrysler 1.7 L Powerplant (Table IV. F-2)
For automatic transmission equipped vehicles, there was no consistent change
in fuel economy, either city or highway, as emissions decreased. As the
emissions decreased each year from 1979 to 1981, the fuel economy dropped in
1980 and then increased in 1981.
iy-44
-------
For manual transmission vehicles, fuel economy increased as emissions
decreased. Thus both the automatic and manual transmission equipped
vehicles exhibited fuel economy increases which coincided with the
incorporation of more advanced emission control technology.
General Motors 1.6 L Powerplant (Table IV. F-3)
There were two versions of tb*e GM 1.6 L engine reviewed, the base or stan-
dard engine and an optional H.O. (high output) version. Fuel economy
results with respect to emissions differed markedly between these two ver-
sions of the engine. For the standard engine with automatic transmission,
there was a significant increase in fuel economy with decreasing emissions.
For the manual transmission equipped version of the standard engine there
was a decrease in fuel economy between 1979 and 1980, only partly
attributable to increased road load horsepower. Again, there was an
increase in fuel economy coinciding with the utilization of an advanced
emission control system. For the H.O. version of the engine, a different
set of results emerged. Both the automatic transmission and manual
transmission versions of this engine exhibited noticeable declines in fuel
economy with decreasing emission levels. It is not obvious from the data
available why the fuel economy versus emissions relation is so different for
the standard and high output versions of this powerplant.
General Motors 2.5 L Powerplant (Table IV. F-4)
For both vehicles equipped with either automatic or manual transmissions,
the significant drops in fuel economy appear to be associated with the
engine being used to power front-wheel drive (FWD) vehicles instead of
rear-wheel drive (RWD) vehicles. This drop in tested fuel economy might not
be indicative of a real difference in the in-use fuel economies. (There is
evidence to the effect that rear-drive vehicles have larger offsets of
actual in-use fuel economy versus dynamometer fuel economy test results than
do front-drive vehicles.*) The slight drop in fuel economy for the RWD
* - Neil South, "1978 to 1980 Ford On-Road Fuel Economy," SAE Paper 810383,
February 1981.
- Schneider, et al., "In-Use Fuel Economy of 1980 Passenger Cars," SAE
Paper 810384, February 1981.
- Section III. D of this report.
IV-47
-------
vehicles equipped with automatic transmissions (from 1979 to 1980) is of the
magnitude expected with a 125 pound increase in test weight. There is a
slight drop in fuel economy for the FWD vehicles equipped with manual
transmissions from 1980 to 1981. For the automatic transmission equipped
FWD vehicles, the increase (from 1980 to 1981) coincides with the use of
more advanced emission control technology on the 1981 models. However,
since the cylinder heads also underwent substantial design changes during
those three years, it is known that other factors are probably influencing
the fuel economy, too.
General Motors 3.8 L Powerplant (Table IV. F-5)
For automatic transmission equipped vehicles, the increase in fuel economy
appears to be attributable primarily to the usage of a lock-up torque
converter on MY 81 vehicles. For manual transmission vehicles the slight
decrease in fuel economy between 1979 and 1980 appears to be partly
attributable to an increase in test weight and rear axle ratio. Again, an
increase in fuel economy coinciding with the adoption of an advanced control
system can be seen.
Summary
After reviewing the comparisons, the only apparent trend (with the exception
of a 2.1% drop in fuel economy associated with the 2.5 L vehicles equipped
with manual transmissions) is an increase in fuel economy from the 1980 to
the 1981 model year, which coincides with the replacement of oxidizing
catalyst systems with more sophisticated 3-way catalyst systems. The
average increase in combined fuel economy (MPG ) that accompanied
replacing an open-loop oxidation catalyst system with a closed-loop 3-way
catalyst system was 5.3%. However, since this "trend" includes increases in
fuel economy as small as 1.1%, we tested this trend by comparing 1980
Federal test vehicles with similar 1980 California vehicles. The 1980
Federal vehicles were all equipped with oxidation catalyst systems while the
California vehicles were all equipped with 3-way catalyst systems.
IV-50
-------
The ten (10) comparisons between similar 1980 MY Federal vehicles and 1980
MY California vehicles (detailed in the previous 5 tables) indicate an
increase in fuel economy with the addition of a 3-way, closed-loop system
occurred in 7 out of the 10 cases. The average increase in fuel economy,
for the 10 cases, which accompanied the utilization of a closed-loop 3-way
catalyst system was 2.6%.
Thus, there appears to be a trend indicating a slight increase in fuel
economy when an open-loop oxidizing catalyst system is replaced with a
closed-loop 3-way catalyst system. However, since the amount of data is
limited and the absolute value of the fuel economy improvement is not large
relative to test variability, all that can be said is that, based on this
engines over time study, there appears to be no identifiable fuel economy
penalty which can be attributed to changes in the Federal emission standard
between 1979 and 1981.
IV-52
-------
IV. G. Sensitivity Coefficient Study
In yet another way to investigate fuel economy/emissions relationships a
sensitivity analysis was undertaken. A sensitivity coefficient can be thought
of as sort of a normalized derivative, and is usually expressed as percent
change in the dependent variable per percent change in independent variable.
Suppose there are two points a and b, with FE , FE. . and if for this
a b
example the.pollutant is HC, then we have HC and HC, .
3. 0
Let FE = (FE + FE )-r2, HC = (HC + HC,)-f 2,
and AFE = FE, - FE , AHC = HC, - HC .
b a' b a
.Then the sensitivity, call it SFEHC, is equal to A FE/FE-5-AHC/HC .
Besides giving an indication of the magnitude of the effect, the algebraic
sign of the sensitivity coefficient can contain useful information. Since the
quantities FE and emissions are positive, the sign of the sensitivity
coefficient is determined by the signs of the deltas. Four cases are
possible, and they are shown below:
AEmissions
AFE
AFE > 0
AFE < 0
AEmissions > 0 AEmissions < 0
S > 0
S < 0
S < 0
S > 0
Two cases are possible for the sensitivity coefficient. Either it is positive
or negative. If it is positive then whenever fuel economy goes up, so do
emissions, and vice versa. If the sensitivity coefficient is negative, then
then when fuel economy goes up, emissions go down, and vice versa.
What was done was to search the CERT/EPA data base (for 1975 to 1982 model
years) for multiple tests of the same vehicle. From the multiple tests the
maximum number of sensitivity coefficients were determined.
IV-53
-------
The results are shown below:
Variables
for
MPG
Pollutant
No. of Tests Sensitivity Coefficient
N Mean Max Min Sigma
MPG Urban HC Urban
1026 +0.001 31.8 -12.8 1.62
MPG Urban CO Urban
MPG Urban NOx Urban
1027 -0.079
6.5 - 8.7 0.75
1027 -0.132 10.5 -16.6 1.38
MPG Hwy
HC Hwy
432 -0.035
3.9 -19.3 1.09
MPG Hwy
CO Hwy
432 -0.060
0.9 - 3.5 0.35
MPG Hwy
NOx Hwy
432 -0.071
8.7 -15.2 1.48
The average sensitivity coefficients are small and the standard deviations are
large. This means that for all practical purposes, the average sensitivity
coefficient is neither positive nor negative, it is zero, meaning that as
often as fuel economy goes up when emissions go up, the opposite case
happens. If anything, the fact that 5 out of the 6 average sensitivity
coefficients were negative would lead one to suspect that the relationship
hinted at was one of "emissions down-fuel economy up". However, what is
concluded is that the sensitivity coefficient study did not discover any
relationship between fuel economy and emissions. If anything, given the data
in the above table, it almost appears as if the results might be from
independent parameters, i.e. two independent parameters randomly varying.
IV-54
-------
Section IV. H. Data Stratification By Emission Control System
As a first step in the analysis, each of the three fuel economies (UMPG,
HMPG, and CMPG) were plotted against each of the three regulated emissions
(FTPHC, FTPCO, and FTPNOX) as well as against the three emissions from the
highway test cycle (FEHC, FECO, and FENOX). The data examined were
restricted to:
1. Test data generated at EPA's laboratory (the CERT/EPA data base) to
eliminate any possible variation due to differences in laboratories,
2. Spark-ignition vehicles (i.e., no Diesels),
3. 1979 through 1981 model year vehicles, in order to make the
analysis more relevant to present and future vehicles, and
4. Non-durability vehicles, since durability vehicles with test data
up to 50,000 miles and without any matching highway test data were
of limited usefulness in this analysis.
However, this first set of 18 graphs (see appendix 4 for similar graphs) did
not indicate any relationship between fuel economy and exhaust emissions.
This result was expected since regressions of fuel economy against each
emission produced very small values of r-squared. (See section IV. A.)
Since the highway fuel economy (HMPG) had a weaker relationship to the
regulated emissions (FTPHC, FTPCO, FTPNOX) than did either the urban fuel
economy (UMPG) or the combined fuel economy (CMPG), subsequent analyses were
restricted to trying to relate UMPG or CMPG to FTPHC, FTPCO, or FTPNOX.
The next step was to stratify the data by the emission control technology
employed. The following eight (8) emission control systems were identified
which are representative of a majority of the systems that are currently in
use as well as the ones which will be used in the near future:
IV-55
-------
Emission Control Technologies
Fuel
Injected
System 1
System 2
System 3
System 4
System 5
System 6
System 7
System 8
No
No
No
No
No
No
Yes
Yes
EGR?
Yes
Yes
Yes
Yes
Yes
Yes
No
Yes
Ox
Cat?
Yes
Yes
Yes
Yes
No
Y/N
Y/N
Y/N
3 Way
Cat?
No
No
No
Yes
Yes
Yes
Yes
Yes
3CL?
No
No
No
Yes
Yes
No
Yes
Yes
Air
Pump?
No
No
Yes
Y/N*
Y/N
Yes
Y/N
Y/N
Pulsating
Air Sys?
Yes
No
No
Y/N
Y/N
No
Y/N
Y/N
* Y/N means either Yes or No.
(Note: While the above eight systems are non-overlapping, they do not
represent all technologies.)
Stratifying the data by emission control technologies increased the number
of graphs by a factor of 8 and accordingly decreased the amount of
variation. These graphs showed a slight tendency for an increase in fuel
economy with a decrease in emission levels; however, since that tendency
might have been a result of ETW or transmission differences, further
analysis was carried out. Since it was possible that much of this variation
was due to differences in the test weight (ETW) of the vehicles, fuel
economy results were plotted against the emissions and stratified by both
ETW and emission control technologies. However, this approach produced too
many graphs, most of which lacked a sufficient number of points upon which
to make a statistically sound analysis.
In order to incorporate both ETW and the emission control technology but
still have enough data per graph to analyze, the data were stratified by
only the technology and then a new variable was plotted. The new variable
was fuel economy multiplied by ETW and divided by 2000 (i.e. ton-miles per
gallon). This new variable was plotted against each of the three urban
emission results. This ton miles per gallon variable is essentially a
measure of efficiency for the vehicle. By using
IV-56
-------
this variable, we were able to reduce much of the variation in fuel economy
caused only by differences in ETW. Stratifying by control system produced a
set of 48 graphs. The graphs for UMPG vs FTPNOX are presented in figures
IV. H-l through IV. H-8. The remaining 40 graphs are in appendix 9. An
additional set of 48 graphs was created by plotting gallons per ton-mile
(i.e., the reciprocal of ton-miles per gallon) against each of the regulated
emissions. These graphs suggested that these variables, which are a measure
of efficiency, are independen: of the emission results. If we assume that
each of these data poirts represents vehicles having acceptable
driveability, then there is little difference among the groups in the
minimal levels of FTPHC and FTPCO attainable. However,.for FTPNOX, the fuel
injected, 3-way, closed-loop vehicles attain 1/3 to 1/2 the emissions of the
carbureted, 3-way vehicles, which attain about 1/2 to 2/3 of the emissions
of the carbureted vehicles without a 3-way catalyst.
A fifth set of 138 graphs related fuel consumption per ton (i.e. gallons per
ton-mile) for the urban and combined test cycles to each of the three
regulated emissions and stratified by transmission class (CA, CM, and LA) as
well as by control system. A similar set of 132 graphs (see appendix 10 for
the graphs which contain at least 12 points), relating ton-miles per gallon
to each of the three regulated emissions and also stratified by transmission
class, was generated. This sixth analysis had six fewer graphs than the
fifth analysis because the sixth analysis did not include any test data
prior to the 1980 model year while the fifth analysis additionally used 1979
data. These two analyses confirmed the conclusions drawn from the third and
fourth sets of graphs; that is, the efficiency of the vehicle (as measured
in ton-miles per gallon) appears to be independent of the emission results.
To assist in analyzing the graphs of ton-miles per gallons versus the urban
emissions, the upper and lower ranges on each graph were sketched in. (see
appendix 10). These curves were drawn by visual inspection and are not
IV-57
-------
SCATTER PLOT <1> SYSITOBU CASES=TAYft:(79-81)
N= 1B4 OUT OF 245 M.UTE.S.2K VS. 3.FTPNOX
UTE.S.2K
55.000 *
Figure IV. H-l
50.000 *
Urban Ton-Milea per Gallon versus
FTP NOx Emissions
Emission Control System 1
45.000 *
40.000 +
« * «
«
« • * «
Ln
00
35.000
30.000
» » «
2 «
* »
> « 2 «2 ft * *
3 »2 «* » 2 * **2 *
« ft
*
« •« « «23
«• « «2 2 ft
« « 0«
2«
25.000
20.000
15.000 »
10.000 *
4
0.
.50000
1.0000
l.bOOO
2.0000 FTPNOX
-------
N= 162 OUT OF 22V 41.UTE.S.2K VS. 3.FTPNOX
UTE.S.2K
55.000 *
Figure IV. H-2
50.000 *
Urban Ton-Hlle* per Gallort veraus
FTP NOx EmiasionB
Emission Control System 2
45.000
40.000 +
35.000
30.000 «
«»»« 2* « «
»» « o
2 * * » *»
« 2* 2 2 ** 2 *
2 2° ° «»«
3 2° '2
« • o o
« «
0 «
00
« »
25.000 »
20.000 *
15.000 *
10.000
0. .50000 1.0000 1.5000 2.0000 FTPNOX
.25000 .75000 1.2500 1.7500 2.2500
-------
N= 176 OUT OF 243 41.UTE.S.2K VS. 3.FTPNOX
UTE.S.2K
55.000 »
Figure IV. 11-4
50.000 «
Urban Ton Miles per Gallon versus
tie NUx Emission!?
Emission Control System 4
45.000 »
40.000 *
I
ON
35.000 *
30.0UO
« o
2
« «. ««»«« » 2
2 « »3«««2 « * ««
« » 2 ** £»«««« o
22242«6 433 * •»
a « ^^>0 oo »
•• 2 3 » » 2* °3 2 *
2* 2° 22 ° ««2 °2 * *
« oo ^°4 »«» o « o
25.000
20.000 +
15.000
10.000 +
4
0.
.50000
.25000
.75000
1.0000
l.SOOO
1.2500
2.0000 FTPNOX
1.7500
-------
SCATTER PLOT <5> SYS1T0885 CASES=TAYK:<79-81)
N= 129 OUT OF 164 41.UTE.S.2K VS. 3.FTPNOX'
UTE.S.2K
55.000 *
Figure IV. H-S
50.000
Urban Ton-Miles per Gallon versus
FTP NOx Emissions
Emission Control System 5
45.000 »
40.000
35.000
30.000 «•
2«
0*00
« «• a
2 »
003 0 00 0 0 0
P 0 00 00 ?0
3
2 «
» «
» «»
« «»
»» 2 2
25.000 +
20.000
15.000
00
0
10.000
^^^*
00
.7
_ _ 2
-------
UTE.S.2K
5S.OOO
N= 65 OUT OF 88 M.UTK.S.2K VS. 3.FTPNOX
Figure IV. H-6
50.000
Urban Ton-Miles per Gallon versus
r'TP NOx Emissions
Emission Control System 6
45.000 *
40.000 »
I
ON
U>
35.000
30.000
«
o
« «
o
«« » 3 «
oo oo o
o o£o » oo
« oo « o
Oi» O »
2<>« o o o
oo «
25.000
20.000 »
15.000 »
10.000 »
0. .50000 1.0000 1.5000 2.0000 FTPNOX
•25000 .75000 1.2500 1.7500 2.2500
-------
SCATTER PLOT <7> SYS1T08I7 CASES=TAYRI<79-81)
N= 145 OUT OF 203 41.UTE.S.2K VS. 3.FTPNOX
UTE.S.2K
55.000 »
Figure IV. H-7
50.000 *
Urban Ton-Miles per Gallon versus
FTP NOx Emissions
Emission Control System 7
45.000
40.000 *
35.000 +
30.000
* ««2 °?3222 « •« »
2 «3» « 2 ««2
* * • 52 2"**2 • • «
« «
o *
« ««
» »
« « •
»«
25.000
20.000 «•
15.000 *
10.000 *
UUMl
frff
ox
on
-------
SCATTEH PLOT <8> SvSlTUBIH CASES=TAYRS(79-81)
N= 55 OUT OF 7t) <• 1 .UTF. .S.2K VS. 3.FTPNOX
UTE.S.2K
55.000 *
Figure IV. H-8
50.000
Urban Ton-Miles per Gallon versus
FTP NOx Emissions
Emission Control System 8
45.000 *
40.000
35.000 *
2
« «
« • »
• « «« «
30.000
25.000 « « 2
20.000 *
» o
o
15.000
10.000
0. .50000 1.0000 1.5000 2.0000 FTPNOX
.?5000 .75000 1.2500 1.7SOO 2.2500
-------
meant to represent a specific functional relationship between fuel economy
and emissions; they are meant only to indicate upper and lower bounds for
any such functional relationship that may exist. The plots were then
compared to determine whether a given system yielded superior performance
(i.e. higher ton-miles per gallon at every level of a given emission). (See
figure Toyota -1 in section V. D. for an idealized relationship between
systems.) The only such instance we detected was that system number 1
produced higher ton-mile per gallon results at given emission levels than
did system number 3. It may be possible that relationship resulted from the
smaller parasitic horsepower loss of the pulsating air system compared to an
air pump system.
There is an obvious problem, however, in using ton-miles per gallon to
analyze the data; that is, the weight of the vehicle is not the only
variable which is changing. To compensate for the diversity in engines,
tires, axles, et cetera, the test data were examined in each of the 24
groupings obtained by stratifying by the 8 emission control technology
systems and by the 3 transmission classes. Within each of the 24 groups,
the mean (average) of each of the following: 1/ETW, dynamometer horsepower
(VDHP), engine displacement (DISP), rated horsepower (RTHP), and N/V ratio
(NSVR) were calculated. We then used the 1981 Cheng Equation (see section
IV. D.) to obtain an "adjusted" fuel economy described by the following
equation:
Adjusted MPG = Actual MPG + A [ mean (ETW"1) - actual "
+ B [ mean (VDHP) - actual (VDHP)]
+ C [ mean (DISP) - actual (DISP)]
+ D [ mean (RTHP) - actual (RTHP)]
+ E [mean (NSVR) - actual (NSVR)]
3-V-&6'
-------
These "adjusted MPG" values are actually the measured fuel economy results
plus (or minus) a correction factor, predicted by the 1981 Cheng Equation,
which approximate the expected results if all the tests within each of those
24 groups had been performed on vehicles of the same test weight, the same
VDHP, the same engine size and power, and the same N/V ratio. We then
plotted these adjusted MPG values against the corresponding regulated
emissions. (See appendix 11.) Care should be taken in evaluating the
results of this procedure. (See the discussion on collinearity in appendix
8.)
In examining the plots of adjusted UMPG versus FTPNOX, it was found that 20
graphs each contained at least 12 points. A (two variable) linear
regression analysis was performed for each of those 20 sets of data. The
analyses indicated that 10 sets of data displayed a tendency for the
adjusted UMPG to decrease as the NOx emissions decreased. The other 10 sets
displayed the opposite (i.e., the tendency for the adjusted UMPG to increase
as the NOx emissions decreased). A similar analysis was performed with
graphs of adjusted CMPG versus FTPNOX. In 19 cases, with at least 12 data
points, 11 exhibited a tendency for decreases in adjusted CMPG as the FTPNOX
decreased, while the remaining 8 exhibited the opposite tendency. Thus, the
results from this approach tend to imply that decreasing the level of FTPNOX
emissions has no consistent effect on fuel economy.
A visual inspection of the 48 graphs of adjusted fuel economy versus FTPNOX
exhibited some patterns which corresponded to figure Toyota-1 (section
V.D.). Specifically, it was found that carbureted vehicles equipped with
3-way catalysts (system numbers 4, 5, and 6) produced higher adjusted fuel
economy (both urban and combined) results than did carbureted vehicles
equipped with air pumps but without 3-way catalysts, for every transmission
grouping. This visual analysis also supported the conclusion made earlier
in this section that, among carbureted vehicles without 3-way catalysts,
those vehicles equipped with pulsating air systems (system number 1) produce
higher fuel economies than similar vehicles equipped with air pumps (system
number 3).
IV-67
-------
Another stratification used was to divide the data based on the FTPHC
values. Using only the FTPHC test values not exceeding 0.60 grams per mile
we grouped the data as follows:
1. HC up to (i.e. less than) 0.20 g/mi
2. HC at least 0.20, but less than 0.30 g/mi
3. HC at least 0.30, but less than 0.40 g/mi
4. HC at least 0.40, but less than 0.50 g/mi, and
5. HC at least 0.50, but less than 0.60 g/mi.
In this last stratification, the attempt was to stratify by both the 8
emission control technology systems and the 5 FTPHC ranges, and then, for
each of those 40 groups, plot urban ton-miles per gallon versus FTPNOX.
Linear regression analysis was performed on each of those 40 groups of
data. Of the 26 sets of data containing at least 12 points, the analysis
indicated that 10 sets of data displayed a tendency for decreasing vehicle
efficiency with decreasing FTPNOX, and the remaining 16 displayed a tendency
for increasing vehicle efficiency with decreasing FTPNOX.
IV-68
-------
Section IV. I. Is There a "Knee" in the Relationship between Tailpipe
Emissions and Fuel Economy?
We will refer to a graph as possessing a "knee" (or "tolerance point") if:
1. To the right of the knee, the graph is almost horizontal, and
2. To the left of the knee, the graph is angled downward sharply.
An idealized sketch of such a graph is:
Since many manufacturers have alluded to the existence of such a curve in
relating fuel economy to emissions (usually FTPNOX), the graphical analyses
in Section IV. H. were investigated for the existence of such a
relationship. While such a "knee" curve might exist, its existence was
hidden by the variation among the test vehicles (i.e., differences in ETW,
VDHP, DISP, RTHP, and NSVR). However, since the graphs of "Adjusted" fuel
economy versus FTPNOX were designed specifically to compensate for those
vehicle differences, those 48 graphs were re-examined.
19-69
-------
There were several graphs for which could be inferred, from a visual
inspection, that decreasing the FTPNOX emissions had a significant effect on
the fuel economy. These particular graphs each seemed to possess a
"tolerance point" with the following property: for FTPNOX emissions higher
than that point, the adjusted fuel economy was unaffected by variations in
the FTPNOX level; however, for FTPNOX emission below that tolerance point,
the adjusted fuel economy exhibited a significant deterioration with
decreasing FTPNOX levels. The groups for which these tolerance points were
found are:
1. Carbureted vehicles with lockup automatic transmissions (LA) and
oxidizing catalyst systems exhibited a tolerance point at 1.0 NOx
for the adjusted UMPG graphs. However, this pattern did not appear
on the adjusted CMPG graphs for these groups.
2. For carbureted vehicles equipped with an oxidation catalyst, a
3-way, closed-loop system (i.e., system number 4), and with a
manual transmission (CM), a tolerance point of 0.8 g/mi NOx
appeared for both the adjusted UMPG and adjusted CMPG.
3. For fuel injected vehicles with EGR and 3-way catalysts (i.e.,
system number 8), equipped with other than a lockup transmission
(i.e., either CA or CM), the adjusted UMPG exhibited a tolerance
point at 0.4 g/mi NOx. This same tolerance point (i.e. 0.4 g/mi
NOx) also appeared for the adjusted CMPG for only the manual trans-
mission.
These groups represent a very small proportion of the total number of
emission control/transmission combinations. Tolerance points were not found
for the majority of the combinations. It is interesting to note that the
above three tolerance values approximately coincided with actual NOx
emission standards (1.0, 0.7, and 0.4 g/mi). However, in order for a test
to pass the applicable emission standard, its FTPNOX level multiplied by the
appropriate deterioration factor (DF) must not exceed the standard. Since
the data used in this section represented low mileage data only, directly
relating these data to existing or future standards is not straightforward.
IV-70
-------
SECTION V.
V. Individual Manufacturer Discussions
V. A. General Motors
1. GM Statements Concerning the Relationship between Emissions and Fuel
Economy
In a paper titled "Automotive Fuel Economy Review,"* GM said that emission
standards penalized fuel economy in the following manner.
While we are meeting the 1980 Fuel Economy Standards, which is
20 miles per gallon, we also have to overcome a penalty
estimated at up to 5 percent imposed by the increased stringency
of the 1980 emission standards. In 1981, when the corporate
average fuel economy must be 22 MPG, another increase in
emission controls severity imposes an estimated 3 percent
penalty on fuel economy as compared with 1979. Introduction of
the new GM C-4 Emissions Control System in 1981 will prevent
this loss from being larger.
In the preceeding years, compromises in spark and carburetor
settings for emission control had reduced GM's new car average
fuel economy to 12.0 miles per gallons (mpg) in 1974. Adoption
of a new emission control system made possible engine operating
improvements that brought average fuel economy up to 15.3 mpg in
1975. The reason was that the use of the oxidizing catalyst to
remove most of the hydrocarbons (HC) and carbon monoxide (CO)
after the exhaust gases have left the engine enabled us to
recalibrate the ignition and carburetion nearer to their optimum
settings. Each year thereafter has seen further increase in our
fuel economy, so the GM sales-weighted average for 1980 is
estimated at 21.4 mpg, an improvement of more than 78% since
1974.
Improved basic engine efficiency also is constantly being sought
in research programs. Engine efficiency will be further
improved by increasing use of electronic engine controls, to
decrease emissions and offset much of the fuel economy penalty
resulting from emission control.
It has been a well established fact .that the Diesel engine is
superior to the gasoline engine in fuel economy, and GM
introduced the Diesel engine as an option for its full size
Oldsmobile in 1978. For a comparable performance, the Diesel
Earl K. Werner, General Motors, "Automotive Fuel Economy
Review", Society of Plastics Engineers National Technical
Conferences, November 7, 1979, pages 2, 5, 6 & 8.
V-l
-------
gives upwards of 25 percent better fuel economy than the
gasoline engine, and if we can overcome the negative effect of
tightening emission standards, we expect to apply the Diesel to
a greater number of our vehicles, further improving our CAFE.
In a September 22, 1980 letter* to the Administrator of the Environmental
Protection Agency, General Motors included a chart which among other things
provided information on the effects of emission standards on fuel economy.
This information can be found in tables GM-1 and GM-2.
Six CO waiver applications from GM were decided between September 5, 1979 and
March 3, 1981. Fuel economy was one of the criteria for granting a CO waiver,
yet GM did not make an issue of fuel economy in their first four CO waiver
applications. However, in their fifth and sixth CO waiver requests (1.8/2.0 L
and 1.6L engines), GM claimed the following:
Compliance with the 3.4 gpm CO standard as compared to a 7.0 gpm
CO standard has the potential of adversely affecting fuel
economy in the range of 2 to 5 percent, citing engine
calibration changes ("retuning") as one of the principal causes.
To further substantiate its position, General Motors embarked on
an engineering development and test program in which a "J" car
1.8L engine was calibrated to meet the 3.4 gpm CO standard and
then recalibrated to meet a 7.0 gpm CO standard. Exhaust
emissions and fuel economy were measured under both conditions
and driveabilty was compared. The results of these tests have
been presented to EPA for the public record. These results show
that a fuel economy penalty of 1 mpg would exist in the EPA
label value if the waiver were denied.
To further substantiate the fuel economy penalty, we have now
completed an investigation involving a matched pairs study of
federal gasoline passenger cars comparing the fuel consumption
improvement from 1980 to 1981 of those passenger cars granted
the CO waiver to the fuel consumption improvement from 1980 to
1981 of those [Federal gasoline] passenger cars not granted the
CO waiver. This study was based on the premise that any
improvement in fuel consumption between the CO and non-CO waiver
groups could be attributed to the CO waiver. The following
parameters were matched: engine displacement, test weight,
transmission, axle ratio and dynamometer horsepower setting.
The results are shown in [table GM-3]:**
* Letter and enclosures to the Honorable Douglas M. Costle, Administrator,
U.S. Environmental Protection Agency from Dr. Betsy Ancker-Johnson,
General Motors, dated September 22, 1980.
** Letter and attachment to Mr. D.M. Costle (Administrator, EPA), from T. M.
Fisher (GM), dated December 8, 1980.
V-2
-------
Table GM-L
I
OJ
GENERAL MOTORS CORPORATION
LIGHT DUTY (PASSENGER CAR) GASOLINE ENGINES
Issued: October 1, 1980
EFFECTS Of EXHAUST AND EVAPORATIVE EMISSION STANDARDS ON HARDWARE, FUEL ECONOMY, FUEL CONSUMPTION, AND ADDITIONAL-FIRST-COST TO CONSUMER
NOTE: The data in this table ate baaed on tests of current CM hardware designed to comply with light duty (passenger car) gasoline emission and fuel
economy standards. Hardware utilized In this effort has an effect on performance In both areas, although Installed primarily for Its contribution
to either emission control or fuel economy. Since the "state of the art" In this effort has changed rapidly, It should be noted that the hardware
and cost estimates represent current technology adjusted to reflect 1981 economic levels and new 1981 average dealer discounts. As a result. It
should be expected that future versions of this table may not be the same as shown here. This table will be updated periodically.
Standards
HC/CU/NO«/Evap,
Applicable/
Most Probable Hardware^
Increased Fuel
Fuel Consumed In Gasoline
Economy Vehicle Life Vehicle
Loss, Billions First Costa. ^
Percent2 of Gallons^ Add'lTotal
1.5/15/2.0/6.0
1979 Federal
Oxld. Conv., ECR or BPEGR, Baseline Baseline
.41/7.0/2.0/6.0 Oxld. Conv., AIR or PAIR, 3 1.0
1980 Federal BPEGR
.41/9.0/1.0/2.0 CCC Sys. (Single Bed), 2-3 Included
1980 California AIR, BPEGR Above
.41/3.4/1.0/2.0 CCC Sys. (Dual-Bed), AIR, Minimum
1981 and Later BPEGR, EST, ISC of 1
Federal
.41/3.4/.41/2.0 CCC Sya. (Dual-Bed), AIR,
Research BPECR, EST, ISC, and T
Objective
2.1
Baseline 185
60 245
475* 6605
540* 725»
7 »
Remarks
For each set of future standards, the tabulated results are
based on comparlaons between vehicles meeting the 1.5/15/2.O/
6.0 standards and comparable vehicles equipped with hardware
designed to meet the future standards.
Dual-bed CCC systems and EST 'received limited use In
California In 1980 to gain field experience with these advanced
technologies. Additional vapor storage capacity and Improved
fuel system seals were required to meet the 1980 California
evaporative standard (same as the 1981 Federal standard).
In general, a dual bed CCC system la necessary to provide the
Improved CO control necessary to meet a 3.4 g/mlle standard.
A limited number of engine fsmllles have been granted a waiver
to a 7.0 g/mlle CO standard for 1981-82.
Meaningful assessments of fuel economy and additional first
cost cannot bo made since control systems have not been dora-
onntrated to meet these standards through the complete EPA
50,000-mile certification durability requirements, selective
enforcement audit and In-use surveillance requirements.
'Hardware Deflnltonsi (a) EGR - Exhaust Gas Reclrculatlon, (b) BPECR - Back Pressure ECR, (c) AIR - Air Injection Reactor, (d) PAIR - Pulse Air
Injection Reactor, (e) CCC System - Computer Command Control System, (f) Single-Bed " 3-Way Catalyst, (g) Dual-Bed - 3-Way Catalyst plus Oxidation
Catalyst, (h) EST • Electronic Spark Timing, and (1) ISC - Idle Speed Control. Other electronic engine controls Day be Included It emission control
or fuel economy benefits are demonstrated.
*Fiu-l Economy! The use of a dlesel engine of comparable performance In place of a gasoline-fueled engine, may Increase the fuel economy of that vehicle
by nbout 25-30Z. Refer to other side for dlesel Information.
Jfucl Consumption! Based on emission certification and fuel economy teats, a fuel economy penalty of 3Z has been determined for 1980 vehicles and IX
estimated fur 1981-1985 vehicles to project the Increase In total fuel consumed during the life of 1980 through 1985 vehicles, In comparison to vehicles
meeting the future fuel economy standards. Total fuel consumed was calculated using the General Motors fuel consumption model. These fuel economy
pcn,iIlies result In an Increase In total fuel consumed of 3.1 billion gallons of fuel during the life of 1980-1985 vehicles.
Hardware Costsi On-going production costs of 1979 gasoline vehicle exhaust emission control systems waa $165, In comparison to uncontrolled vehicle
costs; evaporative emission controls cost an additional $20 for a total gasoline vehicle baseline cost of $185. Additional first costs of future
systems are estimated on the basis of system definitions as of August, 1980. For vehicles equipped with a CCC system, the oxygen sensor may have to be
changed during the useful life of the vehicle, at an additional maintenance cost.
t
^California 1980 Hardware Costi The cost data shown are cslculated on the basis of high volume production and therefore are conservative estimates.
Compared to 1981 and Late/ Federal, 1980 California costa Include compliance testing.
Future Cost Trends! Based on projections of potential reductlona of electronic component costs, first cost to the consumer Is expected to decreaae
in later years from the value given In the table.
ARC 185
9/15/80
-------
Table CM-2
GENERAL MOTORS CORPORATION
LIGHT DUTY (PASSENGER CAR) DIESEL ENGINES
lasued: October 1. 1980
ESTIMATED EFFECTS OF EXHAUST AND EVAPORATIVE EMISSION STANDARDS ON POTENTIAL HARDWARE, AND ADDITIONAL-FIRST-COST TO CONSUMER
NOTEi The projections In this table are based on tests of current CM experimental hardware and advanced research on control concepts which may have
potential for compliance with future emission and fuel economy standards*. NO ADVANCED HARDWARE BEYOND 1982 FEDERAL HAS DEMONSTRATED AN ABILITY TO
ACHIEVE FUTURE STANDARDS. The control hardware listed below represents GH's best ESTIMATE of the TYPE and EXTENT of the hardware that will be necessary
to approach future standards, and should not be Interpreted to mean that the needed technology Is available. This table will be updated periodically to
reflect changes In the "state of the art."
Event
1979
Federal
1980
Federal
1982
California
1984
Federal
Standards
IIC CO
1.5
IS
2
5
NOx
.0
PART-
ICULATES
Applicable/Host
Baseline
Probable
Uncontrolled
Hardware?
Engine
Diesel
First
Add'l
Base-
line
Vehicle
Cost i3
Total
Base-
line
0.41 3.4 2.0
Isle)
1981 0.41 3.4 1.0*
Federal "Waived to l.S
1982 0.41 3.4 1.0* 0.6
Federal *Walved to l.S
O.S4 7.0 1.2
0.41 3.4 1.0*
0.6
0.6
'F.I. Timing, EGR
F.I. Tilling, HOD-EGR
F.I. Timing, HOD-EGR
F.I. Timing, analog
electronic control of
EGR and TCC engagement
Digital electronic control
of ECR, F.I. Timing, rich
fuel llmlter and TCC engage-
ment. Modified F.I. Pump.
SO
SO
SO SO
SO 50
Undetermined3
Undetermined
Remarks
Same engine marketable nationwide In 1979.
1980 California dlesel engine certified to l.S NOx
standard with 100,000 mile ~---6111ty requirement
(See Footnote 1).
•Waived to l.S NOx Is available for 1981-84 with EPA
approval. GH S.7 L dlcsel granted waiver to l.S NOx
for 1981-82.
Redesign of EGR system offsets potential hardware cost
Increase of more stringent NOx standard.
First application of purttculate requirement. No'
specific additional hardware deemed necessary for
compliance.
NOx standard for turbocharged engines - l.S gpm NOx at
100,000 miles. NOx standards were set by CARB emer-
gency action and are subject to confirmation at public
hearing. O.S7 IIC standard reflecta adjustment for low
evap. allowance.1'5 This system would be required to
meet 1983 Federal Standards w/o NOx waiver.
Compliance required 198
-------
Table GM-3 .
Percent Improvement in Fuel
Consumption of 1981 Federal Passenger Cars
Compared to 1980 Federal Passenger Cars**
Classification
Average Adjusted
GPM Improvement
Between 1980 and 1981
% Improvement
From 1980
to 1981
Number of
Combination
Groups Matched
All Federal
Passenger Car
Engines
.0012
2.6%
28
CO Waiver-
Engines
.0022
4.7%
10
Non-CO Waiver
Engines
.0007
1.4%
18
Difference
Between the CO
Waiver Engines
and the Non-CO
Waiver Engines
.0015
3.2%*
* A 95% confidence interval was calculated for the difference of
two means with common unknown variances. The confidence
interval ranges from -1.5% to 7.9%.
These data clearly indicate that those vehicles granted the CO
waiver exhibit a 3.2% increase in fuel economy over those
vehicles without the CO waiver. This finding compared favorably
with both the National Research Council's estimate of 2 to 5
percent and with our experience on the "J" car test program as
well.
** Retyped for clarity.
V-5
-------
Coincidentally, as part of a new application being submitted
today for a 1982 CO waiver on the 1.6L Chevrolet Chevette
engine, we have conducted a test program similar to that
conducted on the "J" car as described above. The data from the
Chevette testing also confirms a fuel economy penalty on the
order of 1 mpg for a car complying with a 3.4 gpm CO standard as
compared to a car complying with a 7.0 gpm standard. These data
are enclosed in the attachment [table GM-4].
All of these data consistently show that, without a waiver, the
"J" car 1.8/2.0 liter engine family will achieve lower fuel
economy levels than it could have achieved were a relaxed CO
standard in effect. Based upon our best estimates, we believe a
fuel economy penalty of the magnitude of 1 mpg on the EPA label
will cause a significant number of potential buyers to either
purchase a competing model of vehicle, or to withdraw from the
new car market place until better fuel economy is -achieved to
help offset the cost of the new car. In either case, the effect
is one of lost sales to General Motors.
In a report to EPA, GM said "a 0.4 g/mi NOx standard would worsen fuel economy
and driveability of gasoline engines, and, at present levels of technology, it
would eliminate the fuel-efficient Diesel..."*
Automotive News said "GM also stressed that averaging and lowering emission
standards may improve fuel economy."**
* General Motors, "1979-1980 Report of General Motors On Advanced Emission
Control System Development Progress," February, 1981, Vol. 1, page 9.
** Automotive News, April 6, 1981, page 49.
V-6
-------
Table GM-4
ATTACHMENT
COMPARISON OF EMISSION AND FUEL ECONOMY RESULTS FOR
DEVELOPMENT DATA CARS EQUIPPED WITH 1.6 LITER ENGINES AND
CALIBRATED TO 3.4 AND 7.0 G/MI CO STANDARDS
Car Calibrated to Car Calibrated to
3.4 g/mi Standard* 7.0 g/mi Standard**
Test. 1 last 2 Avg. Test: 1 Test. 2. Avg.
• Automatic Transmission: Equipped: Cars: (II035A: and IIQ35S)
Emissions (FTP) g/mi
HC. .087 .077 .082 .154 .195 .174
CO .90 1.21 1.06 3.06 4.01 3.54-
NOx .32. .29 .30 ..28 .33 .30
Fuel Economy mpg
City 22.9 23.2 23.05 24.0 23.7 23.35
Highway 27.3 27.9 27.60 30.0 30.2. 30.10
Composite — — 24.90 — — 26.31
Car Calibrated to Car Calibrated to
__. 3.4 g/mi Standard* 7.0 g/jai Standard**
Test 1 Test Z Test 3 Test 4 Avg. Test 1 Test 2. Test. 3 Avg.
Four-Speed Manual Transmission Equipped Cars (IT033A. and IT033B)
Emissions (FTP) g/mi
HC .075 .084 .077 .084 .080 .137 .115 .118 .123
CO .99 1.12. .42, .84 .84 2.26 1.94 1.86 2.02
NOx .31 .33 .24 .26 .28 .52. .36 .35 .41
Fuel Economy mpg
City 24.9 25.2. 25.1 25.3 25.13 26.4 26.0 26.1 26-. Id-
Highway 31.5 33.3 33.. 6. — 33.63. 34.9 -34.5 34.6: 34.66.
Composite — — — — 2S.35 — — — 29.40.
Notes:
* Calibrated using deterioration factors (HC - 1.419, CO - 1.590 and NOx -
1.285) developed on certification durability car 1110.
** Calibrated using deterioration factors (HC - 1.679, CO - 1.490 and NOx -
1.248) developed on certification durability car 1104.
V-7
-------
2. New Technology or Developments which Could Provide a Change in the
Relationship Between Emissions and Fuel- Economy as Compared to 1981 Vehicles
or GM's Estimates
A. The first device discussed is GM's detonation sensor system, which was
described in Society of Automotive Engineers Paper //790173, titled "Energy
Conservation with Increased Compression Ratio and Electronic Knock Control".
The following GM discussion provides; a) insight into the circumstances which
led to the development of this system, b) how it operates, and c) test data
for two vehicles calibrated to meet 0.41 HC, 3.4 CO and 1.0 NOx, one with the
detonation sensor system, and the other, without.
Spark Knock Control has been the focus of extensive research
over the years by both oil companies and engine manufacturers.
Efforts to improve the fuel economy and performance of spark
ignited gasoline engines have been hindered by the occurence of
knock which can be affected by such variables as engine
tolerances, fuels, and vehicle operating conditions. In the
past, spark knock control has been achieved by the use of fuel
additives, fuel blends, or engine design modifications such as
combustion chamber design, cooling system design, or retard of
ignition timing. Currently, spark retard is utilized in many
production engines to reduce the tendency for knock under
adverse driving conditions such as heavy load, high ambient
temperature, or low humidity. Accordingly, the engine may
operate with reduced efficiency under less adverse conditions
when retard is not required to control knock. With the advent
of catalytic converters necessitating the use of unleaded fuels,
the engine designer has limited the compression ratio so that
the engine octane requirement is satisfied by the current
available unleaded fuels.
Historically, raising the compression ratio of an engine has
improved efficiency. However, there has been some concern, in
addition to the concern for higher octane requirement that
efficiency might not be improved when higher compression ratio
engines were recalibrated to meet emission constraints. As a
result, a study (1)* was recently completed at General Motors
which indicated that with catalytic converter emission control
systems, the traditional efficiency gains were possible at the
current Federal Exhaust Emission Standards, (1.5 g/mi HC, 15
g/mi CO, 2.0 g/mi NOx) when raising compression ratio from 8.3:1
to 9.2:1.
References are listed by the number in parentheses at the end of
the excerpts.
-------
This gain was accompanied by an increase in octane requirement,
however, which could not be satisfied with 91 Research Octane
Number (RON) unleaded fuel. Published information from the
petroleum industry indicated that substantial energy losses
occurred in the refinery when producing higher octane unleaded
fuel. As a result, the energy losses in the refinery for making
the higher octane fuels offset the efficiency gains in the
vehicle. Consequently, the conclusion of this study was that
there did not appear to be an incentive for increasing unleaded
fuel octane levels to allow for the use of higher compression
ratio engines with a catalytic converter emission control system.
This paper will describe an electronic closed loop knock control
system that may satisfy the octane requirements of more
efficient higher compression ratio engines, at current
(1.5/15.0/2.0) and 1981 exhaust emissions levels (.41/3.4/1.0),
with 91 octane fuel.
A knock control system (2) is currently being offered by the
Buick Motor Division of General Motors on the Turbocharged V-6
engine. The scope of the paper will, therefore, be limited to
the application of the knock control system to higher
compression ratio engines.
CLOSED LOOP KNOCK CONTROL (CLKC) SYSTEM DESCRIPTION
The knock control system was designed for use on engines with
electronic ignition. It includes a detonation sensor or
accelerometer, electronic controller and modified distributor.
The detonation sensor, is mounted on the intake manifold of the
engine. The location of the sensor is carefully selected to
insure that the knock vibration will be transmitted as equally
as possible from all cylinders.
If no knock is present, the normal ignition timing, determined
by the basic timing, centrifugal advance, and vacuum advance is
passed directly to the distributor. When knock occurs, the
detonation signal is larger than the background noise level and
a retard command is generated that is proportional to the
intensity of knock encountered.
The retard command is used to produce a delayed or retarded
ignition pulse to the distributor. The distributor ignition
module has been modified to accept the retarded ignition pulse.
When knock is reduced to desired levels, the controller restores
the spark advance at a predetermined rate until the normal spark
values are re-established. It is important to note that the
system only retards and does not advance timing to seek out
knock.
-------
It can be noted that the spark is retarded rapidly to control
knock and then readvanced slowly. This is a feature used to
maintain smooth vehicle operation during knock control. More
rapid changes in spark timing can cause uneven engine power
resulting in surge which is observed by the driver as a jerky
forward motion in the vehicle. Throughout the acceleration, the
spark advance is retarded and readvanced while the system seeks
to maintain a commercial knock level. In this example, the
maximum retard for this condition was 11° during the test.
The operation of the knock control system is dependent on the
occurrence of knock. That is, a knocking cycle must occur to
trigger the control system and generate retard. These "knock
control" cycles, which occur just prior to control, may be heard
as random trace to light knock in the engine during the control
mode. It should, therefore, be emphasized that the system does
not eliminate detonation, but controls the knock to commercial
levels. More sensitive control of detonation below these levels
can produce a "false retard" condition when retard occurs
without knock in the engine. False retard should be' avoided as
it can result in a loss of vehicle performance and fuel economy.
The maximum spark retard available using the knock control
system is determined by the physical characteristics of the
distributor and vehicle driveability. For example, the maximum
retard is physically limited by arcing or crossfire in the
distributor cap which can result in a spark occurring in the
wrong cylinder.
Likewise, large changes in spark retard could also contribute to
the surge problem described earlier. The final vehicle spark
calibration with a knock control system must, therefore, be
evaluated under the most severe knocking conditions expected, to
insure that the system has sufficient retard capacity to control
knock while maintaining driveability.
FUEL ECONOMY POTENTIAL AT .41/3.4/1.0 EMISSION STANDARDS - Two
vehicles (4000 Ib. inertia weight) with 305 CID engines and
3-way catalyst emission control systems were built to determine
the fuel economy advantage of a higher compression ratio and a
closed loop knock control at the 1981 Federal emission standards
of .41/3.4/1.0. The engines were built with 8.4:1 and 9.2:1
compression ratios, respectively, and calibrated within the same
percent loss (approximately 1%) from best fuel economy spark
timing. Both vehicles were equipped with closed loop knock
control systems allowing the vehicles to be calibrated without a
knock constraint. Fuel economy gains were therefore determined
for the effect of compression ratio at equivalent emission
levels. Rear axle ratios were selected to equalize the
performance of the cars.
V-10
-------
Emission Results - Emission targets of .25/2.57.60 were
established to allow for emission control system deterioration,
vehicle variability, and test variability. These targets would
not necessarily assure that these vehicles would meet the
Federal Standards at 50,000 miles, however, since durability
testing was not run as part of this program. Figure 13
illustrates that the exhaust emission goals for hydrocarbons,
carbon monoxide, and oxides of nitrogen were met with each
vehicle. In general, the higher compression ratio engine
required more EGR for NOx control and more spark retard during
cold start operation for HC control.
FUEL ECONOMY RESULTS - Figure 14 shows the comparison of fuel
economy on the EPA city (5) and highway (6) schedules for the
8.4 CR and 9.2 CR vehicles. The 9.2 CR vehicle exhibited a 3.7%
advantage in fuel economy on the EPA city test and 5.3% gain on
the EPA highway test. The overall fuel economy improvement for
the composite 55/45 test was 4.3%. These results were obtained
from the average of four tests on each vehicle and were all run
with 91 RON fuel.
The fuel economy tests on the 9.2 CR vehicle were then run with
clear Indolene test fuel (RON 98). This resulted in a small
gain in fuel economy on both the city and highway tests (Figure
24) and the overall gain for the change in compression ratio was
5.9%. The reason for this difference appears to be due to the
reduction in spark retard required to control knock with the
higher octane fuel on the 9.2 CR vehicles.
Figure 15 illustrates the spark retard generated by the 9.2:1
compression ratio engine on 91 and 98 RON fuels during one of
the more severe portions of the EPA city test. The frequency of
occurrence of knock and amount of retard required to control
knock was greater with 91 RON fuel. The increased retard had a
minimal effect on the EPA fuel economy results as the spark was
not retarded over the majority of operating conditions. The 8.4
CR vehicle did not have significant spark retard during the EPA
test; therefore, no change in fuel economy would be expected due
to increasing fuel octane over 91 RON.
Vrll
-------
305-4
CLOSED LOOP CARBURETOR
1975 FEDERAL TEST PROC.
AVG. 4 TESTS
4000* I.W., 10.8 HP
3-WAY CONVERTER - EGR
91 RON FUEL
TAILPIPE (GMS/MII
C.R.
8.4:1
9.2:1
TARGET
AXLE
2.41:1
2.28: 1
STANDARD
HC
.25
.22
.25
.41
CO
2.5
2.4
2.5
3.4
NOx
.52
.63
.CO
1.0
Fig. 13 - The exhaust emissions of Che 8.4 CR
and 9.2 CR vehicles met the target levels; for
the 1981 Federal standards
305-4
CLOSED LOOP CARBURETOR
4000* I.W.. 10.8 H.P.
3 WAY CONVERTER-EGR
AVERAGING OF 4 TESTS
FUEL
(CHEVRON 91)
98 BON
IINOOLENE
CLEAR)
COMPRESSION RATIO
REAR AXLE RATIO
EPA FUEL ECONOMY
CITY
HIGHWAY
COMPOSITE 55/45
CITY
HIGHWAY
COMPOSITE
8.4:1
2.4:1
16.1
22.7
18.5
_
-
9.2:1
2.28:1
16.7
23.9
19.3
16.9
24.4
19.6
\ GAIN
3.7
5.3
4.3
5.0
7.5
5.9
Fig. 14 - The 9.2 CR vehicle exhibited a 4.3%
gain in fuel economy on the EPA composite 55/45
test
VEHICLE SPEED
CHEVRON SI RON
RETARD
10"
MOOLENE CLEAR M RON
Fig. 15 - The amount of retard encountered by
the 9.2 CR vehicle on the EPA City test was
reduced when fuel octane was increased from 91
RON to 98 RON
V-12
-------
The following references are for the preceding quotations from SAE paper
number 790173.
REFERENCES*
1. J.J. Gumbleton, G.W. Niepoth, and J.H. Currie, "Effect of Energy and
Emission Constraints on Compression Ratio", Paper 760826 presented at
SAE, October 1976.
2. T.F. Wallace, "Buick's Turbocharged V-6 Powertrain for 1978", Paper
780413 presented at SAE, March 1978.
5. R.E. Kruse and T.A. Huls, "Development of the Federal Urban Driving
Schedule", Paper 730553 presented at SAE, May 1973.
6. T.C. Austin, K.H. Hellman and C.D. Paulsell, "Passenger Car Fuel Economy
During Non-Urban Driving", Paper 740592 presented at SAE, August 1974.
Only those references which were cited in the previous excerpts were
included in this listing. The original reference numbers were used for
this document.
'Yrl3
-------
B. Another area, which affects the relationship between fuel economy and
emissions, is combustion chamber design. GM described development work
concerning combustion chambers in the Society of Automotive Engineers Paper
#800920, titled "The Attributes of Fast Burning Rates in Engines." The
following excerpts from this paper provide insights into combustion chamber
design and its effect on the relationship between emissions and fuel
economy. The references cited by GM are listed at the end of the excerpts.
THE CONCEPT OF BURNING THE MIXTURE in a spark ignition engine
relatively fast is increasingly making its way into engine
designs throughout the world. Fast burning is certainly not
new, with the higher efficiency or output and fuel octane
advantages being known for decades. In fact, some of the
engines of the 1920's (1) incorporated combustion-chamber design
features promoting fast burning.
In the past, fast burning was curtailed, due to adverse noise
and mechanical failures caused by too great a rate of pressure
development, particularly as compression ratios were increased.
With the introduction of exhaust gas recirculation (EGR) for
oxides of nitrogen control in the early 70's and the lowering of
compression ratios in the mid 70's, the pendulum swung the other
way, and burning rates decreased sharply. The current movement
toward fast burning represents an effort to reestablish, at
least to some degree, the Crates of combustion that once existed
in engines. """
ENGINE EFFICIENCY AND EMISSIONS -
Tuttle and Toepel (10) diagnosed and compared combustion in the
two different combustion chambers shown in Fig. 8. The wedge
chamber, with its off-set spark plug and resulting long flame
travel distance, promotes / relatively long burning times. In
contrast, the compact open chamber, with a more central spark
plug, would be expected to produce faster burning. Each
combustion chamber was instrumented extensively, including the
incorporation of cylinder pressure instrumentation (14-16) to
diagnose heat release and flame motion (17). Testing followed a
• statistical experimental design involving selected combinations
of engine speed, fuel input, air-fuel ratio, EGR rate and spark
timing.
V-14
-------
WEDGE CHAMBER
OPEN CHAMBER
Fig. 8 - Experimental sCudy combustion
chambers (10)
35
34
O 33
2
UJ
S2 32
li.
UJ
31
£ 30
I
UJ
29
MBT
" SPEED - 1900 r / min
- A If RATIO -16:1
-g/kg
OPEN'
INPUT-23.0 mg /cycle v.'EDGE
i i i i i i
8 16
EGR.%
24
Fig. 9 - Efficiency - NOX tradeoff (10)
V-15
-------
Experimental results showed that the open chamber did indeed
have faster burning, as made evident by shorter combustion times
and reduced spark-advance values at MET. Net thermal efficiency
of both chambers is shown in Fig. 9 for various amounts of EGR,
with engine speed, air-fuel ratio, and fuel input constant.
Concentrating first on the curves denoted MBT, which represents
optimum timing data, the open, or faster burning chamber, is
seen to exhibit higher efficiencies. Starting with no EGR, the
efficiency at first increases with the addition of EGR. This
behavior is due to both the higher specific-heat ratio and the
reduced throttling of the more dilute mixture.
Another important advantage of the faster burning chamber is its
greater tolerance to dilution. As shown, EGR rates as high as
28% can be accommodated with the open chamber, versus 22% for
the wedge chamber, before the efficiency drops off
significantly. This capability of going to greater dilution
with fast burning has also been found (18) to occur when the
dilution gas is air rather than EGR.
Also shown in Fig. 9 are lines of constant NO emissions.
Movement vertically downward from the MBT line represents
increasing spark retard from optimum timing. Evident is the
familiar reduction in NO with either spark retard or higher EGR
rate. A comparison of the condition at points A and B shows the
same trend uncovered earlier using the engine simulation -
namely, that faster burning produces higher NO emissions. Note,
however, that by adding a slight amount of EGR to the open
chamber (point C), its efficiency advantage can be maintained
while matching the lower NO value of the slower burning wedge
chamber. This behavior is very significant since it shows that
the judicious selection of spark timing and EGR rate can lead to
both higher efficiency and lower NO emissions in the fast-burn
chamber.
Kuroda et al. (11), in going from single ignition to dual
ignition in the same combustion chamber and introducing varying
amounts of EGR for oxides of nitrogen (NOx) control, also found
that fast burning could achieve both higher efficiency and lower
NOx. Likewise, Thring (13), going from one up to four spark
plugs to increase burning rate in a given combustion chamber,
found the dual improvement with faster burning. As an
additional part of his studies, Thring used alternate design
approaches to achieve the same burning rate, as indicated by the
time in crank-angle degrees to go from burning 10% of the
mixture to burning 90% of the mixture. Included among these
approaches were different combinations of four spark plugs, the
use of ordinary or extended electrode spark plugs, and the use
of a shrouded intake valve to create fluid motions conductive to
faster flame speeds and therefore faster burning rates.
Vrl6.
-------
An important finding was that engine efficiency and NOx
emissions were not unique functions of the burning rate. While
both of these parameters increased as the burning time was
reduced, their magnitudes at a given burning time depended upon
the approach taken to achieve that particular burning time. To
modify slightly a former advertising slogan, "It's not how
short you make it, but how you make it short that is important."
With regard to hydrocarbons, the data of Tuttle and Toepel (10)
shown in Fig. 10 indicate no significant influence of burning
rate on the hydrocarbon emissions index (grams of the HC per
kilogram of fuel). The data do show, however, that operation at
low EGR rates and retarded spark timing is an effective means to
lower HC emissions, primarily due to the resulting higher
exhaust temperatures demonstrated earlier with the engine
simulation.
There are two factors associated with the formation of hydro-
carbons. First, as thermal efficiency increases with faster
burning, the exhaust temperature generally decreases. This
leads to lower HC oxidation rates and therefore higher HC
concentrations in the exhaust port. In fact, Mayo (12) has
found an almost linear variation of HC concentration with
burning time. For a given engine load, however, the higher
efficiency for the faster burning means that less total mass
flow is required through the engine, which may more than
compensate for the higher HC concentration and lead to lower HC
emissions expressed on a mass per unit time, or per vehicle mile
traveled basis. These conflicting trends make it difficult to
establish a general rule regarding the effect of burning rate on
HC emissions. For example, Kuroda et al. (11) have found the HC
emissions on a mass per unit time basis to be lower for fast
burning at a given EGR rate, while Thring (13) has reported no
change in HC emissions at low loads and higher HC emissions with
increased burning at high loads.
As a sidelight, the data of Figs. 9 and 10 illustrate a
dichotomy regarding engine calibration factors. That is, high
efficiency considerations dictate that optimum • timing be
established with moderate EGR rates. Low HC emissions, on the
other hand, favor low EGR rates and retarded timing, while low
NO emissions favor high EGR rates and retarded timing. Managing
these conflicting goals represents a formidable challenge in
engine calibration. Adding to the difficulty of engine cali-
bration is the need to provide good vehicle driveability by
minimizing the cyclic variability of engine power.
SIMULATION STUDIES OF BURNING ENHANCEMENT APPROACHES - Lienesch
(3) has used the engine simulation discussed earlier to examine
the two approaches to faster burning rates. Starting with
measurements of turbulence in the wedge-shaped chamber shown in
Fig. 16 as a baseline condition, the turbulence level was
V-17
-------
o
z
UJ
o
u.
u.
UJ
tr
HI
I
35
34
33
32
31
30
29
OPEN •
WEDGE.
SPUD - 1900r/min
A/F RATIO • 16:1
FUEL INPUT- n.O ng/cycle
EIHC — g/kg
J | I
8
16
EGR. %
24
32
Fig. LO - Efficiency - hydrocarbons
tradeoff (10)
-------
IGNITION POINT
WEDGE CHAMBER
IGNITION POINT
EXTREME OPEN CHAMBER
Fig. 16 - Combustion chambers of
analytical study (3)
v-19
-------
increased analytically by 50%, as might be accomplished
experimentally (35) with a shrouded valve. Use of a
correlation of burning velocity with turbulence developed
experimentally by Groff and Matekunas (35) allowed the
resulting engine performance to be predicted.
As an alternative to this fluid-motions approach, a second
chamber was envisioned in which the cylinder head and piston
crown both have concave spherical shapes and the ignition
point is located in the center of the chamber. While such a
shape is not practical to manufacture, its extreme openness
represents an upper limit to the geometric approaches for
increasing fuel burning rates.
The performance of the standard wedge, increased turbulence
wedge, and extreme open chamber was analyzed for a light
acceleration condition involving a stoichiometric air-fuel
ratio, 10% nominal EGR and MET timing. Results are given in
Table 1 for near-equal NO exhaust concentrations, achieved by
adding 1 to 3% of additional EGR to the fast-burn cases.
Starting with the standard wedge chamber and increasing the
turbulence level resulted in a reduction of the combustion
duration from 43 to 32 crank-angle degrees. However, in the
case of the extreme open chamber, the duration was even
shorter, namely 27 degrees. These variations resulted in a
slight improvement in efficiency for the increased turbulence
case and an even greater improvement in efficiency in going
to the extreme open combustion chamber.
The reason for this higher efficiency is linked to the heat
transfer. Because the high-temperature burned gases in the
extreme open chamber do not contact the chamber wall during
most of the combustion process, the heat losses in this
chamber are significantly lower. In contrast, the wall
contact of the flame in the wedge chamber plus the higher
film coefficient (due to the higher pressure term in the
Woschni (6) correlation) for the faster burn case caused
greater heat-transfer losses. Had an additional increase in
the film coefficient been made to account for the higher
turbulence of this case, these heat losses would have been
even greater. These heat-loss considerations account for the
slightly higher exhaust gas temperature of the extreme open
chamber in spite of its higher efficiency and more complete
expansion.
•VT20.
-------
Table I - Simulation Results-
Parameter
Combustion Duration
Peak Burning Rate
Peak Heat-Transfer Rate
Indicated Thermal Eff.
Exhaust Nitric Oxide
Exhaust Temperature
Standard
Wedge
iCase l;
43
13.2
3.8
34.6
1530
1130
Increased
Turbulence
(Case 2)
32
18.0
4.3
35.1
1530
1115
Extreme
Open
I Case 3)
27
25.3
3.3
37.0 -
1520
1135
Units
°CA
mg/°GA
J/°CA
ppra
K
Retyped for clarity
Vr-21
-------
While these results show the technique of increasing frontal
areas to be more effective than that of increasing the turb-
ulence level, generalizations should not be drawn. First of
all, the heat-transfer model in the engine simulation is not
sophisticated enough to account for details of chamber
geometry and fluid motions. Furthermore, the heat-transfer
correlation (6) may not be applicable for engine designs far
removed from conventional practice. In addition, the
ignition delay computation procedure in the engine simulation
does not account for swirl effects, thereby underestimating
the influence of swirl on the overall time of the combustion
event. Finally, as mentioned before, the extreme open
chamber design represents an upper-limit case which is
impractical to manufacture for many reasons. Within these
constraints, the study does indicate that there may be a
limit to the optimum amount of turbulence, after which
excessive heat losses cause a drop in efficiency.
Some evidence of this situation is indicated by the experi-
mental data of Thring (13) discussed earlier. To review, the
NOx and indicated specific fuel consumption were not a unique
function of combustion duration. When the same combustion
duration was obtained, in one case with increased swirl and
in the other case with multiple spark plugs, the increased
swirl led to greater heat losses, which produced lower NOx
but lower indicated efficiency as well.
CONTEMPORARY ENGINE DESIGNS
The fast-burn approach to engine combustion is rapidly being
adopted in new engine designs throughout the world. A small
sample of some of these new designs is illustrated in Fig. 18
[and 19].
In figure 18, a cross-section of the General Motors 2.8 L V-6
engine designed by Chevrolet (39), shows the chamber to be
somewhat open with a near-central spark plug location.
Another geometric approach to fast burn is shown in Fig. 19,
where the original shallow cavity of a flat-topped piston in
a wedge-shaped chamber was redistributed to provide an offset
cavity. The original compression ratio was maintained. This
offset cavity provides a large frontal area to the
propagating flame while improving the fuel octane requirement
by providing more effective cooling of the end gas. In
experimental studies (17) with this piston, the attributes of
fast burning were realized. This piston-cavity approach has
been employed in the turbocharged versions of the Pontiac 4.9
L V-8 engine (40, 41).
VT22
-------
SUMMARY
Fast burning in engines has been shown to have a number of
attractive attributes. Among these advantages is an improved
tradeoff between efficiency and tfOx emissions and a greater
tolerance to dilution, either with EGR or with excess air.
Resulting from the combined effect of these two traits is the
potential for simultaneously achieving higher efficiency and
lower NOx emissions.
Fast burning also reduces the cyclic variation of engine
power, which, can improve vehicle driveability. Furthermore,
the greater resistance to knock with faster burning can allow
the fuel economy advantages associated with higher compres-
sion ratios to be realized.
A number of different combustion chamber design techniques
exist for accomplishing fast burning. These techniques fall
into the two categories of altering the combustion-chamber
shape and affecting the fluid motions in the chamber. Com-
bining these different techniques into an "optimum" com-
bustion chamber design, while still recognizing all of the
important constraints, offers a challenge as well as an
opportunity for the engine designer.
Finally, the growing number of new engines throughout the
world incorporating fast burning attests to the merits of
this concept.
•VT24
-------
References*
1. J. C. G. Hempson, "The Automobile Engine, 1920-1950," Paper 760605, pre-
sented at SAE West Coast Meeting, San Francisco, August 1976.
3. J. H. Lienesch, "Engine Simulation Identifies Optimal Combustion Chamber
Design." Presented at Seventeenth FISITA International Congress, Hamburg,
West Germany, May 1980.
6. G. Woschni, "A Universally Applicable Equation for the Instantaneous Heat
Transfer Coefficient in the Internal Combustion Engine." SAE Trans-
actions, Vol. 76, 1967, pp. 3065-3083.
10. J. H. Tuttle and R. R. Toepel, "Increased Burning Rates Offer Improved
Fuel Economy-NOx Emissions Trade-Offs in Spark Ignition Engines." Paper
790388, presented at SAE Automotive Engineering Congress, Detroit,
February 1979.
11. H. Kuroda, J. Nakajima, K. Sugihara, Y. Takagi and S. Muranaka, "The Fast
Burn with Heavy EGR, New Approach for Low NOx and Improved Fuel Economy."
SAE Transactions, Vol. 87, 1978, pp. 1-15.
12. J. Mayo, "The Effect of Engine Design Parameters on Combustion Rate in
Spark-Ignited Engines." SAE Transactions, Vol. 84, 1975, pp. 869-888.
13. R. H. Thring, "The Effect of Varying Combustion Rate in Spark Ignited
Engines." Paper 70-387 [sic], presented at SAE Automotive Engineering
Congress, Detroit, February 1979.
14. D. R. Lancaster, R. B. Krieger and J. H. Lienesch, "Measurement and
Analysis of Pressure Data." SAE Transactions, Vol. 84, 1975, pp. 155-172.
15. R. V. Fisher and J. P. Macey, "Digital Data Acquistion with Emphasis on
Measuring Pressure Synchronously with Crank Angle." Paper 750028,
presented at SAE Automotive Engineering Congress, Detroit, February 1975.
16. M. B. Young and J. H. Lienesch, "An Engine Diagnostic Package
(EDPAC)-Software for Analyzing Cylinder Pressure-Time Data." Paper
780967, presented at SAE International Fuels and Lubricants Meeting,
Toronto, November 1978.
17. J. N. Mattavi, E. G. Groff, J. H. Lienesch, F. A. Matekunas and R. N.
Noyes, "Engine Improvements Through Combustion Modeling." Proceedings of
Symposium at General Motors Research Laboratories on Combustion Modeling
in Reciprocating Engines, J. N. Mattavi and C. A. Amann, Editors, Plenum
Press, New York-London, 1980, pp. 537-587.
18. A. A. Quader, "Effects of Spark Location and Combustion Duration on Nitric
Oxide and Hydrocarbon Emissions." SAE Transactions, Vol. 82, 1973, pp.
617-627.
Only those references which were cited in the previous excerpts were
included in this listing. The original reference numbers were used for
this document.
YT25
-------
34. D. R. Lancaster, "Effects of Engine Variables on Turbulence in a Spark
Ignition Engine." SAE Transactions, Vol. 85, 1976, pp. 671-688.
35. E. G. Groff and F. A. Matekunas, "The Nature of Turbulent Flame
Propagation in a Homogeneous Spark-Ignited Engine," Paper 800133,
presented at SAE Automotive Engineering Congress, Detroit, February 1980.
39. D. A. Martens, "The General Motors 2.8 Liter 60° V-6 Engine Designed by
Chevrolet." Paper 790697, presented at SAE Passenger Car Meeting,
Dearborn, June 1979.
40. "Automotive Industries." October 1979, p. 87.
41. "Automotive Design and Development." May 1980, pp. 16-17.
V-26
-------
C. In a SAE Paper (#800794) titled "Controlling Engine Load by Means of Late
Intake-Valve Closing", General Motors described a technique of reducing an
engine's pumping losses in order to improve thermal efficiency, with the
accompanying benefit of considerably reduced NOx emissions. Pumping losses
are defined as the power required to pump the fuel-air mixture into and out of
the cylinder during the intake and exhaust strokes. Pumping losses were
reduced by "delaying intake-valve closing (with respect .to a conventional
intake-valve closing) as a method of controlling the engine load without
incurring the usual part-load throttling losses." The following excerpts more
fully describe the concepts and the results.
CONCEPT
The late intake-valve-closing (LIVC) engine is an engine with
the power output regulated by controlling the crankangle at
which the intake valve closes. An LIVC engine operating at part
load is depicted in Fig. 2. As in the case of a conventional
engine, the intake valve opens just prior to and remains open
throughout the intake stroke of the engine. However, the intake
valve also remains open over a portion of the compression
stroke while the piston pushes part of the cylinder charge back
into the intake manifold. After the intake valve closes, the
remainder of the compression stroke, as well as the expansion
and exhaust strokes, are similar to those of a conventional
engine.
An idealized pressure-volume diagram for an LIVC engine is shown
in Fig. 3. For comparison, an idealized pressure-volume diagram
for a conventional engine is also shown. The pumping losses of
the conventional engine operating at part load are substantial,
as depicted by the shaded area in Fig. 3. In contrast, since
the LIVC engine inducts fresh charge at near atmospheric
pressure (unthrottled), the throttling losses are nearly
eliminated for all load conditions. The trapped cylinder
charge, and therefore power output, is determined by the
effective cylinder volume at the time of the intake-valve
closing Vjvc of Fig. 3. A mechanism that would vary the time
the intake valve closes as a function of speed and desired load
would allow the LIVC engine to achieve the same maximum power as
that of an equal-displacement conventional engine, while
providing lower part-load pumping losses.
V-27
-------
a) Intake Stroke
c) Compression
Stroke (before
Intake Valve
Closes)
b) Start of
Compression Stroke
d) Compression
Stroke (after
Intake Valve
Closes)
Fig. 2 - Depiction of LIVC operation
PRESSURE
LATE INTAKE -
VALVE CLOSING
CONVENTIONAL
THROTTLING
VTDC vivc VBDC
VOLUME
'TDC
VOLUME
'BDC
Fig. 3 - Pressure-volume diagrams
V-28
-------
EXPERIMENTAL EQUIPMENT
Because of the exploratory nature of this investigation, a
variable valve-timing mechanism was not necessary. Rather,
modification of the test engine to LIVC operation was
accomplished by replacing the production camshaft with a splined
shaft. The splined camshaft allowed cams of different dwells at
maximum lift to be installed such that the intake-valve closing
could be incrementally varied. The timings of exhaust-valve
opening and closing and of intake-valve opening were the same as
those of the conventional engine. The intake-and exhaust-valve
opening and closing crankangles for the various intake-valve
dwells investigated are listed in Table 2. Valve lifts are
shown in Fig. 5 for both the conventional engine and an engine
with the intake-valve closing delayed 80 degrees of crankshaft
rotation. An intake-valve dwell of 0°CA corresponds to the
conventional engine.
Table 2-Valve Timing
Intake-Valve Exhaust Valve Intake Valve
Dwell ( °CA) Open Close Open Close
0 (conventional)
60
80
96
88°BBDC
88°BBDC
88°BBDC
88°BBDC
52°ATDC
52 °ATDC
52 °ATDC
52°ATDC
38°BTDC
38°BTDC
38°BTDC
38°BTDC
82 °ABDC
142°ABDC
162°ABDC
178°ABDC
RESULTS
All specific test data reported herein are based on
integrated pressure cards. "Indicated" values represent
power developed within the cylinder during the
compression-expansion portion of the engine cycle, and
"pumping" values represent the negative power required for
the exhaust and intake strokes. The difference between
these two powers is termed the "net" power.
ENGINE LOAD CONTROL - The power output of the LIVC engine
is controlled by the effective cylinder volume* at the time
of the intake-valve closing. Increasing the dwell of the
intake valve at its maximum lift decreases the cylinder
volume at the intake-valve closing, thus decreasing the
amount of fresh charge trapped in the cylinder.
FUEL CONSUMPTION - The fuel-consumption characteristics of
the LIVC engine relative to those of the conventional
engine were analyzed as the additive result of changes in
both the indicated and pumping portions of the engine
The term "effective cylinder volume" is used to denote the
cylinder volume at which the intake valve is effectively closed.
Vr29_
-------
cycle. The change in the pumping portion of the cycle is
presented first. Because the LIVC engine inducts fresh charge
at near atmospheric pressure, the pumping losses of the engine
are expected to be much lower than those of the conventional
engine. Values of pumping mean effective pressure (PMEP) for
both the LIVC and conventional engines are plotted against
fueling level in Fig. 9. As the fueling level of the
conventional engine is decreased from the maximum value, the
PMEP of the engine continually increases. The values of PMEP
shown in Fig. 9 for the LIVC engine are essentially the same as
those of the conventional engine operating at WOT. Thus the
LIVC concept of load control can limit engine "breathing"
without incurring the large throttling losses associated with a
conventional engine operating at part load.
Indicated thermal efficiency is plotted versus fueling level in
Fig. 10 for both the LIVC and the conventional engines. With
the exception of the highest values of fueling level in Fig.
10, the values of indicated thermal efficiency are not MET
spark timing. At the highest values of fueling level, MET
timing could not be attained due to knocking of the engine on
the 97 RON fuel. Under knocking conditions, the indicated
thermal efficiency in Fig. 10 is that corresponding to
borderline knock. Estimated values of indicated thermal
efficiency for non-knocking operation are represented by the
dashed line. For non-knocking operation, the rate of decrease
in indicated thermal efficiency with decreasing fueling level
(load) is significantly greater for the LIVC engine. The cause
of the larger decrease in indicated thermal efficiency for .the
LIVC engine will be addressed later.
Even though the LIVC engine is at a disadvantage when indicated
thermal efficiencies are compared, the lower pumping losses of
this engine result in its having an advantage when net thermal
efficiencies are compared. Net thermal efficiency is based on
the indicated output power less the pumping losses and,
consequently, gives credit for the lower PMEP of the LIVC
engine. Fig. 11 is a plot of net thermal efficiency versus
fueling level and shows the improvement in net thermal
efficiency of the LIVC engine over the conventional engine to
increase with decreasing fueling level.
The data of Fig. 11 are replotted in Fig. 12 as net specific
fuel consumption (NSFC) versus net mean effective pressure
(NMEP). Use of net output (indicated less pumping work) allows
the fuel economy of the LIVC and the conventional engines to be
compared at equal engine load, where equal engine load refers
to equal NMEP. If the engine friction were unchanged between a
conventional engine and one modified for LIVC operation, then
V-31
-------
36
34
32
NET
THERMAL .
EFFICIENCY 30
28
26 -
•^
ESTIMATED VALUE
(NON-KNOCKING)
LIVC
ENGINE
CONVENTIONAL
ENGINE
I
10 20 J 30
FUELING LEVEL (mg/cyclc)
VALUES AT
BORDERLINE
KNOCK
I
40
Fig. II
1 eve 1
- tiet thermal efficiency versus fueling
NET
SPECIFIC
FUEL I
CONSUMPTION 280 h
(g/kW-h) i
270}-
330
320
310
-
•
; \
\
300 1- §\
290 H \;;v\
SYMBOL ENGINE
• CONVEiNTIONAL
A LIVC
A LIVC * THROTTLING
VALUES AT
BORDERLINE KNOCK
' 200 400 600 800 1000
NET MEAN EFFECTIVE PRESSURE (kPa)
Fig. 12 - Specific fuel consumption
30.0
25.0
20.0
NET
SPECIFIC
NOx 15.0
(g/kW-h)
10.0
5.0
SPARK TIMING FOR 1%
LOSS FROM MBT TORQUE
CONVENTIONAL
ENGINE
0.0 l_\_J_
200
LIVC ENGINE
400 600
NMEP (kPa)
800
Fig. 15 - Specific NQx. emissions
V-32
-------
the comparison would also be valid at equal brake load. As
mentioned earlier, because delayed intake-valve closing is
limited to about 96°CA and because this amount of delay does
not provide a range of load control sufficient for vehicle
operation, LIVC must be combined with conventional throttling.
Consequently, also included in Fig. 12 are data for which
combinations of delayed intake-valve closing and conventional
throttling were used to regulate engine load. These data are
indicated by the dashed lines in Fig. 12, which connect data
having a common value of intake-valve dwell The data indicate
the fuel consumption when throttling is used to decrease the
load further from that achieved by unthrottled LIVC operation.
In comparison to the conventional engine, both the throttled and
the unthrottled LIVC engines exhibit improved fuel-consumption
characteristics at part load. Comparing data for engine
operation with combined LIVC and throttling to data for
unthrottled LIVC operation shows the LIVC-plus-throttling method
of load control has a higher NSFC than the unthrottled LIVC
method. These results indicate that unthrottled LIVC should be
used to control engine load for loads which can be regulated by
intake-valve dwells of up to about 96°CA. For lighter engine
loads, which cannot be achieved by unthrottled LIVC operation, a
combination of 96°CA of intake-valve dwell plus conventional
throttling exhibits the lowest fuel consumption.
EXHAUST EMISSIONS - Specific emissions of NOx are shown in Fig.
15 for both the conventional and the LIVC engines. These
emissions are based on net output power. Because NOx emissions
are extremely sensitive to spark timing near the MBT timing and
because MBT spark timing itself is not always well defined,
emissions are not compared at MBT power. Rather, emissions are
compared at a spark timing which resulted in a 1% loss in engine
torque from the MBT output. Emissions are not shown at the
zero-dwell power (WOT power for the conventional engine) because
knocking prevented the estimation of the MBT value of the engine
torque.
At equal NMEP, the NOx emissions of the conventional engine can
be seen in Fig. 15 to be considerably higher than those of the
LIVC engine. At mid-load, the NOx emissions of the the LIVC
engine are about 24% lower than those of the conventional
engine. The higher NOx emissions of the conventional engine are
attributable to its higher peak cylinder pressures and, as
calculations showed, its higher peak cylinder-gas temperatures.
The lower pumping losses of the LIVC engine allow the engine to
have a substantially lower peak pressure while producing the
same part-load NMEP as the conventional engine. The measured
V-33
-------
cylinder-gas pressures were input to a heat-release analysis
[10] which computes the associated peak cylinder-gas
temperature. At mid-load, the peak cylinder-gas temperature of
the conventional engine is higher than that of the LIVC engine
by 70 K.
Specific emissions of unburned hydrocarbons (HC) .are plotted in
Fig. 18. Again, the emissions measurements correspond to a
• spark timing which resulted, in a 1% loss in torque from the MET
output. The HC emissions from the conventional engine and from
the LIVC engine are not too different, particularly at the lower
engine loading. Because exhaust-gas temperature will affect HC
cleanup in a vehicular exhaust system, the measured exhaust
temperatures are compared ... Since the temperatures are
similar for the two engines, post-cylinder HC clean-up should be
no more difficult for the LIVC engine than it is for the
conventional engine.
CONCLUSIONS
1. Load Control - Delayed closing of the intake valve is
limitedtoabout 96 degrees of crankshaft rotation because
greater delays cause large deterioration in indicated thermal
efficiency. This amount of delay does not provide a sufficient
range of load control for vehicular application. Consequently,
the LIVC concept would have to be combined with a
variable-density throttle control.
2. Fuel Economy -
a. The LIVC concept of load control can limit engine
power output without incurring the large throttling losses
associated with a conventional engine operating at part
load.
b. In comparison to the efficiency of the conventional
engine with its fixed compression ratio, the indicated
thermal efficiency of the LIVC engine at part load is
lower, and it decreases more rapidly with decreasing load.
The lower indicated thermal efficiency of the LIVC engine
at part load results from a lower effective compression
ratio and a longer combustion duration.
c. Although the indicated thermal efficiency of the LIVC
engine is lower, the lower pumping losses result in the
LIVC engine's having up to a 6.5% lower net specific fuel
consumption than the conventional engine at the same
light-load operation.
10. R.B. Krieger and G.L. Borman, "The Computation of Apparent Heat
Release for Internal Combustion Engines," ASME Paper 66-WA/DGP4,
1966.
V-34
-------
d. Conversion of a conventional engine to LIVC operation requires
the addition of a control mechanism. Consequently, the fuel savings
offered by the LIVC concept must be diminished by the losses innerent
in any r-uch control device.
e. The current trend toward lower vehicular power/weight ratios
diminishes the fuel-economy benefit of the LIVC concept.
3.
Exhaust Emissions -
a. At spark, timings yielding a 1% loss in torque from the MET
values, the NOx emissions of the LIVC engine are considerably lower
than those of the conventional engine. At mid-load, the reduct.on is
about 24%.
b.
HC emissions of the LIVC engine are similar to those of che con-
vene ional engine,
6.0 r
s.o
NET
SPECIFIC
HC
(g/kW-h)
4.0
LIVC
ENGINE
SPARK TIMING FOR 1%
LOSS FROM MBT TORQUE
-CONVENTIONAL
ENGINE
3.0
-NT
200 30O 400 500 600 700
NMEP (kPa)
Fig. 18 - Specific hydrocarbon emissions
V-35
-------
V. B. Ford
1. Ford Statements Concerning the Relationship between Emissions and Fuel
Economy
A. High Altitude Emission Standards
Ford states that they believe that the implementation of the 1984 high
altitude requirements will reduce their fleet average fuel economy by 2%
to 7%. Ford states that this loss would result from their replacing
fuel efficient configurations by less efficient ones. Ford claims, "The
1984 and Subsequent Model Years High Altitude Control Requirements could
regulate out of existence some of the most fuel-efficient vehicle
configurations because of their inability to simultaneously meet
stringent high altitude standards and yield acceptable performance."*
However, Ford did not provide any data to back up this claim.
B. Effects of Reducing the NOx Standard to 0.4 g/mi
Ford states, "Further reducing NOx emission levels below 1.0 g/mi could,
in some cases, without further design and development of high EGR tol-
erance engines, cause power reduction, tip-in stumbles, lean sags and
surging. These conditions can usually be minimized by reducing EGR
rates and richer air-fuel mixtures. However, rich air-fuel operation is
inconsistent with high fuel economy and low HC and CO objectives."*
Ford further stated that their "California cars will suffer an
additional 8% loss in fuel economy in 1983 [relative to their 1981 model
year counterparts] when the 0.4 NOx emission standard goes into effect
(assuming this NOx standard can be met at all); that will lower CAFE by
0.2 MPG. These losses will occur in spite of the fact that we will be
using sophisticated and very expensive emission control systems —
"Ford Status Report," February 1981, Sections IIA6 and IIJ.
V-36
-------
including three-way catalysts and complex electronics to control fuel
metering, spark timing, and other important parameters of the engine —
to minimize what otherwise would be much larger fuel economy losses."*
While Ford provided no data to substantiate the above claims, they did
submit a summary of computer engine mapping results comparing the fuel
economy of vehicles which meet the research target (0.41 HC, 3.4 CO,
0.41 NOx) versus vehicles whose emissions are totally uncontrolled. The
comparisons used engine maps of vehicles equipped with 2.3 liter and 5.0
liter engines. Ford's results indicated that the unconstrained 2.3
liter vehicles obtained between 11% and 21% better fuel economy than the
controlled versions, and the unconstrained 5.0 liter version produced
between 9% and 12% higher fuel economy than their controlled counter-
parts. Ford obtained between 25.9 mpg and 26.0 mpg combined fuel
economy with their uncontrolled 2.3 liter vehicles, and 17.8 mpg with
their uncontrolled 5.0 liter vehicles.** Unfortunately Ford did not
provide any emissions data or specifics on those vehicles (e.g., type of
transmission, test weight, axle ratio, whether turbocharged, whether
closed-loop, and whether fuel injected or carbureted). Since the Ford
report was submitted in September 1979, the EPA technical staff examined
1980 model year Ford vehicles which were used to generate the fuel
economy label values used by Ford for their 1980 model year vehicles
equipped with either 2.3 or 5.0 liter engines. We found that the 2.3
liter, 2 barrel, manual (both 4- and 5-speed) transmission, open-loop
vehicles produced 28 MPG , and thus exceeded the "best fuel economy"
version in Ford's study. Similarly, the 5.0 liter, 2 barrel or fuel
injected, automatic (3- and 4-speed) transmissions, open or closed-loop
vehicles (with the exception of the LTD, LTD Wagon, Marquis, and Marquis
Wagon) all achieved at least 18 MPG .*** We, therefore, have doubts
concerning the accuracy of Ford's computer engine mapping technique as
an indicator of EPA test results.
* Hearings before the Subcommittee on Energy and Power of the Committee on
Interstate and Foreign Commerce of the House of Representatives, March
13 and 14, 1979.
** "Ford 0.4 NOx Research Objective Program," September 1979, pages 1-2.
*** EPA/DOE 1980 Gas Mileage Guide.
V-37
-------
C. Effects of Alternate "Standards" on Fuel Economy
Ford has studied the effects on fuel economy of a proposed (by Ford)
emissipn standard of 0.39 g/ml non-methane HC, 9.0 g/mi CO, and 1.5 g/ml
NOx compared to the current standard of 0.41 g/mi HC, 3.4 g/mi CO, and
1.0 g/mi NOx. Ford states, "Based on a preliminary analysis, such a
change starting in the 1983 model year could, conservatively, provide an
average benefit of 0.5-0.7 mpg to the fuel economy of the Ford fleet."*
Since the 1983 CAFE standard is 26.0 mpg, Ford apparently believes that
their proposed emission standard will result in a fuel economy improve-
ment of approximately 2% to 3%. Ford provided neither the test data nor
the analysis method used to predict that fuel economy improvement.
D. General Comments By Ford
Ford, in the same report states, "It has not been possible for Ford to
separate the fuel economy penalty attributable to emission controls from
the overall improvement in fuel economy, but the penalty is believed to
be significant."** Ford's statements contradict the computer simula-
tions of Heywood, Higgins, Watts, and Tabaczynski which indicate that
fuel economy can be improved while controlling emissions. For example
simultaneously increasing the amount of EGR and advancing the spark
timing could increase fuel economy without causing the NOx to sig-
nificantly increase.*** The resulting increase in HC could probably be
handled by the catalyst. See figures Ford-1 and Ford-2.
E. General Comments by EPA's Technical Staff
It is quite time consuming to generate test data to analyze since it is
difficult to run more than one city (FTP) test per day on a given
vehicle. Ford eliminates this problem by running either steady-state
* "Ford Status Report," February 1981, Section IIF.
** Ibid., Section III B.
*** Heywood, et. al, "Development and Use of a Cycle Simulation to Predict
SI Engine Efficiency and NOx Emissions," SAE Paper 790291, page 18.
V-38
-------
360
340
0>
u
320
300
40*
6 8 10 12
bsNO. gNO/kW-hr
14
16
Figure Ford - 1*
- Lines of constant percent exhaust gas
recycle, and constant timing relative to MBT, on
a plot of brake specific fuel consumption versus
brake specific nitric oxide emissions. Combus-
tion duration 40°, equivalence ratio 1.0, 1400
rev/rain, bmep 325 kPa
360
340
i? 320
300
I
JL
6 8 10
bsNO. qNO/kW-hr
12
14
Figure Ford - 2
- Lines of constant percent exhaust gas
recycle, and constant timing relative to MBT, on
a plot of brake specific fuel consumption versus
brake specific nitric oxide emissions. Combus-
tion duration 60°, equivalence ratio 1.0, 1400
rev/min, bmep 325 kPa
Ibid., page 18.
V-39
-------
tests or by running what Ford calls a hot-CVS (which is a city test with
the vehicle already warmed-up). Unfortunately, it is very difficult, if
not impossible, to accurately predict both emissions and fuel economy
results for the FTP cycle based only on steady-state or hot-CVS test
results (as shown in appendix 1 of this report). Therefore, Ford's
approach might not be correct for actual FTP results. Also, Ford makes
a number of claims but without providing any data to document those
claims.
2. New Technology or Developments which Could Provide a Change in the Re-
lationship between Emissions and Fuel Economy as Compared to 1981 Vehi-
cles or Ford's Estimates.
A. New Engines
1. PROCO (PROgrammed COmbustion)
The PROCO engine is a direct injection, open chamber stratified
charge engine. Fuel is injected into the combustion chamber during
the compression stroke and ignited by dual spark plugs. Stratifi-
cation is achieved by relatively late injection timing and the
swirl of incoming air, resulting in a rich mixture in the vicinity
of the spark plugs and a very lean mixture around the periphery of
the combustion chamber.
Ford has tested PROCO engines with displacements of 7.5, 6.6, 5.8,
5.0, 3.3, and 2.5 liters.* Ford has found that, "Because the PROCO
engine runs at overall lean air/fuel ratios and the fuel charge is
concentrated in the center of the combustion chamber, thus reducing
fuel contact with the cylinder head and walls, both HC and CO
* "Ford Status Report," February 1981, Sections IIG & IIH,
V-40
-------
emissions are minimized. However, oxidation catalysts will still
be required to meet the 0.41 g/mi hydrocarbons and 3.4 g/mi carbon
monoxide emission standards. The NOx standard of 1.0 g/mi is also
achievable due to this engine's ability to use large amounts of
EGR."* Fortunately, the use of EGR should not lower fuel economy,
as Ford explains, "The erroneous notion that EGR is inherently
detrimental to fuel economy comes from the fact that EGR reduces
the combustion speed. If the engine had the optimum combustion
speed without EGR and measures are not taken to counteract the com-
bustion speed reduction effect of EGR, then the fuel economy will
indeed suffer from EGR. The combustion speed of the PROCO engine,
however, is optimized for the 100% charge dilution condition with
the aid of inlet swirl, squish, and dual ignition in order to take
full advantage of the fuel economy potential inherent in charge
dilution."**
Ford estimates that the PROCO engine, when calibrated to meet the
1981 statutory emissions standards, will produce a 20-25% fuel
economy improvement over a carbureted, conventional gasoline engine
of the same displacement which was also calibrated to the 1981
standards.**
According to Ford, the major problem in mass producing the PROCO
engine "revolves around mass manufacturing of its precision fuel
injection system which requires extremely tight tolerances.
Pending continued successful development and resolution of manu-
facturing open issues, the PROCO may be available for production
use by the mid-1980's."**
Ford states that, "As engine power-to-vehicle weight ratios are re-
duced, it may be necessary to turbocharge the smaller engine to
restore the required performance level. In addition to meeting the
* Nickol, "Automotive Powertrains - Now and into the 1990*s," SAE Paper
801340, October 1980, page 6.
** Scussel, et al, "The Ford PROCO Engine Update," SAE Paper 780699, August
1978, page 4.
V-41
-------
required performance level, it may be necessary to turbocharge in
order to meet the NOx emission control requirements. ...Testing to
date indicates that turbocharging PROCO, with its inherently fast
combustion, will require modifications to the current technology
PROCO combustion system."* In fact Ford says, "The advanced [i.e.
turbocharged] PROCO is not expected to be available prior to
1990."**
In a recent interview with "Ward's Engine Update," Ford Chairman,
Philip Caldwell indicated that the PROCO engine will not be going
into production. Ford will, however, try to incorporate some of
the technological know-how they developed while working on the
PROCO on some other projects that will go into production.***
2. Stirling
The Stirling engine that Ford is investigating is a reciprocating,
closed cycle, continuous external combustion engine with multi-fuel
capabilities. Ford states that, "Using gasoline, the Stirling has
been projected to be capable of achieving up to 40% better economy
than conventional gasoline engines, although experimental vehicles
have fallen far short of this goal."**** These projections, how-
ever, are not based on actual test data.
Ford's work on the Stirling engine had been done under Department
of Energy Contract No. EC-77-C-02-4396. At the conclusion of Task
* "Ford Status Report," February 1981, Section IIH.
** Nickol, "Automotive Powertrains - Now and into the 1990's", SAE Paper
801340, October 1980, page 10.
*** Ward's Engine Update," Volume 6, Number 24, December 15, 1980, pages 5
& 7.
****Nickol, "Automotive Powertrains - Now and into the 1990's," SAE Paper
801340, October 1980, pages 11-12.
V-42
-------
I in September 1978, Ford officially notified the Federal govern-
ment agencies and private firms that it intended to withdraw from
active participation in the development of the Stirling engine as
an automotive power plant.*
3. Gas Turbine
Ford is a subcontractor to AiResearch Manufacturing Company of
Arizona on the current NASA/DOE Advanced Gas Turbine Powertrain
System (Contract No. DEN 3-37). "Ford believes that the ceramic
single shaft turbine has the best potential of any turbine for
attaining low emissions, excellent fuel economy and reasonable
manufacturing cost. The potential of meeting a 0.4 [g/mi] NOx
standard is yet to be determined."*
Ford and AiResearch believe that a 3000 pound car powered by such a
turbine could achieve a fuel economy of 42.8 MPG ; however, a
vehicle is not planned to be available for testing until July 1981
and no test data is currently available.** There was no
speculation as to when this engine might be available in
automobiles.
4. Diesel
Ford is currently developing Diesel engines; however, there is no
indication of when they will be available for production. Until
Ford's Diesel is ready for production, Ford will purchase Diesel
engines from companies such as Bayerische Motoren Werk AG (BMW)***
and Toyo Kogyo (TK).****
* "Ford Status Report," February 1981, Section IIH.
** Rackley, "Advanced Gas Turbine Powertrain System Development Project,"
presented at Automotive Technology Development Contractors Coordination
Meeting - Dearborn, Michigan, November 1980.
*** "Ward's Engine Update," Volume 6, Number 24, December 15, 1980, page 5.
****"Ward's Engine Update," Volume 6, Number 20, October 15, 1980, page 6.
V-43
-------
Ford's Diesel research has not reached a point at which they can
predict fuel economy and emission results. They are currently in-
vestigating turbocharging and both direct and indirect injection
combustion chamber configurations.*
As part of a correlation program with Ford, EPA tested a front-
wheel drive Escort equipped with a manual transmission and a TK
Diesel engine. A total of two tests were performed at EPA and
eight tests at Ford; the average results of those tests are shown
below:
FTP (g/mi) _
MPG
51.4 42.4
50.8 41.7
It should be pointed out that, at both test facilities, the NOx
levels are below 1.0 g/mi and the total particulate levels are
below 0.20 g/mi.** There has been some interest in the fuel
economy of a Diesel at these emission levels. This vehicle was
tested at an ETW of 2750 pounds with a dynamometer horsepower
(VDHP) setting of 6.7 horsepower. Two similar 1981 model year
Diesel vehicles with manual transmissions were identified. The
following table allows a rough comparison of these three vehicles:
Vehicle ETW VDHP MFG., MPG
Lab
EPA
Ford
# of
Tests
2
8
HC
.241
.213
CO
.70
.69
NOx
.74
.74
Part.
.184
.145
MPGU
37.1
36.4
Ford Correlation Diesel 2750 6.7 37.1 51.4 42.4
Chevette Diesel 2500 8.3 40.1 55.1 45.7
VW Dasher Diesel Wagon 2625 6.9 36.0 48.0 40.6
* Nichol, "Automotive Powertrains - Now and Into the 1990fs," SAE
Paper 801230, October 1980, pages 8-9.
** Watson, "Ford - EPA 1981 Light Duty Vehicle Diesel Correlation",
EPA Technical Report No. EPA-AA-EOD-81-3, May 1981.
V-44
-------
B. Continuously Variable Transmission (CVT)
A truly continuously variable transmission is one which permits ratio
variations completely independent of vehicle speed or engine output.
Its major advantage is that it provides the flexibility to run an infi-
nite number of drive ratios between fixed upper and lower limits. Ford
says that the advantages of the CVT are:
1. Smooth vehicle acceleration without shift impulses.
2. Acceleration with the engine operating very close to its
optimum power output, thereby improving performance.
3. Road load engine operation at peak efficiency consistent with
acceptable driveability, thereby improving economy, due to
CVT's ability to optimize its performance within this narrow
range.*
Ford states that, "To date, however, some of the apparent benefits have
been offset by high internal frictional losses. Much work remains,
making the CVT an unlikely prospect for the mid-1980's."**
3. The Effect of Known Technology which Helps Quantify Single or Multiple
Calibrations or Vehicle Description Changes.
A. Automatic Overdrive Transmission (AOD)
The AOD, which Ford began using in the 1980 model year, incorporates
three fuel savings features. First, the overdrive gear (fourth gear)
has a gear ratio of 0.67 to 1.0, thus allowing the engine to operate at
* Nichol, "Automotive Powertrains - Now and Into the 1990's", SAE Paper
801230, October 1980, page 13.
** "Ford Status Report," February 1981, Section IIF.
V-45
-------
a lower speed than it would with a similar automatic transmission which
had a final gear ratio of only 1.0-to-l.O. Secondly, the torque con-
verter clutch directs the flow of engine power to completely bypass the
torque converter and to follow a direct mechanical path to the drive-
shaft, thus eliminating torque converter losses while the transmission
is in fourth gear. Finally, the transmission uses a split torque path
which all'ows a portion of the engine power to bypass the torque con-
verter and follow a mechanical path to the drive shaft, thus eliminating
some of the torque converter losses. While none of those three features
is unique in the automotive industry, Ford claims that the combination
of all three "is an industry 'first.1 Only Ford Motor Company offers
this feature."*
Ford claims, without providing substantiating data, that the overdrive
gear can increase highway fuel economy (MPG,) by up to 6%, that the
torque converter clutch can increase combined fuel economy (MPG )
approximately 4%, and that the split torque principle can improve MPG
approximately 77,.**
Ford pointed out that this system could be improved by incorporating an
electronic control system such that a single microprocessor would
control both engine and transmission functions. Ford states that,
"Potential benefits from such a system could include improved fuel
economy through implementation of a neutral idle control strategy and
increased transmission shift scheduling flexibility permitting the best
compromise of shift timing to attain optimum fuel economy, emissions and
driveability."*** However, Ford did not indicate if they are studying
this possible improvement.
* Dabich, "Ford Motor Company Automatic Overdrive Transmission," SAE Paper
800004, February 1980.
** Nickol, "Automotive Powertrains - Now and into the 1990's," SAE Paper
801230, October 1980, pages 12-13.
*** Ibid.
V-46
-------
B. Weight Reductions
Ford's Metallurgy Department has been investigating the potential for
using high strength steels (ESS) in place of mild steel. Ford states
that "the use of higher strength steels allows for reduced weight in
components that are primarily used in energy absorbing applications. An
example of this is the widespread use of HSS in sidedoor beams with
large weight reduction over original mild steel versions. ...Although
the total weight of steel in future vehicles is somewhat uncertain, it
is possible to project that sheet steel stampings will make up about
forty percent of future vehicle weights. The overall penetration of
high strength steel among the sheet steels is now about 20% in Ford
vehicles; the authors speculate that this penetration will reach 30 to
50% of the total automotive sheet steel by the end of the decade... We
speculatively project the application of high strength sheet steels per
vehicle to approach the 300 to 600 pound range by the end of the
decade."* Ford believes that the use of high strength steels will be a
key feature in "achieving very significant worldwide fuel savings."**
Ford anticipates that the use of high strength steels to replace average
strength steels will result in the savings of 100 pounds per vehicle by
1985 (See figure Ford-3).** See figure Ford-4 for overall weight re-
duction.
C. Reduction of Internal Friction within the Engine
Ford's Fuel and.'Lubricants Department analyzed motored engine friction
data from a Ford 1.6 liter and a Dacsun 2.0 liter engine. Ford found
the frictional losses in the Ford 1.6 liter engine were significantly
higher than in the Datsun 2.0 liter engine. Ford determined that this
* Magee, et.al., "Automotive Sheet Steels for the 1980's," May 1980.
** Magee, et.al., "Factors Influencing Automotive Applications of High
Strength Steels," August 1979.
V-47
-------
6
150
M
t/i
O
z
SU
BS
fr*
cn
z
w
H
es
o
I
co
a
100
WEIGHT
SAVE
/
50
19.75
1980
1985
Figure Ford - 3*
Trends in strength level of HSS used in automobiles and the
increased weight reduction
Ibid.
V-48
-------
Figure Ford - 4*
SAtES MELG'.ITED AVERAGE INERTIA TEST WEIGHT FOR ?ORD BY MODEL YEAR
PASSENGER CARS
4500
4000
1/1
3500
3000
2500 H-
Tentative Programs Under Consideration
1974 1976 1978 1980
1982
MODEL YEAR
198"4 1986
8 1990
4-
99
* ''Ford Status Report," February 1981, Section IIP.
-------
friction could be reduced by redesigning the piston/cylinder interface
and by the use of low friction oils.*
Ford states that their "studies show that reduced friction engines in-
corporating fast burn/lean burn combustion technology offer potential
for about a 12% fuel economy improvement over the original engine.
Although some technological, cost and lead time issues remain to be
resolved, increased usage of this technology is expected throughout the
mid-1980's.**
D. Reducing Parasitic Losses
Ford found that, by replacing the conventional V-belt arrangement on a
1978, 5.0 liter Mustang with a serpentine drive arrangement, "increased
belt flexibility and improved traction resulted in a more efficient
drive system with potential increase of 0.04 mpg."*** An increase of
0.04 mpg on that model amounts to an increase of 0.2%.****
"Fans which deactivate when satisfactory engine cooling exists offer
potential for improved on-road fuel economy which may, to some degree,
be reflected on the EPA test. Electro-mechanically clutched fans have
been used on Ford vehicles such as the Fiesta and Escort/Lynx and are
expected to see more wide spread applications in the future."*****
Variable speed and two-speed accessory drives are being studied for use
on vehicles with low power-to-weight ratios and relatively heavy ac-
cessory loads. These systems will drive air pumps, alternators, power
steering pumps, and air conditioners. For the variable speed system,
*Willermet, "Lubrication Modes and Engine Friction - A Comparison of Motored
Engine Friction Data for the 1.6L Ford and 2.0L Datsun Engines," January
1981, attachment to "Ford Status Report," February 1981.
**"Ford Status Report," February 1981, Section IIF.
***Cassidy, et.al., "Serpentine - Extended Life Accessory Drive," SAE Paper
790699, June 1979.
****EPA/DOE 1978 Gas Mileage Guide.
*****"Ford Status Report," February 1981, Section IIF.
V-50
-------
"[k]ey problem areas relate to durability, packaging and optimization of
accessory performance." For the two-speed system, "ultimate usage is
questionable on a cost/benefit basis."*
E. Supercharging
Ford claims that supercharging by way of a turbocharger can allow that
engine to equal the performance of a naturally aspirated engine which is
50% larger than the turbocharged engine. Also, the turbocharged engine
according to Ford, can provide approximately 9% better fuel economy than
the larger naturally aspirated engine and with no significant impact on
emissions.**
F. Decel/Idle Fuel Shut-Off System
This system shuts off the fuel, and hence the engine, when the engine is
warm and would normally be idling. Ford claims, without submitting any
supporting data, that this system provides a fuel economy improvement of
1% to 2% on vehicles equipped with either manual or automatic overdrive
transmissions when tested on the FTP cycle excluding the cold start
(Ford's HOT-CVS test cycle). Vehicles equipped with conventional auto-
matic transmissions have fuel economy gains of less than 1%. Ford
states these produce higher HC feedgas levels; however, there is no
change in HC tailpipe levels. The problems with this system, according
to Ford, are:
1. Without a speed signal, the system gives poor driveability and
potential stalls.
2. The system incurs substantial cost penalties for additional
hardware (e.g., sensors, solenoids, etc.).*** However, some
*"Ford Status Report," February 1981, Section IIF.
**Nickol, "Automotive Powertrains - Now and into the 1990's," SAE Paper
801340, October 1980, page 5.
***"Ford Status Report," February 1981, Section IIA3f.
V-51
-------
of that hardware would already be present if the vehicles were equipped
with a closed-loop system.
Ford states that, "These problems are expected to be resolvable and pro-
duction incorporation is being studied or actively pursued for several
engines."*
G. Multipoint Electronic Fuel Injection
Ford currently (1981 model year) uses central fuel injection (CFI) on
its fuel injected models. Ford is working to develop a multipoint
electronic fuel injection (EFI) system for use on 4-cylinder engines.
The multipoint EFI system with its four injectors has the potential for
greater control of the fuel metering than does the CFI with only its two
injectors.** If this potential is realized, the multipoint EFI could
result in an improvement in both emissions and fuel economy.
* Ibid., Section IIF.
**Ibid., Section IIA3.
V-52
-------
V. C. Chrysler
1. Chrysler's Statements Concerning the Relationship between Emissions
and Fuel Economy
In CO waiver hearings, Chrysler said that fuel economy, on average, is
degraded by about 2% for vehicles complying with the 3.4 g/mi CO standard, as
compared to a 7.0 g/mi CO standard. According to Chrysler, the 7.0 standard
would allow the use of richer A/F ratios during acceleration, to enhance the
catalyst's NOx reduction capability. This allows the use of EGR and spark
timing calibrations that are more optimum for fuel economy. The fuel economy
benefit varies from one engine family to another depending on the degree to
which EGR and spark timing calibrations have been compromised in order to meet
the NOx standard. Chrysler indicated that, on average, fuel economy is
expected to improve .by 2%, but this was based on their judgment. They had not
run back-to-back tests between vehicles calibrated for 3.4 and 7.0 g/mi CO
standards.*
In a historical review on emission controls, Chrysler commented that with the
introduction of catalysts,
it became necessary to remove lead from gasoline. The
antiknock level of unleaded fuel that could be accomplished
with existing refining technology was and is substantially
below that which existed with the lead antiknock compound
present. As a result, lower compression ratio had to be
used with resulting loss in fuel economy.**
* Transcript of Proceedings, United States Environmental Protection Agency,
In the Matter of: Public Hearing on Applications for Waiver of Carbon
Monoxide Emission Standards for Certain Model Automobiles, Washington,
D.C., October 24, 1980.
** Charles M. Heinen, Chrysler Corp., "We've Done the Job - What's It Worth,"
SAE Paper 801357, October 20-23, 1980, pages 9 & 11.
V-53
-------
This paper made the case that lighter vehicles with electronic engine con-
trols and changes in
mixture preparation and distribution should result in exhaust
emissions before any catalyst of 1.5 HC, 10 CO and 1.5 NOx based
on vehicles already in existence. These systems will be on
vehicles regardless of the emissions requirements because of
their fuel economy benefits. Without a catalyst a moderate
amount of lead would be permissible. The results, of course,
would be a gain of not only fuel economy, but a reduction in the
cost of fuel with all their attendant benefits.
The author admitted that the fuel economy loss is a highly controversial sub-
ject, but referred to a Detroit News Editorial, June 9, 1980, for order of
magnitude figures that indicated that leaded regular fuel would cost 6-7% less
than unleaded fuel and improve fuel economy 5-10%.**
Table Chrysler-1 below was included in the SAE Paper cited above. This infor-
mation was taken by Chrysler from the Annual Report of the Administrator of
the Environmental Protection Agency to the Congress of the United States, "The
Cost of Clean Air and Clean Water," 96th Congress, First Session, Document No.
96-38, December, 1979.
Table Chrysler-1**
Fuel Fuel
Consumption Consumption
Year Penalty* Year Penalty*
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
* millions of
343.64
599.19
1033.89
1556.84
2346.72
3304.32
2954.63
2540.78
2077.18
1851.86
1977 dollars
1978
1979
1980
1981
1982
1983
1984
1985
1986
1587.17
1349.58
1049.19
849.61
524.89
366.24
255.02
151.95
85.44
** Charles M. Heinen, Chrysler Corp., "We've Done the Job - What's It
Worth," SAE Paper 801357, October 20-23, 1980, pages 9 & 11.
V-54
-------
In their "Oxides of Nitrogen Research Plan for Model Year 1979," Chrysler
stated the following:
Chrysler's fuel economy improvement plans are based on
increasingly more stringent emission standards which eventually
reached the level of .41/3.4/1.0. If the 0.4 NOx level should
ever be achievable, and a standard promulgated at that level our
efforts to improve fuel economy would suffer a severe setback.**
Chrysler did not provide test data to support their comments regarding the
relationship between fuel economy and emissions.
2. New Technology or Developments which Could Provide a Change in the
Relationship between Emissions and Fuel Economy as Compared to 1981 Vehicles
or Chrysler's Estimate
The "Progress Report on Chrysler's Efforts to Meet the Federal Emission
Standards for HC, CO, and NOx in 1981 and Subsequent Model Years," dated
February 1981, discussed seven development programs whose objectives were to
improve vehicle fuel economy. These include:
1. Vehicle weight reduction.
2. Power-to-weight reduction.
3. Drive trains.
4. Accessory drives.
5. Engine efficiency.
6. Tires.
7. Lubricants.
** Chrysler Corporation, "Chrysler's Oxides of Nitrogen Research
Plan, for Model Year 1979," March 14, 1979, page 7.
V-55
-------
This section will discuss only the new technology being developed by Chrysler
in the above projects.
Chrysler began development efforts on lock-up torque converters in January of
1973 and entered production for rear wheel drive vehicles in the 1978 model
year. They began development of lock-up torque converters for front wheel
drive vehicles in June of 1976. Chrysler stated, "at the present time, how-
ever, the use of lock-up on front wheel drive cars has been deferred due to
unacceptable driveability at the current state of development."* Chrysler did
not discuss the impact on emissions nor quantify the effect on fuel economy
for this technology.
In their engine efficiency program, Chrysler's "engine mapping and controls
optimization" project seems to be their most significant program to improve
fuel economy and emissions.
This project's objective is "to develop both the hardware and software for
electronic control of spark timing, fuel/air ratio, exhaust gas recirculation,
etc., so that fuel economy is maximized while emissions constraints are met."**
The project includes chassis dynamometer measurement and
computer simulation of engine speed/load requirements during
EPA Urban and Highway driving cycles. Key speed/load
conditions are selected by analyzing resulting histograms. The
engine configuration is then "mapped" (steady-state) on an
engine dynamometer at the "key" conditions for emissions and
fuel consumption. From these data, for any reasonable
emissions constraint, key point operating conditions can be
found which predict minimum fuel usage on a hot '74 EPA cycle.
These operating conditions are then incorporated in a vehicle
and evaluated on a chassis dynamometer.
* Chrysler Corp./'Progress Report on Chryslers Efforts to Meet the Federal
Emissions Standards for HC, CO, and NOx in 1981 and Subsequent Model
Years," February 1981, page F-3.
** Chrysler Corporation, "Oxides of Nitrogen Annual Report for Model Year
1980," December 23, 1980, page 1-9.
V-56
-------
Engine maps have been obtained, under a contract with the
Department of Transportation (DOT), for several versions of
both the 318 CID V8 and 225 CID slant 6 engines, as well as
turbocharged 1.7 liter 4-cylinder engine. Strategies have been
obtained for the base 318 CID and 225 CID engines. Vehicle
tests of these strategies are reported in the section on
Integrated Engine Controls. A strategy for the turbocharged
1.7 liter engine has also been prepared.
Preliminary tests on Car 198 (360 CID) have been made that
demonstrate the potential capabilities of the system and
optimization strategies. Tests with "optimized" strategies for
spark advance and EGR were compared against the standard spark
advance and EGR control system. The "optimum" strategies were
tested by programming schedules into a microprocessor capable
of controlling spark advance and EGR. The fuel/air ratio was
maintained around stoichiometric by an oxygen sensor feedback
Holley 2 bbl. carburetor with Holley electronics. A summary of
these tests follows [table Chrysler-2]*.
20C
Avg. of
4 tests
Avg. of
2 tests
Table Chrysler-2*
Car 198, 360 CID Engine
HOT '74 EPA TEST
Engine Out
Emissions (gm/mi)
HC CO NOx
2.84
3.94
26.9 1.71
21.2
2.48 18.6
2.42 19.9
2.71
2.26
1.66
Fuel.
Economy
MPG
12.3
14.5
14.2
13.9
Comments
Baseline-Mech. spark
adv. and prod. EGR
Schedule to improve
fuel economy
Schedule to improve HC
Schedule to improve NOx
Control strategies have also been run on 318 and 225 CID engines
(Cars 373 and 535, respectively). These cars were modified to
include three-way catalysts and feedback carburetor systems.
Typical results for hot '74 EPA tests are as follows:
* Chrysler Corporation, "Application For Waiver of the 1981-1982 Model Year
Carbon Monoxide (CO) Standard of 3.4 Grams Per Vehicle Mile For Passenger
Cars," July 3, 1979 at Vol. II, pages B5-54, B5-55, & B5-57.
V-57
-------
Table Chrysler-2 (con't)
Car 373, 318 CID Engine
Strategy
As received
Minimum Fuel
Emissions
Constrained
Engine-Out
HC CO
2.7 22.6
3.0 12.5
2.6 10.9
Hot '74
NOx
1.5
3.5
2.1
EPA
HC
.3
.2
.2
Test
Tailpipe
CO
4.6
2.3
1.7
NOx Fuel Economy
1.4 15.5
2.0 18.4
1.3 17.1
Car 535, 225 CID Engine
Hot '74 EPA
Strategy
As received
Minimum Fuel
Emissions
Constrained
Engine -Out
HC CO
NOx
Not Available
3.3 13.4
3.3 13.3
4.0
3.0
test
HC
.4
.4
.3
Tailpipe
CO
7.9
3.1
3.2
NOx Fuel Economy
1.4 19.4
2.2 21.3
.7 20.8
V-58
-------
These data have been processed to identify optimum control
strategies for specific vehicles with 318, 225, and 1.7 liter
engines. The strategies are currently being evaluated. The
results were presented in Volume II of Reference 22 Sec. B-5,
pp. 54-57 [see table Chrysler-2]. Since that time, initial
cold start and warm-up tests have been conducted on car 535.
Preliminary results are:
Table Chrysler-3
Car 535, 225 CID Engine
Cold '75 EPA Test
Engine Out Tailpipe
Urban Corrected
Strategy H£ C0_ NOx H£ C0_ NOx Fuel Economy
Emissions 4.00 20.6 2.13 .34 3.29 .37 17.9
This car was built up with experimental induction and emission
control systems and does not directly compare to the 0.4 NOx
program cars. The significance of the results, then, is that
we have established that the program cars are achieving fuel
economy comparable to the optimum values predicted from
analytic evaluation of comprehensive engine data.
It should be noted that, informally, the systematic
methodology discussed here has already had an evolutionary
effect on our test philosophies and is expected to have
increasingly greater impact. Therefore, all of these
activities will continue except for the DOT project which is
limited by contract.*
Chrysler Corporation, "Oxides of Nitrogen Annual Research Report for Model
Year 1979," October 25, 1979 pages 23-24.
V-59
-------
In addition to emission control systems optimization in engine mapping,
Chrysler's engine efficiency program also includes projects to study the fuel
economy benefits of:
a) Deceleration fuel shut-off
b) Reduced engine friction
c) Combustion chamber designs
d) Intake manifold designs
e) Turbocharged engines
The deceleration fuel shut-off project will restrict fuel flow during closed
throttle deceleration in an attempt to improve fuel economy. Chrysler is now
evaluating the fuel economy and driveability implications of this feature, but
did not discuss or provide data on emissions or fuel economy and did not
indicate when this technology would be used in production.*
For the 1981 model year, Chrysler has introduced lower tension oil rings in
order to reduce engine friction. Chrysler is also evaluating other friction
reducing items for their 2.2 L engine but has not indicated which components
will be evaluated. No emissions or fuel economy data were provided.
Fast burn combustion chambers are being evaluated to develop a design "which
combines the best possible part load efficiencies and dilution tolerance with
high specific power output and reasonable fuel octane requirement."* Swirl
ports and squish-type chambers with reduced flame travel distance are among
those being evaluated. Chrysler did not indicate when the results of this
development would be used in production, but said that the work is primarily
based on a 1981 2.2 L engine. Chrysler said that the 2.2 L engine's combus-
tion chamber "has proven to be a very good one with the exception that W.O.T.
output is limited by valve shrouding in the deep, compact chamber design."*
Emissions and fuel economy data were not provided.
Chrysler Corp., "Progress Report on Chrysler's Efforts to Meet the
Federal Emissions Standards for HC, CO, and NOx in 1981 and Subsequent
Model Years", February, 1981, pages F-7 & F-8.
V-60
-------
In order to improve the power output of the 2.2 L engine, Chrysler has rede-
signed the intake manifold by reducing branch lengths and providing less
restrictive flow paths. "The result was a large gain in mid-range to top-end
power. This was transformed into a low-to-mid-speed torque benefit through
advancing the camshaft timing."* Chrysler plans to introduce the new intake
manifold in the 1982 model year.
Chrysler evaluated the use of turbocharged engines of smaller displacement
than the naturally aspirated engines which they would replace, to increase
fuel economy without a loss in performance. Chrysler concluded that
turbocharged small displacement engines would not be a cost effective method
for improving fuel economy for high volume production. No data were provided.*
Chrysler's tire test program includes rolling resistance tests in an effort
to improve fuel economy. Chrysler stated:
The rolling resistance characteristics of our current
production tires are evaluated through head-to-head coast-down
tests against tires on competitive domestic and foreign
vehicles.
The rolling resistance characteristics of the tires to be
released for production are optimized by selecting only the
best for production. This has resulted in a five percent
reduction in vehicle force required as reported to the EPA
between the 1980 and 1981 model years. Through competition
between tire suppliers, all eventually had tires that were of
about equal rolling resistance and were acceptable from other
release standards.*
Chrysler said that although tire rolling resistance has improved greatly from
1979 to the 1981 model years, they do not feel that there will be any rolling
resistance improvement for the near future until major improvements are made
in design or compounding. Chrysler did not discuss the emissions impact of
their tire test program.
Chrysler Corp., "Progress Report on Chrysler's Efforts to Meet the
Federal Emissions Standards for HC, CO, and NOx in 1981 and Subsequent
Model Years," February, 1981, pages F-7 to F-10.
V-61
-------
A "fuel efficient" lubricant test program included evaluations of engine lub-
ricants and axle lubricants. Four engine oils were tested for fuel economy
improvements. Of these four, "only one, Sun Code X892, showed significant
(1.6-3.2%) fuel economy improvement potential in the 225 CID engine."* The
Sun Code X892 oil is now available commercially as Sunoco "MPG PLUS" oil.
Chrysler said that since this oil includes a friction modifier containing
phosphorus, durability tests would be required to determine the effect on
catalysts and oxygen sensors. Chrysler did not indicate when this oil or
other fuel efficient oils might be used in production. No emissions data were
provided from this testing.
Table Chrysler-4 lists fuel economy improvements associated with the use of
an experimental 75W axle lubricant as compared to an 80W-90 rear axle lub-
ricant. According to Chrysler "additional economy testing of other candidate
oils will be limited due to the future decrease of sales volume of rear wheel
drive vehicles. It is estimated that no fuel economy improvement potential
exists in this area for the corporate front wheel drive vehicles; therefore,
similar tests were not planned. " Although a fuel economy improvement was
shown on all four vehicles "the 75W oil was considered unacceptable due to
differential bearing wear and ring and pinion gear scoring." The effect on
emissions was not discussed.
Table Chrysler-4
Axle Lubricant Test Results
REFERENCE OIL: MS-5644, SAE 80W-90 FACTORY FILL
TEST OIL: O.S. 38623, SAE 75W
CONTAINS SOLUBLE FRICTION MODIFIER
Vehicle Type EPA Composite Economy Effect*
1978 "F" Body (225) 2.0% Improvement
1978 "F" Body (318) 0.6% Improvement
1979 "F" Body (318) 1.2% Improvement
1979 "D" Body 100 1.1% Improvement
*Compared to factory-fill baseline.
Chrysler Corp., "Progress Report on Chryslers Efforts to Meet the Federal
Emissions Standards for HC, CO, and NOx in 1981 and Subsequent Model
Years," February, 1981, pages F-10 & F-ll.
V-62
-------
Chrysler evaluated the feasibility of utilizing an oxidation catalyst system
to meet emission standards of 0.41 HC, 3.4 CO, 1.0 NOx on a vehicle with a 2.2
liter engine. Chrysler's data, listed in table Chrysler-5, shows a 2 mpg
fuel economy loss for the oxidation catalyst system on the FTP.
Table Chrysler-5
Tailpipe
HC CO
r, /wi-? ...
g/mi —
.25 2.1
.19 1.9
-24% -9.5%
-.06 -.2
NOx
.58
.83
+43%
+.25
Engine- Out
HC CO NOx
____-.«> I m •? _.._
g/ mi
2.0 11 1.80
2.2 30 0.80
Fuel
Urban
1'
26.0
24.0
-7.7%
-2.0
Economy
Hwy
mr*
34.4
33.6
-2.3%
-0.8
82 Dual Bed (DB)
Ox Cat (OC)
100 (OC-DB), %
DB
Difference (OC-DB)
In order to comply with the 1.0 g/mi. NOx standard, Chrysler used high EGR
rates and rich A/F ratios. Chrysler said that this type of calibration "leads
to unstable or inconsistant combustion which can produce undesirable vehicle
surge and unsatisfactory performance under certain operating conditions."*
The 30 g/mi. engine-out CO level for the oxidation catalyst system as compared
to the 11 g/mi. CO level for the dual bed system indicates that it is likely
that the oxidation catalyst equipped vehicle was operating with richer A/F
ratios than the dual-bed equipped vehicle. This may explain the 2 mpg loss in
fuel economy. Chrysler also implied that the spark timing and EGR cali-
brations also had an adverse impact on fuel economy, but did not provide their
specifications.
Chrysler Corp., "Progress Report on Chrysler's Efforts to Meet the Federal
Emissions Standards for HC, CO, and NOx in 1981 and Subsequent Model
Years," February, 1981, pages A-l & A-2.
V-63
-------
Chrysler's open chamber Diesel project is intended to determine the feas-
ibility of using the open chamber Diesel engine in passenger cars.
Early tests indicated significant potential for improvement
of the fuel economy/NOx relationship over that of a pre-
chamber engine if fuel injection timing and rate profile
could be precisely controlled• Electromechanical actuators
were developed to provide this control.
Single and multicylinder tests confirmed indications of a
superior fuel economy/NOx relationship. Fuel economy
benefits of 10 to 15% over an equivalent pre-chamber Diesel
were observed, along with some evidence that NOx levels
below 1 gm/mi might be achieved.
However, two major problem areas were identified:
- HC emissions are excessively high in those areas where
NOx control is best.
- Fuel control system reliability is not sufficient to
maintain a productive test program.
An intensive series of tests involving combustion chamber
parameters as well as a variety of fuel system components
has indicated some promise for solving these problems. It
is not yet possible to say, however, that this concept will
achieve the anticipated lower emission levels while re-
taining a substantial fuel economy advantage. Current de-
velopment efforts are directed to the near term solution of
these problems.*
Chrysler Corp., "Progress Report on Chrysler's Efforts to
Meet the Federal Emissions Standards for HC, CO, and NOx in
1981 and Subsequent Model Years", February, 1981, page H-2.
V-64
-------
V.D. Toyota
1. Toyota's Statements Concerning the Relationship Between Emissions and
Fuel Economy
A. Toyota states that "strengthened emission standards affect not only fuel
economy but also all of other factors such as driveability, performance,
durability, cost, etc." Toyota has been complying with the increasingly
stringent emission standards by trading off the above factors except
fuel economy. Toyota states, "Fuel economy cannot be sacrificed because
it shares a large factor in market competition." Accordingly, Toyota is
"optimizing and upgrading emission control systems in order to maintain
or improve economy, which results in increased cost of system."*
Toyota included graphs (which appear on the next page, figures Toyota-1
and Toyota-2) which indicate the effects of various NOx standards (from
2.0 to 0.4 g/mi) on both fuel economy and on costs. These graphs show
that to go from a NOx standard of 1.5 g/mile to one of 0.4 g/mile would
increase cost due to the need to employ a more sophisticated technology;
however,.the new technology would increase fuel economy by as much as 10
to 15%.** A 15% increase in fuel economy for a vehicle rated at 25 MPG
would result in a savings of 260 gallons of fuel during the first 50,000
miles. (At a fuel cost of fcl.SS/gallon, the savings would be $403.) On
the graph, the system designated "p-2" indicates carbureted vehicles
equipped with oxidation catalyst systems; "P-8-2" and "P-8-3" indicate
carbureted vehicles equipped with closed loop 3-way catalyst systems;
and "P-7" and "P-ll" indicate fuel injected vehicles with closed-loop,
3-way catalyst systems.***
*"Toyota Status Report - Efforts to Meet the 1981 and Subsequent Model
Year Light-Duty Vehicle Emission Standards and Other Statutory Requirements,"
page 70.
** IBID., page 71.
*** IBID., page ii.
V-65
-------
20
+ 10
o
c
o
u
0)
(U
3
fc,
-10
-20
2.0
P-7 or P-ll
P-8-2^or P-S-3
1.0 0.70.4
g/mile
Nox. emission...
Figure Toyota - 1* NOx Emission VS. Fuel Economy
c
0)
a
u
LI
O
u
•H
ro
4J
'
-------
2. New Technology or Developments which Could Provide a Change in the
Relationship between Emissions and Fuel Economy Compared to 1981 Vehicles
or Toyota's Estimates.
A. Combustion Chamber Redesign
Toyota confirmed the general tendency for fuel economy to improve with
increased EGR rate and advanced spark timing. When Toyota retarded the
spark timing to decrease NOx, they found the fuel economy penalty was re-
duced by using a high turbulence type combustion chamber in place of a
non-squish type chamber. The high turbulence type chamber, which used a
"twist type piston", also reduced the HC emissions.*
Since the accompanying data were generated by using non-FTP test cycles,
we cannot directly translate their results to probable changes in fuel
economy on either EPA's urban or highway test cycles.
A similar calibration (high turbulence combustion chamber, higher com-
pression ratio, extension of EGR tolerance) is being used for compliance
with the Japanese emission standards. Toyota plans to investigate the
feasibility of using a similar calibration to comply with the U. S. emis-
sion standards. The current problems are insufficient acceleration per-
formance, higher NOx emissions, and insufficient fuel economy im-
provement.**
* Matsumoto. et al, "The Effects of Combustion Chamber Design and Com-
pression Ratio on Emissions, Fuel Economy and Octane Number Requirement,"
SAE Paper 770193, page 13.
** "Toyota Status Report - Efforts to Meet the 1981 and Subsequent Model Year
Light-Duty Vehicle Emission Standards and Other Statutory Requirements,"
page 38.
V-67
-------
Toyota provided emissions and fuel economy data generated with such a
combustion chamber design change. Unfortunately, Toyota did not detail
the changes. The following data were generated on a 3,000 Ib. IW vehicle
equipped with a five-speed manual transmission and a 2.4 liter, 4
cylinder engine* (the description of this vehicle is similar to that of
the Toyota Celica):
Compression Fuel Economy
Ratio HC (gpm) CO (gpm) NOx (gpm) Gain (%)
Before Modification 8.4
0.100
1.65
0.55 (Baseline)
After Modification
9.0
0.104
1.65
0.56
5.5
B. Digitally Controlled EFI
Toyota introduced the digitally controlled EFI system into the Japanese
market in August 1980. Use of this system, Toyota states, "results in
the improvements in performance, driveability, fuel economy, and relia-
bility" by controlling the "amount of fuel, ignition timing, and idle
speed." Toyota' did not submit any data to substantiate the claims.
Toyota plans to phase this system into the U.S. market beginning in the
1983 model year.**
* IBID., pages 64-65.
** IBID., page 37.
V-68
-------
C. Single Point Injection (SPI)
Toyota says that they have been investigating the basic characteristics
and "potentialities of performance and emission reduction of the SPI
system for several years." Toyota says that they hope to improve the fuel
metering accuracy over the current conventional carburetors or EFI systems
by using a SPI system.
Toyota stated that, with the SPI system, they are currently experiencing
the following problems:
1. "Unbalanced distribution of A/F mixture among the cylinders,
2. Large fluctuation of transient A/F ratio, and
3. Insufficient startability."
"After making a comparison among SPI systems and current carburetors or
EFI systems," Toyota says they, will decide the policy of how to adopt the
SPI systems to their engines.* SPI is, according to Toyota, more ex-
pensive and more fuel efficient than convential carburetors, but SPI is
less expensive and less fuel efficient than EFI.**
* IBID, pages 34-36.
** "Toyota NOx Research Program - 1979 Annual Report Period Plan,'
September 1979, page 15.
V-69
-------
3. The Effect of Known Technology which Helps Quantify Single or Multiple
Calibrations or Vehicle Description Changes
A. Vehicle Weight Reduction Programs
Toyota plans to reduce vehicle weight by "rational design, minia-
turization, and material substitution." Some models will be lightened,
and others will be replaced by new and lighter car lines during the 1980
through 1985 model years.*
The lightening of vehicles will improve fuel economy with no adverse
effect on emissions.
B. Transmissions
1. Manual Transmission
To lower engine speed, Toyota adopted wide ratio transmissions and its
fuel economy gain was about 2% for the 1981 models. Toyota plans to
review the current gear ratios to gain additional fuel economy
improvements for the 1983 models. Toyota did not state whether they
anticipate any effects on emissions.
2. Automatic Transmission
To lower engine speed and eliminate slippage losses, Toyota plans, for the
1982 model year, to equip its Corolla and pickup truck with a four-speed
automatic transmission with overdrive and its Celica Supra with an auto-
matic transmission with a lockup torque converter. Toyota expects a fuel
economy improvement of 2% to 4% with the automatic four-speed and an im-
provement of 1% to 2% with the lockup.
"Toyota Status Report - Efforts to Meet the 1981 and Subsequent Model Year
Light-Duty Vehicle Emission Standards and Other Statutory Requirements,"
pages 57-58. v_?()
-------
Toyota plans to improve the lockup torque converter by allowing the lockup
function to operate in both second and third gears. They anticipate an
additional 1% increase in fuel economy due to this improvement. Toyota
plans to introduce this improved transmission on its 1983 model year
cars.* Toyota did not state whether they anticipate any effects on emis-
sions.
C. Toyota currently (1981 model year) has a 6 cylinder, 168 CID (gasoline)
engine equipped with feedback EFI and 3-way catalyst (engine family number
BTY2.8V5HB4). Of the 15 valid (non-void) FTP tests performed on the seven
certification data vehicles in that family:
1. No HC emission level exceeded 0.226 g/mi (without DF),
2. No CO emission level exceeded 1.42 g/mi (without DF), and
3. No NOx emission level exceeded 0.32 g/mi (without DF).
Applying the applicable deterioration factors (HC = 1.385, C0= 1.385, and
NOx = 1.275), we note that every FTP would meet the 0.41 HC, 3.4 CO, 0.41
NOx research target level. However, the durability vehicle line-crossed
by generating a 50,000 mile calculated NOx value of 0.425 g/mi, which
exceeds 0.41 g/mi.** Considering how close the durability vehicle came to
being acceptable for use at the NOx research target, it is reasonable to
assume that Toyota could achieve the research target using the technology
which they currently employ on their 1981 model year Celica Supra,
Cressida, and Cressida station wagon. Those models have EPA fuel economy
*Ibid, page 60.
**EPA/MFR test data base as of February 10, 1981
V-71
-------
values of:*
Car Line Transmission _
Celica Supra M5 21 29 24
Celica Supra A4 22 29 25
Cressida A4 22 29 25
Cressida Wagon A4 22 29 25
If we compare these models' MPG with those of other 1981 model year
cars, we see the fleet average MPGc for all 3,000 Ib IW cars is 24.85
mpg.** Thus, we conclude that Toyota is not suffering a fuel economy
penalty on these models compared to the competition as a result of the low
emission levels.
*EPA/DOE 1981 Gas Mileage Guide
** Foster, Murrell, Loos, "Light Duty Automotive Fuel Economy ... Trends
Through 1981,", SAE Paper 810386, February, 1981, page 16.
V-72
-------
Section V. E. Nissan (Datsun)
1. Nissan Statements Concerning the Relationship between Emissions and Fuel
Economy
This section summarizes Nissan's statements regarding the relationship
between fuel economy arid emissions.
In CO waiver hearing testimony*, Nissan said that they expected the fuel
economy of their 1981 model year vehicles to improve about 15% to 25% over
that of their 1979 model year Federal vehicles. Nissan also said that an
air pump may be necessary to meet the 3.4 CO standard, but might not be
necessary to meet a 7.0 CO standard. They felt that the increased parasitic
horsepower due to the use of an air pump, could cause a one to two percent
decrease in fuel economy.
In a status report,** dated, March 1981, Nissan said the following with
respect to the relationship between emission standards and fuel economy.
We have been proceeding with our research work on an oxidation
catalyst system and on a 3-way catalyst system in an effort to
develop a system that will meet the 0.4 g/mile NOx standard.
The results that we have obtained with these two systems to date
are shown in [table Nissan-1] and [table Nissan-2] and in
[figure Nissan-1]. The specifications of the systems that
produced these results are shown in [table Nissan-3] through
[table Nissan-6].
In order for the oxidation catalyst system to clear the 0.4
g/mile NOx standard, heavy EGR is required, but this has an
adverse effect on combustion stability. Using a richer air-fuel
ratio, to improve combustion stability lowers fuel economy by
more than 10%, in comparision with the fuel economy that can be
obtained in clearing the 1.0 g/mile NOx standard.
* Transcript of Proceedings, Environmental Protection Agency, "In
the matter of; Public Hearing on Applications for Waiver of the
1981 and 1982 CO Emission Standard for Light Duty Vehicles,"
Washington, D.C., September 12, 1979, pages 91 and 92.
** "Research and Development Program Status Report for 1981 and
Subsequent Emission Standards"; Nissan Motor Co., Ltd; March,
1981, pages 260 and 261.
V-73
-------
Figure Nissan - 1
RELATIONSHIP BETWEEN NQx AND FUEL ECONOMY IN
OXIDATION CAT. SYSTEM AND THREE WAY CAT. SYSTEM
ENGINE OUT NO*
CATAL/ST OUTNOx
OXI. CATALYST
. SYSTEM
•
o
3WAY CATALYST
SYSTEM
A ' •
A
30r
o 28
Q.
>- 26
o
o
LJ 24
UJ
ID
U.
22
I
IW=2750Lbs
3Way Catalyst System
Oxidation Catalyst System
05 UO 15
"75 FTP NOx (g/mile)
2JO
V-76
-------
Table Nissan - 3
Vehicle Specifications
1. Vehicle ID. No
2. Vehicle model
3. Model year
4. Inertia weight
5. Engine
6. Transmission
7. Axle ratio
8. N/V ratio
9.-Fuel metering system
8D-968
Datsun 510
Experimental Vehicle
•2,750 Ibs
L4, 191CID
M4
3.545
53.3
Carburetter
10. Exh. emission control system 2 plug+EGR+EAI_OX.CAT.
11. Catalyst
(1) Type
(2) Substrate construction
(3) Size Cinches) •
(4) Location
12. EGR
CD Type
(2) EGR Rate
Oxidation Catalyst
.Monolith
Oval Width 6.68 in Height 3.18 in
leneth 5.65 in
JJnder floor
WT
30% C2600 rpm x" -300mmHg-)•
V-77
-------
Table Nissan - 4
Vehicle Specifications
1. Vehicle ID. No.
2. Vehicle model
3. Model year
4. Inertia weight
5. Engine
6. Transmission
7. Axle ratio
8. N/V ratio
9. Fuel metering system
8D-972
Datsun 510
Experimental Vehicle
-2,750 Ibs
L4, 191 CID
M4
3.545
53.3
Carburetter
10. Exh. emission control system 2 plug+EGR+EAI+OX.CAT,
11. Catalyst
CD Type
Substrate construction
(3) Size (inches)
(4) Location
12. EGR
(1J Type
(2) EGR Rate
Oxdation Catalyst
.Monolith
Oval Width 6.68 in Height 3.18 in
Length 5.65 in
Under floor
WT
20% (2600 rpm x -300 mmHg)
V-78
-------
Table Nissan - 5
Vehicle Soecifications
1. Vehicle ID. No.
2. Vehicle model
3. Model year
4. Inertia weight
S. Engine
6. Transmission
7. Axle ratio
8. N/V ratio
9; Fuel metering system
YD-020
Datsun 510
Experimental Vehicle
2,750 Ibs
L4, 191 CID
M4
3.545
53.3
ECC
10. Exh. emission control system ECC+EGR+TWC+CL
11. Catalyst
(1) Type
(2) Substrate construction
(3) Size Cinches)
(4) Location
12. EGR
(1) Type
(2) EGR Rate'
3 Way Catalyst
.Monolith
Oval Width 6.68 in Height 3.18 in
Length 5.65 in
Under floor
VVT
20% (2600rpm x -300mmHg)
V-79
-------
Table Nissan - 6
Vehicle Specifications
1. Vehicle ID. No..
2. Vehicle model
3. Model year
4. Inertia weight
5. Engine
6. Transmission
7. Axle ratio
8. N/V ratio
9. Fuel metering system
YD-001
Datsun 510
Experimental Vehicle
2,750 Ibs
L4, 191 CID
M4
3.545
53.3
Carburetter
10. Exh. emission control system 2 plug+EGR+EAI+OX.CAT.
11. Catalyst
(1) Type
(2) Substrate construction
(3) Size (inches)
'4) Location
12. EGR
CD Type
(2) EGR Rate
Oxdation Catalyst
Monolith
Oval Width 6.68 in
Length 5.65 in
Under floor
Height 3.1'
I
VVT
25% (2600rpm x -300mmHg)
V-80
-------
On the other hand, the 3-way catalyst system divides the work of
reducing NOx emissions between EGR and the catalyst. In this
system the NOx level that should be cleared by EGR is 1-1.5
g/mile, and thus the necessary EGR rate is not as heavy as in
the oxidation catalyst system which uses only EGR to reduce NOx
emissions. Consequently, the EGR rate in the 3-way catalyst
system has no adverse effect on fuel economy.
As [figure Nissan-1] shows, however, the fuel economy of the
3-way catalyst system in clearing the 1.5 g/mile NOx standard is
4% worse than that afforded by the oxidation catalyst system.
The reason is that since the air-fuel ratio in the 3^way
catalyst system is controlled to the stoichiometric air-fuel
ratio, a reduction in fuel consumption cannot be obtained by
making the mixture leaner.
Irrespective of the durability, driveability and cost of each
system, the fuel economy afforded by the 3-way catalyst system
in clearing the HC 0.41, CO 3.4 and NOx 0.4 g/mile standards is
roughly equal to that of the oxidation catalyst in meeting the
standards of HC 0.41, CO 3.4 and NOx 1.0 g/mile. Even if the
compression ratio is increased or if improvements are made in
combustion performance, with the oxidation catalyst system there
is no way to avoid a decrease in fuel economy of at least 5%.
2. New Technology or Developments which Could Provide a Change in the
Relationship between Emissions and Fuel Economy as Compared to 1981
Vehicles or Nissan's Estimates
In their status report* on emission control systems, Nissan discussed four
areas in which they were doing research and development work in order to
improve fuel economy. These included:
1. Accessory drive program;
2. Engine efficiency improvement program;
3. Tire test program; and.
4. Lubricant test program.
The following are excerpts from Nissan's discussion of the accessory drive and
engine efficiency programs. Nissan did not discuss the impact on emissions as
a result of their tire and lubricant test programs.
II.F.3 Accessory Drive Programs
* Ibid., pages 251 to 253.
V-81
-------
The air-conditioner compressor, the power steering pump, the
alternator and the water pump are included in the category of
engine assisted equipment. For the 280ZX model, we studied what
effect the air-conditioner compressor and the power steering
pump have on catalyst inlet emissions and fuel economy.
In contrast to engines without compressors, engines equipped
with them showed an overall worsening of 10 to 12% for HC and CO
emissions, 30% for NOx and 11% for fuel economy.
II.F.4 Engine Efficiency Improvement Program
Improving engine efficiency is important not only in achieving
better fuel economy but also in reducing emissions. As a basic
step toward attaining these goals, the research to upgrade
engine efficiency is a vital part of our overall research and
development activities.
In our efforts to improve engine efficiency, we have carried out
basic research on raising the engine compression ratio and on
lowering the pumping loss. We have also been proceeding with
applied research of the engine efficiency improvement program to
other projects such as "0.4 gpm .NOx Research Program." The
following is our status of this research work.
In addition we are presently making every effort to develop a
new engine which will achieve a reduction in mechanical friction
loss.
(1) High Compression Ratio Engine
As we explained in our 0.4 g/mile NOx Research Program report
(September 1979 to January 1981), we found that improved
combustion by the high compression ratio made it possible to
expand the region of stable combustion under heavy EGR
conditions and to lessen partial-load fuel consumption at the
equivalent NOx level. At the low engine speed WOT condition,
however, it is necessary to retard the ignition timing to
prevent engine knocking, and this results in lowering the engine
torque by approximately 10%.
To solve this problem we have been making a concerted effort to
develop the technology needed for raising the engine mechanical
octane number. These findings indicate that when engine torque
and fuel economy are taken into account, a combustion pattern
having a short heat generation period and slow initial
combustion is effective in suppressing knocking.
The fuel economy level of an experimental vehicle -mounted with
10:1 compression ratio engine is about 5% better than that with
8.5:1 compression ratio engine. However, it is estimated that
this improvement in fuel economy will be lowered somewhat, when
the engine is actually used on production vehicles and
countermeasures are implemented to reduce the increased HC
V-82
-------
emissions and to compensate for the deterioration in
driveability due to the decreased torque in low engine speed
ranges.
Nissan has shown a significant improvement in fuel economy, 27% over the last
three years, while achieving the required emissions performance, for their 119
CID engine. Most of this improvement is attributable to the introduction of
new cylinder heads which speed-up the burning rate within the combustion
chamber*. The engine which utilizes the new cylinder head is referred to as
the NAPS-Z engine. The new cylinder head incorporates the following
features:
1. Compact hemispherical combustion chamber;
2. Dual spark plugs; and
3. Redesigned intake porting to induce swirl and turbulence.
The net result of these changes is to improve the burning rate of the fuel by
about 50% if there is no exhaust gas recirculation, and about 65% with the
intake charge diluted 20% with recirculated exhaust gas, as shown in figure
Nissan-2. Major improvements in combustion stability, with heavy EGR, were
made, permitting dramatic improvements in both economy and emissions*, (see
figure Nissan-3).
An equal contribution, to improved economy and emissions, was obtained by
optimizing the control of ignition timing, fuel-air ratio and EGR.
Basically, this was accomplished by mapping the steady-state response of
the engine, in terms of economy and emissions, as functions of fuel-air ratio,
ignition timing, and EGR rate, at selected speed-load points representing the
EPA Urban Cycle.
Detailed refinements to the engine, to improve mixture distribution, and
to enhance performance during cold-start and warm-up, account for the rest of
the progress made by Nissan.
The Fast Burn with Heavy EGR, New Approach for Low NOx and Improved Fuel
Economy; SAE paper no. 780006. Figures 11, and 15, and pages 8 and 10
respectively
V-83
-------
Figure Nissan - 2
100
z
o
g
<
-------
In the case of the 2.8 liter (168 CID) engine used in the Datsun 280 series
cars, the same progress with respect to emission control has taken p.'.ace,
however, the improvement in fuel economy is small.
Nissan employs a microprocessor to control, electronically, the following
parameters:
1. Fuel/air ratio;
2. Spark timing;
3. Exhaust gas recirculation;
4. Air intake at idle.
In addition to emissions standards, heavy emphasis was placed on drive-
ability.* The system is described schematically in figure Nissan-4. It will
be noted that this engine uses a "three-way" catalyst for exhaust after
treatment, rather than the simple oxidation catalyst used on the smaller
engines.
In the case of the' smaller Nissan engines of 1.2, 1.4, and 1.5 liter capacity
(75.5, 85.0, and 90.8 CID), the picture is qualitatively similar to that of
the 2.8 liter engine, i.e., the emissions standards have been met with no
noteworthy effect on fuel economy. Nissan has recently announced that this
family of smaller engines will be modified to incorporate the fast-burn
features, found on the 119 CID engine, for the Japanese market in 1982.
Nissan ECCS (Microprocessor Control System) Boosts Fuel Economy " First
International Automotive Fuel Economy Research Conference: Oct. 31 to Nov
2, 1979, pages 212 to 219.
V-85
-------
Figure Nissan - 4
EGR CONTROL VALVE
THROTTLE CHAMBER
AIR PLOW AIR CLEANER
METER
POWER
TRANSISTOR-' ( I
IGNITION COIL!
.THROTTLs VALVc
(SWITCH
CRANKSHAFT
SENSOR
u III! -T\
WI-
THER MOSTA
HOUSING
VSXHAUST
TEMP SENSOR
WATER TEMP SENSOR
ICRANKSHAP
rn I—T EXHAUST TEMP
^H*) 0 MONITOR LJGHT
V NEUTRAL SWITCH
VEHICLE SPEED SENSOR
Schematic of Electronic Concentrated Engine Control System
V-86
-------
V. F. Honda
1. Honda's Statements Concerning the Relationship between Emissions and
Fuel Economy
Honda stated that a basic characteristic of their CVCC engine is low spe-
cific fuel consumption (i.e., high fuel economy) and low NOx emissions when
bench tested.* Honda also reported that the use of EGR on their CVCC engine
could:**
1. Decrease NOx emissions but increase HC emissions.
2. Increase fuel economy, and
3. Improve driveability.
Honda controls the increase in HC emissions by using a catalyst.** Thus,
with the use of EGR and a catalyst, Honda says that they are able to control
emissions without adversely affecting their fuel economy.
2. New Technology or Developments which Could Provide a Change in the
Relationship between Emissions and Fuel Economy as Compared to 1981
Vehicles or Honda's Estimates.
A. New Engines
1. While Honda has not supplied EPA with any. development data on a
three-cylinder engine, "Automotive News" reports that Honda is
planning to introduce in Japan, during autunn of 1981, a small car
powered by such an engine. That report speculates that the
*"Honda's NOx Research Program - 1979 Annual Research Period Plan," March
25, 1979, page 10.
**"Honda's Effort and Progress in Meeting the 1981 and Subsequent Model Year
Emission Standards," March 20, 1981, page 7.
V-87
-------
new car will be based upon the Civic but will be smaller. * Honda
has neither confirmed nor denied these reports.
2. Diesel Engine
Honda is continuing to conduct basic research and to analyze avail-
able information on Diesel engines.** Honda has been bench testing
a 4 cylinder, 4 cycle, 2.2 liter Diesel engine.*** They have not,
however, provided EPA with any of their conclusions or test results.
B. Three-Way Catalysts
Honda's primary candidate for a system to meet the 1983 California emis-
sions standards is their stratified charge engine (CVCC) equipped with
EGR, a closed-loop carburetor, and a three-way catalyst (system D).
Honda has experimented with two other systems, one is similar to their
1981 models (i.e., CVCC with EGR, an- open-loop carburetor, and oxidation
catalyst); the second system uses a conventional engine with EGR, a
closed-loop carburetor, and three-way catalyst. However, Honda found
that their primary system was superior to the other two in driveability,
fuel economy, and emissions. This system has been tested in a 2125
pound car equipped with a manual five-speed transmission and a 1.5 liter
engine. Honda reports that this vehicle met the research target levels
(0.41 HC, 3.4 CO, 0.41 NOx); however, no emission or fuel economy data
was provided.
*"Automotive News," March 23, 1981, page 30.
**"Honda's Effort and Progress in Meeting the 1981 and Subsequent Model Year
Emission Standards," March 20, 1981, page 19.
***"Honda's Emission Control Status Report," January 13, 1978, page 0.4-15.
V-88
-------
"However," Honda states, "in view of the urgent need for complying with
the 1983 California standard, development and refinement work for a
modified system (system Dl) has been initiated based upon 1981 model
production hardware."* Honda has not provided EPA with a description of
this modified system; although, the typical emission results that Honda
provided on these two systems** leads the EPA technical staff to believe
that their primary system is superior to the modified (Dl) system for
emissions. (See table Honda-1.)
Honda states that they are continuing to refine their primary system.*
While Honda claims to have confirmed the durability of their new three-
way catalyst up to 50,000 miles, they are still experiencing problems
with cars equipped with the automatic transmissions. Honda reports,
"Automatic transmission models equipped with the system D currently fail
the applicable . CO standard because their high gear ratio causes the
carburetor secondary passages to frequently operate. A change of the
gear ratio results in a reduction of fuel economy."***
Honda's current work on three-way catalysts centers on the following
four projects:****
1. Development of the three-way catalyst for both systems D and Dl,
2. Evaluation of Pt/Rh ratio,
3. Study of substituting Pd for Pt, and
4. Study of improving catalyst durability at elevated temperature.
* "Honda's Effort and Progress in Meeting the 1981 Standards," March 20,
1981, pages 37-38.
** Ibid., Table 14.
*** Ibid., pages 44-46.
****Ibid., page 55.
V-89
-------
System D
MODEL
ENGINE (1)
TRANS.
TEST WT.
(Ibs)
N/V
LA-4
Emiss.
(g/mi)
CO
HC
NOx
CIVIC
1.5
5MT
2125
41.0
1.84
0.254
0.312
System D,
CIVIC CIVIC CIVIC CIVIC ACCORD ACCORD
1.3 1.3 1.5 1.5 1.8 1.8
5MT 3AT 5MT 3AT 5MT 3AT
2000 2125 2125 2125 2625 2625
50.7 60.3 43.3 58.7 48.7 56.0
2.72 2.54 2.51 3.04 3.81 4.28
0.29 0.32 0.32 0.32 0.28 0.24
0.29 0.31 0.28 0.28 0.21 0.23
Table Honda - 1*
TYPICAL EMISSION TEST RESULTS OBTAINED BY
USING AGED CATALYSTS
* Ibid., Table 14.
V-90
-------
3. The Effects of Known Technology which Helps Quantify Single or Multiple
Calibrations or Vehicle Description Changes
A. Engine Improvements
Honda is continuing to study variations of their stratified charge
combustion chamber. Honda reports that they have evaluated several
such systems. One type, called a "branched conduit" system, is
similar to Honda's CVCC system except that the auxiliary combustion
chamber is connected to the main combustion chamber by a relatively
long torch passage oriented toward the approximate center of the
chamber. Honda states that this system "is able to further reduce
exhaust emissions, HC and NOx, in particular, with no adverse
effect on fuel consumption under the wide range of engine operating
conditions."* Honda also reported test results from other auxilia-
ry stratified charge combustion chamber engines identified simply
as Type 1, Type 2, and Type 3. Honda did not describe these
systems in detail; however, their test results indicate that these
systems can yield emissions and fuel economy results superior
(during some operating conditions) to Honda's standard CVCC
engine.** Some of these features have already been incorporated
into Honda's 1981 models.
B. EGR
Incorporation of EGR on the CVCC system, according to Honda,
enables them to increase the compression ratio and advance the
spark timing, thus resulting in improved fuel economy and drive-
ability.*** For the 1981 model year, Honda modified the
* S.Yagi, et al., "A New Combustion System in the Three-Valve Statified
Charged Engine, "SAE Paper 790439, February, 1979, page 1.
** "Honda's Emission Control Status Report," January, 1978, pages 80-26 to
80-28.
*** "Honda's Effort and Progress in Meeting the 1981 and Subsequent Model
Year Emission Standards," March 20, 1981, pages 29-32.
V-91
-------
back-pressure EGR, that they used on their 1980 California models,
to allow for more recirculation of exhaust gas; this increased
capacity of the EGR valve was combined with an improved stratified
charge combustion chamber design. Honda states that, "This EGR
system [combined with the aforementioned improved CVCC engine] con-
tributed to improved fuel economy and driveability by:
1. The combination of a high EGR rate and near stoichiometric
engine operation at high loads, in order to improve drive-
ability while reducing NOx emissions, and
2. The combination of CVCC's lean combustion characteristics and
a low EGR rate at low loads, in order to improve fuel economy
and driveability."*
C. Oxidation Catalyst
Honda reports that, "The increase in fuel economy and reduction in
NOx emissions, achieved through the use of the improved combustion
chamber design, are accompanied by an increase in HC emissions, the
control of which is done by an improvement in oxidation catalyst.
Specifically, HC conversion efficiency was improved by 2% through
an increase in the specific surface area of substrates, that is, by
the use of 400 cell substrates as compared with 300 cell for 1980
system."**
D. Engine Friction Reduction
For 1982, Honda reports, "An additional fuel economy gain is sought
by reducing mechanical friction due to piston rings or oil seals,
and also by controlling idle speed by engine loads."***
* Ibid. , page 7.
** Ibid., pages 7-8.
*** Ibid., pages 30-32.
V-92
-------
E. Carburetor
Honda hopes to minimize the fuel penalty during decelerations by
replacing the throttle opener with a deceleration control device.*
F. Reduction of Aerodynamic Drag and Vehicle Weight
"For some [1982] models, "Honda states, "minor modifications of
body are made in order to reduce aerodynamic drag as well as
vehicle weight, such as adopting high strength steel and plastics."*
G. General Comments
The EPA technical staff attempted to evaluate the effects on
emissions and fuel economy of adding EGR and an oxidation catalyst
to Honda's CVCC engine. The most dramatic change was with the
non-California version of the 81 C.I.D. (1.3 liter) Civic. The
results of all tests performed at EPA's laboratory and Honda's
laboratory, are shown on the following chart:**
* Ibid., pages 30-32.
** EPA/MFR Test Data Base as of February 10, 1981.
V-93
-------
Manual 5-Speed Manual 4-Speed
Percent Percent
1980 1981 Change 1980 1981 Change
HC
CO
NOx
MPG
u
MPG,
n
MPG
c
0.256
3.84
1.20
27.75
3B.40
31.71
0.182
1.18
0.75
33.56
44.65
37.78
-28.9%
-69.3%
-37.5%
+20.9%
+16.3%
+19.1%
0.220
3.08
1.21
27.30
35.65
30.52
0.189
1.35
0.80
33.33
41.93
36.72
-14.1%
-56.2%
-33.9%
+22.1%
+17.6%
+20.3%
Thus, Honda was able to substantially reduce emissions while at the same
time substantially increasing fuel economy. Honda was also able to
increase the rated horsepower (from 55 to 60 HP) by raising the com-
pression ratio from 7.9 to 8.8.
The changes in fuel economy from 1980 to 1981 with Honda's other engines
were less dramatic and less consistent. Honda's Accord equipped with a
1.8 liter engine and manual 5-speed transmission actually lost about 3%
in MPG from 1980 to 1981.*
c
i
Using approximations for Honda's sales-weighted fuel economy (CAFE) (the
official values are not yet available), we found that Honda was able to
increase its CAFE from 1980 to 1981 by 3% (from 30.0 to 30.9 mpg), even
though the average weight increased by 2%.** However, compared to
similar 1981 model year vehicles of all other manufacturers, Honda's
vehicles had an average poorer fuel economy (by 3 to 4%) in each inertia
weight class, as shown on the following table:
* Ibid.
** Foster, Murrell, Loos, "Light Duty Automotive Fuel Economy... Trends
through 1981, "SAE Paper number 810386, February 1981, pages 14-16.
V-94
-------
Table Honda-2*
Comparison of Honda's 1981 Fuel Economy
Versus the Entire 1981 Fleet By Weight
Inertia Honda's Avg. of All Percent Diff.
Weight Avg. MPG Car's MPG (Relative to Fleet)
~ ~ " C t ~
2000 35.81 36.96 -3.1%
2250 33.13 34.47 -3.9%
2500 28.11 29.37 -4.3%
Ibid.
V-95
-------
V. G. Volkswagen
I. Volkswagen's Statements Concerning the Relationship between Emissions
and Fuel Economy
A. VW Study
Volkswagenwerk AG (VW) investigated the impact of emission
standards on fuel economy. This program was conducted under a
contract between VW and the U.S. Department of Transportation -
Transportation Systems Center. The following results of that study
were reported by VW in SAE Paper number 790230, titled "Impact of
Emission Standards on Fuel Economy and Consumer Attributes."
In addition to testing uncontrolled vehicles, VW also tested
vehicles targeted to meet the following four standards:
Emissions (g/mi)
Standard HC_ C0_ NOx
Federal 1976 1.5 15. 3.1
California 1976 0.9 9.0 2.0
Federal 1981 0.41 3.4 1.0
Research Goal 0.41 3.4 0.4
Each of the. five (5) groups included a 2250 pound inertia weight
Rabbit equipped with a 1.3 liter engine 'never sold in the U.S.)> a
2250 and a 2500 pound Rabbit each equipped with a 1.6 liter engine,
and a 3000 pound Audi. The Audis used in the uncontrolled case and
with the 1976 standards were equipped with 1.6 liter, 4-cylinder
engines; the ones used with the 1981 standard and the research goal
were equipped with 2.2 liter, 5-cylinder engines.
The technologies that the VW engineers chose to use were:
V-96
-------
Standard Technology
Uncontrolled Carburetor
1976 Federal Carburetor, Oxidation Catalyst
1976 California Carburetor, Oxidation Catalyst
1981 Federal Fuel Injection, 3-Way, Closed-loop
Research Goal . Fuel Injection, 3-Way, Closed-loop
VW then tested the uncontrolled vehicles as well as those equipped
with oxidation catalysts and carburetors. VW found that no
additional modifications were required of the catalyst equipped
cars in order to meet the 1976 standards.
Before testing the fuel injected vehicles, VW set engineering goals
for the emissions that were more stringent than the standards.
Those more stringent goals were "formulated to ensure that once
they had been attained at the end of the development period, all
the mass production vehicles made along the same lines will meet
the exhaust emission standards, no matter when or where they are
tested within the initial 50,000 miles of their service life."* VW
chose as goals for HC and CO values that were one-half the
respective standards; and for NOx, one-fourth of the applicable NOx
standard.
The fuel injected vehicles then were modified and tested until the
emissions, driveability, and fuel economy were considered
satisfactory. As a criterion for satisfactory fuel economy, VW
used the fuel economy of the uncontrolled version as the goal. In
addition to changes in ignition timing, the modifications included
in some cases adding a proportional EGR system and a secondary air
pump.
H. Getting, "Impact of Emission Standards on Fuel Economy and Consumer
Attributes," SAE Paper No. 790230, February, 1979, page 2.
V-97
-------
The results of the testing indicates that the Audi vehicles
suffered a substantial loss in fuel economy in meeting either the
1981 standards or the research goal. The loss was more than 20%
relative to those meeting the 1976 standards. VW believes that the
loss in fuel economy is primarily "due to the fact that friction in
these engines is higher because of the fifth cylinder."*
The three types of Rabbits (1.6 liter at 2250 and 2500 pounds and
1.3 liter at 2250 pounds) exhibited two consistent fuel econony
trends:
1. All the Rabbits which were targeted to meet the research
standard suffered a loss in fuel economy of between 4.2% to
10.9% relative to the corresponding Rabbits which were
targeted to meet the 1981 Federal standards.
2. All the Rabbits which were targeted to meet the 1981 Federal
standards displayed a fuel economy improvement of between 2.1%
to 7.5% relative to the corresponding Rabbits which were
targeted to meet either of the 1976 emission standards.
VW speculated that the reason the vehicles targeted to meet the
1981 standards exceeded the fuel economy performance of similar
vehicles which met far less stringent standards was due to "the
introduction of new technological features, such as lambda control
and K-Jetronic fuel injection."*
VW then studied these vehicles with respect to unregulated
emissions, noise, startability, and performance. VW concluded that:
1. "[T]he emission control concepts featuring K-Jetronic
fuel injection and 3-way catalysts are capable of meeting
the advanced standards, furnishing good fuel economy, low
HCN emissions, and a stable SO, emission rate."**
* Ibid., pages 5-6.
** Ibid, page 7.
V-98
-------
2. "[T]he noise emissions of modifications 13 through 16 [i.e.,
.those vehicles which were designed to meet the 1981 Federal
emission standards] are lower than those of other vehicles
meeting less stringent exhaust emission standards."*
3. The "sophisticated concepts introduced to meet advanced
emission standards are superior to conventional concepts
regarding their startability and driveability."*
4. "[LJeaving aside the 5-cylinder engine, the performance of the
vehicles which meet the advanced standards is slightly
inferior to that of the others. This is because gradeability
is related to maximum torque in a practically linear function,
and by some sort of accident it was the carburetor-type
vehicles which always had the highest torque. There is no way
of explaining this by the emission control concept."**
Finally, VW analyzed the effects on the cost of the engine of those
vehicles which met the more stringent emission standards. They
determined, that in order to meet either the 1981 Federal standards
or the research goal (the components are the same for both
standards), the cost of the engine (not the entire vehicle) would
be increased by between 64% and 89% over the corresponding
uncontrolled vehicles, and by between 38% and 56% over the
corresponding vehicles meeting the 1976 Federal standards.***
B. EPA Comments on the VW Study
VW assumed that a closed-loop, three-way catalyst, fuel injected
system was necessary to meet the 1981 Federal standards; however, a
number of open-loop, carbureted vehicles have been certified as
meeting those standards.
* Ibid., page 8.
** Ibid., page 9.
*** Ibid., page 11.
V-99
-------
The engineering goals (of 1/2 the applicable HC and CO standards
and .of 1/4 the applicable NOx standard) were conservative. The
deterioration factors associated with the 1981 model year,
closed-loop, 3-way catalyst, fuel injected, gasoline powered
Federal vehicles equipped with either 1.7 liter, 4-cylinder VW
engines or 2.2 liter, 5-cylinder Audi engines are all 1.000.
Thus, VW's selection of the required technology and engineering
(emission) goals was possibly more conservative than necessary.
However, VW did show, that, for an increase in cost, the fuel
economy, noise, and driveability could all be improved while at the
same time substantially reducing the emissions. :
2. New Technology or Developments which Could Provide a Change in the
Relationship between Emissions and Fuel Economy as Compared to 1981
Vehicles or VW's Estimates.
A. Formula E
VW will be introducing a package of several features that will be
called "Formula E" (for Economy). This package will consist of one
or more (depending on the model) of:
1. Signal to Shift
VW is developing an electronic device that will signal the driver
when to shift for optimal fuel economy at various speeds and
loads. This option has been available on VWs in Europe since
December, 1980. It is anticipated that this option will soon be
available on some U.S. Rabbits.*
EPA is currently testing such vehicles, but not enough data has
been generated to assess the effect of this option. However, VW
* "Wards Engine Update," March 15, 1981, page 6,
V-100
-------
reported to EPA that, "Experimental results show that
observing the electronically given shifting signals can lead
to fuel savings of 6 to 10% in the urban cycle, depending on
engine and gearbox adjustment. This is due simply to correct
'engine management' and does not entail any diminuation of
performance. Practical driving results confirm these
findings."*
2. Automatic Engine Shut-Off
VW is testing a system that will switch off the engine during
idling and deceleration. VW claims this system will increase
urban fuel economy. VW states, "This [system] would mean a
25% increase in spark ignition engine efficiency and an 18%
increase in Diesel engine efficiency."**
The engine is restarted from kinetic energy which was stored
in a spinning flywheel that turns the crank. A second clutch
disconnects the flywheel when the car stops or coasts.
Simultaneously, the computer shuts off the ignition and fuel
flow. When the driver steps on the gas, the second clutch
recouples the flywheel to the crankshaft, fuel is injected,
and the spark plugs fire.
VW engineer Paulus Heidemeyer was quoted as saying, "Shutting
off the engine when its power isn't needed can cut fuel
consumption by 20-25% in city and suburban driving, when
frequent stops and starts are made."* A test vehicle was
reported to have yielded higher fuel economy on the FTP cycle
and produced lower emissions than a similar vehicle with a
conventional power train.*
* "Status Report of Volkswagenwerk AG and Audi NSU," May 1981, page 191.
** Ibid., page 186.
*** Norbye, "VW's Stop/Start 21st-century Car," Popular Science, July 1980,
pages 76-77.
V-101
-------
A drawback of this system is that there is no provision for
.running accessories while the engine is stopped. Thus, it
might not be suitable for cars equipped with air conditioning,
power steering, or power brakes.*
3. Overdrive Top Gear
VW has redesigned and renamed their gearboxes. The 4-speed
manuals will be called "3 4- E" (i.e., 3-speed plus an economy
gear), and the 5-speed manuals will be called "4 + E." The
gear ratios within the transmission will be changed so that
the maximum speed is reached in the next to last gear, with
the top gear becoming an economy overdrive gear. VW claims,
"In normal operation, the savings amount to 10 to 13% on 3 + E
gearboxes and 5 to 6% on 4 + E gearboxes."** VW made no
predictions for fuel savings on either EPA's urban or highway
test cycles.
4. Aerodynamic Aids and Weight Reduction
While reducing neither aerodynamic drag nor weight is new,
these concepts will be incorporated in the Formula E
package.***
B. Combustion Chamber Redesign and Knock Sensing
VW is developing a combustion chamber which provides a higher
compression ratio, an increase of 2 to 3 units.**** VW is working
on a knock limit sensing system in conjunction with the development
of high compression concepts. By using a computer to closely match
the spark advance to the knock limit, VW hopes to improve
efficiency.*****
*Norbye, "VW's Stop/Start 21st-century Car," Popular Science, July 1980,
pages 76-77.
**"Status Report of Volkswagenwerk AG and Audi NSU," May 1981, page 189.
***Ibid., page 186.
****Ibid., page 168.
*****Ibid., page 161. V-102
-------
C. Three-Cylinder Diesel
VW has built and is testing a 1.2 liter, direct injection, Diesel
engine with supercharger. VW believes that direct injection
could possibly provide a "10-15% increase in fuel economy over
what is already a miser, the swirl chamber Diesel."* Even though
VW has a 1.3 liter, 4-cylinder engine already available in
Europe, they decided to make this engine a 3-cylinder instead of
4, in order to reduce frictional losses.*
VW is investigating the advantages of a turbocharger versus a
mechanically driven supercharger versus natural aspiration.*
*Ibid., page 31.
V-103
-------
V. H. Toyo Kogyo (Mazda)
1. Toyo Kogyo's Statements Concerning the Relationship between Emissions
and Fuel Economy
Toyo Kogyo (TK) did not state that there was a well defined relationship
between emissions and fuel economy. However, TK did state that meeting the
1983 California standards (0.41 HC, 7.0 CO, 0.4 NOx) will result in a fuel
economy penalty even though the emission control system they are developing
to meet those standards has a "potential to reduce the fuel consumption and
exhaust emission."* TK did not quantify the amount of the expected loss in
fuel economy, nor did they provide EPA with data to substantiate their claim
of a loss in fuel economy.
2. New Technology or Developments which Could Provide a Change in the
Relationship between Emissions and Fuel Economy as Compared to 1981
Vehicles or TK's Estimates.
A. New Engines
1. Rotary Engine
TK is the only manufacturer of automotive rotary (Wankel) engines
sold in the U.S. TK began using a 3-way catalyst with an open-loop
carburetor and an air pump on their rotary models for the 1981
model year. They intend to use this same system for 1982. For
1983, TK plans to use a closed-loop carburetor with a 3-way
catalyst, an air pump, and EGR.** TK provided no estimates or test
data on this new system.
TK reported that they are studying a supercharged version of their
rotary.*** Industry sources speculate that the supercharged
version of the twin-rotor Wankel engine will offer more horsepower,
greater low-end torque, and improved fuel economy.*** The
* "Mazda Status Report," February 1981, pages 64-65.
** "Mazda Status Report," February 1981, page 3.
*** "Ward's Engine Update," Volume 6, Number 12, June 25, 1980, page 6.
V-104
-------
speculation is that this vehicle would not be available prior to
the 1983 model year.*
2. Diesel Engine
With the 1981 model year, TK introduced its first passenger car
Diesel engine In TK's Japanese Luce series. This A cylinder, 2.2
liter engine is available only in TK's Japanese market; however,
industry sources believe that it is similar to the engines that TK
has agreed to sell to Ford in the next year or so for the U.S.
market.** TK claims that this Diesel vehicle has a fuel economy of
51.7 mpg when tested at a steady 37.5 mph**(not an EPA test cycle).
TK reported that work on a direct-injection combustion system for
their Diesel engines is a project, second in importance only to
TK's work in turbocharging small engines.***
TK has not provided EPA with any emissions or fuel economy data on
their Diesel powered vehicles. Some emission and fuel economy
results from a TK engine in a Ford vehicle are shown in section
V.B, Ford.
3. Small (2- and 3-Cylinder) Engines
TK reports that they are investigating the use of A-, 3-, and
2-cylinder engines for use in A-passenger, 2-passenger, and 2+2
small cars. The engine displacements range to below 1 liter (61
CID).**** "Automotive News" reports that TK will begin producing
small cars, powered by a 3-cylinder engine, for sale in the U.S.
beginning with the 1985 model year. These cars would be sold under
the Mazda, Ford, and Mercury nameplates. The report predicts an
urban fuel economy for these cars of between AO to 50 mpg.**
* "Automotive News," November 17, 1980, pages 2 and AA.
**" "Ward's Engine Update," Volume 6, Number 20, October 15, 1980, page 6.
*** "Ward's Engine Update," Volume 6, Number 17, September 1, 1980, page 6.
****"Ward's Engine Update," Volume 6, Number 17, September 1, 1980, page 6.
V-105
-------
Another report indicates that this car will be front-wheel drive,
and that the engine displacement will be between 1 and 1.2 liters
(61 to 74 CID).* TK did not submit any data on this vehicle to EPA.
B. Closed-Loop Carburetor
TK plans to employ a closed-loop carburetor, an air pump, EGR, and a dual
3-way catalyst on all of their passenger cars beginning with the 1983 model
year.** This system has been tested by TK in a vehicle with 2250 pound test
weight and equipped with a 1.5 liter engine and manual transmission. TK
states that neither the fuel economy results nor the dfiveability obtained
thus far with this system (see table TK-1) met their goal.***
Table TK-1
Comparison of Fuel Economy and Emissions of the
Closed-Loop Test Vehicle to the Fuel Economy and
Emissions of a Comparable 1981 Model Year Open-Loop Vehicle.
HC CO NOx MPG MPG, MPG
u n c
Closed-Loop
Test Vehicle*** 0.22 4.0 0.31 33.4
1981 TK Data
Vehicle**** 0.224 1.03 0.49 34.5 41.6 37.4
* "Ward's Engine Update," Volume 7, Number 2, January 15, 1981, pages 1
and 4.
** "Mazda Status Report," February, 1981, page 3.
*** Ibid, pages 66-67.
****"EPA Test Car List, 1981 Second Edition."
V-106
-------
Even though this one test vehicle experienced a drop in urban fuel
economy of 3.2% (relative to its 1981 model year counterpart), the
effect of this system on overall fuel economy cannot be determined since
TK did not submit highway test data (MPG ).
TK believes that the loss in fuel economy and driveability is, "due to
the insufficient controlling performance for the air/fuel ratio and EGR,
the insufficient catalytic efficiency, the choking-up of the EGR passage
by the carbon and so forth. And thus, the refinement of the control
system of the mixture and EGR, the combustion improvement, the catalyst
screening and so forth are considered important."* TK says they plan to
improve those components during the 1982 model year based on the results
of the analysis done during 1981.**
C. Combustion Chamber Improvements
TK reports:
"We have taken an approach to mainly accelerate combustion in the
engine with the help of increased swirl and squish, and have
achieved the excellent results through the MSH (Masked Seat Head)
method."
"We intend to continue pursuing the improvement of fuel economy
still further by means of modifying the combustion chamber config-
uration in the future."***
"In addition, the following items will be considered, too:
To make the combustion chamber more compact.
To make the compression ratio higher.
Development of [a]DIS (Dual Induction System).
The selection of spark plug location."****
* "Mazda Status Report," February, 1981, page 67.
** Ibid., page 33.
*** Ibid., page 70.
****Ibid., page 69
V-107
-------
D. Electronic-Controlled EGR
Backpressure type EGR provides increased control over the type not using
backpressure to simultaneously reduce NOx and to stabilize drive-
ability. TK plans to develop an electronic-controlled EGR system which
will provide greater control accuracy of EGR than the backpressure type
provides.*
3. The Effect of Known Technology Which Helps Quantify Single or Multiple
Calibrations or Vehicle Description Changes.
TK plans to gradually replace the main moving parts of their engines (such
as rotors for rotary engines and connecting rods for conventional engines)
with lighter weight parts. TK says that this will decrease frictional
losses and thus increase engine efficiency.**
TK also plans on studying four other systems that can affect fuel economy:**
1. Using pulse air instead of an air pump,
2. Fuel injection,
3. Electronic ignition, and
4. Knock controller.
For the 1981 model year, TK increased the fuel economy of their GLC and RX-7
models over the corresponding 1980 models. (See table TK-2.)
*Ibid., page 69.
**Ibid., page 33.
V-108
-------
Table TK-2*
Fuel Economy of 1980 and 1981 Mazda Vehicles
1980
MPG
MFC.
MPG
MPG
1981
MPG.
MPG
GLC
A3-Trans
MA -Trans
M5-Trans
27
30
29
34
42
39
30 ,
34
33
30
35
35
37
43
45
32
38
39
RX-7
A3-Trans
M5-Trans
16
17
24
28
19
20
19
21
24
30
21
24
TK claims that they improved the fuel economy of their conventional engine
(used in the GLC) by:**
1. Change in emission control system
Although we had been using a single catalyst + air
injection + EGR up to the 1980 year models, we adopted on
the 1981 year models a dual catalyst + air injection +
EGR so as to improve the fuel economy and meet the more
stringent NOx standard.
2. Vehicle weight reduction
By changing the GLC to a front wheel drive model, we were
able to reduce the weight from 2,250 Ibs to 2,125 Ibs.
*
**
EPA/DOE Gas Mileage Guide.
"Mazda Status Report" February, 1981, page 32,
V-109
-------
3. Power-to weight reduction and drivetrain programs
The GLC engine was increased in piston displacement from
86.4 CID to 90.9 CID, thus having adequate power/weight
ratio. So, N/V ratio was changed from 57.0 to 54.1 in
top gear and from 47.1 to 43.1 in overtop, thus making
this engine a high speed type, thereby improving the fuel
economy.
4. Accessory drive program
The cooling fan was changed to a motor driven type, hence
reducing the propulsion loss of the fan.
TK says that the improved fuel economy of their rotary engine (used in the
RX-7) resulted from:*
1. Engine efficiency
We improved fuel economy and reduced emissions by
improving the gas sealing, modifying the engine (the
ignition system and intake system in particular),
operating the engine at a leaner mixture, making the
combustion more stable and optimizing the catalyst
system. In order to optimize all three systems, we made
studies on reduction in raw HC, improvement in thermal
efficiency, best design of the catalytic converter,
selection of a catalyst best suited to the rotary engine,
reduction in the temperature of the exhaust gas entering
the catalyst, and adjustment of the air/fuel mixture.
2. Reduction in vehicle air resistance
The shape of the front grille of the car was redesigned
so as to reduce the air resistance of the car body, and
the load to be applied to a chassis-dynamometer was
lowered from 8.0 hp/50 MPH for 1980 to 7.1 hp/50 MPH for
1981.
3. Vehicle weight reduction
We reduced the weight of the car body, engine and their
related parts, thereby lowering the test weight from
2,750 Ibs to 2,625 Ibs.
Ibid., page 34.
V-110
-------
The EPA technical staff has generated approximations of TK's sales-
weighted fuel economy value (the official values are not yet available),
and we found that TK's fleet average fuel economy increased from 26.3
mpg in 1980 to 31.1 for their 1981 models (an increase of 18%).* Com-
paring, by weight, TK's 1981 vehicles with conventional (not rotary)
engines to the average of all 1981 model year vehicles, yields the fol-
lowing table:
Table TK-3**
Comparison of Toyo Kogyo's 1981 Fuel
Economy Versus the
Inertia
Weight
2000
2250
2750
By
TK's
Avg.
38
34
30
Weight
MPGr
.59
.52
.75
Entire 1981 Fleet
(Excluding Rotary
Avg.
Car's
36.
34.
27.
of All
MPG,,
96
47
17
Engines)
Percent
(Relative
+4.
+0.
+13.
Diff.
to Fleet)
4%
1%
2%
Thus TK was able to increase the fuel economy of its conventional vehi-
cles so that they were more fuel efficient than most other similar
vehicles. TK's rotary engine vehicles average 24.27 MPG compared to
29.37 MPG for all 2500 pound inertia weight cars.** These data are
somewhat biased against the Mazda vehicles since their equivalent test
weight is 2625 pounds and they are being compared to a group of vehi-
cles which includes lighter vehicles.
* Foster, Murrell, Loos, "Light Duty Automotive Fuel Economy ... Trends
through 1981," SAE Paper number 810386, February 1981, pages 14-16.
** Ibid.
V-lll
-------
V. I. American Motors
1. American Motors' Statements Concerning the Relationship between
Emissions and Fuel Economy
Quoting their latest status report,* "American Motors is vendor dependent
with regard to emission control systems. These include catalytic
converters, air injection pumps, ignition systems, carburetors, EGR systems,
and electronic control modules." Because of this dependency, and the
necessity of prior financial commitment to the vendors, AMC's development
program has been limited in scope to evaluation of emissions control
components, evaporative emissions control, driveability, and fuel economy.
Their emissions control program is one of evaluation of prototype, vendor-
supplied, control hardware installed on suitably modified production cars.
"Once a particular emission control strategy, and associated components have
been selected, there is an inherent relationship between the achieveable
fuel economy and the emissions standards as a result of the control system
constraints. Consequently, there has been no separate reportable effort or
data in the area of fuel economy improvement."
With respect to a 0.41 HC, 3.4 CO, 0.4 NOx standard, they stated "AM has
been unable to demonstrate, even on a research vehicle that these standards
can be met with any control system planned for 1981 through 1983." Work is
planned to investigate different fuel systems, catalyst configurations, and
refinement of control strategies. Confidential treatment was requested for
the details of AMC's production vehicle development program.
American Motors Corporation Efforts and Progress to Meet 1981 and
Subsequent Model Year Light-Duty Vehicle Emission Standards; March, 1981.
V-112
-------
General Comments
A substantial body of test data was provided in the status report; however,
the vehicle descriptions and the details of the test configurations were so
sketchy that the EPA technical staff could not draw conclusions from the AMC
efforts. It is possible to consider the recent (1979-1981) history of AMC
cars with the 258-2V engine (table AMC-1), as tested by EPA. This is the
only passenger car engine common to model years 1979, 1980, and 1981. Since
1979, control of HC and CO emissions has been improved significantly (up to
66% and 60% respectively), and urban fuel economy has improved 5 to 10%.
Some improvement in NOx control was observed for the manual transmission
cars, but not for those with automatic transmissions. No major change in
weight occurred, other than that attributable to the finer test weight in-
crements used since 1980. There is not a demonstratable adverse effect on
fuel economy attributable to the improvements in emission control.
In the case of the 151-2V engine, which entered production in 1980, there
have been no changes in emissions control requirements for 1981, because of
waivers. Consequently, as shown in table AMC-2, there have been no major
changes in the emissions and fuel economy performance. .
V-113
-------
Table AMC-1
Eniasiona and Fuel Economy of AMC Cars
258-2V Engine
HC EMISSIONS
Model
Year Trans. Mean Max Mi.n
1979 A3 .565 .61 .50
M4 .468 .52 .40
1980 L3 .27 .33 .25
M4 .2165 .27 .183
<19B1 L3 .193 .280 .132
£ H4 .186 .204 .171
HC EMISSIONS
Model (g/ral)
Year Trans Mean Max Mi.n
1980 A3 .201 .232 .170
M4 .255 .260 .250
1981 A3 .097
M4 .1593 .189 .124
CO EMISSIONS
(g/n)ll.
Mean Max
8.7 9.5
4.8 4.9
4.55 6.65
2.36 4.0
3.450 4.71
2.653 5.27
Emissions
Min
7.9
4.7
3.0
1.38
0.79
2.12
Table
and Fuel
151-2V
Mean
1.125
1.645
1.30
1.347
1.11
.865
AMC-2
Economy
Engine
CO EMISSIONS
(g/nU)
Mean Max
2.335 2.65
3.80 4.7
1.13
2.637 3.86
Min
2.02
2.8
—
1.31
Mean
1.505
1.18
1.59
1.787
NOx EMISSIONS
(e/ni)
Max Min
1.29 0.96
1.73 1.56
1.38 1.16
1.38 1.29
1.24 0.92
1.21 0.81
of AMC Cars
NOx EMISSIONS
(g/ml)
Max Min
1.69 1.32
1.21 1.15
—
1.90 1.70
URBAN ECONOMY
; Tost
Mean Max Min UVl^ln
20.5 21 20 3125
22 22 22 293H
20 — -- 3250 sample
22.67 2958
-------
Appendix 1
Prediction of Federal Test Procedure
Results from Hot Emissions Data
-------
Appendix 1
Prediction of Federal Test Procedure
Data from Hot Emissions Data
The test procedures used to measure exhaust emissions from new motor
vehicles are described in Part 86 of the Code of Federal Regulations. The
Federal Test Procedure is commonly referred to as the FTP. The FTP
specifies the test fuel, driving schedule, vehicle preparation, test site
ambient conditions and other details which are intended to give the test
results a high degree of scientific accuracy.
A brief discussion of the exhaust emission measurement portion of the FTP is
required at this point. The exhaust emissions of a light-duty vehicle are
measured while the vehicle is operated on a chassis dynamometer.
The vehicle exhaust gas is collected while the vehicle is being operated on
the dynamometer. The exhaust gas stream is diluted with ambient air and a
sample of this dilute mixture is collected during each phase for analysis.
The dynamometer run is divided into three phases. A composite gas sample is
collected in a separate bag for each phase. Bag one contains the sample
collected during the "transient" phase of the cold start test which is the
first 505 seconds of the driving schedule. The second bag sample is
collected from the 505th second of the driving schedule until its end at
1372 seconds. This is the "stabilized" phase of the test. Bag three is
collected from the time the vehicle is restarted following the ten minute
hot soak until the 505th second of the driving schedule. The three bags are
sometimes referred to as the cold transient (ct) bag, the stabilized (s)
bag, and the hot transient (ht) bag.
The bag samples are analyzed and calculations are performed to determine the
grams of emissions per test phase. The final reported test results are then
computed using the grams per test phase data from the bag samples and the
following equation:*
*Code of Federal Regulations, Part 86.144-78 As of July 1, 1979.
1-1
-------
Ywm = 0.43 ((Yct + Y8)/(Dct + Ds)) + 0.57((Yht + Ys)/(Dht + Ds))
Where: Y = Weighted mass emissions of each pollutant, i.e., HC, CO, NOx
or C02, in grams per vehicle mile.
Y = Mass emissions as calculated from Bag 1,
in grams per test phase.
Y. = Mass emissions as calculated from Bag 3,
in grams per test phase.
Y = Mass emissions as calculated from Bag 2,
s
in grams per test phase.
D = The measured driving distance during Bag 1,
in miles.
D, = The measured driving distance during Bag 3,
in miles.
D = The measured driving distance during Bag 2,
in miles.
The final value of the measured emissions is obviously a function of the
design of the test procedure. Emissions values determined using another test
procedure or a modified version of the FTP, may or may not be related to the
FTP emissions.
Automotive research and development programs will often utilize an
abbreviated form of the FTP. For example, if a researcher was only
interested in a vehicle's exhaust emissions, the portion of the FTP which
measures evaporative emissions may be omitted. Another common abbreviated
form of the FTP is called a "hot" FTP. The implied meaning of a "hot" FTP is
that the 12 to 36 hour soak period before the dynamometer run was deleted.
Instead the dynamometer run is started with the vehicle in a "warmed up" or
1-2
-------
"hot" condition. Elimination of the many hours long soak period prior to the
start of the dynamometer run allows many "hot" FTP's to be run in an eight
hour working day compared to only a part of a single emission test in
accordance with the FTP.
The cold transient and stabilized phases of the dynamometer run are performed
with a warmed-up vehicle. The driving schedule will be identical for both
the "hot" FTP and Bags 1 and 2 of the official FTP. However, because of the
difference in thermal preconditioning between these two test procedures, the
final emission values may be vastly different.
FTP equivalent emissions and fuel economy results are sometimes calculated
from steady-state engine data. These calculated projections are often (or
should be) qualified by a statement like the following: .
It should be noted that these calculations do not account
for the warm-up portion of the FTP drive cycle, the engine
transients, or... *
This statement above should alert users of such data that the possibility
exists that the calculated emissions and fuel economy may be different from
actual FTP results. SAE paper number 800396, "The Exhaust Emission and Fuel
Consumption Characteristics of an Engine During Warmup - A Vehicle Study",
said the following in reference to gasoline fueled vehicles:
The relative contribution of pollutants emitted during
vehicle warmup has been magnified because the outstanding
performance of contemporary catalytic converters has nearly
eliminated the pollutants emitted from fully warm vehicles.
Trella, Thomas J., "Fuel Economy Potential of Diesel and
Spark Ignition - Powered Vehicles in the 1980s" SAE 810514.
1-3
-------
Many vehicle components influence the exhaust emissions as a function of the
vehicle s thermal state. For example, all, or nearly all, model year 1981
gasoline fueled light-duty vehicles utilize a catalytic converter in their
emission control system. The conversion efficiency of a catalyst increases
with higher temperature. Exhaust emissions are reduced with greater
conversion efficiency. Diesel engines in production now do not have any
catalytic devices to treat the exhaust.
Another design feature of gasoline fueled vehicles which effects exhaust
emissions as a function of thermal pre-conditioning is the start-up enrichment
system. This cold start enrichment and the initial inactivity of the catalyst
during Bag 1 of the FTP contribute a large percentage of the total CO and HC
measured in a FTP.
Emission test data from certification testing were analyzed to determine what,
if any, relationships exist between Bag 1 and Bag 3 emissions. Figure 1-1
presents a summary of the selection criteria for the main data base which was
analyzed. Additionally, this figure shows how the main data base was divided
into ten subgroups.
Correlation matrices were generated for the ten data sets shown in figure
1-1. These matrices are presented in table 1-3 to table 1-12. Some
correlation coefficients (r values) have been compiled from these matrices and
listed in tables 1-1 and 1-2 for easier reference in the following discussion.
These correlation coefficients could provide some guidance in developing an
emission test procedure which is based on an abbreviated form of the FTP.
Conversely, these data may provide some insight into the relationship between
a so-called "hot FTP" and an actual FTP.
The data of table 1-1 seem to substantiate the following observations:
1. Bag 3 NOx correlates to some extent (r > .85) to the FTP NOx for all
cases. The r is greater when NOx is not catalytically controlled.
1-4
-------
Figure 1-1
Data Base Description
Test Active Years = 79, 80 and 81
Sales Classes = 49 States, 50 States and California
Vehicle Type = Certification Data and Fuel Economy Data
Test Type = Certification Data Test Emissions and Fuel Economy
Test Procedure = Three Bag 1975 FTP
Certification Test Disposition = Passed Used for Certification
Fuel Economy Disposition = Used for Fuel Economy
Active
Year =
79
Active
Year =
81
Active
Year =
80
Fuel Type = Gasoline
[Fuel Type = Diesel
Fuel Injection = Yes
Injection =
^Converter Type = Oxidation
Converter Type = Three-way
Converter Type = Three-way + Oxidation
1-5
-------
e I1
Correlation Coefficients (r)
1981 Active Year
Variable
BjHC vs. FTPHC
BjCO vs. FTPCO
BjNOX vs. FTPNOx
Gasoline Diesel
.7238 .3519
.8700 .9430
.8035 .9121
Oxidation Cat. 3-Way Cat. 3-W +O.C. Fuel Injected (Non-Diesel)
.7357 .9162 .6705 .9129
.8719 .9120 .8309 .9339
,8746 .7546 .6890 .8517
Not Fuel Injected
.6937
.8518
.7766
B3HC vs. FTPHC .6744 .5228 .7662
B3CO vs. FTPCO .5965 .9043 .6115
B3NOx vs. FTPNOx .9286 .9105 .9757
.5433
.6260
.8859
.6426
.6067
.8593
.5322
.3788
.9217
.7133
.6132
.9130
BjHC vs. FTPHC
vs. FTPCO
vs. FTPNOx
Active Year 1979. All
.6527
.8015
.7644
Active Year 1980. All
.3346
.8623
.8661
Active Year 1981. All
.3512
.8767
.7800
B3HC vs. FTPHC
B3CO vs. FTPCO
vs. FTPNOx
.8578
.7789
.9011
.3764
.6614
.9353
.5439
.5957
.9333
Table 1-2
Correlation Coefficients (r)
1981 Active Year
Variable
B3HC vs.
B3CO vs. BjCO
B3NOx vs
Gasoline
.0985
.1927
.7401
Diesel
.6457
.8656
.9365
Oxidation Cat.
.1929
.2047
.8669
3-Way Cat.
.2916
.3029
.6986
3-W +. O.C.
.0152
.1305
.5192
Fuel Injected (Hon Diesel)
.2817
.1081
.7654
Not Fuel Injected
.1140
.1739
.6816
B3HC vs.
B3CO vs.
B3NOx vs.
iHC
Active Year 1979, All
.4826
.4342
.7465
Active Year 1980. All
T5S52
.2616
.8485
Active Year 1981. All
.1927
.2181
.7687
-------
2. Bag 1 CO correlates to the FTP CO better than Bag 3 CO for all
cases. The Bag 1 CO correlation coefficients exceed .80 for all
cases compared to a minimum Bag 3 CO r of .38. Bag 3 CO appears
useful for prediction of FTP emissions for Diesels only.
3. Model year 1981 Diesel vehicles have very good correlation (r > .90)
of Bag 1 and Bag 3 to the FTP for both CO and NOx emissions. Bag 1
CO and NOx correlation coefficients are slightly greater than the
corresponding Bag 3 values.
4. The correlation coefficients for Bag 1 HC vs. FTP HC and Bag 3 HC vs.
FTP HC range from .33 to greater than .91. Active year 1981 vehicles
equipped with three-way catalysts and/or non-Diesel fuel injected
engines appear unique from the other subgroups, having Bag 1 HC vs.
FTP HC r values which indicate reasonably good correlation (r ^> .91).
In summary, the correlation coefficients analyzed indicate that 'accurate
prediction of FTP equivalent emissions from test data which don't reflect
operation over the cold transient phase is doubtful. However, FTP equivalent
emissions can be predicted with reasonable confidence for some individual
pollutants and certain specific subgroups of vehicles.
Based on the lack of correlation between Bag 1 emissions and Bag 3 emissions
shown in table 1-2 it can be concluded that there is effectively no way to
predict Bag 1 emissions, even assuming that one could computer-model-calculate
Bag 3 emissions accurately. Even if one further assumes that
computer-model-calculated MPG values are accurate, any relationship between
predicted emission levels (or emission standards) and MPG based on the use of
hot data only is probably worthless.
1-17
-------
Appendix 2
Calculation Methodology for Fuel
Economy Change Allocation
-------
Appendix 2
Calculation Methodology for
Fuel Economy Change Allocation
The procedure for computing fleet fuel economy changes due to specific
factors, such as system optimization and weight mix shifts, involves the
construction of matched sets of data from a base fleet (e.g. 1978) and a new
fleet (e.g. 1979), and calculation of intermediate sales-weighted fleet fuel
economy values for the matched sets. Depending on the degree of matching,
the data sets being compared include only certain known changes between the
sets, and hence the calculated intermediate fleet MPG values reflect the
fuel economy effects of only those specific changes in fleet makeup.
CALCULATION OF DIFFERENCES DUE TO SYSTEM OPTIMIZATION: To determine the
differences in fuel economy between the 1978 and 1979 cars due to system
optimization, it was necessary to limit the comparison to nominally ident-
ical vehicles. For each manufacturer it was established which 1978 and 1979
models were identical in terms of weight, displacement, and transmission
type.. When this was established a new set of sales fractions was
calculated, based on 1978 sales estimates, using only those combinations
which were carried over from 1978 to 1979. Two sales-weighted fuel economy
values (SWMPG) were calculated using the equation below. One calculation
using 1978 model MPG values and 1978 carryover sales fractions, and one
using the 1979 model MPG values, also with 1978 carryover sales fractions.
SWMPG
Jfi(l/MPG1)
where fi = sales fraction for stratum ig whose MPG is
The difference between the two values reflects the change in fuel economy
due to what we have called system optimization. Since the weights, dis-
placements, transmissions - and their sales distributions - are matched, any
difference in fuel economy is due to other factors. The main factors which
could be contributing to such a system optimization change in fuel economy
are:
2-1
-------
o Emission control system design changes;
o Engine design and/or calibration changes;
o Changes in transmission efficiency, shift scheduling, or gear ratios;
o Axle ratio changes;
o Changes in test procedure which influence fuel economy.
DIFFERENCES DUE TO TRANSMISSION MIX SHIFTS: In the analysis of fuel economy
changes due to system optimization, any IW/CID/transmission combination not
common to both years was eliminated from consideration, and the sales
distribution of those combinations that were carried over was held at the
1978 mix. If the calculation is repeated using only weight/displacement
combinations as the determinants for model year carryover, those
IW/CID/transmission combinations that are not common to both sets of data
are not "sifted out", but remain in their respective data bases; also, each
of the data bases retains its own sales split between automatics and manuals
within the carryover IW/CID combinations.
Again,, two SWMPG values are calculated using the same equation, wherein the
first MPG is the harmonic mean sales-weighted fuel economy of each manu-
facturer's 1978 models in IW/CID class 1, and the second MPGi is the fuel
economy of his 1979 models in IW/CID class i. Both of these SWMPG values
are based on the same mix of the IW/CID classes (the 1978 mix), so the dif-
ference between the two is due to system optimization plus all changes in
transmission mix.
DIFFERENCES DUE TO ENGINE MIX SHIFTS: Similarly, by sifting for carryover
at only the weight class level, all differences in the IW/CID structures of
the fleets are allowed to remain. The difference between the two SWMPG
2-2
-------
values calculated on this basis is thus due to system optimization, trans-
mission mix shifts, and shifts in the mix of engine displacements*.
DIFFERENCES DUE TO WEIGHT MIX SHIFTS: The bottom-line SWMPG values cal-
culated from the full, unperturbed data bases, each with its own sales mix,
includes all of the above effects plus the effect of non-carryover weight
classes and the 1979 redistribution of sales among carryover weight
classes*
Table 2-1 summarizes the above calculation methodology, and figure 2-1 shows
a diagram of the relationship between the various calculated SWMPG values.
Since the methodology is suitable for a comparison between any two vehicle
sets (49-States vs. California, cars vs. trucks, manufacturer X vs. Y,
etc.), table 2-1 and figure 2-1 are notated for the general case rather than
the year-to-year case.
Table 2-2 illustrates the equations for separation of individual factors
from the combined effects discussed above.
*This also includes shifts in the mix of engine standards/systems; Federal
vs. California and Spark vs.. Diesel.
-------
Table 2-1
Method for Constructing Fuel Economy
Comparisons between Two Vehicle Groups
Configuration
Determinants
IW/CID/Trana-
aisslon Type
Vehicle Croup "A"
Vehicle Group "B"
MFC Sales Fleet MFC
Base(apgi) Base
-------
Table 2-2
of Specific Facto
Percent Change In
Fuel Economy Due To:
Systems Optimization
Calculated By;
x 100
Transmission Mix Shifts
^ FEBAICT\
' ' X
x 100
Engine Mix Shifts
• FEBAIC \ .1
"" FE^IC ) ~ \
x 100
Weight Mix Shifts
x 100
All Changes Combined
x 100
2-5
-------
Appendix 3
Variability Estimates for Emissions
and Economy over the FTP
-------
To Karl Hellnan
From John Foster
Re Test to test variability
25 March, 1981
Using, similar procedure and the same data as my sensitivity stuay
(see my memo to you of 12 March), I have looked into the differences
between multiple tests on the san;e vehicles.
For each multiply-tested car I calculated the mean fuel economy,
L-iC, CO, and i;0x 5 for both city and highway test cycles. I also found
the standard deviations of these 8 means, whicn are measures or their
test-to-test variability. The 1027 vehicles were taen avera^ea to
find tiieir uiean standard deviation. lie re are the results:
VARIABLE
l.FECITY
2-iiCCITY
3.COCITY
4.NOXCITY
5.FEHWY
6 .HCHWY
7.COiiUY
8.NGXHWY
it
1027
1027
1027
1027
432
430
361
431
ai;iL.u:i
0 .
0.
0.
o.
0.
0.
0.
o.
DIFFE;
3.
1 .
54
1.
7.
RENCE
5360
7460
.477
2800
5660
.43500
11
1.
.709
6720
.i£;u;
.46217
•OfaoGbd
1.0577
.12346
1.003S
.015156
.33817 .
.18613
(AVERAGE)
STD DLV
.4*b,o
.10265
2.0976
.13782
1.C633
.029851
.88500
.22276
Data base is CERT/EPA for model years 1975 to 1982.
3-1
-------
Appendix 4
Two Variable Linear Regression Plots
of Emissions versus Economy
(Data from all Model Years Available
and No Stratification of the Data)
-------
61. U.D. Jose is the urban fuel economy "delta-residual" using the Bascunana
equation.
62. U.D. Dill is the urban fuel economy "delta-residual" using the Murrell
equation.
63. U.D. J.P. is the urban fuel economy "delta-residual" using the Cheng
equation.
64. U.K. Jose is the urban fuel economy "ratio-residual" using the Bascunana
equation.
65. U.K. Dill is the urban fuel economy "ratio-residual" using the Murrell
equation.
66. U.K. J.P. is the urban fuel economy "ratio-residual" using the Cheng
equation.
When U is replaced by H, urban fuel economy becomes highway fuel economy.
When U is replaced by C, urban fuel economy becomes composite fuel
economy.
6-1
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 61.UD.JOSE N= 265 OUT OF 571
SOURCE OF SUM SQRS MEAN SOR F-STAT
REGRESSION
ERROR
TOTAL
1
263
264
8.8443
1005.4
1014.3
8.8443
3.8230
MULT R= .09338 R-SOR= .00872 SE= 1.9552
2.3135
SIGNIF
. 1295
VARIABLE
CONSTANT
1.FTPHC
PARTIAL
COEFF
STD ERROR T-STAT
.63058
-.09338 -2.2958
.33410
1.5094
1.8874
-1.5210
SIGNIF
.0602
. 1295
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 62.UD.DILL N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
.17837
REGRESSION
ERROR
TOTAL
1 .55990
263 825.55
264 826. 1 1
.55990
3.1390
MULT R= .02603 R-SQR= .OO068 SE= 1.7717
SIGNIF
.6731
VARIABLE
CONSTANT
1.FTPHC
PARTIAL
-.02603
COEFF
.21263
.57764
STD ERROR T-STAT
.30274
1.3677
.70235
-.42234
SIGNIF
.4831
.6731
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 63.UD.J.P. N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT SIGNIF
.91198 -1 .7629
REGRESSION
ERROR
TOTAL
1 .31441
263 906.71
264 907.03
.31441
3.4476
MULT R= .01862 R-SQR= .00035 SE= 1.8568
VARIABLE
CONSTANT
1.FTPHC
PARTIAL
COEFF
STD ERROR T-STAT
.16228
-.01862 -.43287
.31728
1.4334
.51147
-.30199
SIGNIF
.6095
.7629
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 64.UR.dOSE N= 265 OUT OF 571
SOURCE DF SUM SORS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1
263
264
. 16764 -1
1.9207
1.9374
.16764 -1
.73029 -2
2.2956
SIGNIF
. 1309
MULT R= .09302 R-SQR= .O0865 SE= .85457 -1
VARIABLE PARTIAL COEFF STD ERROR T-STAT SIGNIF
CONSTANT 1.0262 .14603 -1 70.279 0.
1.FTPHC -.09302 -.99954 -1 .65971 -1 -1.5151 .1309
LEAST SQUARES REGRESSION
ON
I
ANALYSIS OF VARIANCE OF 65.UR.DILL N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT SIGNIF
REGRESSION
ERROR
TOTAL
1
263
264
.43762 -3
1.6806
1 .6811
.43762 -3
.63903 -2
MULT R= .01613 R-SQR= .OOO26 SE= .79939 -1
.68481 -1 .7938
VARIABLE
CONSTANT
1.FTPHC
PARTIAL
COEFF
STD ERROR
T-STAT
SIGNIF
1.OO78 .1366O -1 73.779 0.
-.01613 -.16149 -1 .61711 -1 -.26169 .7938
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 66.UR.J.P. N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT SIGNIF
.77083 -1 .7815
REGRESSION
ERROR
TOTAL
1 .53910 -3 .53910 -3
263 1.8394 .69938 -2
264 1.8399
MULT R= .01712 R-SQR= .00029 SE= .83629 -1
VARIABLE
CONSTANT
1.FTPHC
PARTIAL
COEFF
STD ERROR T-STAT
SIGNIF
1.0077 .14290 -1 70.518 0.
-.01712 -.17924 -1 .64559 -1 -.27764 .7815
-------
I
•P-
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 71.HD.JOSE N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
SIGNIF
REGRESSION 1
ERROR 263
TOTAL 264
MULT R= .
VARIABLE
CONSTANT
1 .FTPHC
LEAST SQUARES
05387 R-SQR=
PARTIAL
.O5387
REGRESSION
ANALYSIS OF VARIANCE OF 72.
SOURCE
DF
REGRESSION ' 1
ERROR 263
TOTAL 264
MULT R=
VARIABLE
CONSTANT
1 .FTPHC
LEAST SQUARES
.08581 R-SQR=
PARTIAL
.08581
REGRESSION
ANALYSIS OF VARIANCE OF 73.
SOURCE
DF
REGRESSION 1
ERROR 263
TOTAL 264
MULT R=
VARIABLE
CONSTANT
1 .FTPHC
.07590 R-SQR=
PARTIAL
.07590
6.4400
2212.6
2219. 1
.00290 SE=
COEFF
-. 16283
1 .9591
HD.DILL N=
SUM SQRS
15.056
2029.8
2044.8
.OO736 SE=
COEFF
-.61347
2.9955
HD.J.P. N=
SUM SQRS
12.890
2224.9
2237.7
.00576 SE=
COEFF
-.52557
2.7716
6 . 4400
8 . 4 1 30
2 . 9005
STD ERROR
.49563
2.2391
265 OUT OF
MEAN SQR
15.056
7.7177
2.7781
STD ERROR
.47471
2. 1446
265 OUT OF
MEAN SQR
12.890
8.4595
2 . 9085
STD ERROR
.49700
2.2453
.76549
T-STAT
-.32854
.87492
571
F-STAT
1 . 9509
T-STAT
-1 .2923
1 .3967
571
F-STAT
1.5237
T-STAT
1 .0575
1 .2344
.3824
SIGNIF
.7428
.3824
SIGNIF
. 1637
SIGNIF
. 1974
. 1637
SIGNIF
.2182
SIGNIF
.2913
.2182
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 74.HR.JOSE N= 265 OUT OF 571
SOURCE OF SUM SQRS MEAN SQR F-STAT
.57969
REGRESSION
ERROR
TOTAL
1 .47380 -2 .47380 -2
263 2.1496 .81734 -2
264 2.1543
MULT R= .04690 R-SQR= .OO220 SE= .90407 -1
SIGNIF
.4471
VARIABLE
CONSTANT
1.FTPHC
PARTIAL
.04690
COEFF
STD ERROR T-STAT
.99616 . 15448 -1 64.483
.53138 -1 .69792 -1 .76137
SIGNIF
O.
.4471
LEAST SQUARES REGRESSION
I
Ul
ANALYSIS OF VARIANCE OF 75.HR.DILL N= 265 OUT OF 571
SOURCE OF SUM SQRS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1 . 12233 -1
263 2.0592
264 2.0715
.12233 -1
.78297 -2
1.5624
MULT R= .O7685 R-SQR= .00591 SE= .88486 -1
SIGNIF
.2124
VARIABLE
CONSTANT
1.FTPHC
PARTIAL
.07685
COEFF
.9827O
.85383 -1
STD ERROR
T-STAT
.15120 -1 64.993
.68309 -1 1.2500
SIGNIF
0.
.2124
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 76.HR.d.P. N= 265 OUT OF 571
SOURCE OF SUM SQRS MEAN SQR F-STAT
1.3160
REGRESSION
ERROR
TOTAL
1 .11410 -1
263 2.2803
264 2.2917
.11410 -1
.86703 -2
MULT R= .07056 R-SQR= .O0498 SE= .93114 -1
SIGNIF
.2524
VARIABLE
CONSTANT
1.FTPHC
PARTIAL
.07O56
COEFF
STD ERROR T-STAT
SIGNIF
.98630 .15911 -1 61.988 0.
.82461 -1 .71882 -1 1.1472 .2524
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 81.CD.JOSE N= 265 OUT OF 571
SOURCE
OF
REGRESSION 1
ERROR 263
TOTAL 264
MULT R= .
VARIABLE
CONSTANT
1 .FTPHC
LEAST SQUARES
02937 R-SQR=
PARTIAL
-.02937
REGRESSION
ANALYSIS OF VARIANCE OF 82.
SOURCE
OF
REGRESSION 1
ERROR 263
TOTAL 264
MULT R=
VARIABLE
CONSTANT
1 .FTPHC
LEAST SQUARES
.03956 R-SQR=
PARTIAL
.03956
REGRESSION
ANALYSIS OF VARIANCE OF 83.
SOURCE
OF
REGRESSION 1
ERROR 263
TOTAL 264
MULT R=
VARIABLE
CONSTANT
1 .FTPHC
.03070 R-SQR=
PARTIAL
.0307O
SUM SQRS
.97346
1127.6
1 128.6
.00086 SE
COEFF
.24461
- .76166
CD. DILL N
SUM SQRS
1 .4807
944.80
946.28
.00156 SE
COEFF
-. 18803
.93937
MEAN SQR
.97346
4.2876
= 2.O7O7
STD ERROR
.35383
1 .5985
= 265 OUT OF
MEAN SQR
1 . 4807
3.5924
= 1.8954
STD ERROR
.32387
1 .4632
CD.J.P. N= 265 OUT OF
SUM SQRS
1 .0476
1110.3
1111.3
.00094 SE
COEFF
-. 15913
.79015
MEAN SQR
1 .0476
4.2216
= 2.0547
STD ERROR
.35109
1 .5861
F-STAT
.22704
T-STAT
.69132
-.47649
571
F-STAT
.41217
T-STAT
-.58056
.642O1
571
F-STAT
.24816
T-STAT
-.45324
.49816
SIGNIF
.6341
SIGNIF
.4900
.6341
SIGNIF
.5214
SIGNIF
.5620
.5214
SIGNIF
.6188
SIGNIF
.6507
.6188
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 84.CR.dOSE N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1
263
264
. 17618 -2
1.6411
1 .6428
. 17618 -2
.62397 -2
MULT R= .03275 R-SQR= .OO107 SE= .78992 -1
.28235
SIGNIF
.5956
VARIABLE
CONSTANT
1.FTPHC
PARTIAL COEFF STD ERROR T-STAT SIGNIF
1.0099 .13498 -1 74.816 0.
-.03275 -.32403 -1 .60980 -1 -.53137 .5956
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 85.CR.DILL N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1 .27560 -2 .27560 -2
263 1.4919 .56725 -2
264 1.4946
MULT R= .04294 R-SQR= .00184 SE= .75316 -1
.48586
SIGNIF
.4864
VARIABLE
CONSTANT
1.FTPHC
PARTIAL
.04294
COEFF
STD ERROR T-STAT
.993O2 .12870 -1 77.160
.40527 -1 .58142 -1 .69703
SIGNIF
0.
.4864
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 86.CR.J.P. N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
.22573
REGRESSION
ERROR
TOTAL
1 . 14246 -2 . 14246 -2
263 1.6598 .63112 -2
264 1.6613
MULT R= .02928 R-SQR= .00086 SE= .79443 -1
SIGNIF
.6351
VARIABLE
CONSTANT
t.FTPHC
PARTIAL
.02928
COEFF
STD ERROR T-STAT
SIGNIF
.99579 .13575 -1 73.355 0.
.29137 -1 .61328 -1 .47511 .6351
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 61.UD.dOSE N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SOR F-STAT
REGRESSION
ERROR
TOTAL
1 33.425
263 980.86
264 1014.3
33.425
3.7295
MULT R= .18153 R-SQR= .O3295 SE= 1.9312
8.9623
SIGNIF
.0030
VARIABLE
CONSTANT
2.FTPCO
PARTIAL
COEFF
.79283
-.18153 -.29846
STD ERROR T-STAT SIGNIF
.24345 3.2566 .0013
.99697 -1 -2.9937 .003O
LEAST SQUARES REGRESSION
00
ANALYSIS OF VARIANCE OF 62.UD.DILL N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
5.8904
REGRESSION
ERROR
TOTAL
1 18.097
263 808.02
264 826.11
18.097
3.0723
MULT R= .14801 R-SQR= .02191 SE= 1.7528
SIGNIF
.0159
VARIABLE
CONSTANT
2.FTPCO
PARTIAL
COEFF
.56163
-.14801 -.21961
STD ERROR T-STAT SIGNIF
.22097 2.5417 .0116
.90487 -1 -2.4270 .0159
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 63.UD.J.P. N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
5.9366
REGRESSION
ERROR
TOTAL
1 20.022
263 887.00
264 907.03
20.022
3.3726
MULT R= .14857 R-SQR= .02207 SE= 1.8365
SIGNIF
.0155
VARIABLE
CONSTANT
2.FTPCO
PARTIAL
COEFF
.56545
-.14857 -.23100
STD ERROR T-STAT SIGNIF
.23151 2.4424 .0152
.948O7 -1 -2.4365 .O155
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 64.UR.JOSE N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SOR F-STAT
REGRESSION
ERROR
TOTAL
1
263
264
.37122 -1
1.90O3
1.9374
.37122 -1 5.1377
.72255 -2
MULT R= .13842 R-SOR= .O1916 SE= .85003 -1
SIGNIF
.0242
VARIABLE
CONSTANT
2.FTPCO
PARTIAL
COEFF
STD ERROR T-STAT
1.0268 .10716 -1 95.822
-.13842 -.99466 -2 .43882 -2 -2.2666
SIGNIF
0.
.O242
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 65.UR.DILL N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1
263
264
. 18843 -1
1.6622
1.68 11
.18843 -1 2.9814
.63203 -2
MULT R= .10587 R-SQR= .01121 SE= .79500 -1
SIGNIF
.0854
VARIABLE
PARTIAL
COEFF
STD ERROR T-STAT
CONSTANT 1.O196 .10022 -1 1O1.73
2.FTPCO -.10587 -.70865 -2 .41042 -2 -1.7267
SIGNIF
0.
. O854
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 66.UR.J.P. N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
3.7364
REGRESSION
ERROR
TOTAL
1
263
264
.25773 -1
1.8141
1.8399
.25773 -1
.68978 -2
MULT R= .11835 R-SQR= .01401 SE= .83053 -1
SIGNIF
.0543
VARIABLE
PARTIAL
COEFF
STD ERROR T-STAT
CONSTANT 1.0217 .10470 -1 97.582
2.FTPCO -.11835 -.82878 -2 .42876 -2 -1.9330
SIGNIF
0.
.0543
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 71.HD.JOSE N= 265 OUT OF 571
SOURCE OF SUM SQRS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1 14.171
263 2204.9
264 2219.1
14.171
8.3836
MULT R= .07991 R-SOR= .00639 SE= 2.8954
1.6903
SIGNIF
. 1947
VARIABLE
CONSTANT
2.FTPCO
PARTIAL
COEFF
STD ERROR T-STAT
.65622
-.07991 -.19434
.36501
.14948
1.7978
-1.3001
SIGNIF
.0734
. 1947
LEAST SQUARES REGRESSION
I
M
O
ANALYSIS OF VARIANCE OF 72.HD.DILL N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
.50399
REGRESSION
ERROR
TOTAL
1 3.911O
263 204O.9
264 2044.8
3.91 10
7.7601
MULT R= .O4373 R-SQR= .00191 SE= 2.7857
SIGNIF
.4784
VARIABLE
CONSTANT
2.FTPCO
PARTIAL COEFF
.22295
-.O4373 -.102O9
STD ERROR
.35118
.14381
T-STAT
.63487
-.7O992
SIGNIF
.5261
.4784
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 73.HD.J.P. N=
SOURCE DF SUM SQRS
REGRESSION
ERROR
TOTAL
1 26.006
263 2211.7
264 2237.7
265 OUT OF 571
MEAN SQR F-STAT
26.OO6 3.0924
8.4097
SIGNIF
.0798
MULT R= .10780 R-SQR= .01162 SE= 2.8999
VARIABLE
CONSTANT
2.FTPCO
PARTIAL
COEFF
.60830
-.1078O -.26327
STD ERROR
.36558
.14971
T-STAT
1.6639
-1.7585
SIGNIF
.0973
.0798
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 74.HR.JOSE N= 265 OUT OF 571
SOURCE OF SUM SQRS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1 .75769 -2
263 2.1468
264 2.1543
.75769 -2
.81626 -2
MULT R= .05930 R-SQR= .OO352 SE= .90347 -1
.92825
SIGNIF
.3362
VARIABLE
CONSTANT
2.FTPCO
PARTIAL
COEFF
STD ERROR T-STAT
1.0167 .11390 -1 89.267
-.0593O -.44937 -2 .46641 -2 -.96346
SIGNIF
0.
.3362
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 75.HR.DILL N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
. 17866
REGRESSION
ERROR
TOTAL
1 .14062 -2 .14062 -2
263 2.0700 .78709 -2
264 2.0715
MULT R= .O2605 R-SQR= .O0068 SE= .88718 -1
SIGNIF
.6729
VARIABLE
CONSTANT
2.FTPCO
PARTIAL COEFF STD ERROR T-STAT SIGNIF
1.OO45 .11184 -1 89.811 O.
-.02605 -.19359 -2 .45800 -2 -.42268 .6729
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 76.HR.J.P. N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1 .19343 -1
263 2.2724
264 2.2917
.19343 -1 2.2387
.86401 -2
MULT R= .09187 R-SQR= .00844 SE= .92952 -1
SIGNIF
. 1358
VARIABLE PARTIAL COEFF STD ERROR T-STAT SIGNIF
. CONSTANT 1.0186 .11718 -1 86.929 0.
2.FTPCO -.09187 -.71799 -2 .47986 -2 -1.4962 .1358
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 81.CD.JOSE N= 265 OUT OF 571
SOURCE OF SUM SQRS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1
263
264
24.568
1104.1
1128.6
24.568
4.1979
MULT R= .14754 R-SQR= .02177 SE= 2.0489
5.8523
SIGNIF
.0162
VARIABLE
CONSTANT
2.FTPCO
PARTIAL
COEFF
.63293
-.14754 -.25588
STD ERROR
.25829
.10577
T-STAT
2.4504
-2.4192
SIGNIF
.O149
.O162
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 82.CD.DILL N= 265 OUT OF 571
SOURCE OF SUM SQRS MEAN SQR F-STAT
2.7074
REGRESSION
ERROR
TOTAL
1 9.642O
263 936.64
264 946.28
9.6420
3.5614
MULT R= .10094 R-SQR= .01019 SE= 1.8872
SIGNIF
. 1011
VARIABLE
CONSTANT
2.FTPCO
PARTIAL
COEFF
.34783
-.10094 -.16030
STD ERROR T-STAT SIGNIF
.23790 1.4621 .1449
.97424 -1 -1.6454 .1011
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 83.CD.J.P. N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
4.8007
REGRESSION
ERROR
TOTAL
1 19.922
263 1091.4
264 1111.3
19.922
4.1498
MULT R= .13389 R-SQR= .O1793 SE= 2.0371
SIGNIF
.0293
VARIABLE
CONSTANT
2.FTPCO
PARTIAL
COEFF
STD ERROR T-STAT
.49543
-.13389 -.23042
.25681
.10517
1.9292
-2.1910
SIGNIF
.O548
.0293
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 84.CR.JOSE N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SOR F-STAT
REGRESSION
ERROR
TOTAL
1 .19776 -1 .19776 -1 3.2O45
263 1.6230 .61713 -2
264 1.6428
MULT R= .10972 R-SQR= .O1204 SE= .78557 -1
SIGNIF
.0746
VARIABLE
CONSTANT
2.FTPCO
PARTIAL
COEFF
STD ERROR T-STAT
1.0186 .99033 -2 102.86
-.10972 -.72597 -2 .4O555 -2 -1.7901
SIGNIF
0.
.O746
LEAST SQUARES REGRESSION
I
M
CO
ANALYSIS OF VARIANCE OF 85.CR.DILL N=
SOURCE DF SUM SQRS
REGRESSION
ERROR
TOTAL
1
263
264
.62646 -2
1.4883
1.4946
265 OUT_OF^571
MEAN SQR F-STAT ... SIGNIF
.62646 -2 1.l670 .2937
.56591 -2
MULT R= .06474 R-SQR= .OO419 SE= .75227 -1
VARIABLE
PARTIAL
COEFF
STD ERROR T-STAT
CONSTANT 1.O1O1 .94835 -2 1O6.51
2.FTPCO -.06474 -.40860 -2 .38836 -2 -1.O521
SIGNIF
O.
.2937
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 86.CR.J.P. N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1
263
264
.19604 -1
1 .6417
1.6613
.19604 -1 3.1407
.62421 -2
MULT R= .1O863 R-SQR= .01180 SE= .79007 -1
SIGNIF
.0775
VARIABLE PARTIAL COEFF STD ERROR T-STAT SIGNIF
CONSTANT 1.0172 .99599 -2 102.13 0.
2.FTPCO -.10863 -.72282 -2 .4O787 -2 -1.7722 .0775
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 61.UD.JOSE N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SOR F-STAT
9.3206
REGRESSION
ERROR
TOTAL
1 34.716
263 979.57
264 1O14.3
34.716
3.7246
SIGNIF
.0025
MULT R= .18500 R-SOR= .03423 SE= 1.9299
VARIABLE
CONSTANT
3.FTPNOX
PARTIAL
COEFF
-.62211
.1850O 1.4213
STD ERROR
.28121
.46554
T-STAT
-2.2123
3.053O
SIGNIF
.0278
.0025
LEAST SQUARES REGRESSION
I
(-•
4>-
ANALYSIS OF VARIANCE OF 62.UD.DILL N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
3.2744
REGRESSION
ERROR
TOTAL
1 10.159
263 815.95
264 826.11
1O.159
3.1025
MULT R= .11089 R-SQR= .01230 SE= 1.7614
SIGNIF
.0715
VARIABLE
CONSTANT
3.FTPNOX
PARTIAL
COEFF
-.32780
.11089 .76884
STD ERROR
.25665
.42489
T-STAT
-1.2772
1.8095
SIGNIF
.2O26
.0715
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 63.UD.J.P. N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
5.9862
REGRESSION
ERROR
TOTAL
1 20.186
263 886.84
264 907.03
20.186
3.3720
MULT R= .14918 R-SQR= .02225 SE= 1.8363
SIGNIF
.0151
VARIABLE
CONSTANT
3.FTPNOX
PARTIAL
COEFF
-.52076
.14918 1.0838
STD ERROR
.26757
.44296
T-STAT
-1.9463
2.4467
SIGNIF
.0527
.0151
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 64.UR.JOSE N= 265 OUT OF 571
o>
I
H
Ui
SOURCE
OF
REGRESSION 1
ERROR 263
TOTAL 264
MULT R= .
VARIABLE
CONSTANT
3.FTPNOX
LEAST SQUARES
19650 R-SQR=
PARTIAL
. 1965O
REGRESSION
ANALYSIS OF VARIANCE OF 65.
SOURCE
DF
REGRESSION 1
ERROR 263
TOTAL 264
MULT R= .
VARIABLE
CONSTANT
3.FTPNOX
LEAST SQUARES
. 10237 R-SQR=
PARTIAL
. 10237
REGRESSION
ANALYSIS OF VARIANCE OF 66.
SOURCE
DF
REGRESSION 1
ERROR 263
TOTAL 264
MULT R= .
VARIABLE
CONSTANT
3.FTPNOX
15052 R-SQR=
PARTIAL
. 15052
SUM SQRS
.74805 -1
1 .8626
1 .9374
.03861 SE=
COEFF
.96947
.65976 -1
UR.DILL N=
SUM SQRS
. 17617 -1
1 .6635
1 .681 1
.O1O48 SE=
COEFF
.98693
.32017 -1
UR.J.P. N=
SUM SQRS
.41685 -1
1 .7982
1 .8399
.O2266 SE=
COEFF
.97704
.49250 -1
MEAN SQR
.74805 -1
.7O822 -2
.84156 -1
STD ERROR
. 12262 -1
.2O300 -1
265 OUT OF
MEAN SQR
. 17617 -1
.63250 -2
.79530 -1
STD ERROR
. 11588 -1
. 19184 -1
265 OUT OF
MEAN SQR
.41685 -1
.68373 -2
.82688 -1
STD ERROR
. 12048 -1
. 19946 -1
F-STAT
1O. 562
T-STAT
79.O61
3 . 2500
571
F-STAT
2.7853
T-STAT
85. 167
1 .6689
571
F-STAT
6.0967
T-STAT
8 1 . 093
2.4692
SIGNIF
.OO13
SIGNIF
0.
.OO13
SIGNIF
.0963
SIGNIF
0.
.O963
SIGNIF
.0142
SIGNIF
0.
.0142
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 71.HD.JOSE N= 265 OUT OF 571
SOURCE DF SUM SORS MEAN SOR F-STAT
REGRESSION
ERROR
TOTAL
1
263
264
53.998
2165. 1
2219.1
53.998
8.2322
MULT R= .15599 R-SQR= .02433 SE= 2.8692
6 5594
SIGNIF
.0110
VARIABLE
CONSTANT
3.FTPNOX
PARTIAL
COEFF
STD ERROR T-STAT
-.72910
. 15599 1 .7726
.41806
.6921 1
-1 .7440
2.561 1
SIGNIF
.0823
.0110
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 72.HD.DILL N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
3.3185
REGRESSION
ERROR
TOTAL
1 25.480
263 2019.3
264 2044.8
25.480
7.6781
MULT R= .11163 R-SQR= .01246 SE= 2.7709
SIGNIF
.0696
VARIABLE
CONSTANT
3.FTPNOX
PARTIAL
COEFF
-.66170
. 1 1163 1 .2176
STD ERROR
.40375
.66841
T-STAT
-1.6389
1 .8217
SIGNIF
. 1O24
.0696
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 73.HD.J.P. N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
3.6665
REGRESSION
ERROR
TOTAL
1 30.767
263 2207.0
264 2237.7
3O.767
8.3916
MULT R= .11726 R-SQR= .01375 SE= 2.8968
SIGNIF
.0566
VARIABLE
CONSTANT
3.FTPNOX
PARTIAL
COEFF
STD ERROR T-STAT
-.68598
. 1 1726 1 .3380
.42209
.69878
-1.6252
1.9148
SIGNIF
. 1053
.0566
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 74.HR.JOSE N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SOR F-STAT
REGRESSION
ERROR
TOTAL
1 .58894 -1 .58894 -1 7.3918
263 2.0954 .79674 -2
264 2.1543
SIGNIF
.OO70
MULT R= .16534 R-SQR= .O2734 SE= .89261 -1
VARIABLE PARTIAL COEFF STD ERROR T-STAT SIGNIF
CONSTANT .97507 .13006 -1 74.970 0.
3.FTPNOX .16534 .58540 -1 .21532 -1 2.7188 .OO70
LEAST SQUARES REGRESSION
I
M
~J
ANALYSIS OF VARIANCE OF 75.HR.DILL N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1 .21948 -1 .21948 -1 2.8165
263 2.0495 .77928 -2
264 2.0715
MULT R= .1O294 R-SQR= .01060 SE= .88277 -1
VARIABLE
CONSTANT
3.FTPNOX
PARTIAL
.1O294
COEFF
STD ERROR T-STAT
.98O76 .12863 -1 76.248
.35737 -1 .21294 -1 1.6782
SIGNIF
.0945
SIGNIF
O.
.O945
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 76.HR.J.P. N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
3.3644
REGRESSION
ERROR
TOTAL
1 .28946 -1
263 2.2628
264 2.2917
.28946 -1
.86036 -2
MULT R= .11239 R-SQR= .01263 SE= .92756 -1
SIGNIF
.0678
VARIABLE
CONSTANT
3.FTPNOX
PARTIAL
.11239
COEFF
STD ERROR T-STAT
.98O85 .13515 -1 72.573
.41040 -1 .22375 -1 1 .8342
SIGNIF
0.
.0678
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 81.CD.JOSE N= 265 OUT OF 571
SOURCE DF SUM SORS MEAN SQR F-STAT
SIGNIF
I
h-1
oo
REGRESSION 1
ERROR 263
TOTAL 264
MULT R= .
VARIABLE
CONSTANT
3.FTPNOX
LEAST SQUARES
19399 R-SQR=
PARTIAL
. 19399
REGRESSION
ANALYSIS OF VARIANCE OF 82.
SOURCE
DF
REGRESSION 1
ERROR 263
TOTAL 264
MULT R=
VARIABLE
CONSTANT
3.FTPNOX
LEAST SQUARES
. 10682 R-SQR=
PARTIAL
. 10682
REGRESSION
ANALYSIS OF VARIANCE OF 83.
SOURCE
DF
REGRESSION 1
ERROR 263
TOTAL 264
MULT R=
VARIABLE
CONSTANT
3.FTPNOX
.12181 R-SQR=
PARTIAL
. 12181
42.470
1086. 1
1 128.6
.03763 SE=
42.470
4. 1298
2.0322
COEFF STD ERROR
-.77378
1 .5720
CD. DILL N=
SUM SQRS
10.797
935.48
946.28
.01141 SE=
COEFF
-.42815
.79262
CD.d.P. N=
SUM SQRS
16.489
1094.8
1111.3
.01484 SE=
COEFF
-.53245
.97954
.29611
.49O22
265 OUT OF
MEAN SQR
10.797
3.557O
1 .8860
STD ERROR
.27481
.45495
265 OUT OF
MEAN SQR
16.489
4. 1629
2.0403
STD ERROR
.29729
.49217
1O. 284
T-STAT
-2.6131
3.2068
571
F-STAT
3.0354
T-STAT
-1 .5580
1 .7422
571
F-STAT
3.9610
T-STAT
-1 .7910
1 .9902
.0015
SIGNIF
.0095
.0015
SIGNIF
.0826
SIGNIF
. 1204
.0826
SIGNIF
.0476
SIGNIF
.0744
.0476
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 84.CR.JOSE N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SOR F-STAT
REGRESSION
ERROR
TOTAL
1
263
264
.69217 -1
1.5736
1.6428
.69217 -1
.59833 -2
11.568
SIGNIF
.O008
MULT R= .20526 R-SQR= .04213 SE= .77352 -1
VARIABLE PARTIAL COEFF STD ERROR T-STAT SIGNIF
CONSTANT .96840 .11271-1 85.921 0.
3.FTPNOX .20526 .63464 -1 .18659 -1 3.4012 .0008
LEAST SQUARES REGRESSION
I
M
VO
ANALYSIS OF VARIANCE OF 85.CR.DILL N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1 .12375 -1 .12375 -1 2.1958
263 1.4822 .56359 -2
264 1.4946
MULT R= .09099 R-SQR= .00828 SE= .75O73 -1
SIGNIF
. 1396
VARIABLE
CONSTANT
3.FTPNOX
PARTIAL
COEFF
STD ERROR T-STAT
.98669 .10939 -1 9O.202
.09099 .26835 -1 .18109-1 1.4818
SIGNIF
0.
. 1396
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 86.CR.d.P. N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
3.5239
REGRESSION
ERROR
TOTAL
1
263
264
.21965 -1
1.6393
1 .6613
.21965 -1
.62331 -2
MULT R= .11499 R-SQR= .01322 SE= .78950 -1
SIGNIF
.0616
VARIABLE
CONSTANT
3.FTPNOX
PARTIAL
.11499
COEFF
STD ERROR T-STAT
SIGNIF
.98222 .11504 -1 85.383 0.
.35751 -1 .19045 -1 1.8772 .0616
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 61.UD.dOSE N= 265 OUT OF 571
SOURCE DF SUM SORS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1 16.233
263 998.05
264 1014.3
16.233
3.7949
MULT R= .12651 R-SQR= .01600 SE= 1.9480
4.2776
SIGNIF
.0396
VARIABLE
CONSTANT
4.FEHC
PARTIAL
COEFF
STO ERROR T-STAT
.43330
-.12651 -6.1138
.17958
2.9561
2.4129
-2.0682
SIGNIF
.0165
.0396
LEAST SQUARES REGRESSION
I
IS5
O
ANALYSIS OF VARIANCE OF 62.UD.DILL N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
1.3045
REGRESSION
ERROR
TOTAL
1 4.0773
263 822.03
264 826.11
4.0773
3.1256
MULT R= .07025 R-SQR= .00494 SE= 1.7679
SIGNIF
.2544
VARIABLE
CONSTANT
4.FEHC
PARTIAL
COEFF
.2321 1
-.07025 -3.0641
STD ERROR
.16297
2.6828
T-STAT
1.4242
-1.1421
SIGNIF
. 1556
.2544
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 63.UD.J.P. N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
2.S563
REGRESSION
ERROR
TOTAL
1 8.7314
263 898.29
264 907.O3
8.7314
3.4156
MULT R= .09811 R-SQR= .O0963 SE= 1.8481
SIGNIF
.1111
VARIABLE
CONSTANT
4.FEHC
PARTIAL
COEFF
STD ERROR T-STAT
.27596
-.09811 -4.4839
.17037
2.8045
1.6198
-1.5989
SIGNIF
. 1065
.1111
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 64.UR.JOSE N= 265 OUT OF 571
SOURCE OF SUM SQRS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1 .27433 -1
263 1.9100
264 1.9374
.27433 -1 3.7774
.72623 -2
MULT R= .11899 R-SOR= .O1416 SE= .85219 -1
SIGNIF
.0530
VARIABLE
CONSTANT
4.FEHC
PARTIAL
COEFF
1.0170
-.11899 -.25134
STD ERROR T-STAT SIGNIF
.78558 -2 129.46 0.
.12932 -1.9436 .0530
LEAST SQUARES REGRESSION
I
to
ANALYSIS OF VARIANCE OF 65.UR.DILL N= 265 OUT OF 571
SOURCE OF SUM SQRS MEAN SQR F-STAT
.49329
REGRESSION
ERROR
TOTAL
1
263
264
.31472 -2
1.6779
1 .6811
.31472 -2
.6380O -2
MULT R= .04327 R-SQR= .00187 SE= .79875 -1
SIGNIF
.4831
VARIABLE
CONSTANT
4.FEHC
PARTIAL
COEFF
STO ERROR
T-STAT
1.0083 .73631 -2 136.94
-.04327 -.85130 -1 .12121 -.70235
SIGNIF
0.
.4831
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 66.UR.J.P. N= 265 OUT OF 571
SOURCE OF SUM SQRS MEAN SQR F-STAT
1.4074
REGRESSION
ERROR
TOTAL
1
263
264
.97938 -2
1.8301
1.8399
.97938 -2
.69586 -2
MULT R= .07296 R-SQR= .00532 SE= .83418 -1
SIGNIF
.2366
VARIABLE
CONSTANT
4.FEHC
PARTIAL
COEFF
1.0108
-.O7296 -.IS017
STD ERROR T-STAT SIGNIF
.76898 -2 131.45 0.
.12658 -1.1864 .2366
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 71.HD.JOSE N= 265 OUT OF 571
SOURCE OF SUM SQRS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1 12.415
263 2206.6
264 2219.1
12.415
8.3903
MULT R= .07480 R-SQR= .00559 SE= 2.8966
1.4797
SIGNIF
.2249
VARIABLE
CONSTANT
4.FEHC
PARTIAL
COEFF
STD ERROR T-STAT
.48399
-.07480 -5.3468
.26702
4.3955
1.8126
-1.2164
SIGNIF
.07 10
.2249
LEAST SQUARES REGRESSION
I
N3
1-0
ANALYSIS OF VARIANCE OF 72.HD.DILL N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
.86819
REGRESSION
ERROR
TOTAL
1 6.7280
263 2038.1
264 2044.8
6.7280
7.7494
MULT R= .05736 R-SQR= .00329 SE= 2.7838
VARIABLE
CONSTANT
4.FEHC
PARTIAL COEFF
.18353
-.05736 -3.9360
STD ERROR T-STAT
.25662
4.2243
.71517
-.93177
SIGNIF
.3523
SIGNIF
.4751
.3523
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 73.HD.J.P. N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
.87503
REGRESSION
ERROR
TOTAL
1 7.4205
263 2230.3
264 2237.7
7.4205
8.4803
MULT R= .O5759 R-SQR= .OO332 SE= 2.9121
SIGNIF
.3504
VARIABLE
CONSTANT
4.FEHC
PARTIAL
COEFF
STD ERROR T-STAT
.23414
-.05759 -4.1337
.26845
4.4190
.87219
-.93543
SIGNIF
.3839
.3504
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 74.HR.JOSE N= 265 OUT OF 571
SOURCE DF SUM SORS MEAN SQR F-STAT
1.3604
REGRESSION
ERROR
TOTAL
1 .11O86 -1
263 2.1432
264 2.1543
.11086 -1
.81492 -2
MULT R= .O7173 R-SQR= .OO515 SE= .9O273 -1
SIGNIF
.2445
VARIABLE
CONSTANT
4.FEHC
PARTIAL
COEFF
STD ERROR T-STAT
1.0144 .83217 -2 121.89
-.07173 -.15977 .13699 -1.1663
SIGNIF
0.
.2445
LEAST SQUARES REGRESSION
ON
I
OJ
ANALYSIS OF VARIANCE OF 75.HR.DILL N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1 .55034 -2 .55034 -2 .70059
263 2.O659 .78553 -2
264 2.0715
MULT R= .05154 R-SQR= .00266 SE= .88630 -1
SIGNIF
. 4O33
VARIABLE
CONSTANT
4.FEHC
PARTIAL
COEFF
1.OO54
-.05154 -.11257
STD ERROR T-STAT SIGNIF
.817O2 -2 123.O6 0.
.13449 -.83701 .4O33
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 76.HR.J.P. N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
.48631
REGRESSION
ERROR
TOTAL
1 .42297 -2
263 2.2875
264 2.2917
.42297 -2
.86976 -2
MULT R= .04296 R-SQR= .00185 SE= .93261 -1
SIGNIF
.4862
VARIABLE
CONSTANT
4.FEHC
PARTIAL
COEFF
STD ERROR T-STAT
1.0078 .85971 -2 117.23
-.04296 -.98690 -1 .14152 -.69736
SIGNIF
0.
.4862
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 81.CD.JOSE N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SOR F-STAT
3.0669
REGRESSION
ERROR
TOTAL
1 13.009
263 1115.6
264 1128.6
13.009
4.2419
MULT R= .10736 R-SQR= .01153 SE= 2.0596
SIGNIF
.0811
VARIABLE
CONSTANT
4.FEHC
PARTIAL
COEFF
STD ERROR T-STAT
.33519
-.10736 -5.4733
.18986
3.1253
1.7655
-1.7513
SIGNIF
.O786
.081 1
LEAST SQUARES REGRESSION
I
ho
ANALYSIS OF VARIANCE OF 82.CD.DILL N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
.84147
REGRESSION
ERROR
TOTAL
1 3.0180
263 943.26
264 946.28
3.0180
3.5866
MULT R= .05647 R-SQR= .00319 SE= 1.8938
SIGNIF
.3598
VARIABLE
CONSTANT
4.FEHC
PARTIAL
COEFF
STD ERROR T-STAT
.12540
-.05647 -2.6362
.17458
2.8738
.71833
-.91732
SIGNIF
.4732
.3598
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 83.CD.J.P. N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
1.2558
REGRESSION
ERROR
TOTAL
1 5.2814
263 1106.0
264 1111.3
5.2814
4.2055
MULT R= .06894 R-SQR= .00475 SE= 2.05O7
SIGNIF
.2635
VARIABLE
CONSTANT
4.FEHC
PARTIAL
COEFF
STD ERROR T-STAT
.16203
-.06894 -3.4873
.18904
3. 1119
.85712
-1.1206
SIGNIF
.3922
.2635
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 84.CR.JOSE N= 265 OUT OF 571
SOURCE DF SUM SORS MEAN SOR F-STAT
REGRESSION
ERROR
TOTAL
1 .16863 -1 .16863 -1 2.7277
263 1.6260 .61823 -2
264 1.6428
MULT R= .10132 R-SOR= .01026 SE= .78628 -1
SIGNIF
.O998
VARIABLE
CONSTANT
4.FEHC
PARTIAL
COEFF
1.0121
-.10132 -.19706
STD ERROR T-STAT SIGNIF
.72482 -2 139.63 0.
.11931 -1.6516 .0998
LEAST SQUARES REGRESSION
Cf-
I
NJ
Ln
ANALYSIS OF VARIANCE OF 85.CR.DILL N= 265 OUT OF 571
SOURCE DF SUM SORS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1 .15659 -2 .15659 -2 .27583
263 1.4930 .56770 -2
264 1.4946
MULT R= .O3237 R-SQR= .O0105 SE= .75346 -1
SIGNIF
.5999
VARIABLE
CONSTANT
4.FEHC
PARTIAL
COEFF
STD ERROR
T-STAT
1.0041 .69456 -2 144.57
-.03237 -.60048 -1 .11433 -.52520
SIGNIF
O.
.5999
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 86.CR.J.P. N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
.37139
REGRESSION
ERROR
TOTAL
1
263
264
.23426 -2
1.6589
1.6613
.23426 -2
.63077 -2
MULT R= .03755 R-SQR= .00141 SE= .79421 -1
SIGNIF
.5428
VARIABLE
CONSTANT
4.FEHC
PARTIAL
COEFF
STD ERROR T-STAT
1.0051 .73213 -2 137.29
-.O3755 -.73446 -1 .12052 -.60942
SIGNIF
0.
.5428
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 61.UD.JOSE N= 265 OUT OF 571
SOURCE DF SUM SORS MEAN SOR F-STAT
REGRESSION
ERROR
TOTAL
1 25.244
263 989.04
264 1014.3
25.244
3.7606
MULT R= .15776 R-SQR= .02489 SE= 1.9392
6.7129
SIGNIF
.O1O1
VARIABLE
CONSTANT
5.FECO
PARTIAL
COEFF
STD ERROR T-STAT
.31744
-.15776 -.34475
.13437
.13306
2.3624
-2.5909
SIGNIF
.0189
.0101
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 62.UD.DILL N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
3.3809
REGRESSION
ERROR
TOTAL
1 10.485
263 815.63
264 826. 1 1
10.485
3.1012
MULT R= .11266 R-SOR= .01269 SE= 1.7610
SIGNIF
.0671
VARIABLE
CONSTANT
5.FECO
PARTIAL
COEFF
. 19712
-.11266 -.22218
STD ERROR
.12202
.12084
T-STAT
1 .6154
-1 .8387
SIGNIF
. 1O74
.0671
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 63.UD.J.P. N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
4.9056
REGRESSION
ERROR
TOTAL
1 16.608
263 890.42
264 907.03
16.608
3.3856
MULT R= .13532 R-SQR= .O1831 SE= 1.8400
SIGNIF
.0276
VARIABLE
CONSTANT
5.FECO
PARTIAL
COEFF
STD ERROR T-STAT
.20350
-.13532 -.27963
.12749
.12625
1.5962
-2.2149
SIGNIF
.1116
.0276
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 64.DR.JOSE N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SOR F-STAT
6.1497
REGRESSION
ERROR
TOTAL
1
263
264
.44267 -1
1.8932
1.9374
.44267 -1
.71983 -2
MULT R= .15116 R-SOR= .02285 SE= .84843 -1
SIGNIF
.0138
VARIABLE
CONSTANT
5.FECO
PARTIAL
COEFF
STD ERROR
T-STAT
1.0123 .58788 -2 172.20
-.15116 -.14437 -1 .58216 -2 -2.4799
SIGNIF
0.
.O13B
LEAST SQUARES REGRESSION
a-.
I
NJ
--J
ANALYSIS OF VARIANCE OF 65.UR.DILL N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1
263
264
. 16692 -1
1.6644
1.6811
.16692 -1 2.6377
.63285 -2
MULT R= .09965 R-SQR= .00993 SE= .79552 -1
VARIABLE
PARTIAL
COEFF
STD ERROR T-STAT
CONSTANT 1.0086 .55122 -2 182.98
5.FECO -.09965 -.88652 -2 .54586 -2 -1.6241
SIGNIF
.1056
SIGNIF
0.
. 1056
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 66.UR.J.P. N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1
263
264
.30796 -1
1.8091
1.8399
.30796 -1 4.4770
.68787 -2
MULT R= .12937 R-SQR= .O1674 SE= .82938 -1
SIGNIF
.0353
VARIABLE
CONSTANT
5.FECO
PARTIAL COEFF STD ERROR T-STAT SIGNIF
1.0096 .57468 -2 175.69 0.
-.12937 -.12041 -1 .569O9 -2 -2.1159 .0353
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 71.HD.JOSE N= 265 OUT OF 571
SOURCE OF SUM SORS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1 43.878
263 2175.2
264 2219.1
43.878
8.2706
MULT R= .14062 R-SQR= .01977 SE= 2.8759
5.3053
SIGNIF
.0220
VARIABLE
CONSTANT
5.FECO
PARTIAL
COEFF
STD ERROR T-STAT
.45415
-.14O62 -.45452
.19927
.19733
2.2791
-2.3033
SIGNIF
.0235
.0220
LEAST SQUARES REGRESSION
Ov
00
ANALYSIS OF VARIANCE OF 72.HD.DILL N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
3.0506
REGRESSION
ERROR
TOTAL
1 23.446
263 2021.4
264 2O44.8
23.446
7.6858
MULT R= .10708 R-SQR= .011,47 SE= 2.7723
VARIABLE
CONSTANT
5.FECO
PARTIAL
COEFF
.16046
-.1O7O8 -.33225
STD ERROR
.19210
.19023
T-STAT
.83532
-1.7466
SIGNIF
.0819
SIGNIF
.4O43
.0819
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 73.HD.J.P. N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
2.4492
REGRESSION
ERROR
TOTAL
1 20.647
263 2217.1
264 2237.7
20.647
8.4300
MULT R= .09605 R-SQR= .00923 SE= 2.9035
SIGNIF
. 1188
VARIABLE
CONSTANT
5.FECO
PARTIAL
COEFF
. 19256
-.09605 -.31178
STD ERROR
.20118
.19922
T-STAT
.95715
-1.565O
SIGNIF
.3394
. 1188
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 74.HR.JOSE N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SOR F-STAT
REGRESSION
ERROR
TOTAL
1 .36739 -1
263 2.1176
264 2.1543
.36739 -1 4.5629
.80517 -2
MULT R= .13059 R-SOR= .01705 SE= .89731 -1
SIGNIF
.0336
VARIABLE
CONSTANT
5.FECO
PARTIAL
COEFF
STD ERROR T-STAT
1.0133 .62175 -2 162.97
-.13059 -.13152 -1 .61570 -2 -2.1361
SIGNIF
0.
.0336
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 75.HR.DILL N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1 . 18341 -1
263 2.O531
264 2.0715
.18341 -1 2.3494
.78065 -2
REGRESSION
ERROR
TOTAL
1 .11 194 -1
263 2.2805
264 2.2917
.11194 -1
.86711 -2
SIGNIF
. 1265
MULT R= .0941O R-SQR= .00885 SE= .88354 -1
VARIABLE PARTIAL COEFF STD ERROR T-STAT SIGNIF
CONSTANT 1.0047 .61221 -2 164.11 0.
5.FECO -.O9410 -.92926 -2 .60626 -2 -1.5328 .1265
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 76.HR.J.P. N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
1 .291O
SIGNIF
.2569
MULT R= .06989 R-SQR= .00488 SE= .93119 -1
VARIABLE
PARTIAL
COEFF
STD ERROR
T-STAT
CONSTANT 1.0067 .64522 -2 156.03
5.FECO -.06989 -.72599 -2 .63895 -2 -1.1362
SIGNIF
0.
.2569
-------
LEAST SQUARES REGRESSION
,
-I
LO
O
SOURCE
REGRESSION
ERROR
TOTAL
MULT R= . 15807
DF
1
263
264
R-SQR=
VARIABLE PARTIAL
CONSTANT
5.FECO
15807
SUM SQRS
28.200 .
110O.4
1128.6
.02499 SE=
MEAN SQR
28 . 200
4. 1841
2.0455
COEFF STD ERROR
.25751
- .36438
. 14173
. 14036
F-STAT
6.7398
T-STAT
1 .8168
-2.5961
SIGNIF
.O10O
SIGNIF
.0704
.0100
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE
SOURCE
REGRESSION
ERROR
TOTAL
MULT R= . 10563
OF 82.
DF
1
263
264
R-SQR=
VARIABLE PARTIAL
CONSTANT
5.FECO
1O563
CD. DILL N=
SUM SQRS
10.559
935.72
946.28
.011 16 SE=
COEFF
. 1 1016
-.22296
265 OUT OF
MEAN SQR
10.559
3.5579
1 .8862
STD ERROR
. 13070
. 12943
571
F-STAT
2.9677
T-STAT
.84287
-1 .7227
SIGNIF
.0861
SIGNIF
.4001
.0861
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE
SOURCE
REGRESSION
ERROR
TOTAL
MULT R= . 10944
OF 83.
DF
1
263
264
R-SQR=
VARIABLE PARTIAL
CONSTANT
5.FECO
10944
CD.J.P. N=
SUM SQRS
13.310
1098.0
1111.3
.01198 SE=
COEFF
. 12102
- .25033
265 OUT OF
MEAN SQR
13.31O
4. 1750
2.0433
STD ERROR
. 14158
. 14020
571
F-STAT
3. 188CT
T-STAT
.85480
-1 .7855
SIGNIF
.0753
SIGNIF
.3934
.0753
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 84.CR.JOSE N= 265 OUT OF 571
SOURCE DF SUM SORS MEAN SOR F-STAT
REGRESSION
ERROR
TOTAL
1
263
264
.36751 -1
1.6061
1.6428
.36751 -1 6.0182
.61067 -2
MULT R= .14957 R-SQR= .O2237 SE= .78145 -1
SIGNIF
.0148
VARIABLE
CONSTANT
5.FECO
PARTIAL
COEFF
STD ERROR T-STAT
1.0093 .54147 -2 186.40
-.14957 -.13154 -1 .5362O -2 -2.4532
SIGNIF
0.
.0148
LEAST SQUARES REGRESSION
I
OJ
ANALYSIS OF VARIANCE OF 85.CR.DILL N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1
263
264
. 11913 -1
1.4827
1.4946
.11913 -1 2.1131
.56376 -2
MULT R= .08928 R-SQR= .00797 SE= .75084 -1
SIGNIF
. 1472
VARIABLE
PARTIAL
COEFF
STD ERROR
T-STAT
CONSTANT 1.OO49 .52O26 -2 193.15
5.FECO -.08928 -.74891 -2 .51520 -2 -1.4536
SIGNIF
0.
. 1472
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 86.CR.J.P. N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1
263
264
.13848 -1
1.6474
1.6613
.13848 -1 2.2107
.62639 -2
MULT R= .09130 R-SQR= .00834 SE= .79145 -1
SIGNIF
. 1383
VARIABLE
PARTIAL
COEFF
STD ERROR
T-STAT
CONSTANT 1.0056 .54840-2 183.37
5.FECO -.09130 -.80745 -2 .54306 -2 -1.4868
SIGNIF
0.
. 1383
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 61.UD.JOSE N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SOR F-STAT
REGRESSION
ERROR
TOTAL
1 21.292
263 992.99
264 1014.3
21 .292
3.7756
MULT R= .14489 R-SQR= .02099 SE= 1.9431
5.6394
SIGNIF
.0183
VARIABLE
CONSTANT
6.FENOX
PARTIAL
COEFF
STD ERROR T-STAT
-.19055
.14489 .77445
.18865
.32612
-1.0100
2.3747
SIGNIF
.3134
.0183
LEAST SQUARES REGRESSION
I
UJ
Ni
ANALYSIS OF VARIANCE OF 62.UD.DILL N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
1.4278
REGRESSION
ERROR
TOTAL
1 4.4607
263 821.65
264 826.11
4.4607
3.1242
MULT R= .07348 R-SQR= .00540 SE= 1.7675
SIGNIF
.2332
VARIABLE
CONSTANT
6.FENOX
PARTIAL
.07348
COEFF
-.65472 -1
.35448
STD ERROR T-STAT
.17161
.29666
-.38152
1.1949
SIGNIF
.7031
.2332
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 63.UD.J.P. N=
SOURCE DF SUM SQRS
REGRESSION
ERROR
TOTAL
1 5.6669
263 901.36
264 907.03
265 OUT OF 571
MEAN SQR F-STAT SIGNIF
5.6669 1.6535 .1996
3.4272
MULT R= .07904 R-SQR= .00625 SE= 1.8513
VARIABLE
CONSTANT
6.FENOX
PARTIAL
COEFF
-.10611
.07904 .39954
STD ERROR
.17974
.31071
T-STAT
-.59036
1.2859
SIGNIF
.5555
. 1996
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 64.UR.JOSE N= 265 OUT OF 571
I
LO
OJ
SOURCE
OF
REGRESSION 1
ERROR 263
TOTAL 264
MULT R = .
VARIABLE
CONSTANT
6.FENOX
LEAST SQUARES
12174 R-SQR=
PARTIAL
. 12174
REGRESSION
ANALYSIS OF VARIANCE OF 65.
SOURCE
DF
REGRESSION 1
ERROR 263
TOTAL 264
MULT R=
VARIABLE
CONSTANT
6.FENOX
LEAST SQUARES
.03043 R-SQR=
PARTIAL
.03043
REGRESSION
ANALYSIS OF VARIANCE OF 66.
SOURCE
DF
REGRESSION 1
ERROR 263
TOTAL 264
MULT R= .
VARIABLE
CONSTANT
6.FENOX
.04459 R-SQR=
PARTIAL
.04459
SUM SORS
.28712 -1
1 .9087
1 .9374
.O1482 SE=
COEFF
.99286
.28439 -1
UR.DILL N=
SUM SQRS
. 15571 -2
1 .6795
1 .6811
.00093 SE=
COEFF
1 .0015
.66229 -2
UR.d.P. N=
SUM SQRS
.36590 -2
1 .8362
1 .8399
.00199 SE=
COEFF
.99947
.10152 -1
MEAN SQR
.28712 -1
.72575 -2
.85191 -1
STO ERROR
.82711 -2
. 14298 -1
265 OUT OF
MEAN SQR
. 15571 -2
.63860 -2
.79913 -1
STD ERROR
.77586 -2
. 13412 -1
265 OUT OF
MEAN SQR
.36590 -2
.69819 -2
.83558 -1
STD ERROR
.81126 -2
. 14024 -1
F-STAT
3.9562
T-STAT
120.04
1 .989O
571
F-STAT
.24383
T-STAT
129.08
.49379
571
F-STAT
.52407
T-STAT
123.20
.72393
SIGNIF
.0477
SIGNIF
O.
.O477
SIGNIF
.6219
SIGNIF
0.
.6219
SIGNIF
.4698
SIGNIF
0.
.4698
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 71.HD.JOSE N= 265 OUT OF 571
SOURCE OF SUM SQRS MEAN SOR F-STAT
2.6805
REGRESSION
ERROR
TOTAL
1 22.388
263 2196.7
264 2219.1
22.388
8.3524
MULT R= .10044 R-SOR= .O1009 SE= 2.8900
SIGNIF
. 1028
VARIABLE
CONSTANT
6.FENOX
PARTIAL
COEFF
STD ERROR T-STAT
-.11393
.10O44 .79414
.28059
.48505
-.40604
1.6372
SIGNIF
. 685O
. 1028
LEAST SQUARES REGRESSION
I
U>
-P-
ANALYSIS OF VARIANCE OF 72.HD.DILL N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
1.0692
REGRESSION
ERROR
TOTAL
1 8.2792
263 2036.5
264 2044.8
8.2792
7.7435
MULT R= .06363 R-SQR= .O0405 SE= 2.7827
SIGNIF
.3021
VARIABLE
CONSTANT
6.FENOX
PARTIAL
COEFF
STD ERROR T-STAT
-.21109
.06363 .48293
.27017
.46704
-.78131
1.0340
SIGNIF
.4353
.3021
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 73.HD.J.P. N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
2.0487
REGRESSION
ERROR
TOTAL
1 17.296
263 2220.4
264 2237.7
17.296
8.4428
MULT R= .O8792 R-SQR= .O0773 SE= 2.9056
SIGNIF
. 1535
VARIABLE
CONSTANT
6.FENOX
PARTIAL
.08792
COEFF
- .26578
.69801
STD ERROR T-STAT
.28211
.48767
-.94212
1.4313
SIGNIF
.347O
. 1535
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 74.HR.JOSE N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1 .21243 -1
263 2.1331
264 2.1543
.21243 -1 2.6192
.81106 -2
MULT R= .09930 R-SQR= .OO986 SE= .90059 -1
SIGNIF
. 1068
VARIABLE
CONSTANT
6.FENOX
PARTIAL
.09930
COEFF
STD ERROR T-STAT
.99617 .87437 -2 113.93
.24462 -1 .15115 -1 1.6184
SIGNIF
O.
. 1068
LEAST SQUARES REGRESSION
(^
U>
ANALYSIS OF VARIANCE OF 75.HR.DILL N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
.75167
REGRESSION
ERROR
TOTAL
1
263
264
.59035 -2
2.0655
2.O715
.59035 -2
.78538 -2
MULT R= .05338 R-SQR= .OO285 SE= .88622 -1
SIGNIF
.3867
VARIABLE PARTIAL COEFF STD ERROR T-STAT SIGNIF
CONSTANT .99456 .86O42 -2 115.59 O.
6.FENOX .05338 .12896 -1 .14874 -1 .86699 .3867
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 76.HR.d.P. N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
1.4430
REGRESSION
ERROR
TOTAL
1 .. 12505 -1
263 2.2792
264 2.2917
.12505 -1
.86661 -2
SIGNIF
.2307
MULT R= .07387 R-SQR= .00546 SE= .93092 -1
VARIABLE
CONSTANT
6.FENOX
PARTIAL
.07387
COEFF
STD ERROR T-STAT
SIGNIF
.99492 .90382 -2 110.08 0.
.18769 -1 .15624 -1 1.2013 .2307
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 81.CD.JOSE N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1
263
264
20. 133
1108.5
1 128.6
2O. 133
4.2148
MULT R= .13356 R-SQR= .01784 SE= 2.O530
4.7768
SIGNIF
.0297
VARIABLE
CONSTANT
6.FENOX
PARTIAL COEFF
-.25007
STD ERROR T-STAT
. 13356
.75308
. 19932
.34457
-1.2546
2.1856
SIGNIF
.2107
.0297
LEAST SQUARES REGRESSION
I
OJ
ANALYSIS OF VARIANCE OF 82.CD.DILL N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1 2.8629
263 943.42
264 946.28
2.8629
3.5871
MULT R= .05500 R-SQR= .00303 SE= 1.8940
.79811
SIGNIF
.3725
VARIABLE
CONSTANT
6.FENOX
PARTIAL
COEFF
-. 12121
.05500 .28398
STD ERROR T-STAT
. 18388
.31788
-.65918
.89337
SIGNIF
.5104
.3725
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 83.CD.J.P. N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
1.003O
REGRESSION
ERROR
TOTAL
1 4.2221
263 11O7.1
264 1111.3
4.2221
4 . 2095
MULT R= .06164 R-SQR= .OO380 SE= 2.0517
SIGNIF
.3175
VARIABLE
CONSTANT
6.FENOX
PARTIAL
.06164
COEFF
-.15041
.34487
STD ERROR
.19920
.34435
T-STAT
-.75508
1.0015
SIGNIF
.4509
.3175
-------
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 84.CR.dOSE N= 265 OUT OF 571
SOURCE OF SUM SORS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1 .22586 -1 .22586 -1 3.6662
263 1.6202 .61606 -2
264 1.6428
MULT R= .11725 R-SQR= .01375 SE= .78489 -1
SIGNIF
.O566
VARIABLE
CONSTANT
6.FENOX
PARTIAL
.11725
COEFF
.99186
.25223 -1
STD ERROR T-STAT
.76205 -2
.13173 -1
130.16
1.9147
SIGNIF
0.
.0566
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 85.CR.DILL N= 265 OUT OF 571
SOURCE OF SUM SQRS MEAN SQR F-STAT SIGNIF
.81606 -1 .7754
REGRESSION
ERROR
TOTAL
1
263
264
.46362 -3
1 .4941
1.4946
.46362 -3
.56812 -2
MULT R= .01761 R-SQR= .OOO31 SE= .75374 -1
VARIABLE
CONSTANT
6.FENOX
PARTIAL
COEFF
STD ERROR T-STAT
SIGNIF
.99977 .73179 -2 136.62 0.
.01761 .36138 -2 .12650 -1 .28567 .7754
LEAST SQUARES REGRESSION
ANALYSIS OF VARIANCE OF 86.CR.J.P. N= 265 OUT OF 571
SOURCE DF SUM SQRS MEAN SQR F-STAT
REGRESSION
ERROR
TOTAL
1
263
264
. 11715 -2
1.6601
1.6613
.11715 -2 .18560
.63121 -2
MULT R= .02656 R-SQR= .00071 SE= .79449 -1
SIGNIF
.6670
VARIABLE
CONSTANT
6.FENOX
PARTIAL
.02656
COEFF
STD ERROR T-STAT
SIGNIF
.99923 .77136 -2 129.54 0.
.57446 -2 .13334 -1 .43081 .6670
-------
Appendix 7
Plots of Emissions and Fuel Economy
from Special Engines over Time
-------
Two variable plots of the ratio of the Combined Fuel Economy of
each year divided by the Combined Fuel Economy of the
corresponding base year (usually 1979) versus each of FTPHC,
FTPCO, FTPNOX, and FENOX.
Details of the data used may be found in Tables IV. F-l through
IV. F-5 in Section IV. F.
7-1
-------
FDRD'5 2.3 LITER ENGINE OVER TIME
—i
i
1.20
LIB
m
h
w
Q.
x
\
m
Q.
l.BB
B.H0
B.BB
A URBHN MPE CRUTD TRRNB>
CD HIEHWRY MPE CRUTD TRRN5>
^ CDMHINED MPB CRUTD TRRNS>
YE
B
BI.B
-------
FDRD'5 2.3 LITER ENEINE DVER TIME
i
U)
I .20
I .IB
m
r>
m
CL
z
\
m
a.
I .00
0.30
0.B0
A URBRN MPB CMRN TRHN5>
ID HIBHWRY MPE CMRN TRRN5>
^ CDMBINEI> MPE CMHN TRflN5>
MI?C>EL YE
0
BI.0
-------
COMBINED MPG VS. URBRN HC
FDRD'5 2.3 LITER ENBINE OVER TIME
LIB
01
h
IH
Q.
1.00
\
in
a,
z
0.30
0.00
B-'BHC B'20C
A NDN DETERIDRRTED CRUTD TRRN3
Q DETERIDRHTED CRUTQ TRRN3
0.H0
0.50
-------
COMBINED MPE V5. U
FDRD'5 2.3 LITER ENEI
vj
2R
N
DV
HC
ER
T I ME
I.IB
m
[L
z
\
IB
DL
Z
1.00
0.30
0.00
0.20
A NDN DETER IORRTCD CMRN TRHN3
Q DETERJ DRHTED- C MRN TRRN 3
I.BB
1.20
-------
COMBINED MPE V5. URBRN CD
FDRt>'5 2.3 LITER ENGINE DVER TIME
I.IB
m
CL
z
\
LD
EL
1.00
0.30
A NDN DETER I DRRTEI> C RLJTD TRRN3
El DETERIDRRTED C RUTD TRHN 3
0.0 1.0 2.0
B.0 3.0 10.0
-------
COMBINED MPG V5. URBRN CD
FDRD'B 2.3 LITER ENGINE DVER TIME
—i
i
HI
h
ID
CL
\
LU
Q.
Z
I. IB
I.BH
0.30
A NDN DETERIDRRTED CMRN TRRN3
0 DETER IDRRTED CMHN TRRN3
0.00
1.00
CD
H.00
S. 00
-------
COMBINED MPG V5. URBRN NDX
FDRD'5 2.3 LITER ENGINE DVER TIME
i
CO
I.IB
m
m
DL
z
\
m
a.
z
1.00
0.90
0.00 B.2B
A NDN DETERIDRRTED C RLJTD TRRN
ID DETERIDRRTED C RUTD TRRN 3
VB.BB JIJHB J-BB I.2B I .HB I.EB
-------
COMBINED MPE V5. URBRN NDX
FDRD'5 2.3 LITER ENGINE DVER TIME
i
VD
m
h
in
DL
51
\
in
EL
1.10
1.00
0.30
A NDN DETER IDRRTED CMRN TRRN3
El DETERIDRRTED C MRN TRRN 3
0.0B 0.20
X
0.G0
I.H0 I.E0
-------
CDMBINED'MPE V5. HIGHWRY NDX
FORD'S 2.3 LITER ENEINE OVER TIME
—i
i
m
h
ID
D.
Z
\
m
a.
z
LIB
1.00
0.30
RUTDMHT1C TRRN5MI 551DN
MRNUHL TRRN5MI551ON
0.00
0.50
NDX
1.00 1.50
C B/M 1 >
2.00
2.50
-------
CHRYSLER'S 1.7 LITER ENGINE OVER TIME
1.20
LJ
Ul
[E 1.10
tfl
I
ID
0.
\ 1.00
ID
EL
Z
0.90
A URBRN MPB
Q HIBHWRY MPB
0 COMBINED MPB
0
BI.0
-------
CHRYSLER'S 1.7 LITER ENGINE OVER TIME
1.20
LJ
in
c
to
i
LIB
Q.
E I.BS
\
ID
Q.
Z I
H.3S
A URBRN MPB CMRN. TRRN>
H HIBHMRY MPB (MRN. TRRN>
0 COMBINED MPB
BI.H
-------
CHYR5LER'5 1.7 LITER ENGINE DVER TIME
1.15
u
Ul
IE
m
i
m
Q.
E
\
HI
Q.
E
I.IB
1.05
1.00
B.95
A URBRN MPE CMHN. TRRN3
^ COMBINED MPE CMRN.CDCO TRRN
Mdt>HEL YE
BI.B
-------
CDMB
CHRYSLER'5
NED MPB VS. URBRN HC
1.7 LITER ENEINE OVER T I ME
I
M
*-
I.IB
01
EL
z
\
in
Q.
51
1.00
0.90
0.00
A NDN DETER IDRRTED CRUTD TRRN3
Q DETERIQRRTED CRUTO TRRN3
0.10 0.20
HC C E/M I >
0.30
0.H0
-------
CDMB
CHRYSLER'5
NED MPE VS. URBRN HC
1.7 LITER ENEINE OVER TIME
i.is
a
a:
to
i
LD
0.
z
\
If]
Q_
£
-
a.as
NON OCTERIDRRTCD CMRN. TRRN3
B DETER IDRRTED CMRN. TRRN3
+ NON DET. CMRN. TRRN3
X DET. CMRN.Cai>> TRRN 3
E.BB
B. IB,
HC
B.MB
B.SB
-------
CDMB
CHRYSLER'5
NED MPE VS. URBRN CD
1.7 LITER ENGINE OVER
T I ME
I.IB
Dl
h
ID
D.
X
\
ID
H
Z
I.BB
B.9B
A NDN DETER IDRRTED CHUTO TRRN3
Q DETER 1DRRTED CRUTD TRRN3
B.BB
B.BB
B.BB
-------
COMBINED MPE VS. URBRN CD
LITER ENEINE OVER TIME
CHRYSLER'5 1.7
1.20
i
M
—I
LJ
Ul
01 I.IB
ffl
1
LQ
D.
\ I.BB
ID
Q.
Z
0.50
A NDN DETER IORRTED CMRN. TRHN3
Q DETER IDRRTED CMHN. TRRN3
+ NDN DET. CMHN.COD) TRRN3
X DET. CMRN.
TRRN3
0.0
2.0
IB.B
12.0
IH.0.
-------
COMBINED MPG VS. URBflN NDX
•CHRYSLER'5 1.7 LITER ENGINE OVER
T I ME
I.IB
i
M
oo
01
h
LD
Q.
E
\
ID
CL
E
I.BB
B.9B
A NUN DETERIORATED C RUTd TORN 3
B DETER I PRRTEI> CRUTQ TRRN 3
B.BB B.2B B
I.H0 I.EB I.BB
-------
COMBINED MPE VS. LJREflN NDX
CHRYSLER'S 1.7 LITER ENGINE DVER
T I ME
1.20
LJ
in
a:
CD
I
in
[L
z
\
in
IL
z
LIB
1.00
B.90
A NDN DETER IDRRTED C MRN. TRRN3
H £>ETER I DRRTED C MRN . TRRN 3
+ NON DCT. CMRN.
TRRN3
X DET. CMHN.
1.75 2.00
-------
COMBINED
CHRY5LER'5 I
MPE VS.
7 LITER
HIGHWRY NDX
ENE I NE DVER T I ME
1.20
y
in
CE LIB
i
Q.
2:
\ I.BB
in
0.
B.9B
RUTDMRTIC TRRN5MI 55IDN
MRNURU TRRNEMISEIDN
MRNURU
TRRNSMI 55 I ON
B.BB
H.BB
5.BB
-------
EMf5 I.E LITER ENGINE OVER TIME
1.3B
1.20
01
r*
ID »•«•
[L
z
\
in
Q.
z
B.S0
I.HH
B.BB
A URHRN MPE CRUTO TRRN>
Q HIBHHRY MPE CRUTD TRRN>
• COMBINED MPE
B
BI.B
-------
EM'S I.E LITER ENGINE DVER TIME
LIB
i
t-o
CT I.BB
LD
EL
\
ID
Q. B<9B
z
B.BB
LJRBHN MPB
HIBHMMY MPB CMHNURL TRRN>
COMB I NED MPB
-------
EM
COMBINED MPE V5. URBRN HC
5 I.E LITER ENGINE OVER T I ME
1.20
i
N3
U)
01
I. IB
in
[L
z I-BB
\
ID
EL
^- 0.30
H.BB
A NDN DCT. CRUTP TRRhOSTD ENB3
Q DCT. CHUTO TRHN;STI> ENB 3
+ NON DCT. CRUTP TRRhO HO ENB3
X DET. CRUTD TRRN/ HO ENB3
0.00
0,
0.B0
1.00
-------
COMBINED MPE VS. URBRN HC
S I.E LITER ENGINE DVER TIME
LIB
—i
K>
01
h
m
n
£
\
in
Q.
I.BB
B.HB
B.BB
A NON DET. CMRN TRHN^ BTD ENB 3
13 DET. CMRN TRRFOETD ENB 3
+ NDN DE.T. CMF1N TRRN^ HD ENE 3
X DET. CMHN TRRN^ HD ENB3
B.BB
B.EB
B.BB
-------
COMBINED MPE VS. LJRBRN CD
EM'5 I.B LITER ENEINE DVER TIME
1.20
I.IB
I
to
01
h
IU
Q.
Z
\
I.BB
D.
2- B.HB
H.BH
A NDN DCT. CHUTO TRRN/BTD EN IS 3
El DCT. CRUTO TRHNsBTD CNB D
+ NON PET. CRUTO TRRN^ HD ENB3
X DCT. CRUTD TRRN/ HO ENB3
B.B 2.B
IB.B
12. B
IH.B
-------
COMBINED MPB VS. URBRN CD
EM«5 I.E LITER ENEINE OVER TIME
I.IB
I
N)
01
|\
ID
EL
£
\
LD
D.
B.BB
B.B
NDN DE:T. CMRN TRRN^STD ENBI
H J>CT. C MHN TRHN^ STD CNQ 3
2'B
CD
B.B
IB.B
-------
COMBINED MPE VS. URBRN CD
EM'5 I.B LITER ENGINE OVER TIME
1.15
01
r*
13
IL
n
\
ID
DL
z
l.BB
B.B5
B.B
A NDN DET. CMF1N TRRN; HO ENQ3
El OCT. CMHN TRRN^ HO CNB3
B.B
1B.B
-------
CDMB NED MPG V5. URBRN NDX
EM '5 I.ELI TER ENE I NE DVER T I ME
I .30
i
hO
co
m
r-
LU
a.
z
\
LU
[L
Z
i. is
i .00
0.BS
0.70
A
\
A NDN DET.
Q DET. CRUTD TRRN^ 5TI> ENB>
+ NDN DET.
X DET. CRUTD TRRN; HD ENB>
0.00
0.S0
NDX
i .00
C EXM
I .50
2.00
-------
CDMB NED MPE-V5. URBRN NDX
EM ' 5 I- . E LI TER ENE I NE DVER T 1 ME
! . IS
I
ro
m
CL
z:
\
ID
[L
X
1.00
0.BS
0.70
A NDN DET. CMRN TRRN^STD ENB3
ED DET. C URN TRRN^STD ENB 3
+ NDN DET. CMRN TRRN^ HD ENB3
X DET. CMRN TRRN; HD ENS 3
0 . S0
NOX
I ,00
C feXM
I . S0
2.00
-------
CDMB
EM ' 5 I
NED MPG VS. H EHNRY NDX
. E LI TER ENE I NE OVER T I ME
I .20.
1
u>
o
Dl
I . IB
I .00
ID
CL
Z
\
LD
Q.
^- 0.30
0.B0
RUTD
RUTD
TRRNS .
TIRRNS.
< STD ENB . >
C HP ENE . >
0.00
"sfftx
I .50
2.00
-------
CDMB
EM ' 5 I
NED MPE VS. H GHWRY NDX
. B LI TER ENE I NE OVER T I ME
I .20
1 . 10
m
m
\L
n
\
LU
CL
I .00
0.a0
A MHNURL TRRNS
El MRNURL TRRNS
-------
EM'5 2.S L FWD ENGINE OVER TIME
LO
N>
m
ID
0.
£
\
ID
CL
1.20
1.15
I.IB
1.0S
I.0H
0.B5
0.30
B0.B
MDDEL YERR
A URBRN MPH CRUTD. TRRN5
Q HIBHWRY MPB
<> CDMB1NCD MPB
BI.0
-------
EM'S 2.5 L FND ENGINE OVER TIME
i
U)
OJ
m
ID
a.
\
in
Q.
1.15
1.10
I.B5
I.HH
B.35
0.90
H.B5
A LJRBRN MPB CMRN. TRRN>
tD HIGHMRY MPB
<> COMBINED MPE
BB.B
MODEL YERR
BI.B
-------
COMBINED MPE VS. URBRN HC
EM'5 2.5 L FWD ENBINE OVER TIME
m
m
z
\
LD
CL
Z
1.05
I.B0
0.95
0.SB
B.BS
A MDN-DETER. CRUTD TRHN>
LD DETER IDRRTCD CRUTD TRRN>
B.BB B.BK B.IB 0.15 B.2B 0.25 D.3D 0.35 B.HD
HC CE/MI
-------
COMBINED MPE V5. URBRN HC
GM ' 5 2.S L FWD ENGINE OVER TIME
H
m
13
\
13
CL
£
I.0S
1.00
0.30
B.BS
A NDN-DETER.
tH DETER I DRRTEJ> CMRN TRRN>
B.0B
B.CK
0.10
H.IS
0.20
B.25
0.3H
a.as
0.40
HC C E/M I
-------
COMBINED MPE V5. URBRN CD
BM'5 2.5 L FWD ENEINE OVER TIME
H
ED
in
1 CL
£
\
a.
LO
1.0S
1.00
D.3S
0.90
0.BS
NC1N-DETER. COUTD TRRN>
DE:TCR i DRRTED CRUTD TRRN>
H.D0 o.sn
1.00 \.SX 2.00 2.SO
CD < E/M I >
3.00 3.SB H.00
-------
COMBINED MPE VS. LJRBRN CD
EM • 5 2.5 L FWI> ENGINE OVER TIME
r
H
ra
IE
o.
n
\
in
n
z:
I.US
1.00
0.35
0.30
0.BS
A NON-DETER.
Q DETERIORRTCD CMRN TRRN)
0.00 0.5B
1.00
1.50 2.00 2.50 3.00 3.50 H.00 M.50 5.00
CD
-------
COMBINED MPG Y5. URBRN NDX
EM'S 2.5 L FWI> ENE I NE OVER TIME
t.us
CO.
oo'
s
ra
ID
u
>
sV"
\
LD
DL
Z
t.BH
B.as
B.BB
B.BS
—Q
A HIJTDMRTIC TRRNSMISQIDN
E) MHNUHL TRRNSMIBSIDN
DF*S RRE i.0>
B.BB 0.20 B.MB B.EB B.BB I.BB
Nnx c B/Mi >
t.MB I.EH
-------
COMBINED MPE.YB. HIGHWRY NDX
EM'5 2.5 L FWD ENGINE OVER TIME
s
ra
13
£CL
\
13
DL
1.0S
f .00
0.9S
H.H0
0.BS
Q -••
A RUTDMRTIC TRHNSMISSION
Q MRNURL TRRNSMI SB ION
H.BH
a.an 0.B0 H.SH
NDX CE/M1>
1.20
l.SB
-------
EM'S 3.H LITER ENGINE DVER TIME
i
4>
o
m
1.20
1.10
ID
Q.
Z
\
ID
Q. I'00
Z
0.30
A LJRBIRN MPE
<> COMBINED MPB < HUTQ TRRN5>
7H.0
MODEL
B0.0
YERR
BI.0
-------
EM ' 5 3
LITER ENGINE OVER TIME
!.
m
h
LD
EL
Z
\
LD
H
Z
1.00
0.90
LJRBRN MPB C MRNURL TRRN5 3
HIEHWRY MPB CMRN TRRNS3
COMBINED MPB CMRN TRRNS3
0.B0
79.0
MDDEL
B0.0
YERR
BI.0
-------
COMBINED MPE VS. URBRN HC
EM'S 3.B
_ I TE
R ENEINE
DVER T I ME
I .20
" '•'-
in
CL
Z
\
CL I'M
Z
E.3B
H.00
A NDN DETER IDRRTED CHUTO TRHNS3
Q DETER IDRRTED CRUTD TRRN53
0.20
HC
0.M0
0.E0
C E/M I
0.H0
t.BB
-------
cm
EM ' 5 3
IB IN
1 . B
ED
L 1 T
j
PE V5.
R ENE 1 N
JR
BR
DV
N'HC
ER TIME
I
4>
OJ
I. IB
m
m
DL
z
\
ID
D.
Z
0.90
0.B0
0.00
A NQN D-ETER I DRRTED- C MRN TRRN 3
El DETER 1ORHTED C MHN TRRN 3
0.B0
1.00
-------
COMBINED MPE V5. URBRN CD
L I TER ENE I NE DVER T I ME
EM'5 3.B
m
LD
Q.
£
\
LD
EL
£
I.2B
LIB
1.00
H.90
0.00
NPN DETER IDRRTED CRUTD TRRN3
OETERIORHTED CRUTD TRRN3
1.00
M I V
£.00
B.00
-------
COMBINED MPE VS. URBRN CD
EM'S. 3 . B
LITER ENEINE OVER TIME
1 .IB
01
I .BE!
i
i>
Ul
in
Q-
z
\
LU
[L B'30
Z
0.B0
0.00 1.00 2
NON £>ETER I nRHTEJ> C MRN TRHN
DETERIDRHTED CMHN TRHN3
E.00 7.00 B.00
-------
CDMB NED MPE V5. URBRN NDX
EM'S 3.B LITER ENGINE OVER TIME
I.2B r.
OT ,.,0
CL
\
IJ]
H '•»
z
0.30
A NDN DETER inRRTED C RUTD TRRI>
ID DETERIDRRTED C RUTQ TRRN 3
NPN DETERIDRRTED CMRN TRNRr
DETER 1ORRTED C MRN TRRN 3
H.00
0.S0
NDX
1. 00
I.S0
C E/M I
-------
COMB NED-MPE V5-. HIEHNRY NOX
EM ' 5 3.B LITER ENEINE DVER TIME
-o
I
m
h
m
n
•z
\
LU
[L
Z
1 .20
1.10
I.0B
0.90
0.B0
A HUTDMHTIC TRRNSMlSSinN
CD MHNUHL TRRNSMI5SIDN
0.00 0.20 0.
1.20 I.H0
I.E0
-------
Appendix 8
Multicollinearity Discussion
-------
Appendix 8 - Multicollinearity Discussion
A. Introduction
This appendix discusses the regression equations presented in Section IV and
the impact that multicollinearity had on these equations.
Interpretation of the multiple regression equation depends
implicitly on the assumption that the explanatory variables are
not strongly interrelated. It is usual to interpret a
regression coefficient as measuring the change in the response
variable when the corresponding explanatory variable is
increased by one unit and all other explanatory variables are
held constant. This interpretation may not be valid if there
are strong linear relationships among the explanatory
variables. It is always conceptually possible to increase the
value of one variable in an estimated regression equation while
holding the others constant. However, there may be no
information about the result of such a manipulation in the
estimation data. Moreover, it may be impossible to change one
variable while holding all others constant in the process being
studied. When these conditions exist, simple interpretation of
the regression coefficient as a marginal effect is lost.
When there is a complete absence of linear relationship among
the explanatory variables, they are said to be orthogonal. In
most regression applications the explanatory variables are not
orthogonal. Usually the lack of orthogonality is not serious
enough to affect the analysis. However, in some situations the
explanatory variables are so strongly interrelated that the
regression results are ambiguous.
The condition of severe nonorthogonality is also referred to as
the problem of collinear data, or multicollinearity. It is
recommended that one should be extremely cautious about any and
all substantive conclusions based on a regression analysis in
the presence of multicollinearity.*
In order to address the concerns discussed above, the three regression
equations, Cheng, Murrell and Bascunana, (see Section IV. D) were investigated
for the presence of multicollinearity.
Multicollinearity is a question of degree and not of kind. The
meaningful distinction is not between the presence and the
absence of multicollinearity, but between its various degrees.**
* Chatterjee and Price, Regression Analysis By Example, John Wiley & Sons,
1977, pages 143 and 144.
** Jan Kmenta, Elements of Econometrics, Macmillan Publishing Co., 1971, p.
380.
8-1
-------
Given that some multicollinearity almost always exists, the
question is, At what point does the degree of
multicollinearity cease to be "normal" and become
"harmful"? This question has not been satisfactorily
resolved. According . to one criterion sometimes used in
practice, multicollinearity is regarded as harmful if at,
say, the 5% level of significance, the value of the F
statistic is significantly different from zero but none of
the t statistics for the regression coefficients (other
than the regression constant) is. In this case we would
reject the hypothesis that there is no relationship between
YI on one side and Xj[2» ^3 XIK on the other
side, but we would not reject the hypothesis that any one
of the explanatory variables is irrelevant in influencing
Yi- Such a situation indicates that the separate
influence of each of the explanatory variables is weak
relative to their joint influence on Yi« This is
symptomatic of a high degree of multicollinearity, which
prevents us from disentangling the separate influences of
the explanatory variables*.
The objectivity of this test makes it a desirable method for the detection of
multicollinearity, but as Kmenta alluded to in his following statement, there
are other harmful effects caused by multicollinearity which may be present,
but not brought to light by the use of this method.
The disadvantage of this criterion is that it is too strong
in the sense that multicollinearity is considered as
harmful only when all of the influences of the explanatory
variables on Y cannot be disentangled.*
With respect to the detection of multicollinearity, Chatterjee and Price
stated** the following:
Multicollinearity is associated with unstable estimated
regression coefficients. This situation results from the
presence of strong linear relationships among the
explanatory variables. It is not a problem of
misspecification. Therefore, the empirical investigation
of problems that result from a collinear data set should
begin only after the model has been satisfactorily
specified. However, there may be some indications of
multicollinearity that are encountered during the process
of adding, deleting, and transforming variables or data
points in search of the good model. Indications of
multicollinearity that appear as instability in the
estimated coefficients are as follows:
* Kmenta, op. cit., p 390.
** Chatterjee and Price, op. cit., p. 155-156.
8-2
-------
1. large changes in the estimated coefficients when a variable is
added or deleted,
2. large changes in the coefficients when a data point is altered or
droppped.
Once the residual plots indicate that the model has been satisfactorily
specified, multicollinearity may be present if
3. the algebraic signs of the estimated coefficients do not conform
to prior expectations or
4. 'coefficients of variables that are expected to be important have
large standard errors.
The presence of multicollinearity is also indicated by the size of the
correlation coefficients that exist among the explanatory variables. A
large correlation between a pair of explanatory variables indicates a
strong linear relationship between those two variables. [A table of
correlation coefficients (r) will be referred to as a correlation
matrix.]
The source of multicollinearity may be more subtle than a simple
relationship between two variables. A linear relation can involve many
of the explanatory variables. It may not be possible to detect such a
relationship with a simple correlation coefficient.
B.
Discussion
I. Cheng Equation
MPG = A(ETW ) + B(VDHP)- + C(DISP) + D(RTHP)
+ E(N/V) + F
urban
highway composite
A
B
C
D
E
F
70,077
0.2908
-0.0044
-0.0291
-0.1047
4.0439
80,599
-0.4058
-0.0301
-0.0271
-0.4356
34.546
77,444
0.0458
-0.0084
-0.0366
-0.2187
13.560
1. The following observations, with respect to urban fuel economy, suggest
that multicollinearity is present:
8-3
-------
a. The positive sign for coefficient B for urban and composite fuel
economy is counter-intuitive. It means that for this equation, urban
and composite fuel economy increase as dynamometer horsepower increases.
b. Displacement (DISP) was expected to be an important variable for its
effect on fuel economy, yet for this equation it was not statistically
significant for predicting urban fuel economy at a 90% confidence
2
level. Its partial correlation (r) was only -0.077 (r ••• .006); in
other words, it only explained 0.6% of the variation in ruel economy
(see tables 8-1 to 8-3). The t-statistic for displacement was -1.5562
with an attained significance of 0.1204. When a variable which was
expected to be important is found to not be statistically significant,
the cause of the paradox could be multicollinearity.
c. The correlation coefficients (r) for the variables (see table 8-1) are
as high as 0.89. This indicates that there is a high degree of
correlation between some of the explanatory variables.
d. The standard error for displacement (.0028), relative to the magnitude
of the coefficient, is large.
2. Conclusion; Multicollinearity is present in the Cheng equation used to
predict urban fuel economy. There is probably multicollinearity in the other
two versions also. This would tend to imply caution in the use the equation,
but since the equation was (1) used for a residual analysis and (2) used on
the same data from which it was generated, the appropriate caution in the use
of the equation has been employed.
8-4
-------
II. Murrell Equation
MPG = A(CID x N/V)~°'8 + B (ETW)~°*67
+ C(RTHP/ETW) + D(RTHP/CID)
+ E[(CMPR°*4 -1)/CMPR°'4] + F
Coeff. urban highway composite
A =
B
C
D
E
F
7501
5764
-47.36
-5.025
16.20
-17.03
29,878
3,956
227.2
-16.80
-1.843
-10.69
14,989
5602
36.28
-10.09
-1.700
-8.915
Tables 8-4, 8-5, and 8-6 show the correlation matrices for this equation.
Multicollinearity probably exists.
III. Bascunana Equation
MPG = A[ (ETW)a (DISP)b (N/V)C ]
Coeff. urban highway composite
A =
a =
b =
c =
141,482
-0.8659
-0.1889
-0.2403
959,401
-0.7840
-0.3301
-0.6450
436,918
-0.8831
-0.2323
-0.4086
Tables 8-7, 8-8, and 8-9 show the correlation matrices,
Multicollinearity is present.
-------
Table 8-4
Correlation Matrix for the Nurrell Variables - FTP Data
Variable
UHPG 1.00
-0.67
ETU .93 1.00
RTHP
-.41 -.28 1.00
-------
Table 8-7
Correlation Matrix for the Bascunana Variables - FTP Data
Variable
UHPG 1.00
0°
i a
G ETW -.94 1.00
b
DISP -.90 .96 1.00
c
(N/V) .65 -,75 -.82 1.00
a~~" b c
UMPG ETU DISP (N/V)
-------
Table 8-8
Correlation Hatrix for the Bascunana Variables - HFET Data
CO
£. Variable
OJ
HHPG 1.00
a
ETU -.87 1.00
b
DISP -.83 .94 1.00
c
(N/V) .47 -.75 -.82 1.00
a b c
HHPG ETM DISP (N/V)
-------
oo
I
Table 8-9
Correlation Matrix for the Bascnnana Variables - Composite Data
Variable
CMPG 1.00
a
ETU -.92 1.00
b
DISP -.68 .95 1.00
• c
(N/V) .58 -.76 -.81 1.00
a b c
CMPG ETU DISP (N/V)
-------
C. Summary
The multicollinearity present in the three equations implies that caution
should be used in their application. Especially sensitive applications would
be those involving predictions of the effect of individual variables or their
use to to predict MPG from vehicles whose design parameters were not related
in the same way as those design parameters are in the MY81 data base.
However, such applications were not made for this report. The use of the
equations, "played back" on the same data set from which their coefficients
were based, is appropriate for the residual analysis that was done in section
IV. For the use of the Cheng equation to generate "adjusted MPG" in section
IV, it is assumed that the equation's coefficients developed from MY81 data
are adequate for use on the MY79 and MY80 data bases also.
8-15
-------
Appendix 9
Plots of Ton-Miles per Gallon versus
Emissions by Emission Control System
-------
Two variable linear regression plots of Ton-Miles per Gallon
for Urban Fuel Economy (UMPGETW) and Combined Fuel Economy
(CMPGETW), each plotted agains FTPHC, FTPCO, and FTPNOX.
Data from EPA/CERT Data Base, restricted to non-durability,
non-Diesel vehicles-of test active years 1979, 1980, and 1981
were used.
Stratification by emission control system. (See Section IV. H.)
-------
Appendix 10
Plots of Ton-Miles per Gallon versus
Emissions by Emission Control System and
Transmission Type
-------
Two variable linear regression plots of Ton-Miles per Gallon for
Urban Fuel Economy (UMPGETW) and Combined Fuel Economy
(CMPGETW), each plotted against FTPHC, FTPCO, and FTPNOX.
Data from EPA/CERT Data Base, restricted to non-durability,
non-Diesel vehicles of test active years 1980 and 1981 was used.
Stratification by emission control system and by transmission
type (CA, CM, and LA). (See Section IV. H.)
Only graphs containing at least 12 points are included.
10-2
-------
Appendix 11
Adjusted Fuel Economy versus Emissions
by Emission Control System and Transmission Type
-------
Two variable linear regression plots of "Adjusted Urban Fuel
Economy" (UMPGMINU) and "Adjusted Combined Fuel Economy"
(CMPGMINU), each plotted against FTPNOX.
Data from EPA/CERT Data Base, restricted to non-durability,
non-Diesel vehicles of test active year 1979 to 1981 was used.
Stratification by emission control system and by transmission
type (CA, CM, and LA). (See Section IV. H.)
11-1
-------
SCATTER PLOT VTUN:CO««Y11TO»:I N> TO OUT OP TO I4.UMPGMINU VI. l.PTPHDK
UMPOMINU
IT.OOO «
K.OOO •
at.000 »
Jt.ooo «
11.ooo *
21.000 •
• ft • • ft ft
• « • ft ft ft
4 • t • s • *
» • •
• ft ft • ft
IV. OOO * Jftftftftftftftft ••
ft 2 • • •
• • * • • •
4 ft ft ft ft
2ft ft ft ft •
•
11.000 «
.tOOOO 1.OOOO 1 .BOOO 2.OOOO PTPMOX
.2IOOO .TtOOO 1.2500 1.7100 2.J5OO
SCATTER PLOT VTRN:CM'SYS1T0>:> Nft IT OUT OP IT 1«.UM»CMINU VS. I.PTPNOX
UMPCM1NU
3T.OOO 4
14.OOO 4
31.000 4
2A.OOO 4
22.OOO 4
K.OOO 4
1C.OOO •
11.OOO 4
10.000 4
.5OOOO ' 1.OOOO t.BOOO 2.OOOO P1PNOX
.2SOOO. .75000 1.2500 1.T5OO 2.2SOO
11-8