United States
Environmental Protection
Agency
Motor Vehicle Emission Lab
2565 Plymouth Rd.
Ann Arbor, Michigan 48105
EPA 460/3-80-010
&EPA
Air
September 1980
Passenger Car Fuel Economy:
EPA and Road
-------
This report is issued by the Environmental Protection Agency to disseminate
technical data. Copies are available free of charge to Federal employees, current
contractors and grantees, and nonprofit organizations — in limited quantities —
from the Library, Motor Vehicle Emission Laboratory, Ann Arbor, Michigan 48105,
or, for a fee, from the National Technical Information Service, 5285 Port Royal Road,
Springfield, Virginia 22161.
If you desire to:
• Comment on this report, or
• Submit data or information on this subject for future EPA reference,
Please communicate IN WRITING to:
Director, Emission Control Technology Division
U.S. Environmental Protection Agency
2565 Plymouth Road
Ann Arbor, Michigan 48105
ACKNOWLEDGEMENTS: Cynthia Ferris for manuscript typing; Peter Thorne for
cover artwork.
-------
PASSENGER CAR FUEL ECONOMY: EPA AND ROAD
-4 REPORT TO THE CONGRESS-
in response to
THE NATIONAL ENERGY CONSERVATION POLICY ACT OF 1978,
PUBLIC LAW 95-619, TITLE IV, PART 1, SECTION 404
Manuscript completed
January 1980
prepared by
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF AIR, NOISE, AND RADIATION
OFFICE OF MOBILE SOURCE AIR POLLUTION CONTROL
EMISSION CONTROL TECHNOLOGY DIVISION
TECHNOLOGY ASSESSMENT AND CHARACTERIZATION BRANCH
Principal Author: Dillard Mwrell
-------
CONTENTS
Page
I. Introduction and Executive Summary 4
II. Background 12
A. EPA MPG: A Comparison Yardstick 16
B. EPA MPG: An Absolute Yardstick? 20
III. In-Use Data 24
A. Methods of Data Analysis and Presentation 24
B. Summary of In-Use Data 27
C. Representativeness of the Data Sample 30
1. Imported Cars 31
2. Fleet and Consumer-Driven Cars 31
3. Odometer Mileage Nonuniformities 36
D. Time Trends in the MPG Shortfall 39
IV. Fuel Economy Influences 41
A. Overview 42
1. Vehicle Slip 46
2. Road Slip 48
3. Vehicle Design Features 49
4. Technical Summary 54
B. Vehicle Slip 57
1. Sources of Vehicle Slip Data 58
2. Odometer Mileage 66
3. MPG Tilt 68
4. Production Slip 73
5. Vehicle Condition (Test) 84
6. Summary Findings: Vehicle Slip 98
C. Road Slip 101
1. The Travel Environment 103
2. Travel Characteristics 126
3. Vehicle Condition (Road) 175
4. Simulation Variance 191
5. Summary Findings: Road Slip 214
D. Fuel Economy Effects in Combination 216
1. Mathematical Implications 216
2. Engine Map Considerations 217
3. Actual Examples 220
-------
V. For The Fuel Demand Analyst 223
A. The Past Revisited 224
B. The Future 225
C. Vehicle Age Effect 234
VI. Consumer Adjustment of EPA MPG 237
A. Questionnaire Approach 238
B. Adjustment Formula Approaches 243
VII. Public Comment 251
Appendices 277
-------
I. INTRODUCTION AND EXECUTIVE SUMMARY
This report responds to requirements of the National Energy Conservation
Policy Act of 1978, Public Law 95-619, at Title IV, Part 1:
Title IV - Energy Efficiency of Certain
Products and Processes
Part 1 - Energy Efficiency Standards
for Automobiles
"SEC. 404. STUDY.
Within six months after the date of the enactment of
this Act, the Environmental Protection Agency, in consul-
tation with the Secretary of Energy and the Secretary of
Transportation and after an opportunity for public comment,
shall submit to the Congress a detailed report on the degree
to which fuel economy estimates required to be used in new car
fuel economy labeling and in the annual fuel economy mileage
guide required under section 506 of the Motor Vehicle Informa-
tion and Cost Savings Act (15 U.S.C. 2006) provide a realistic
estimate of average fuel economy likely to be achieved by the
driving public. Such report shall include such recommendations
as the Environmental Protection Agency deems appropriate based
on the report and written findings or conclusions stated
therein, other than recommendations concerning changes or
alterations in the testing and calculation procedures and
methods measuring fuel economy under such Act as utilized by
the Environmental Protection Agency for model year 1975
passenger automobiles. Nothing in this section shall authorize
such agency to make any changes or alterations in such procedures
and methods in effect for such model year for measuring automobile
fuel economy."
-------
Pertinent to this report also is the Joint Explanatory Statement of the
Committee of Conference, Senate Report 95-1294, at the same Title and Part:
"Seat-ion 404. Study.
Section 404 requires the EPA Administrator to conduct a
study which compares the mileage estimates for automobiles
derived from EPA test procedures with the actual performance
by these automobiles. The EPA Administrator must report his
findings to the Congress. The report should be sufficiently
detailed so that consumers will be able to better evaluate the
fuel efficiency of the automobiles they intend to purchase and
should include the comments of the Secretary of Energy and
Transportation. For example, the report should discuss the
deviation from EPA published mileage estimates which are
caused by particular driving habits or by the addition of
particular optional equipment. The report should therefore
not be merely the percentage by which all automobiles in the
aggregate deviate in actual performance from EPA mileage
estimates.
The EPA Administrator is not to focus his study on
possible changes of EPA test procedures used for testing
automobile fuel efficiency, and the report is not to contain
recommendations in this area. The conferees have no intention
of authorizing any change in the test procedures as established
for model year 1975, and the statutory language specifically
prohibits such changes. The test procedures required to be
used under EPCA are those for the 2975 model year, and they
are not to be amended. The conferees recognize that any
change in these test procedures would effectively change the
fleet average mileage standards in EPCA. Such a 'change in
the rules' for the testing of automobiles, except by statute,
is therefore, prohibited. "
OBJECTIVES OF THIS REPORT
In accordance w.ith Section 404 and the Conference Report, this report
has two objectives:
1. To determine the degree to which "EPA MPG" figures used in fuel
economy labels and gas mileage guides provide realistic estimates
of average in-use fuel economy.
2. To provide a technical basis for revising, as necessary, the label
and guide MPG figures to better agree with average in-use fuel
economy and, further, to provide information on the degree to which
specific in-use fuel economy influences can cause departures from
the standardized label and guide fuel economy figures.
-------
CONCLUSIONS
Conclusion 1. On the average, fuel economy label and mileage guide
figures have been higher than in-use fuel economy since 1976.
Discussion:
The published EPA fuel economy figures vary somewhat from model
year to model year. As shown in the table following, the 1974 gas
mileage guide* included a single figure based on the 1972 urban
emission test procedure. The average EPA fuel economy based on
this figure significantly understated in-use fuel economy, an
observation which played a strong role in the addition of a non-
urban test to the Federal fuel economy information program.
The 1975 labels and guides included two fuel economy figures, a
"City" value based on the [slightly-modified] urban emission test,
and a "Highway" value based on the new non-urban fuel economy test.
There was either an average 1% road MPG overage, or a 29% road
shortfall, depending upon which of the two published EPA numbers is
used for the comparison.
When the Congress, via the 1975 Energy Policy & Conservation Act,
established the "combined City-Highway" MPG figure as the com-
pliance value for the Average Fuel Economy Standards, this figure
was added to the 1976 labels and guides, and presented more pro-
minently than the City and Highway figures. For the 1976-78
models, there was a 19-20% road MPG shortfall relative to the
combined MPG figure. In light of this, the label and guide pro-
tocols were revised to return to the City figure alone, beginning
with the 1979 model year.
Comparisons: Fleet Average Road MPG
versus Fleet Average EPA MPG
Model
Year
1974
1975
1976
1977
1978
1979
Road EPA City Road EPA Combined
MPG MPG Comparison City-Hwy MPG
13.2
13.8
14.1
14.7
15.8
16.9
11.5 +15%
(1972 test)
12.0 +10%
(1975 test)
13.7 +U
15.2 -7%
16.0 -8%
17.0 -7%
17.6 -4%
14.2
15.8
17.5
18.3
19.6
20.1
Road
-7%
-127.
-16%
EPA Hwy Road
MPG Comparison
18.2
24.3
-27%
19.5
21.3
22.3
24.1
-29%
-34%
-34%
-34%
-31%
Denotes the MPG Values Used
in the Labels and Guides.
There were no fuel economy labels in 1974.
-------
The foregoing data and discussion lead to three sub-conclusions:
1A. In-use fleet fuel economy has varied from 10% above the EPA "1975
Test" City MPG, to 8% below it, over the six model years beginning
with 1974. For 1979, the most recent year studied, average in-use
MPG is 4% lower than this EPA figure.
IB. The Combined City-Highway fuel economy figure has always overstated
average in-use MPG. The same is true of the Highway figure, but
the overstatement has been significantly larger. The Highway
figure was never intended to, and certainly does not, represent
average in-use fuel economy.
1C. By any measure (City, Highway, or Combined MPG) the fleet average
EPA fuel economy figures grew increasingly optimistic with respect
to road experience through 1977, after which EPA-to-road shortfalls
have been decreasing.
Note: The 1979 in-use data is predominantly that
of one manufacturer, Ford Motor Co.
Conclusion 2. For higher-MPG cars, i.e., those most likely to dominate
the fleet in the future, road shortfalls have improved recently
from 1974-75 levels, following an initial worsening in 1976 and
1977. For lower-MPG cars, i.e., those disappearing year by year
from the fleet, road shortfalls worsened through 1978, and have
stabilized or perhaps improved slightly in model year 1979. This
is illustrated in the following figure.
FIGURE ES-I. Model Year Trends in Fuel Economy Shortfalls (Percent MPG Difference)
1979
-------
Conclusion 3. As of model year 1979, the Congress' EPCA fuel economy
standards implied a cumulative improvement in road MPG of 33% over
that of 1974. The actual improvement in road MPG has been 28%, as
illustrated in the next table.
Fleet Fuel Economy Changes, Road MPG
Model Target Cumulative Actual Cumulative
Year Road MPG3 Change from 1974 Road MPG Change from 1974
1975
1976
1977
1978
1979
1980
*a
Based on
increase
14.1
15.0
15.9
16.8
17.6
18.5
6.7%
13.3%
20.0%
26.7%
33.3%
40.0%
road fuel economy of 13.2 MPG
of 0.88 MPG per year toward a
13.8 4.5%
14.1 6.7%
14.7 11.3%
15.8 19.5%
16.9 27.5%
for 1974, and a straight-line
cumulative 40% improvement by 1980.
3A. The 33% implied improvement in EPA MPG by 1979 has been exceeded.
Fleet Fuel Economy Changes, EPA MPG
Model
Year
1975
1976
1977
1978
1979
1980
Target Cumulative
EPA MPGa Change from 1974
14.8 6.7%
15.7 13.3%
16.7 20.0%
17.6(Std=18.0) 26.7%
18.5(Std=19.0) 33.3%
19.5(Std=20.0) 40.0%
Actual Cumulative
EPA MPG Change from 1974
15.8 11.1%
17.5 22.9%
18.3 28.9%
19.6 37.7%
20.1 41.5%
22.4 57.4%
EPA 55/45 basis; Based on an EPA fuel economy of 13.9 MPG for 1974
(The best estimate at the time of announcement of fuel economy improve-
ment goals), and a straight-line increase of 0.93 MPG per year toward
a cumulative 40% improvement by 1980.
3Based on the current EPA 55/45 MPG estimate for 1974, 14.2 MPG.
-------
3B. In model years 1975 through 1977, 67%, 54%, and 60% of intended an-
nual road fuel savings occurred; in the model years 1978 and 1979,
77% and 87% of intended yearly fuel savings were realized (calcu-
lated assuming a constant fleet size for all model years).
3C. Achievement of the road fuel economy targets would have corres-
ponded to a reduction in U.S. passenger car fuel consumption by
approximately 7.4 billion gallons during 1979, compared to 1974;
however, due to the shortfall, the actual reduction in road fuel
consumption was less, approximately 5.4 billion gallons during
1979.
In terms of barrels per day (B/D) the target savings are 485,000 B/D and
the actual savings were 351,000 B/D.
Conclusion 4. Three broad categories of factors are responsible for
the difference between EPA MPG and average in-use MPG:
The travel environment: weather and road conditions;
0 Representativeness of EPA test vehicles and test procedures;
0 Owner travel and driving habits and vehicle maintenance.
Discussion:
For the model years studied, each of these three categories of
factors has roughly the same relative contribution to the total
shortfall. Analysis of the shortfall-producing potential of the
many individual MPG influences shows that their aggregate effect is
more than enough to explain the observed in-use shortfalls.
Any effort directed toward minimizing or eliminating in-use MPG
shortfalls must recognize, as inescapable fact, that no one sector—
government or industry or the public—has control over all of the
factors which determine in-use fuel economy. The weather and the
driving public can easily combine to produce better or worse in-use
MPG than would be indicated by the "best, most realistic" test,
whatever that is.
-------
10
RECOMMENDATIONS
Recommendation 1. The DOE should continue the acquisition and analysis
of data on in-use fuel economy, making such data available to EPA
as a matter of routine. Additional data are needed on light trucks
and vans, imported vehicles, and consumer-driven vehicles.
Recommendation 2. All government sectors which can influence adver-
tising should encourage consistency between advertising usage of
MPG figures and those values used for the label and gas mileage
guide system. Currently, use in advertising of the Highway MPG
value, which does not appear in the label or in the guide, may be
resulting in consumer overoptimism and confusion.
Recommendation 3. The DOE and the EPA should continue to pursue their
campaign to expand public awareness of the Federal fuel economy
information program (as mandated by EPCA, the Energy Policy and
Conservation Act, PL 94-163), incorporating both public input and
motivation/media techniques into the program in order to promote
the widespread availability of not only improved average MPG numbers,
but also, information with which consumers can:
° adjust MPG figures to better predict their own personal
fuel economy, and
purchase, maintain, and operate vehicles so as to reduce
fuel consumption.
-------
11
CONTENT OF THE BALANCE OT THE REPORT
Section II - Background
Presents a discussion of the role and evolution of fuel economy figures,
and considers the use of the EPA fuel economy values as relative and
absolute measures.
Section III - In-Use Data
Discusses the in-use data used for this report, describes analysis
techniques used, treats the representativeness of the data bases, and
considers the trend in MPG shortfall with successive model years.
Section IV - Fuel Economy Influences
Introduces and compartmentalizes the factors which influence vehicle
fuel economy, and analyzes these factors, arriving at estimates of their
relative contribution to differences between EPA and in-use fuel economy.
Section V - For the Fuel Demand Analyst
Discusses considerations important for fuel demand forecasting, and
illustrates techniques for estimating past, current, and future model
years' average road MPG.
Section VI - Consumer Adjustment of EPA MPG
Discusses and evaluates methods for adjusting the EPA fuel economy
values for individual consumer vehicle usage characteristics.
Section VII -_ Public Comment
Summarizes the comments received during the preparation of this report.
-------
12
II. BACKGROUND
Page
The Free Market Period . . 12
EPA MPG: A Comparison Yardstick 16
The Voluntary Improvement Period 16
The Mandatory MPG Improvement Period 18
EPA MPG: An Absolute Yardstick? 20
Modification of the EPA MPG System 23
Fuel economy figures have historically served two purposes: as input
data for fuel demand forecasters, and as information for consumers.
Demand forecasters are primarily concerned with the cumulative fuel
consumption behavior of all vehicles in service. They need to know, for
each past, current and future model year in the fleet, the respective
number of vehicles, average miles driven per vehicle, and average vehicle
MPG. Individual vehicle fuel economy variances due to the multiplicity
of vehicle characteristics, and driving patterns and conditions, are of
lesser consequence to demand forecasting as long as the average values
are known.
Consumers are concerned with relative fuel economy capability between
various models, for purposes of comparison shopping, and also with
information for predicting their own absolute fuel economy and fuel
costs.
Recent history encompasses three distinct periods involving fuel economy
information, its basis, and its uses:
The "Free Market" period: prior to 1975;
0 The "Voluntary MPG Improvement" period: 1975 through 1977;
0 The "Mandatory MPG Improvement" period: 1978 and later.
In the Free Market period, vehicles' fuel economy characteristics were
determined by (a) design choices made by the vehicle manufacturers in
-------
13
response to marketing considerations, with some influence from gov-
ernment safety and emission control requirements, and of course by (b)
the drivers of the vehicles. The only basis for fuel demand projections
was knowledge of historical trends and extrapolations of those trends
into the future.
For demand forecasters, the only available MPG figure even approaching
reliability was the ratio of estimated total U.S. vehicle-miles traveled
to estimated total gallons consumed, calculated annually by the Federal
Highway Administration for the nation and for the individual states.
Those consumers who were concerned with fuel economy had to make do with
promotional data (advertising) and independently derived data appearing
in trade and consumer publications. The availability of such data was
sporadic, and the number of "tests" for measuring fuel economy was at
least as large as the number of organizations publishing "MPG figures".
A growing concern for fuel conservation was reflected in a number of
events occuring between late 1972 and early 1974:
9
° In November 1972 , EPA published fuel economy data from tests
performed for emissions surveillance and certification of cars from
the 1957 through 1973 model years;
0 In an energy message to the Congress on April 18, 1973, the President
called for a voluntary fuel economy labeling program for autos,
along with energy-consuming home appliances. Responsibility for
the auto fuel economy labeling program was assigned to EPA, in
cooperation with the Department of Commerce and the Council on
Environmental Quality;
Various computations of annual miles traveled differ by at least 14%
and by as much as 33.5% (see U.S. Department of Energy, "Vehicle-Miles
of Travel Statistics, Lifetime Vehicle-Miles of Travel, and Current
State Methods of Estimating Vehicle-Miles of Travel", Oak Ridge National
Laboratory Report ORNL/TM-6327, February 1979).
o
U.S. Environmental Protection Agency, "Fuel Economy and Emission
Control, November 1972.
-------
14
Responding to the President's April message, EPA published emis-
3
sions certification MPG results in May 1973 for the 1973 models ,
and in November 1973 for the 1974 models ; protocols for the use of
emissions test MPG values in the fuel economy labeling program were
published in August 1974 . Auto manufacturers representing some
95% of auto sales in the U.S. agreed to participate in the program;
In an energy message to the Nation on June 29, 1973, the President
asked consumers to reduce fuel consumption by five percent;
A group of Mideast oil-producing nations imposed an embargo on oil
exports to the U.S. in late October 1973;
Auto manufacturers accounting for some 95% of U.S. auto sales
agreed in early 1974 to a voluntary program to achieve—by 1980—a
40% fuel economy improvement over 1974 levels;
In parallel with the voluntary (EPA MPG) labeling program, a dra-
matic rise In fuel economy advertising claims occurred. An exten-
sive review of auto advertising by the Federal Trade Commission
revealed that in the first three months of 1974, 61% of all auto
ads made some sort of gas mileage claim, an increase of 243% over
the equivalent period in 1973. In response to this concern over
fuel economy advertising practices, and to a petition filed in
April 1974 by Consumers Union, the FTC initiated proceedings to
develop a Trade Regulation Rule pertaining to auto fuel economy
promotion and advertising. In its rulemaking notice the Com-
mission said:
Federal Register 38, at 10868.
4Ibid., a 30495.
5Ibid., at 22944.
Federal Register 39, at 34382.
-------
15
"...Not only had the sheer number of such claims multiplied enormously,
but the specificity of the claims also rose. Advertisements in
which a specific miles per gallon figure was claimed increased from
5% of the ads run in the September to December 1973 period, to
7.6%, 12.9% and 35% of those run in January, February and March of
1974 respectively.
This increase in specific claims was accompanied by a proliferation
of ads in which the test method on which the claim purportedly was
based was referred to or described to some degree in the advertisement.
Only one advertisement citing the test method used was counted
between September 1972 and November 1973, but in March of this year
[1974], 35% of the automobile advertisements made some reference to
a specific test method said to have generated the claimed fuel
economy data.
The utility to consumers of the test information appearing in the
ads is open to serious question. The tests are not comparable.
Some tests used were conducted on interstate highways at or near
the speed limit; others were on test tracks, at varying speeds,
still others were simply termed "city", "suburban" or "highway"
tests, without further description. Test drivers ran the range
from professional drivers, to employees of the manufacturer, to
celebrities.... Other advertisements failed to note the average
speed of the car tested, or the number of stops per mile, or the
degree to which the car was warmed up. The disclosures were not
sufficient to enable the consumer to determine the relevance of the
claimed fuel economy figures to his own likely experience with the
advertised car and they confirmed that variations in the tests
render comparison by the consumer of competing mileage claims
impossible.
These difficulties cannot be entirely overcome solely through...
independent organizations.... Consumers Union, Road and Track
Magazine, and numerous other interested groups have all issued
mileage reports on different automobiles which differed from
advertising claims made for the autos, from one another's figures
and from figures published by EPA, due to the differences in test
methods.
The general confusion resulting from this state of affairs has been
the subject of several recent articles in the popular press inclu-
ding one entitled, 'Gas Mileage: Whom Do You Believe?' in the
April 1974 issue of Consumer Reports."
[Emphasis added]
-------
16
The issue of fuel economy test methods was also examined in detail in
1974 by the Society of Automotive Engineers (SAE), the EPA, and by other
segments of the Government.
A. EPA MPG; A Comparison Yardstick
The net result of all these developments was confirmation of the EPA
emissions test procedure as the sole official yardstick for auto fuel
economy advertising, and for the voluntary fuel economy improvement and
labeling programs. One of the more important considerations leading to
this decision was recognition of the fact that vehicles tested for fuel
economy must be verified capable of meeting emission standards—this is
the only way of assuring that the MPG figures correspond to vehicles
which can legally be sold in the U.S.
In this Voluntary Improvement period, demand forecasters now had a con-
sistent basis for assigning fuel economy values to new-car fleets;
toward this goal, EPA began publishing sales-weighted fleet MPG figures. '
Forecasters also had a fixed target (1980 MPG = 1.40 x 1974 MPG) for
estimating the MPG characteristics of near-term future U.S. fleets.
Consumers, too, had a consistent basis for comparing the relative fuel
economy capabilities of various models tested under one uniform set of
conditions. Consumers were advised that the figures came from an emissions
test, and were reminded of the comparative nature of the EPA MPG figures,
and of the significant effect of vehicle operation upon their own fuel
9 if)
economy . The first EPA Gas Mileage Guide pointed out:
House Committee on Government Operations, "Conservation and Efficient
Use of Energy", Report 93-1635, December 1974, at 110.
Q
U.S. Environmental Protection Agency, "A Report on Automobile Fuel
Economy", October 1973.
9Ibid.
U.S. Environmental Protection Agency, "1974 Gas Mileage Guide for Car
Buyers", February 1974.
-------
17
"The EPA Test which produced this information is a suburban/
urban cycle that is 7.5 miles long. It is patterned after the
conditions the average driver encounters going from home to work.
A dynamometer was used by professional drivers to insure that the
results were scientifically accurate and comparable. In that way
it is possible to make a fair comparison of fuel economy of all
vehicles tested, because every vehicle was tested in exactly the
same way.
That does not mean, however, that you as a driver will get the same
fuel economy that was obtained on our tests. There are many factors
that affect the fuel economy of individual cars. For example, the
length of your trip and your personal driving habits have a major
impact on fuel economy.... this list is primarily useful to the new
car buyer for comparisons of fuel economy of available vehicles."
[Emphasis added]
1974 saw a growing concern that the EPA emission test which was the
basis for the Label and Guide numbers reflected only urban driving,
12
whereas it was reported that nearly 45% of annual vehicle miles accu-
mulated by personal passenger vehicles was traveled on non-urban roads.
EPA responded to these apparently valid concerns by developing a second,
13
non-urban driving cycle , which was incorporated into the voluntary MPG
14
program
Beginning with model year 1975, the Labels and Guides, by this time
jointly sponsored by the EPA and the DOE (then the Federal Energy Admin-
istration) included, along with the "City" MPG figure, a "Highway" MPG
value.
U.S. Government Accounting Office, "Review of the Automobile Fuel
Economy Testing and Labeling Program", Report to the House Conservation
and National Resources Committee, August 1974.
12
U.S. Department of Transportation, Federal Highway Administration,
News Release FHwA 08-74, January 1974.
13
Kruse and Paulsell, "Development of a Highway Driving Cycle for Fuel
Economy Measurements", U.S. Environmental Protection Agency, March 1974;
Austin, Hellman and Paulsell, "Passenger Car Fuel Economy During Non-
Urban Driving", SAE Paper 740592, August 1974.
14
Federal Register 39, at 36893.
Public Law 94-163, December 1975.
-------
18
The Mandatory MPG Improvement period began with the passage of the
Energy Policy and Conservation Act of 1975 , wherein the Congress made
the Label and Guide information program mandatory, and set fuel economy
standards for model years 1978 through 1985, as follows:
Model Year 1978 18.0 miles per gallon
1979 19.0 miles per gallon
1980 20.0 miles per gallon
1985 27.5 miles per gallon
and
thereafter
The Act specifies that compliance with these standards shall be on the
basis of each manufacturer's sales-weighted corporate average combined
EPA City-Highway MPG figure . A third ourpose for the use of fuel eco-
nomy figures, namely promulgation and enforcement of fuel economy stand-
ards, was thus created.
The Act directed the Secretary of Transportation to prescribe standards
for 1981 through 1984, and authorized him to amend the standard for (and
subsequent to) 1985 to any level between 26.0 and 27.5 MPG. In 1977,
DOT set the 1981 - 1984 standards at:
Model Year 1981 22.0 miles per gallon
1982 24.0 miles per gallon
1983 26.0 miles per gallon
1984 27.0 miles per gallon
and retained the 27.5 MPG standard for post-1984 models.
The chart following depicts the trends in EPA (55/45) fuel economy
within the three distinct periods.
The mileage-weighted consumption average of the EPA City and Highway
figures, proportioned 55% City and 45% Highway. This composite, or
"55/45" fuel economy is calculated from:
Composite consumption: GPM = 0.55 x City GPM + 0.45 x Hay GPU
OK>: Composite fuel economy: MPG =
O.SS 0.45
City MPG Htiy MPG
7Federal Register _42, at 33534.
-------
19
Fuel Economy Scenarios and New-Car Fleet Fuel Economy Trends
(Combined City-Highway MPG)
Model Year
1985
1984
1983
1982
1981
1980
1979
1978
1977
1976
1975
1974
1973
1972
1971
1970
1969
1968
1957-1967
(No Target)
1
14.2
14.2
14.5
14.4
14.8
14.7
14.7
14.9
Target: 40%
Improvement over 1974
(18.5) (31.1)
t t
17.2
16.6
14.8
13.2
27.7
25.4
23.3
22.2
(Domestics) (Imports)
Voluntary
Improvement
Targets
(Standards)
27.5 mpg
27
26
24
22
20
19
18
21.2
19.2
18.7
28.2
26.4
26.8C
(Domestics) (Imports
Mandatory
Improvement
a30% improvement over 1974.
25% improvement over 1974.
°Note imports' fleet MPG reversal in 1978, when standards replaced voluntary
improvement.
-------
20
In the free market environment, fleet fuel economy did not improve; the
4.6% drop from pre-1968 (pre-emission control) models to 1974 has been
1 ft
shown to be predominantly due to increases in vehicle weight.
In the three years following the establishment of the voluntary im-
provement program, steady progress in average new-car EPA MPG values was
clearly in evidence. By 1977, the fleet average "Combined 55/45" figure
reached 18.3 MPG, a 29% increase over the 1974 level of 14.2 MPG (55/45).
Thus, in terms of EPA fuel economy ratings, more than 2/3 of the vol-
untary 40% improvement goal had been reached by 1977, with three years
remaining to the 1980 target.
Since the setting of mandatory standards, improvement in EPA MPG has
continued for the U.S. domestic auto manufacturers, but has stalled for
19
the foreign manufacturers . Those manufacturers whose average fuel
economy exceeds the standards can legally decrease their fleet fuel
economy, as long as they still meet the standards. In terms of fuel
consumption, the average fuel consumed per vehicle-mile for these manufac-
turers can increase, which could be considered somewhat counterproductive
given the mandate for reduced fuel consumption for the entire vehicle
fleet.
B. EPA MPG: An Absolute Yardstick?
Early in the voluntary improvement period, data began to emerge which
suggested that passenger cars were not, on the average, achieving the
EPA 55/45 numbers in actual use.
1 ft
Murrell, "Light-Duty Automotive Fuel Economy... Trends through 1979",
SAE paper 790225, February 1979.
19
Of the top-selling ten foreign manufacturers, six showed downturns in
estimated average MPG from 1977 to 1979:
1977 1979 1977 1979
Toyota 29 24 Fiat 24 25
Nissan (Datsun) 27 26 British Leyland 23 21
Volkswagen 29 31 Toyo Kogyo (Mazda) 29 28
Honda 36 31 Mercedes-Benz 19 20
Fuji (Subaru) 30 29 Volvo 20 21
-------
21
20
In August 1975, EPA reported the results of tests on six 1975 pro-
duction cars. On the average, these cars' fuel economy, when tested on
the dynamometer, was within 1% of the EPA rating for their counterpart
EPA certification prototype cars, but when tested on a test track,
(using the same EPA driving cycles), average MPG fell short of the EPA
ratings by about 6% (the average shortfall was about 7% for the City
test and 5% for the Highway test). This study also confirmed that the
EPA Highway ratings were achievable in actual use: every test car met
or surpassed the JJ1PA Highway MPG value in the 150-mile road trip to the
test track site, when driven in adherence to the 55 MPH speed limit.
21
A report by West and others , also in August 1975, presented data from
dynamometer and track tests of over 100 model year 1975 production cars
from twelve manufacturers. The track tests used SAE procedures which
were not the same as the EPA procedures, but the dynamometer tests
employed the 1975 EPA procedures. These tests showed an average shortfall
of about 8% between EPA certification car MPG and that of the production
cars. All tests were conducted on vehicles that had relatively low
odometer mileage, 2000 miles.
In September 1975, General Motors made available to EPA the results of a
22
nationwide customer fuel economy survey , wherein some 2600 private
owners of new 1975 GM cars furnished data on miles driven and gallons of
fuel purchased over a nominal one-month period of vehicle use. The GM
data showed an overall average MPG shortfall of about 13% from the EPA
55/45 figure. On the average, road MPG agreed almost exactly with the
EPA City figure. The data also revealed very large customer-to-customer
variances in MPG: for nominally identical copies of the same model, in-
use fuel economy varied from 42% to 145% of the EPA 55/45 rating.
20
Austin, "Passenger Car Fuel Economy—Dynamometer vs. Track vs. Road",
Report 76-1, Technology Assessment and Evaluation Branch, ECTD, EPA,
August 1975.
21
West, et^ al_, "A Technical Report of the 1975 Union 76 Fuel Economy
Tests", SAE paper 750670, August 1975.
22
Unpublished data.
-------
22
These observations raised serious concerns for both the fuel demand
forecaster and the consumer, and resulted in the initiation or inten-
sification of a number of investigations, in and out of the Government,
to answer three basic questions:
1. Is there one number, such as a fixed percentage, for the
average in-use MPG difference, that is applicable to past
models and to future models, or is there a year-to-year trend?
2. What are the causes of any MPG difference?
3. What factors contribute to the wide variance observed in the
in-use fuel economy of nominally identical cars, and can they
be quantified in a way usable by individual drivers to better
evaluate and predict their own fuel economy?
A large amount of data generated since 1975 related to the first question
has been collected, systematized, and analyzed by the DOE. Interim
23
results of their analyses of these data were reported in 1978 and
24
1979 . Information related to the other two questions has been
gleaned from a host of sources far too numerous to list here: some in
the public domain and some furnished by interested parties during the
course of EPA's studies.
Sections III and IV of this report include our detailed analysis of all
of these data.
23
McNutt et al, "A Comparison of Fuel Economy Results from EPA
Tests and Actual In-Use Experience, 1974-1977 Model Year Cars", SAE
paper 780037, February 1978.
24
McNutt and Dulla, "Factors Influencing Automotive Fuel Demands", SAE
paper 790226, February 1979.
-------
23
Early in the course of these technical studies, it was recognized that
an early, temporary modification to the presentation of EPA MPG numbers
was both possible and desirable, pending the completion of the detailed
analyses. Various options for this short-term measure were tendered in
25 26
a Notice of Proposed Rulemaking , and an interim Rule was promulgated
in May of 1978. The interim Rule specified that—beginning with the
1979 model year and continuing until a better data base could permit a
more satisfactory solution—the Label/Guide program would use only the
EPA City, or "Estimated" MPG figure. The Highway and Combined 55/45 MPG
figures were dropped from the MPG Information Program, although use of
the Highway figure in advertising has continued. The following quotes
from the interim Rule emphasize the temporary nature of this situation:
"... this is clearly not the final action that EPA will be taking
to improve the program..."
"EPA believes that the action adopted for 1979...is an interim.
limited action involving the continued use of one of the estimates
now in use."
"The action taken for 1979 is not likely to satisfy many of those
who commented on the NPRM; indeed, it does not satisfy EPA in the
sense that a better solution may be found for future model years."
"EPA considered and still is favorably disposed to promoting a more
realistic range of in-use mileage, in view of...the fact that no
one value fully characterizes a car's fuel economy."
This report constitutes the primary technical basis upon which we are
formulating a more permanent system. There is a good chance that the
findings herein, together with continuing technical studies and admin-
istrative tradeoffs now being explored, can lead to finalization of MPG
labeling Ruletuakings and Gas Mileage Guide protocols for the 1982 models.
25Federal Register 4^, at 6817.
26Ibid.t at 21412.
-------
24
III. IN-USE DATA
Methods of Data Analysis and Presentation •
Summary of In-Use Data
Representativeness of the Data Sample • • •
Imported Cars
Fleet and Consumer-Driven Cars . . • •
Odometer Mileage Nonuniformities • • •
Time Trends in the MPG Shortfall
Page
. 24
. 27
. 30
. 31
39
A. Methods of Data Analysis and Presentation
For much of this section, comparisons between in-use and EPA fuel economy
will be presented graphically. The standard grid for these presentations
is shown below. The reference EPA fuel economy is scaled horizontally,
and the corresponding in-use ("road") fuel economy is scaled vertically.
The grid is split by a diagonal line along which road MPG equals EPA MPG.
Above this diagonal, road MPG exceeds the EPA value [as in point (A)],
and below the line, road MPG is less than the EPA value [as in point (B)].
FIGURE I. Road Fual Economy vs. EPA Fuel Economy
30
25
i
J 20
I
IS
10
Road MPG
Lower than
EPA MPG
10
IS 20
EPA Put) Economy, MPG
25
-------
25
For groups of cars, several treatments of the data are possible in
drawing comparisons between road and EPA fuel economy; no single method
is "the correct" method. Each of the treatments illustrated in the
following example is used at one time or another in this report.
Consider the hypothetical group of cars listed in the table, of which
three exceeded their respective EPA figures on the road, one achieved
its EPA value exactly, and eight fell short of their EPA numbers.
Car
A
3
C
D
E
F
G
ii
I
J
K
L
Engine
4 cyl.
4 cyl.
4 cyl.
6 cyl.
6 cyl.
6 cyl.
6 cyl.
6 cyl.
8 cyl.
8 cyl.
8 cyl.
8 cyl.
EPA
MPG
27
26
24
22
22
21
19
18
15
14
13
12
Road
MPG
24
22
23
21
20
19
20
16
15
13
15
13
4 cyl. subfleet:
Avg. EPA =25.6
Avg. Road =23.0
6 cyl. subfleet:
Avg. EPA =20.3
Avg. Road =19.0
8 cyl. subfleet:
Avg. EPA = 13.4
Avg. Road = 13.9
Total Fleet
Avg. EPA =18.1
Avg. Road = 17.6
Difference = -10.2%
(-2.6 MPG)
Difference = -6.3%
(-1.3 MPG)
Difference = +3.7%
(+0.5 MPG)
Difference = -2.8%
(-0.5 MPG)
It can be seen that this fleet's average on-the-road MPG fell short of
27
its fleet average EPA figure bv 2.8%. The excess fuel consumed on the
road by this fleet is related to this figure:
Road Gallons/Mile
EPA Gallons/Mile
I/Road MPG
1/EPA MPG
EPA MPG
Road MPG
= 1.028
The data can also be stratified into subsets, as in this example by
engine type, and the respective in-use MPG performance of each subset
examined separately. Graphically, the comparison may be presented as in
the next figure, the left hand plot showing each data point, or as in
one of the right hand plots, where the individual data are collapsed into:
27
The fleet average MPG figures are "harmonic" averages; See Appendix A.
-------
26
one point representing the fleet, one best-fit ("regression") line
through the data, or into points or regression lines representing specific
subsets of data.
FIGURE 2. Examples of Data Stratification
25
1
IS
10
_ A
A = 8cyl.
• = 6cyl.
D = 4cyl.
10 15 20 25
EPA MPG
30
25
IS
10
i
10
IS
20 25
EPA MPG
-
"
IS
10
'
6cyl. <-
Average '
'
8cyl.
Average,
30
4cyl.
Regression
^
t^^.
Regression
8cyl
Regression
_L
10 IS 20 25
EPA MPG
30
There are many orations for constructing regression lines through fuel
economy data (Appendix B); generally, the two types that have been
adopted by DOE will be shown as the boundaries of a "regression band".
Most of the on-road fuel economy data used for this study comes from
cars from model years 1975 through 1978. The primary measure of EPA
fuel economy for these model years was the combined ("55/45") MPG
figure and, unless noted otherwise, this is the figure referred to
throughout the report as "EPA MPG".
-------
27
B. Summary of In-Use Data
The graph below summarizes the road-to-EPA comparison for all of the
data furnished to EPA by DOE as of early summer 1979. There is an in-
use fleet MPG shortfall of 14.4% for all of the post-1974 data treated
X
in the aggregate; fleet average EPA fuel economy for the in-use sample
is 18.1 MPG, whereas these cars averaged 15.5 MPG in actual use.
FIGURE 3. Aggregate DOE In-Use Data
= Average EPA and Road values
For DOE Data, 1975-1978
20
EPA 55/45 MPG
The average shortfall for the aggregated DOE data is worse for higher-
MPG cars, not only from the standpoint of absolute MPG differential, but
on a percentage basis: there is less than 10% average shortfall for
cars with EPA ratings below 13 MPG, while cars with EPA ratings above
25 MPG show an average in-use MPG deficiency between 20% and 30%.
-------
28
The two DOE regression equations for their 1975-1978 data set are given
below; the "fuel economy" regression is seen to be a statistically
better representation of the data than the "fuel consumption" regres-
sion. These two equations form the boundaries of the regression band in
the previous figure.
Fuel Economy Regression;
MPG , = 0.664 (MPG..-.J + 3.492
road EPA
Correlation Coefficient (r) = .81
Fuel Consumption Regression:
l/MPGpoad = O.B?8(l/MPGEpA) + 0.017
Correlation Coefficient (r) = .76
Subdivision of the aggregate data into various strata shows results
which are generally in good agreement with the regression band. This
verifies that regression analysis of the total data sample (which is
dominated by lower-MPG cars) does produce a curve which accurately
reflects the average in-use shortfalls of higher-MPG cars.
The five-graph composite figure illustrates these stratifications.
-------
29
FIGURE 4. 30
Stratified Analyses
of DOE Raw Data
25
y>
, ,-
By Model Year .'
1975 Shortfall 4,000 Lb.
IS 20 25
EPA MPG
-------
30
Certain findings emerge from this analysis of the raw data, essentially
independent of how the data are stratified:
There is a trend in both EPA and road MPG from 1974 through 1978,
and the five-year trend is upward;
There is a disjoint year-to-year trend in the fleet average road-
to-EPA shortfall: the fleet shortfall worsened consistently from
1975 through 1977, then improved with the 1978 models. Stated
another way, fleet MPG changes are:
Change in
EPA MPG
+34%
+1.2%
+6.1%
Change in
Road MPG
+18%
-1.2%
+6.6%
1974 to 1976
1976 to 1977
1977 to 1973
The disjoint shortfall trend is most apparent for the highest-MPG
engine size and vehicle weight strata, wherein road MPG decreased
consistently from 1975 to 1977, while EPA MPG was increasing. The
average shortfall for high-MPG 1978 cars is clearly better than
1976 or 1977.
Every subdivision of the data, whether by vehicle technical specifi-
cation, manufacturer, or model year, shows some MPG shortfall
on the road: every data point representing the average for each
stratum falls below the road/EPA equivalence line.
C. Representativeness of the Data Sample
As pointed out above, these conclusions apply to the DOE data sample
(12,000 cars), which is approximately 3/100 of 1% of the entire vehicle
population on the road for these model years. In view of this relatively
small sample size, the representativeness of the sample must be considered
before projecting the sample's conclusions onto the total road population.
-------
31
The sample is known to have three representational shortcomings:
Imported cars are, for all practical purposes, not included in
the samnle, whereas imports account for some 16% of new-car
? Q
sales in the U.S. for 1975-1978 ;
The majority of the data sample consists of rental or company
_f_lee_t cars, whereas fleet cars £
f
all new-car sales for 1975-1978"
_f_le_et cars, whereas fleet cars account for less than 14% of
,29,30
0 Odometer mileages are progressively higher for earlier model
years, so year-to-year comparisons are not "new-car to new-
car" comparisons.
1. Imported Cars
Until more road MPG data are available on imported cars, we cannot
assess whether or not their exclusion biases the sample results. It
must be pointed out that many of these vehicles are in the higher EPA
MPG strata, and can affect the nature of the overall shortfall rela-
tionship.
2. Fleet and Consumer-Driven Cars
Regarding the disproportionately high number of fleet cars in the
31
sample, it has been noted that cars sometimes show greater shortfalls
in fleet use than in consumer use.
28
Murrell, Op_. Cit. (18), at 25.
29
Shonka, "Characteristics of Automotive Fleets in the United States
1966-1977", Oak Ridge National Laboratory Report ORNL/TM-6449, September
1978.
Bobit Publications, Automotive Fleet, April 1979.
31McNutt, Op_. Cit. (23), at 15.
-------
32
A patent example of this is the data in the following table, for a
popular high-MFC Subcompact car. The "consumer use" figures come from a
customer survey 32 by the manufacturer, and the "fleet use" figures from
the fleet records 33 Of a telephone company, as furnished to DOE. The
differences between the consumer and fleet on-road MPG averages, and
(especially) minimums, are quite significant. In the absence of detailed
data on the usage patterns for these vehicles, one can only theorize
how Subcompact cars could be operated in fleet service to produce road
fuel economies less than five miles per gallon.
EPA 55/45 MPG
Road MPG in Consumer Use:
Number of Cars
Max. Road MPG
Min. Road MPG
Avg. Road MPG
Avg. Shortfall
1976 1977
28.7 29.7
(Summer) (Winter)
20
27.9
17.1
21.5
25%
12
24.8
15.8
20.6
28%
1978
28.0
(Summer)
56
30.2
15.3
22.7
19%
Road MPG in Fleet Use:
Number of Cars
Max. Road MPG
Min. Road MPG
Avg. Road MPG
Avg. Shortfall
48
25.2
4.8
15.0
48%
65
31.1
3.7
15.3
48%
10
26.1
2.8
18.0
36%
Whatever the nature of the fleet car data, if its MPG characteristics
are different from those of consumer cars, fleet car overemphasis in the
sample can produce an unrepresentative picture of the actual vehicle
population.
^"Unpublished, furnished by manufacturer.
33Unpublished, furnished by DOE.
-------
33
Fortunately, it is possible to arrive at an improved estimate of the
actual population from the standpoint of consumer/fleet proportioning:
DOE has provided consumer and fleet regression analyses separately, and
the consumer/fleet proportions — as a function of EPA MPG level — can
be inferred from data in the literature.
The next figure depicts DOE's consumer and fleet data through 1978, and
also new data (unpublished) on 1979 models, received from Ford Motor Co.
and General Motors. The consumption regression equations for all of
these data are:
Consumer Use
1974 DOE: 1/t-ffG = 0.692(1/MPGE) + .025
1975
1976
1977
1978
1/MPG = 0. 878 ( 1/MPG J + .015
1/MPG = 0. 601 ( 1/MPG J + .056
r t,
1/MPG = 0.676 ( 1/MPG E) + .030
1/14PG = 1. 154 ( 1/MPG J + .003
V Hi
Fleet Use
1/MPG = 0.89K 1/MPG J + .011
T b
1/MPG = 0.697(1/MPG } + .028
T* £4
1/MPGr = 0.972(1/MPGE) +. .010
1/MPG = 0.794(1/MPGE) + .023
1/MPG = 1. 244(1/MPGE) + .001
1979 Ford: 1/MPG = 1.251 (1/MPGJ - .004 GM: 1/MPG = 0.819(1/MPGJ + .023
r E i> z
30
20
10
10
FIGURE 5. Road Fuel Economy, Consumer and Fleet Use
30, , r-
I
Consumer Use
Doe Data
Other Data
T
74 75
76 78 77
I
25
i
"D
J
20
15
10
Fleet Use
^— Doe Data
...... Other Data
t t ttt
75 7677 78
J I
15
20
EPA MPG
25
30
10
15
20
EPA MPG
25
30
-------
34
The proportions of fleet cars in the total new-car population for model
years 1974 through 1978 are given in the following table:
•3
Contribution of Fleet Cars to
Total New-car Population
1974b
1975b
1976b
1977b
1978°
Fraction
of Cars
12.2%
11.1%
11.4%
13.4%
12.6%
Fraction of Vehicle
Miles Traveled
19.7%
20.5%
18.3%
22.7%
21.6%
(VMT)
f\
Fleets of 10 or more cars.
Shonka, Op. Cit. (29).
CBobit, Op. Cit. (30).
Because of higher average annual mileage driven by fleet cars, the VMT
contribution of these cars is almost double their numerical contribution.
30
The next table, based on data from Bobit , indicates that fleets include
proportionately much fewer small cars than the general vehicle popula-
tion.
Car Size Differences, Fleet Cars vs. Consumer Cars
(Model Year 1978, millions of cars)
Consumers Cars
Fleet Cars
All Cars (Total)
Fleet Cars, % of Total 8.6% 15.5%
Compact
or Smaller
4.5
0.4
4.9
Midsize
or Larger
5.9
1.1
7.0
Small Cars,
% of Total
43.2%
28.2%
41.3%
-------
Combining the above data and our own fuel economy and vehicle size data,
the VMT proportions between fleet cars and consumer cars, by EPA MPG
level, are estimated as follows:
Fleet Car %VMT/Consumer Car %VMT,
by Model Year, at Selected EPA MPG Levels
EPA MFC 1974 1975/76 1977 1978
12
16
20
24
27.5
26/74
20/28
15/85
9/91
4/96
30/70
25/75
20/80
15/85
11/89
32/68
27/73
23/77
18/82
14/86
34/66
29/71
24/76
18/82
14/86
The actual population fuel economy characteristics can be derived by
using road vs. EPA fuel economy values from the separate fleet and
consumer regression equations, and weighting them according to the above
VMT proportions. For example: in 1975, an EPA !!PG value of 20
corresponds to a fleet car road MPG of 15.9, and a consumer car road
MPG of 17. The relative %VMT values from the preceding table are 20
and 80, so the overall weighted road MPG is:
.20 .80
= 16.8 MPG
15.9 17
This is the road MPG value calculated for an EPA MPG of 20 in 1975.
The resulting combined fleet/consumer-weighted comparison curves appear
in the next figure. The 1978 VMT weightings were assumed applicable to
the 1979 consumer and fleet data. It is important to note that the
1979 curves are representative of only Ford and GM vehicles, and with
the relative consumer/fleet weightings used, the consumer data (Ford)
predominates. It is unfortunate that data from consumer driving of
other manufacturers' 1979 vehicles are not available.
-------
36
FIGURE 6. Consumer/Fleet Weighted Road MPG
25
20
IS
10
DOE Data
Other Data
10
IS 20
EPA MPG
25
30
3. Odometer Mileage Nonuniformities
The overall average odometer readings for the DOE data cars are estimated
to be 28,800 for the 1974 models, 17,300 for the 1975's; 11,000 for
1976's; 10,300 for 1977's, and 4,000 for 1978's. The average odometer
reading for the 1979 Ford data is 5,320.
*2 /
Based on driver survey data furnished to EPA by General Motors,
larger cars accumulate mileage faster than smaller cars, the differential
rate being 160 extra miles per year per 100 extra pounds of vehicle
weight. These figures permit an estimate of the distribution of average
odometer mileages among the model years, vehicle weights, and corres-
ponding EPA MPG levels in the DOE data sample.
A relationship between odometer mileage and relative fuel economy has
been determined by EPA for dynamometer-measured EPA fuel economy.
34.
35,
Unpublished; also see references 50 and 51, subsequent.
Murrell, 0_p_. Cit. (18).
-------
37
This relationship accounts for vehicle initial break-in, and — for new
cars, before "second car" reduced usage patterns begin to affect fuel
economy — should hold for in-use fuel economy behavior. The relation-
ship specifically relates fuel economy at odometer mileage "M" to fuel
economy at 4,000 miles, abbreviated "AK":
MPG(M)
MPG(4K)
= 0.846 + 0.0186 ln(M)
where ln(M) represents the natural logarithm of "M".
All of the above combine to give the following matrix of odometer mileages
and odometer adjustment factors for the in-use data; when multiplied by
these adjustment factors, the in-use data are all normalized to 4,000-mile
equivalent MFC values.
Odometer Mileage Estimates and 4,000-Mile
Adjustment Factors, In-Use Data Base
EPA = 12: Odometer
4,000-mile factor
EPA = 16: Odometer
4,000-mile factor
EPA = 20: Odometer
4,000-mile factor
EPA = 24: Odometer
4,000-mile factor
EPA = 27.5: Odometer
4,000-mile factor
1974
30,800
0.963
27,300
0.965
25,900
0.966
24,800
0.967
23,700
0.968
1975
19,100
0.971
17,100
0.973
16,000
0.975
15,100
0.976
14,600
0.976
1976
12,700
0.979
11,300
0.981
10,500
0.982
9,800
0.983
9,400
0.984
1977
12,200
0.980
10,800
0.982
10,000
0.983
9,400
0.984
9,100
0.985
1978
4,800
0.996
4,300
0.999
4,000
1.000
3,700
1.001
3,600
1.002
1979
6,400
0.991
5,700
0.993
5,300
0.995
4,900
0.996
4,800
0.996
The odometer - corrected values are plotted in the next figure. These
curves are bounded by:
(Upper) EPA MPG _< 20,
EPA MPG > 20,
(Lower) EPA MPG _< 17.5,
EPA MPG > 17.5,
MPG = 0.599(MPGE) + 4.614
MPG^ = 0.864(MPGE) - 0.685
MPG = 0.856(MPGJ - 0.544
T b
MPG = 0.315 (MPG J + 8.638
r &
-------
38
FIGURE 7. Road MPG, Consumer/Fleet Weighted and Odometer Corrected to 4,000 Miles
25
10
15 20
EPA MPG
The bounding relationships are linear equations fitted to the outer edges
of the available fleet average data. As such, they should be viewed as
boundaries only, and other calculations made with these curves (e.g.,
extrapolations to higher EPA MPG levels) should be performed judiciously.
The two equations:
(High) MPG = 0.864 (MPG J - 0.685 and
r a
(Low)
MPG = 0. 215 (MPGJ + 8.638
r hi
bound the historical shortfall relationship for high-MPG vehicles.
While some individual vehicles can be expected to perform outside these
bounds, we would expect that future average MPG levels will fall within
the bounds. With these caveats, some implications of the shortfall band
can be inferred:
0 Vehicle fleets with an average EPA fuel economy of 27.5 MPG could
be estimated to average between 17.3 MPG and 23.1 MPG in actual use.
0 An EPA MPG level of at least 32.6 MPG would be necessary to achieve
27.5 MPG on the road.
-------
39
D. Time Trends in the MPG Shortfall
Tentative conclusions on the shortfall pattern as a function of time, or
model year, were drawn earlier based on the raw data from DOE and other
sources. Having adjusted these data for consumer-fleet weightings and a
common odometer mileage basis, we may now draw consistent comparisons
between the model years. The next figure illustrates these uniform-
basis MPG shortfalls versus model year for five specific EPA 55/45 MPG
levels; the shortfall for each year's entire fleet is also shown.
0 For low-MPG cars, e.g. EPA=12 and EPA=16, there is an increasing
trend in road shortfall with successive model years up through
1978, with the shortfall decreasing slightly in 1979;
0 For the higher-MPG cars, e.g. EPA=20 and above, the road shortfall
reached a maximum in 1976 and has decreased with every model year
since;
For the overall fleets, i.e. each year's sales-weighted combination
of all EPA MPG's, the road MPG shortfall has been relatively stable
from 1976 through 1979, remaining between 3.2 MPG and 3.8 MPG.
FIGURE 8. Model Year Trends in Fuel Economy Shortfalls (MPG Difference)
1974
1975
1976 1977
Model Year
1978
1979
-------
When these shortfall trends are viewed on a percentage basis, as shown
in the next figure, the implications are similar to the above conclusions,
with the additional important observation that:
0 For the two most recent model years, 1978 and 1979, the
percentage shortfall is very nearly constant over the entire
range of EPA MPG levels, i.e. the 1978-79 models do not seem
to share the "worse shortfall for higher-MPG cars" pattern
seen in the 1974-1977 models.
FIGURE 9. Model Year Trends in Fuel Economy Shortfalls (Percent MPG Difference)
1979
-------
IV. FUEL ECONOMY INFLUENCES
Page
Overview 42
Vehicle Slip 45
Production Slip 46
Vehicle Condition (Test) 47
Road Slip . 48
Travel Environment 48
Travel Characteristics 48
Vehicle Condition (Road) 48
Simulation Variance 48
Vehicle Design Features • • • • .49
Technical Summary 54
Vehicle Slip Influences 54
Road Slip Influences 55
Model Year Differences 56
-------
42
IV. FUELECONOMY INFLUENCES
A. Overview
Two basic potential sources of road-to-EPA fuel economy differences are
the vehicles themselves, and the conditions under which the vehicles are
operated in actual use. If an in-use vehicle, when tested on a dynamo-
meter using the EPA procedures, achieves the same fuel economy as did
36
its EPA test counterpart, there is no net vehicle-related MPG slip ; any
road MPG shortfall (or overage) must then be due solely to in-use operating
conditions which differ from those of the EPA tests. Conversely, if an
in-use car's MPG on the dynamometer is different from the EPA test car
value, some portion of its in-use MPG discrepancy is due to fuel economy
performance differences between the EPA prototype test car and the in-
use vehicle as brought in for testing.
We have chosen to apportion the "overall slip" into two parts which
are defined as "vehicle slip" and "road slip". The definitions are
given below:
Overall Slip = In-use Car Road MPG
EPA MPG
Vehicle Slip = Jn-uae Cay Duno MPG
EPA MPG
Road Slip = In-nee Car Road MPG
In-use Car Dyno MPG
Note that overall slip equals vehicle slip multiplied by road slip.
The data supplied by DOE included information on dynamometer tests of
in-use cars, from which "vehicle slip" can be evaluated, and on road MPG
of in-use cars that had also been dynamometer-tested, from which "road
slip" can be evaluated. The DOE data cover model years 1975 - 1977.
0£
"slip" is used herein in the neutral sense: it can be either upward
or downward.
-------
43
The tables below summarize the DOE data. Whether analyzed in terras of
fuel economy or fuel consumption, each model year shows some shortfall
due to vehicle slip, and an always-larger shortfall due to road slip.
It can be concluded from these data that about 2/3 of the observed in-
use fleet shortfall is due to vehicle operating conditions, with the re-
maining 1/3 attributable to the in-use vehicles themselves.
Vehicle Slip, DOE Data
(Approximately 4000 Cars)
1975 - Economy Average
- Consumption Average
1976 - Economy Average
- Consumption Average
1977 - Economy Average
- Consumption Average
EPA MPG
In-Use Car MPG
(Dyno Tests)
16.5
15.4
19.8
18.5
19.5
18.9
16.2
15.1
18.5
17.5
18.3
17.5
Vehicle Slip
.98
.98
.93
.95
.94
.93
Road Slip, DOE Data
(Approximately 400 Cars)
1975 - Economy Average
- Consumption Average
1976 - Economy Average
- Consumption Average
1977 - Economy Average
- Consumption Average
In-Use Car
Dyno MPG
15.5
14.9
17.7
16.9
16.7
16.1
In-Use Car
Road MPG
14.
13.
15.9
15.1
14.7
14.5
Road Slip
.92
.92
.90
.89
.88
.90
On a fleet basis as shown, the data suggest that road slip was rel-
atively constant over these three years (a reasonable possibility) , but
that vehicle slip may be growing worse with time (a disturbing possibility)
Having introduced the concepts of vehicle slip and road slip above, com-
ponent elements of those slip factors can be identified. The next
diagram illustrates the hardware and operational influences on EPA and
in-use MPG.
-------
44
FIGURE 10. Vehicle Slip, Road Slip, and Overall Slip
The following table lists the major elements and subelements which can
cause vehicle slip. Influences which occur prior to acquisition of
vehicles by their ultimate users are categorized as "production slip",
and influences coming into play after acquisition are grouped together
as "vehicle condition" items. The sources of credit or blame for each
influence are also given.
Fuel Economy Influences
Associated with Vehicle Slip
Vehicle Slip Influence
Responsibility
a. Production Slip
Administrative Variance
Hardware Variance
b. Vehicle Condition (Test)
Engine Tune
Engine Response to Fuel Properties
Sampling Bias
EPA, Manufacturer
Buyer, Transporter, Dealer,
Manufacturer
Owner, Tuner
Owner, Tuner, Fuel Refiners
Owner
-------
45
The next table lists factors related to road slip. Included are effects
that are strictly operational, and other factors whose MPG effects are
seen only on the road.
Fuel Economy Influences
Associated with Road Slip
Road Slip Influence
Travel Environment
Ambient Temperature
Barometric Pressure/Altitude
Wind and Aerodynamics
Road Gradient
Road Surface
Road Curvature
Responsibility
Uncontrollable
Uncontrollable
Uncontrollable
Uncontrollable, Route selection
Uncontrollable, Route selection
Uncontrollable, Route selection
b. Travel Characteristics
Vehicle Speed
Traffic Volume Effects
Trip Length/Vehicle Warmup
Acceleration Intensity
Driver, Uncontrollable
Uncontrollable, Driver
Driver
Driver
c. Vehicle Condition (Road)
Wheel Mechanical Condition
Tire Pressure
Vehicle Weight Load
Owner
Owner, Driver
Driver
d. Simulation Variance
Dynamometer Loading
Tire/Dynamometer Interaction
Weight Class Distributions
Manual Transmissions
Power Accessories
Vehicle Cooling
Metric Slip
EPA
EPA
EPA, Manufacturer
EPA, Manufacturer
EPA
EPA
EPA, Manufacturer, Owner/Driver
-------
46
The elements in the preceding tables are discussed briefly below. Detailed
analyses of the MPG effects of the various elements appear in Sections B
and C.
1. Vehicle Slip
a. Production Slip - production slip exists if a production vehicle
differs in fuel economy performance from the "vehicle" corresponding to
the specific EPA fuel economy number on the EPA/DOE Gas Mileage Label and
in the Gas Mileage Guide. Two opportunities for production slip occur:
one associated with the vehicles in the EPA test fleet and the manner in
which test results make their way into the Labels and Guides, and the
other associated with the production vehicle hardware.
Test vehicles are selected nearly a year before new model introduction;
the car configurations to be tested are selected based on several factors,
among them the auto manufacturers' intended model offerings and estimated
sales distributions. For a given model, the "EPA fuel economy number" is,
more often than not, the sales-weighted average for several test cars, all
representing the same nameplate but differing to some extent in detail
specifications. Any misassumption by either EPA or the manufacturers as
to configuration selections and sales weightings has the potential for
making the model MPG unrepresentative (either high or low) as an average .
Moreover, every car configuration within a given model can have a specific
EPA MPG value different from the average MPG published for that model.
With regard to production hardware, differences from the EPA test cars
can and do result from buyer/dealer agreements on new-car specifications.
As an example, if less than one-third of the cars of a given model are
forecast to be sold with air conditioning, the EPA test car is not
tested to simulate the MPG effect of air conditioning. Every buyer/dealer
agreement to purchase an air-conditioned version of that model creates
production hardware that is different from the test car. The same
applies to other power-consuming accessories and a host of convenience
options such as roof racks, outside mirrors and the like.
-------
47
In addition, the necessity of assembling the test cars prior to new-
model production, by definition, forces the use of non-production
hardware in the EPA tests. The manufacturers generally assure EPA that
the EPA test vehicles include only componentry representative of the
expected average of the yet-to-be-built production hardware. When
vehicles are actually built, transported to dealers, prepped and finally
sold to customers, MPG variances are bound to occur due to the processes
involved. Anything that is mass produced will show a distribution in
any measurable parameter, including—for cars—fuel economv. The
dispersion of this distribution and the relationship of the mean/median/mode
or some other stasticial parameter to the EPA MPG value are important
considerations when evaluating production slip. Resources allocated to
EPA have thus far prohibited any significant post-production verification
of test car representativeness. At any rate, some production cars will
inevitably be inferior, and some superior, to the EPA cars simply due to
tolerance stackups in manufacturing, transport, and dealer nrenaration.
b. Vehicle Condition (Test) - once a vehicle is in the owner's
hands, a number of changes can occur which affect its fuel economy. For
some of these changes, the fuel economy effect shows up in dynamometer
tests of in-use vehicles: since vehicle slip is defined in terms of the
dynamometer testing of in-use cars (using the EPA procedures), these
changes are properly included as ingredients of vehicle slip. Other
changes whose fuel economy influence appears only on the road, and not
on the dynamometer, are road slip effects.
Most vehicle condition items which contribute to vehicle MPG slip are
within the control of the vehicle owner. Items whose effects show up
entirely in dynamometer tests include the state-of-tune of the engine;
and response of the engine to in-use fuel properties.
There is also the distinct probability of a sampling bias in connection
with in-use dynamometer test programs. The vast majority of testing in
these programs has been and is being conducted using cars volunteered for
the tests by their owners. Any malmaintained or tampered vehicles which are
not being volunteered, for whatever reasons of owner reluctance, represent
a thus far unmeasured source of vehicle-condition MPG slin.
-------
48
2. Road Slip
a. Travel Environment - this, and the remaining MPG slip elements,
relate to departures from the standard conditions specified for the EPA
tests; their fuel economy effects appear in road operation. Prevailing
weather and road conditions can vary over a wide range, sometimes with
large effects on vehicle MPG. These factors are not generally under the
control of the driver, except in the sense of route selection.
b. Travel Characteristics - the dynamics of vehicle movement have
pronounced MPG influences. While somewhat dependent upon trip routes and
traffic, the speed and acceleration characteristics of vehicle travel are
largely a matter of driver choice and technique.
c. Vehicle Condition (Road) - Particulars of a vehicle's mechanical
state which have little or no influence on dynamometer test MPG, but which
do effect road MPG, include high vehicle weight loads, the mechanical
condition of the vehicle wheels, and tire pressure.
d. Simulation Variance - As contrasted with in-use conditions which
are clearly different from those of the EPA test, this factor refers to
imprecise duplication by the test of those real-world conditions that it
does attempt to simulate, primarily due to testing and facility capabilities.
Where such necessary test compromises do not agree exactly with road experience,
any MPG shortfall or overage would contribute to road slip. Since the
measurements used to calculate EPA MPG and Road MPG are not the same, some
part of the overall slip may stem from metric differences. The way in
which EPA measures MPG and the way in which in-use MPG is measured can
contribute to dispersion in the overall slip.
-------
49
3. Vehicle Design Features
In addition to the factors described above, fuel economy is of
course dependent on many aspects of vehicle design. It is pertinent to
summarize here the results of MPG sensitivity studies performed on data
from the 1975-1978 EPA test car fleets.
It should be pointed out that these design features are not necessarily
causes of vehicle slip or road slip. Rather, an understanding of these
design features is necessary to estimate the relative magnitude of
various shortfall parameters in the remainder of the report.
The design factors studied were vehicle weight, engine displacement, N/V
37
ratio , and transmission type; the studies consisted of comparing the
fuel economies of vehicles matched in all design specifications except
the one being evaluated. For weight, displacement, and N/V, the MPG
dependences are expressed in terms of sensitivity coefficients:
where: Sp = percent change in MPG per percent change in the
variable "F";
MPG = average MPG for the two states of the variable "F";
F = average value of F, 1/2(F + F ).
37
N/V ratio is defined as the quotient of engine speed in rpm divided by
vehicle speed in mph measured in the highest (i.e., lowest numerical
ratio) transmission gear. In other words, N/V is a measure of engine
revolutions per unit vehicle speed. N/V is related to rear axle ratio
and wheel radius according to:
Ne ,0 ,,,„•» Ne AR
x rr- x AR = 14.01 x — x —
.
60 x 2irr No No r
where: Ne
rr~ = transmission top gear output speed ratio
= 1.0 for many cars but less than 1.0 for those
cars equipped with overdrive;
r = wheel radius, in feet; and
AR = rear axle ratio.
-------
50
For transmission type, the MPG influence is expressed as a slip factor, the
ratio of automatic transmission MPG to manual transmission MPG. ^or each of
these design factors, an average of about 90 comparisons were made for each
of the four model years, with an average of about five cars involved in
each comparison. Therefore, each sensitivity coefficient and transmission
slip factor is based on the behavior of some 1800 cars.
The four-year fleet average sensitivity factors (sample-weighted) are listed
below, in order of decreasing EPA City MPG effect:
Average Fuel Economy Sensitivity Coefficients,
1975-78 EPA Test Cars
EPA EPA
City Highway
Engine Displacement -0.589 -0.578
Vehicle Test Weight -0.388 -0.485
N/V Ratio -0.347 -0.603
Transmission Type 0.981 0.882
The negative signs of the displacement, weight, and N/V coefficients indicate
that fuel economy decreases as these design parameters increase, and vice
versa. To illustrate the use of these sensitivity coefficients, a 10%
increase in engine displacement causes a 10 x (-0.589) = 5.89% decrease in
EPA City MPG; a 20% decrease in N/V causes a -20 x (-.603) = 12.06% gain in
EPA Highway MPG. For the transmission factor, automatic transmission City
MPG is 0.981 x manual City MPG, a 1.9% loss; manual transmission Highway
MPG is I/.882 x automatic Highway MPG, a 13.4% gain.
In the basic sensitivity analysis, the weight changes were accompanied by
changes in dynamometer road load setting as well, so some of the observed
MPG variations were due to these road load changes. Using our best current
estimates of MPG sensitivity to 50 mph road load, and road load sensitivity
to weight, the MPG sensitivity to weight changes alone can be determined.
-------
51
These values are given below.
Sensitivity Effects of Weight and
50 mph Road Load Horsepower
EPA EPA
City_ Highway
Sensitivity of MPG to RLHP -0.163 -0.330
Sensitivity of RLHP to Weight S=0. 642-3. 93(10~
Sensitivity of MPG to Weight -0.311 -0.319
The studies did not reveal any consistent patterns with respect to in-
dividual manufacturers, i.e. no specific manufacturer's cars were found
consistently more sensitive to any of the design factors than those of
other manufacturers.
Similarly, no significant trends in the sensitivities were found as a
function of model year, with one notable exception: manual transmissions'
apparent superiority over automatics for the EPA City test have been growing
consistently. In 1975, there was no average manual-to-automatic City MPG
difference; in 1976, manuals' City MPG was an average of 1.1% better, in
1977 3.1% better, and in 1978 3.4% better. This is believed to relate to
changing specifications for shift point scheduling of the manuals, and will
be discussed in a later section.
These sensitivities are not constant across the range of the design variables
in the fleet. Linear regressions were run on the MPG sensitivity of each
variable with respect to itself, and transmission slip was regressed against
vehicle weight. As shown in the next composite figure, vehicles with
larger engines are more sensitive to displacement changes, heavier cars are
less sensitive to weight changes, cars with higher N/V ratios (typically
the smaller cars) are much more sensitive to N/V changes, and smaller cars
show greater automatic transmission MPG losses relative to manuals.
-------
52
-.80
g ~ 70
'y
u ~ 60
X.
bJ
I -.50
0 -.40
-.30
FIGURE 11 (a). MPG Sensitivity to Engine Displacement
I
I
ISO 250 350
Displacement, Cu. In.
-.80
•?. -70
— Avg.SH - -.578
8
u
-.60
I -50
X
rt
*> -.40
±
-.30
I
I
150 250 350
Displacement, Cu. In.
-.70
-.60
U
-.40
-.30
- 20
-.10
FIGURE ll(b). WPG Sensitivity to Vehicle Weight
Avg Sc = - 388
I
I
2.500 3.500 4,500
Inertia Weight. Lbs.
-.70
-.60
C
£ -.50
U
f -.40
0)
! "*
5 -.20
-.10
2,500 3.500 4,500
Inertia Weight, Lbs.
-------
53
-1.25
FIGURE II (c). MPG Sensitivity to N/V Ratio
-1.25
I.IO
1.00
0.90 -
0.80
FIGURE ll(d). MPG Sensitivity to Transmission Type
_ I
2,500 3,500
Inertia Weight, Lbs.
4,500
I.IO
.00
090
0.80
1 T
Avg. RH = .882
T
2,500 3,500 4,500
Inertia Weight, Lbs.
-------
54
4. Technical Summary
In the body of this report, the fuel economy influences identified in
Section 2. above are examined, arriving at quantitative estimates of
their MPG effects. This Technical Summary pulls together the results of
these separate analyses, serving as a preview and also as an illustration
of the influences' relative and cumulative magnitudes.
In the next two tables, two MPG effects are listed for each influence:
the estimated fleetwide MPG effect, i.e. the average effect of that
influence on all cars combined; and an "example" individual car effect.
In other words, the fuel economy effect of an ambient temperature example
value of 20°F, on a typical individual car exposed to that specific
temperature, is listed as a 13% MPG loss; the fleetwide effect of all
temperatures, as distributed year-round among all cars, is listed as a
5.3% MPG loss.
The conditions of the "EPA 55/45" combined test are the reference values
for these analyses: where a condition is the same as the EPA combined
test, there is by definition no MPG effect. The example conditions were
chosen somewhat arbitrarily: there can be other values for each influence
different from these example values, some better for fuel economy, some
worse.
Fuel Economy Effects of Vehicle Slip Influences
Influence
Production Slip:
Administrative variance
Hardware variance
Vehicle Condition (Test):
Engine tune
Engine/fuel response
Brake drag
a
Wheel alignment
Fleetwide Relative Example Example
Effect MPG Effect Basis
0.1%
1.5%
1.0%
2.3%
0.2%
0.1%
0
0
0
0
0
0
.999
.985
.990
.977
.998
.999
+15%
+25%
+10% i
-25% 1
+ 2% i
- 8% 1
- 2%
- 1%
3 a car
3 a car
3 a car
3o car
Estimated
Estimated
aDrive wheels only (those in contact with dynamometer rolls).
-------
55
Fuel Economy Effects of Road Slip Influences
Influence
Travel Environment:
Temperature
Altitude
Wind
Road gradient
Road surface
Road curvature
Travel Characteristics:
Trip length
Average speed
Cold start
Acceleration intensity
Vehicle Condition (road):
Brake drag
Wheel alignment
Tire switching
Tire pressure
Weight load
Simulation Variance:
Dynamometer load
Tire effects
Weight classification
Manual transmissions
Power accessories
Vehicle cooling
Metric slip
Fleetwide
Effect
-5 . 3%
- 0.1%
-2.3%
-1.9%
-4 . 2%
-o.i%a
+O.U
+10.6%
-0.7%
-11.8%
-0.3%
-0 . 3%
-0.4%
-3.3%
-0.4%
-2.7%
-5.1%
-1.0%
-1.8%
c
c
c
Relative
MPG
O.y47
0.999
0.977
0.981
0.958
0.999
1.008
1.106
0.993
0.882
0.997
0.997
0.996
0.967
0.996
0.973
0.949
0.990
0.982
1.000
1.000
1.000
Example
Effect
-13%
+ 2%
- 6%
-25%
-25%
-25%
-10%
-15%
-25%
-15%
-20%
- 5%
-10%
- 4%
- 6%
-20%
-15%
- 9%
- 5%
- 7%
- 9%
c
- 4%
Example
Basis
o
20 F.
5000 feet .
20 mph wind (360°) .
1°L grade (up/down) .
Several possibilities .
1000 central angle/mile.
Four-mile trip .
20 mph vs 27 mph stop & go,
70 mph cruise vs 55 mph.
Four-mile trip.
"Hard" vs "Easy" accel.
Estimated.
% inch misalignment .
Radial => non-radial.
15 PST. vs 26 PSI
Towing camping trailer.
50% underload at 20 mph.
Non-radial.
Misclassif ied by 1 class.
Four-year average shortfal
Air conditioning, 90°F.
-
Estimated.
Minimum penalty; probably worse
Non-drive wheels
Too close to call
-------
56
As shown in Sec. IV.A, the average measured vehicle slip from all available
in-use data sources is approximately -5%, so our accounting of vehicle
slip influences is quantitatively reasonable. Average measured road slip
is approximately -10%, making a calculated cumulative road loss of
27,5% appear overly pessimistic. We expect that most readers, upon
examination of the analyses, will not find our estimates to be pessimistic
by a factor of three across the board; instead, we conclude that the
multiple fuel economy influences seen on the road do not combine simply
as in the calculation above. Admitting to a bit of far-fetchedness,
we might offer two analogies: that of two harsh liquids, and acid and an
alkali, combining to produce not a doubly destructive liquid but a rela-
tively benign saline solution; or that of two 60 decibel noises combining
into not a 120 decibel noise but one of 63 decibels. The subject of
combined MPG influences is discussed briefly in a later section; this
area in itself needs considerable additional study.
Model year differences were investigated for some of the fuel economy
influences, and are listed in the next table. These may not be the only
effects with year-to-year variation; however, the yearly cumulative
effect of these items does parallel the overall fleet shortfall trends
shown earlier (Section III.D.).
Model Year Trends in Certain MPG Influences
Model Year:
Influence 1974 1975 1976 1977 1978 1979
Administrative variance — +1.3% -1.1% -1.6% +0.9%
Hardware variance +0.6% -7.5% -8.0% +3.3% +1.4% +1.2%
Tire type malsimulation — -2.0% -1.8% -1.7% -1.6% a
Weight classification (-1.0%)b (-1.0%) (-1.0%) -1.0% (-1.0%) (-1.
Manual transmissions — +0.1% -1.0% -1.9% -2.5% a
*a
Simulation variances reduced or eliminated in the 1979/80 EPA test procedures.
Only Model Year 1977 analyzed; other 1974-1979 years believed to be similar.
-------
57
FUEL ECONOMY INFLUENCES (Cont'd.)
Page
Vehicle Slip 58
Sources of Vehicle Slip Data 58
EPA Emission Factors Program 58
California Assembly Line Tests 60
DuPont Fuel Economy Fleet 61
Mobil Oil Co. Fuel Economy Tests . 61
Southern California Auto Club Fleet 62
Union Oil Co. Fuel Economy Tests 62
EPA Dynamometer vs. Track Project 63
EPA Restorative Maintenance Program • 64
General Motors Production Car Tests 64
EPA Subcompact Car Test Project 64
EPA Selective Enforcement Audit 65
Odometer Mileage 66
w
MPG Tilt 68
Production Slip 73
Administrative Variance 73
Definition of Vehicle Configuration ...... 73
Hardware Variance 78
EPA Audit Data 80
Vehicle Condition (Teat) > 8*
Engine Tune 8*
Malfunctions 85
Mixed Results of Tuneups ........... 87
Engine Response to Fuel Properties . . ... .90
Fuel Density . . . 90
Fuel Octane Rating 93
Knock Sensors 95
Fuel Volatility 96
Additives . . . .97
Summary Findings: Vehicle Slip .......... 98
-------
58
B. Vehicle Slip
The subject of vehicle slip has been under study within EPA since 1976.
Eleven significant sources of 'iri-'use'vehicle'dynambmet'er test data which
have been reviewed are listed below. Some of these sources are also included
in the DOE data base.
Vehicle Slip
Data Analyzed and Model Years Covered
1974 1975 1976 1977 1978 1979
EPA Emission Factors X X X
California Assembly Line X X X
DuPont Fleet X X X .,
Mobil Oil Co. Fleet X X XX
Southern Calif. Auto Club X
Union Oil Co. Tests X .
EPA Dynamometer vs. Track X X
EPA Restorative Maintenance X , X X
General Motors Tests X X X
EPA, Subcompact Car Tests X; *
EPA Selective Enforcement Audit X X X
1. Sources of Vehicle Slip Data
a. EPA Emission Factors Program - Under the Clean Air Act, EPA is
responsible for surveying air pollution sources and quantifying their
emission characteristics. The resulting emission factors are published
by EPA for emission source inventorying, leading to the development of
regional air quality models and control strategies. Administered by EPA's
Emission Control Technology Division in Ann Arbor, Michigan, the EPA
Emission Factors Program involves testing of privately-owned cars using
the same dynamometer test procedures as are used for EPA's emission
certification and fuel economy labeling programs.
The tests are conducted under contract by independent: test labs in a
number of metropolitan areas; the data analyzed came from the fiscal year
1974 and 1975 EF programs, in which tests were conducted in Los Angeles,
Phoenix, Denver, Chicago, Houston, St. Louis, and Washington, D.C.
-------
The test fleet for each of the sites comprises essentially the same mix
of model years and vehicle configurations. All of the test vehicles are
privately-owned, and are tested in the as-received condition. Most
models are represented by more than one test car. The size of this data
set is as follows:
59
Model Year
Number of Vehicles
Domestic Import Total
Number of Model
Configurations
1974
1975
1976
424
798
490
87
191
107
511
989
597
165
355
194
The figure below is an example of the data for one of the more "well-
populated" vehicle configurations. The graph shows the fuel economy
values plotted against odometer mileage, and a linear regression line
through the data. The harmonic mean MPG for the set is also shown, as
is the EPA certification MPG.
FIGURE
MANUFACTURER
TOTAL NUMBER
24
VIN
1S87H5N56
1C29H5R43
1037H5B43
1037H5D4]
1H57H5R47
1C29H5R46
1087H5N60
1H57H5A51
1H57H5R44
1H57H5K40
1H57H5R50
1D37H5B40
1S87H5NS3
1H57HSK41
1H57H5K41
1Q87H5N59
1037H5R45
1029H5R48
1C37H5R42
1SB7H5N5S
1S87H5N59
1H57H5B45
1H57H5B49
1D37H5R43
VEH*
5250
5266
5254
5020
5257
501fl
5249
5252
5253
5262
5255
5266
5259
5255
5252
5254
5018
5252
5018
5249
5249
5253
5018
5019
12. Example of EPA Emission Factors Program Fuel Economy Data
DIVISION
CID WEIGHT C»RB TRANS
350 4000 2 1
Cert Family: 10123
HARMONIC AVERAGE- CertMPc'- 1418
URBAN FUEL ECONOMY Cert MPG - 14 18
13.5
SITE MILEAGE
CHIC 7699
STLO 14206
CHIC 6870
STLO 34843
STLO 5984
STLO 11592
STLO 2845
HOUS 2259
HOUS 6941
STLO 21953
CHIC 1439
CHIC 9744
STLO 19814
STLO 2714
CHIC 77H5
STLO 4138
HOUS 8455
STLO 14343
CHIC 4756
CHIC 4717
HOUS 5677
STLO 7818
WASH 12517
STLO 40900
URBAN 16
12.6
14.5
13.1
14.8
13.9
13.7 M, 15
12.9 Q;
12.2 *
13.2 g-
15.8 0
13.8 o
13.3 lii 14
14.4 "5
13.5 if
12.5 fc
13.0 «•
12.6 R
14.9 13
13.4
13.5
12.9
13.0
13.5
14,0 n
1 1 B 1
^
•~~ ^^
^^
^p
" • S
(Cert) • ^^
A ^m ^jp
— _ _^^$^ ~
• g ^^**
• t S+
^^^ A
^^ f
— m^^^ —
• e i
St. Louis
^ • Other Sites
• A Cert. Value
1 1 1
|
0 10 20 30 40
Miles, 1,000
-------
60
b. California Assembly Line Tests - The California Air Resources
Board (CARB) program for assembly line testing of production cars,
authorized under Title 13 of the California Administrative Code, has as
its objective the emission testing of new vehicles, with primary attention
given to those engine families which just barely meet the California
emission levels in emission certification tests.
The cars are tested new, prior to being put into everyday use; but are
not literally right off the assembly line: the CARB permits each manu-
facturer to specify the mileage accumulation it considers necessary to
overcome "green engine" emissions variability, and pre-test mileage is
accumulated using a CARB-approved durability driving schedule. The
manufacturer-specified mileage accumulation schedules vary from 20 to
120 miles, so the cars are still quite new compared to certification EPA
prototype cars, which are aged 4000 miles prior to the EPA tests.
The California Title 13 data base covers 270 cars, as follows:
Number of Vehicles
Number of Model
Model Year Domestic Import Total Configurations
1974 59 5 64 31
1975 94 30 124 50
1976 79 3 82 41
The test data consist of three replicate EPA City tests for each
car. Highway cycle tests were not run. Minimum mileage on any test car
was 31 and maximum mileage was 153.
59
94
79
5
30
3
64
124
82
-------
61
38
c. DuPont Fuel Economy Fleet - Since 1970, the DuPont Petroleum
Laboratory annually has purchased and tested small fleets of cars to
determine year-to-year changes in fuel economy and acceleration performance,
Each of the 1974-76 fleets consisted of 20 to 23 cars from the four
major domestic manufacturers; small, intermediate, and large cars were
included, with model selection approximating the U.S. sales distribution
for these manufacturers. At the time of the DuPont dynamometer tests,
all the cars for each model year had been driven at least 3000 miles,
and most had accumulated more than 5000 miles. All cars were tuned to
manufacturer's specifications for the tests. Fuel consumption was
measured gravimetrically for both the City and Highway tests, and also
by the EPA exhaust carbon balance method for the City tests. Fourteen
pairs of the 1974 and 1975 model test cars were also monitored for on-
road fuel economy in several thousand miles of consumer use following
the DuPont dynamometer tests.
d. Mobil Oil Co. Fuel Economy Tests - Similar to the DuPont
tests, the Mobil Research and Development Laboratory conducts an annual
test program on small fleets (up to 11) of company cars. Mobil's
primary test objectives are related to vehicle driveability, octane
requirements, and emission levels, but EPA City and Highway fuel economies
30
are measured in the emissions tests. The Mobil data cover 37 cars
from GM, Ford, and Chrysler, plus one AMC car and three imports.
Each of the cars was driven in consumer use by a number of Mobil
employees—as many as seventeen different drivers per car—and the re-
sulting road fuel economy data recorded.
38
Cantwell, et al, "Demographic and Engineering Factors Affecting
Gasoline Utilization", American Petroleum Institute Report 17-76,
May 1976.
Unpublished.
-------
62
For some model years, the dynamometer tests were conducted at low
odometer mileage, while for others the vehicles had been in service for
several thousand miles prior to the tests.
e. Southern California Auto Club Fleet - The Automotive Engineering
Department of the Club (an American Automobile Association affiliate)
regularly evaluates club-owned vehicles for various aspects of technology,
40
including exhaust emissions and fuel economy. Dynamometer tests
yielded carbon balance and gravimetric EPA City and Highway MPG values
for 40 1975-model GM and Chrysler cars. The cars were tested as received,
essentially new; average accumulated mileage was 43 miles at the time
of the tests.
In-service fuel economies for 92 cars of the same configurations as
the test cars were also determined.
41
f. Union Oil Co. Fuel Economy Tests - These tests were conducted
by the Union Oil Company to determine the fuel economy characteristics
of many of the more popular domestic and imported car models. The
models chosen for the tests were those judged most popular by Union Oil
based on 1974 production figures. All cars were purchased new specifically
for the test program. All were 1975 models.
Each car was broken in for a minimum of 2000 miles, and subjected
to dynamometer tests using the 1975 EPA City and Highway procedures to
verify that its emissions were within the levels of the 1975 emission
42
standards
2100 miles.
42
standards . Maximum mileage allowed at the start of track testing was
40
Appleby, et al, "Comparisons of Exhaust Emissions and Fuel Consumption
Characteristics—1974 and 1975 California Automobiles", SAE Paper
760581, August 1976.
41West, et_ al, Op. Git. (21)
/ 0
Some cars failed to meet the standards on the first test, and had to
be retuned to manufacturers' specifications and retested.
-------
63
The dynamometer test data base from the Union Oil tests includes EPA
City and Highway figures for each of 106 production cars: 86 Domestic
and 20 Import.
Track tests were based on the SAE J1082 Road Test Procedure, which
describes urban, suburban, and interstate driving cycles to be performed
on a test track or road course.
g. EPA Dynamometer vs. Track Project - This was a 3-year EPA in-
house test project aimed primarily at comparing the fuel economy of
production cars as tested per EPA dynamometer procedures with the fuel
economy they achieve when driven outdoors, on a test track, using the
same EPA driving cycles. In the first phase of the project, six 1975
cars were tested at the EPA Ann Arbor lab and on the Ohio Transportation
Research Center test track, using the standard EPA City and Highway
A3 44
cycles . In the second phase , seven 1976 cars were tested at the
same two sites, and additional tests were run to evaluate the fuel
economy effects of vehicle warmup, tire type (radial vs. bias-belted),
modifications of the standard EPA Highway cycle, air conditioner operation,
open-windows driving, and road curvature.
Odometer mileage for the Phase I cars varied from 3700 to 12,700 and
averaged 8300 miles. For the Phase II cars, it averaged 5400 miles and
ranged from 3600 to 10,200.
All cars were tuned to manufacturer's specifications prior to testing;
dynamometer tests included, along with carbon balance fuel measurement,
the use of fuel flowmeters for direct comparison with the track tests.
Each car in Phase I received single EPA City and Highway tests, at both
test sites; each Phase II car received triplicate tests at both sites.
43Austin, Op. Cit. (20)
44
Unpublished
-------
64
h. EPA Restorative Maintenance Program - because more than half
of all 1975 and 1976 cars received in the Emission Factors (EF) program
were failing at least one of the three emission standards, EPA initiated
the Restorative Maintenance (RM) Program to identify possible causes of
the emissions malperformance. A second objective was to quantify Individual
and combined impacts of maladjustment and disablement on exhaust emissions
and also on fuel economy.
The 1975-76 RM cars, manufacturered by Chrysler, Ford, and GM were
privately owned vehicles less than 12 months old, and were tested by
independent test labs in Chicago, Detroit, and Washington D.C. The 300
45
1975-76 cars had odometer mileages ranging from 700 to 14,800, averaging
some 8000 miles. The 1977 RM program involved 81 cars with 2300
average accumulated miles, tested in Denver, Detroit, and Los Angeles.
The cars were tested as received, and again after each of three pro-
gressive maintenance steps, culminating in a complete tune-up. EPA City
and Highway tests, and several "short cycle" tests were performed.
i. General Motors Production Car Tests - GM furnished to EPA the
results of their review of fuel economy data from several dynamometer
47
test programs on production cars . The GM data included zero-mile pro-
duction line audit data as well as data on broken-in vehicles (4000 to
10,000 miles) from all five of GM's Divisions.
48
j. EPA Subcompact Car Test Project - The purpose of this project
was specifically to evaluate vehicle slip on low-mileage Subcompact cars.
Bernard and Pratt, "An Evaluation of Restorative Maintenance on Exhaust
Emissions of 1975-1976 Model Year In-Use Automobiles", EPA Report EPA-
460/3-77-021, December 1977.
46
White, "An Evaluation of Restorative Maintenance on Exhaust Emissions
from In-Use Automobiles", SAE paper 780082, February 1978.
Unpublished.
Hutchins, "An Evaluation of the Fuel Economy Performance of Thirty-One
1977 Production Vehicles Relative to their Certification Vehicle Counterparts",
Report 77-18, Technology Assessment and Evaluation Branch, ECTD, EPA,
January 1978.
-------
65
Subcompact cars were selected because (1) with their higher fuel economies,
there was greater potential for detecting any differences between produc-
tion car and certification car fuel economies, and (2) higher-MPG cars
had been reported as having larger in-use MPG shortfalls.
Eleven models representing the fuel economy leaders of the 1977 Subcompact
class were selected. Generally each specific car selected was the fuel
economy leader within that individual manufacturer's model line. Thus,
the project was directed toward the highest fuel economy vehicles in the
Gas Mileage Guide and Was not designed to be representative of the wide
range of model offerings (and fuel economies) in the Guide.
Five of the models were Domestically-produced, five were Imports, and
one was manufactured overseas for Domestic retailing. For each model,
two privately-owned cars and one manufacturer-furnished car were tested.
All vehicles had between 3200 and 8800 miles of accumulated mileage
(average 5400), and were tuned to manufacturer's specifications prior to
the tests. Manufacturer representatives were invited to, and did,
participate in the check-in inspection of all test cars. Triplicate
runs of the EPA City and Highway tests were performed on each vehicle.
k. EPA Selective Enforcement Audit - EPA's Enforcement arm conducts
a continuing program to monitor the emissions of new, very low-mileage
production cars. Odometer mileages range from zero to 4,000, with most
49
cars tested at about 100 miles. Test data include single EPA City
tests on multiple-car samples of individual models from GM, Ford,
Chrysler, American Motors, and eight foreign manufacturers. The size of
this data base is as follows:
Number of Vehicles Number Q{ ^^
Import Total Configurations
215 20
43 266 27
27 246 26
49
Unpublished.
Model Year
1977
1978
1979
Domestic
215
223
219
-------
66
2. Odometer Mileage
The fleet average vehicle slips for the data sources above vary
considerably, from a low of .80 (a 20% shortfall) to a high of 1.06 (a
6% overage). However, the variation is not random. For instance, the
lowest (worst) vehicle slip value comes from a very low mileage fleet,
and the highest (best) value comes from the fleet with the highest
odometer mileages. When the fleet vehicle slips are arrayed by model
year, in order of increasing mileage, it becomes clear that there is a
definite relation between vehicle slip and odometer mileage:
Model
Year
1974
1975
1976
1977
1978
1979
Data Source
California Assembly Line
DuPont
Mobil Oil
EPA Emission Factors
Southern California Auto Club
California Assembly Line
Union Oil
DuPont
EPA Restorative Maintenance
EPA Dynamometer/Track
Mobil Oil
EPA Emission Factors
California Assembly Line
Mobil Oil
DuPont
EPA Dynamometer/Track
General Motors
EPA Restorative Maintenance
EPA Emission Factors
General Motors Audit
Mobil Oil
EPA Enforcement Audit
EPA Restorative Maintenance
EPA Subcompact Cars
General Motors
General Motors Audit
EPA Enforcement Audit
EPA Enforcement Audit
Average
Odometer Miles
VLOO
>5,000
12,000
27,400
43
^100
2,000
>5,000
8,000
8,300
10,000
13,300
^100
4,000
>5,000
5,400
^7,000
8,000
11,500
•U)
^0
116
2,300
5,400
^7,000
^0
177
428
Vehicle
Slip
.94
.95
1.00
1.06
.80
.86
.92
.93
.98
.99
1.04
.97
.86
.90
.91
.98
.98
.98
.96
.88
,89
.91
.95
.96
.99
.88
.94
.94
-------
67
The tabulated vehicle slip and odometer mileage values are depicted in
the following figure. Based on the source average data points alone,
one might conclude that post-1974 vehicle slip, and hence production car
dynamometer fuel economy, are rising at such a rate as to double by the
time some 50,000 miles have been accumulated!
Fortunately, the EPA Emission Factors data bases are large enough, and
have a wide enough range of odometer mileages, to permit a more careful
analysis of the odometer influence. These analyses were made for the
1974 and 1975 Emission Factors fleets, and do in fact lead to conclu-
sions that are more palatable. The "best-fit" curves for 1974 and 1975
are drawn on the figure, and indicate that indeed there is a sharp rise
in vehicle slip (and in absolute fuel economy) at very low mileage,
FIGURE 13. Relation Between Vehicle Slip and Odometer Mileage
1.10
1.05 -
1.00 -
.80
10.000
20.000
30.000
Odometer Miles
40.000
50.000
60.000
-------
68
but this rapid rate of increase is only temporary: it begins to level
off above 4,000 miles. In these tests, wherein the vehicles are all oper-
ated the same way regardless of age, we see no evidence of a point where
fuel economy peaks and declines thereafter.
The other significant finding from the Emission Factors data is that,
for any mileage, vehicle slip is worse for the 1975 models than for the
1974's. The break-even point — i.e., the point at which average
production car MPG is equal to the EPA value — is only 2,600 miles for
1974 models, but is not reached until some 38,000 miles for the 1975's.
3. MPG Tilt
Since the DOE data suggest that vehicle slip is worse for higher-MPG
cars, those EPA data sources which presented car-by-car data were
analyzed for this effect. Again, the results show considerable variance
from source to source but here, too, we find a dependence on odometer
mileage. Figure 14 shows the vehicle slip patterns for the data sources
that include broken-in cars (those with odometer mileages of 4,000 or
more). Individual car data points are given for two of the sources, for
illustration. It will be noted that the severe MPG-level dependences
inferred by some studies involving only a few cars are not fully supported
by the larger EPA and DOE data bases. All of these sources do show some
worsening of vehicle slip for higher-MPG cars.
In Figure 15, however, which isolates the very low-mileage cars, there
is no such worsening of vehicle slip with MPG level — with the exception
of the 1974 California Assembly Line and 1978 EPA audit data. It thus
appears that the "MPG tilt" in vehicle slip is a phenomenon that appears
only after vehicle break-in.
-------
69
1.20
FIGURE 14. Vehicle Slip vs. EPA MPG
Data Sources with 4,000 or More Odometer Miles
I
T
1.10 -
1.00
(I 00)
•^^.m ^ ™ ^^^m •• »^
••^.i^*«"^...IF75 OOE7? ~ f
^•^J^ri""-":::.-.-.-.*!
A ^*~.^-^ ^^OE 77 EF"76*
.90
80
70
10
IS
25 30
EPA 55/45 MPG
35
45
1.10
FIGURE 15. Vehicle Slip vs. EPA MPG
Data Sources with Less than 4,000 Odometer Miles
I
I
1.00
(I 00)
Union Oil 75 II.OOOHi I
.90
.80
70
I
10
15
20
25 30
EPA MPG
35
40
45
-------
70
Now, there are certain known operational and maintenance differences '
between cars, as a function of vehicle and/or engine size, which can
affect fuel economy after break-in. These differences also correlate
with MPG level, since car and engine sizes are closely related to fuel
economy. However, these observations do not fully explain the vehicle
slip behavior of the specific data sources discussed above.
It is also well known that automatic transmissions predominate among the
larger, lower-MPG cars, while manual transmissions are more common to
the smaller, higher-MPG cars. If cars with manual transmissions charac-
teristically had worse vehicle slips than those with automatics at the
same EPA MPG level, this shift in transmission mix would handily explain
the MPG tilt in vehicle slip. Our analysis of the 1974-76 Emission
Factors data and the 1977 Subcompact car data, however, show that it is
the automatics which generally have worse vehicle slips, as illustrated
in Figure 16. If anything, the transmission mix shift would have the
effect of flattening the tilt, not of causing it.
A 1973 Canadian driver survey,(Canada Department of the Environment,
"Canadian Automobile Driver Survey", Report EPS 3-AP-73-10, October
1973) indicates that smaller engines are tuned up more frequently than
larger ones. The survey also reveals that larger vehicles accumulate
more miles annually than smaller cars.
The 1974-76 EPA Emission Factors data show higher average odometer
readings for heavier cars. For each of these three model years, each
additional 100 pounds of vehicle weight corresponds to 100 to 200 addi-
tional miles driven per year.
-------
71
FIGURE 16. MPG Tilt in Vehicle Slip Factor vs. Transmission Type
I 30
1.20
Q.
i i 10
1.00
0.90
Minull
1974-76 Data from
EPA Emission Factors
1977 Data from
EPA Subcompact Car
Test Project
-------
72
Our findings on odometer mileage and MPG tilt effects are summarized
below:
0 Dynamometer fuel economy increases with vehicle odometer mileage,
with the rate of increase being quite high at very low mileage and
leveling off after 4000 or 5000 miles;
0 For 1974 models, only those cars with low odometer mileages show an
MPG shortfall due to vehicle slip: after some 2500 miles, the average
1974 vehicle slip rises above 1.0, i.e., average production car
dynamometer MPG becomes greater than EPA MPG;
0 For post-1974 models, vehicle slips (at comparable mileages) are
generally worse than those of 1974 cars. However, there is no clear
pattern at all that vehicle slips for post-1975 models may be continu-
ally getting worse year by year;
0 Vehicle slip does not vary with MPG level (i.e., worse at higher MPG)
for very new cars; MPG tilt is a phenomenon peculiar to broken-in
vehicles. The various data sources examined are not in agreement as
to just how strong this dependence is, but they all support the
conclusion that there is some MPG dependence. The data bases containing
the largest number of cars generally show less vehicle slip MPG tilt
than those with only a few cars.
-------
73
4. Production Slip
Any discrepancy between EPA fuel economy ratings and the fuel economy
measured on newly-manufactured cars is referred to herin as "production
slip". Viewing the "production" of vehicles in the broad sense, administrative
factors which influence MPG are included along with influences that are
strictly of a manufacturing nature.
a. Administrative Variance - Since in-use MPG shortfall analyses
virtually always use the EPA/DOE Gas Mileage Guide/Fuel Economy Label
MPG numbers as the comparison basis, it is important to review briefly
how these numbers are generated.
Approximately a year before the scheduled start of a new model year, the
vehicle manufacturers specify to EPA all of the vehicle types they
intend to market, in accordance with regulations for emission certification
52
(40 CFR Part 86). Vehicle configurations to be tested for emission
certification are selected on the basis of several factors, most notably
projected sales volumes and design features to which emissions are
sensitive. Within each group of similar configurations, for example,
attention is given to selecting for test some configurations most likely
to be worst-case emitters. When these cars are tested for emission
certification, their EPA city MPG values are determined; in addition,
EPA highway tests are run for fuel economy purposes.
Besides the emission certification vehicles, additional ("fuel economy
data") vehicles are tested over both EPA cycles to further fill out the
fuel economy data base. Although not official emission data vehicles,
52
A configuration is a unique combination of basic engine, engine code,
inertia weight class, transmission type (manual or automatic), number of
forward gears, and axle ratio; "basic engine" means a unique combination
of combustion type (spark ignition, or compression, i.e. Diesel, ignition),
emission standard (49 states or California), number of cylinders, dis-
placement, fuel system (number of carburetor barrels or fuel injection),
and catalyst usage; "engine code" means a unique combination of emission
control system, auxiliary emission control devices, and specific set of
carburetion and timing calibrations, within a basic engine; inertia
weight class is the weight class to which a vehicle is assigned based on
its loaded vehicle weight.
-------
74
fuel economy data cars must demonstrate compliance with emission stand-
ards, including the application of emission deterioration factors.
The generation of published fuel economy values from these test data
proceeds according to the fuel economy regulations (40 CFR Part 600), as
follows: the test data from all configurations sharing a common basic
engine, inertia weight, and transmission class are combined (harmonically
sales-weighted) to establish MPG values for that configuration group, or
"base level"; next, the MPG values for all base levels constituting a
given "model type" (combination of nameplate, basic engine, and transmission
class) are combined to arrive at the MPG rating for that model type. It
is the model type MPG values, rounded to the nearest whole mile per
gallon, that appear on the Labels and in the Guides. The following
figure illustrates the overall process schematically.
FIGURE 17. Fuel Economy Calculation Flow: from Test Data through Corporate Average
Model Trpes
(Calculated)
Label/Guide
Values.
Rounded
Bite Levels
(Calculated)
Configurations
(Tests)
= Untested )
-------
The next figure is an example of four configurations building into a
base level; the figure includes MPG values and sales volumes, and shows
how they are used in the base level calculations.
75
FIGURE 18. Example of Configurations Combining into a Base Level
Harmonic Salet—Weighted Average:
(City MPG)
Configuration (a)
140 CIO. 3 bfe
49 SUMS
3 M I «,.!«
2 nOLbs
H8k.HJ920.Co 1110
(ll.Tot SalH Sites (blToi Sale-
* "MPG'ibj'
Configuration (b)
)«CID. Ihbl
J Of I Axk
2.7MLH
UK, CUM. H JO 08. Co 31 Bi
Configuration (c)
I40CID. Ibbl
EnfiK Codt E If
rl-4 Transmission
1.75
Axk
DLb*
MPG C 10 II. H 11 75. Co 24 iS
Configuration (d)
I4DCID. 2UW
Engine Codt E 3f
M-* Trintrmuor
) 26 I A
-------
76
FIGURE 19. Example of Base Levels Combining Into Two Model Types
:s
00
Base Level (2)
HOCID 2bbl.
49 Stiui
M-4 Tranimiiiion
2.750 Lb«
19 3 1/29 87/22 96 MPG
Sa
27.
^
Since the Label and Guide MPG values are determined and published very
early in the model year, any changes to the configuration MPG's (via
running changes or configuration deletions or additions) or variation
from the sales volume forecasts can result in model type average MPG's
different from the early Label/Guide values. The magnitude of any MPG
slip due to such changes has not been evaluated at the model type level
for prior model years. We have examined fleet level MPG changes due
only to sales mix shifts (actual vs. projected) for model years 1975
53
through 1977 , with the following results:
53.
Murrell, Op. Cit. (18)
-------
77
Change from Projected-sales MPG to Actual-sales MPG
(Total Fleet)
1975
+1.3%
1976
-1.1%
1977
-1.6%
These comparisons apply only to changes in the sales mix, and include the
effects of sales reproportioning among the manufacturers. End-of-year Corporate
Average Fuel Economy (CAFE) calculations made for compliance with the fuel
economy standards will provide a much better basis for evaluating the combined
slip effects of sales shifts and MPG changes, beginning with model year 1978.
Our preliminary estimate of these combined effects for 1978 (with 15 manu-
facturers, accounting for 96% of sales, evaluated) is +0.9%. "Running changes",
which can cause differences between pre-model year and post-model year fuel
economy, are receiving close scrutiny in these evaluations.
Since individual consumers buy individual cars, not base levels or model
types, inherent individual slips from the listed model type MPG are bound to
occur. The previous examples can be used to illustrate the possible magnitude
of such individual car slips. The table below lists the MPG slips for each of
the four configurations in the example's base level(2), with respect to each
model type that incorporates those configurations. The effect of round-off of
the model type MPG is also illustrated.
55/45 Administrative Slip,
Example Base Level (2)
Rounded Basis:
Model "A" Model "B"
Unrounded Basis:
Model "A" Model "B"
Config.
Config.
Config.
Config.
(a)
(b)
(c)
(d)
.967
.950
1.023
.835
1.009
.992
1.067
.871
.986
.969
1.043
.851
1.002
.985
1.060
.865
MPG Slip, all
configs, sales- .957
wtd harmonic
MPG Slip, Base .957
Level
.998
.998
.976
.976
.9-92
.992
-------
78
Some individual cars' EPA test MPG in this example are seen to vary from
the model type average values by 10% or more. While the example is
hypothetical, the assigned MPG values are not atypical.
Naturally, the sales-weighted average administrative slip for the entire
set of model types is 1.0 (no average shortfall or overage), as shown in the
next table.
55/45 Administrative Slip
(Model Type)
Rounded: Unrounded:
Model "A" Model "B" Model "A" Model "B"
Base level(1)
Base level(2)
Base level(3)
All Base Levels .981 1.006 1.000 1.000
^^ 1.000
I.o02
.957
.936
1.045
.998
.977
1.022
.976
.955
1.038
.992
.971
By definition, administrative slip for all model types using the same
base levels must average 1.00. With some individual cars deviating from
Label/Guide model type value by significant amounts (on the order of +
10%) however, a sales shift — or a sampling bias which overemphasizes
configurations with large administrative slips—could easily create an
average administrative slip of several percent in either direction for
that sample.
b. Hardware Variance - A precise measure of hardware variance can
come only from comparison of the dynamometer test fuel economy of new,
low-mileage production cars against the low mileage MPG of those specific
EPA Certification cars which exactly match the configurations of the
production cars. Only in this way can the effects of mileage accumulation,
vehicle modification, vehicle usage, and other post-manufacturing treatment
(or mistreatment) of the cars be prevented from clouding the comparison.
The following table illustrates the hardware variances, under this
definition, from sources of low-mileage dynamometer test data presented
earlier.
-------
79
Production Hardware MPG slip, by Model year
Model Year
1974
1975
1976
1977
1978
1979
Data Source
Calif. Assembly Line
Sou. Calif. Auto Club
Calif. Assembly Line
Union Oil - Calif. Cars
1975 Average, Calif. Cars
Union Oil - 49 States Cars
Calif. Assembly Line
Mobil Oil
GM Audit (8 models)
EPA Audit
1977 Average
GM Audit (6 models)
EPA Audit
1978 Average
EPA Audit
Number
of cars
64
7
266
246
0.984
1.015
Hardware
Slip
1.006
40
124
13
81
82
11
?
215
0.869
0.921
0.924
0.994
0.985
1.037
0.909
0.928
0.920
1.033
1.014
1.012
These data suggest U.S. average hardware shortfalls of about 8% in 1975
and 1976, but no hardware shortfalls for 1974 or 1977-79. Since most of
the 1974-76 data are from California cars, it may not be accurate to
assume that the indicated slips apply to all cars of those model years,
although the 1975 non-California cars do show shortfalls comparable to
the California cars.
As shown in the next figure, the EPA audit MFC's on average compare very
well with zero-mile certification data for domestic cars and low-MPG
imports. For imports above 20 MPG (EPA City), however, the production
car data are clustered around a line some 2.5 MPG (10%) below the certi-
fication values. The consumption regression curve for all three model
years is within 5% of equality with the Certification data over the
entire MPG range, and lies above the equality line up to 22 MPG.
-------
80
FIGURE 20. MPG Comparison: New Production Cars vs. EPA Prototypes
(EPA Audit Data, EPA City Test)
30
25
20
15
10
I
I
I
I
I
Average Odometer:
• 1977: 116 miles
A 1978: 177 miles
• 1979: 428 miles
(Data points represent domestic
models; only max. and min.
values are shown for each model
for clarity.)
Envelope for all
Import Model Data
(Average Odometer = 10 miles)
- - - - Curve Fit: •=-
+0.0037
10 15 20
EPA Certification Test Car Zero-Mile MPG
25
30
When the EPA Audit data are stratified by manufacturer, the results given
in the next table are obtained. These data reflect comparisons of pro-
54
duction car MPG's with Certification car MPG's interpolated to exactly
the same odometer readings, model-by-model, as the audit data cars (i.e.,
the most accurate method of comparison). For all 727 cars, a 2% production
overage in MPG is indicated; the domestic makes show a 3% overage, and the
import cars show a 7% hardware shortfall.
54
Using each model's zero-mile and 4000-mile Certification MPG values.
-------
81
Summary of EPA Audit Data, by Manufacturer
Manufacturer
AMC
Chrysler
Ford
GM
Year(s)
1077
77-78-79
77-78-79
77_7P,_7q
f?umber
Models
1
13
17
33
Number
Cars
27
147
167
316
Fiat
Honda
Mercedes-Benz
Porsche
Renault
Saab
Subaru
Toyota
Volkswagen
1979
1979
1978
1979
1978
1973
1979
1978
1978
1
1
1
1
1
1
1
1
1
f,
8
9
7
6
9
6
8
11
Production Hardware Slip
(.Sample-Wtd)
1.121
1.018
1.044
1.021
Domestic Subtotal 64 657 1.030
Import Subtotal 9 70 0.931
Overall Average 1.020
The next figure illustrates the MPG distribution curves for several of
the models audited. Those models depicted are the ones with either the
maximum or minimum dispersions, or best or worst average slips, for
their EPA MPG range; all of the models shown had a test sample size of
at least eight. For any EPA MPG level, the widest spreads are approx-
imately + 2 MPG. On a percentage basis, a maximum spread of + 2 MPG
represents a + 20% variation for 10 MPG cars, and a + 7% variation for
30 MPG cars; in terms of fuel consumption, a 20% MPG error at 10 MPG
corresponds to an error of 0.02 gallons per mile, while a 7% MPG error
at 30 MPG is a consumption error of only 0.002 GPM.
-------
82
FIGURE 21. New Production Car MPG Distributions
x 1978
\ Mercedes
• Diesel
f' (Zero mi.)
18 20 22 24 26
EPA Test Car City MPG @ Production Car Mileage
-------
81
Summary of EPA Audit Data, by Manufacturer
Manufacturer
AMC
Chrysler
Ford
GM
Year(s)
1077
77-79-79
77-78-79
77-78-70
Number
Models
I
13
17
33
Number
Cars
27
147
167
316
Domestic Subtotal
64
657
Production Hardware Slip
(.Sample-Wtd)
1.030
Fiat
Honda
Mercedes-Benz
Porsche
Renault
Saab
Subaru
Toyota
Volkswagen
1970
1979
1978
1970
1978
1973
1970
1978
1973
1
1
1
1
1
1
1
1
1
6
8
9
7
6
9
6
8
11
Import Subtotal
70
0.929
0.905
0.863
1.047
0.863
1.031
0.890
0.971
0.871
0.031
Ove ra 11 A ve r age
1.020
The next figure illustrates the MPG distribution curves for several of
the models audited. Those models depicted are the ones with either the
maximum or minimum dispersions, or best or worst average slips, for
their EPA MPG range; all of the models shown had a test sample size of
at least eight. For any EPA MPG level, the widest spreads are approx-
imately + 2 MPG. On a percentage basis, a maximum spread of + 2 MPG
represents a + 20% variation for 10 MPG cars, and a + 7% variation for
30 MPG cars; in terms of fuel consumption, a 20% MPG error at 10 MPG
corresponds to an error of 0.02 gallons per mile, while a 7% MPG error
at 30 MPG is a consumption error of only 0.002 GPM.
-------
82
FIGURE 21. New Production Car MPG Distributions
18 20 22 24 26
EPA Test Car City MPG @ Production Car Mileage
28
30
-------
83
Since Selective Enforcement Audit testing is a continuing effort, additional
data which has been acquired since the publication cutoff date has not been
evaluated. However, we have no reason to believe that the S.E.A. data
reported here is not typical of the now-larger data base.
All of the above apply to hardware variance with respect to EPA City
MPG; Highway tests were run by only two of the data sources, Union Oil
(1975) and Mobil Oil (1977). Hardware variances for the Highway cycle
can be derived from these two sources, but only as an estimate, since
zero-mile highway MPG data were not available for the certification
cars. These estimated Highway slips are shown below, and are reasonably
in line with the values presented earlier for the City MPG values.
Production Hardware Slip, EPA Highway Test
49 States California All
Union Oil (1975)
Domestic .956 .959 .957
Import .973 .908 .965
All .959 .950 .958
Mobil Oil (1977)
Domestic .993
Import .990
All .992
-------
84
5. Vehicle Condition (Test)
This section examines those fuel economy influences for which available
data is limited to MPG effects measured in dynamometer tests.
a. Engine Tune - the effects of engines' "state of tune" on fuel
economy have been widely investigated, the consensus being that engines
out of adjustment with respect to manufacturer specifications suffer
fuel economy penalties. Studies of EPA emissions surveillance data on
pre-1974 models have shown an average improvement of 6% in MPG for tuned
cars, compared to their "as-received" condition. For later model cars
the overall effect of maintenance can be estimated by comparing in-use
to EPA vehicle slip factors from initial (as-received) tests and tests
after maintenance, from th<
resulting MPG changes are:
after maintenance, from the EPA Restorative Maintenance data ' . The
Fuel Economy Change Resulting
from Vehicle Tuneup
City Highway
300 1975-76 models +1.9% +0.5%
81 1977 models +0.1% -0.8%
The aggregate maintenance effect on MPG for all of these RM cars was a
net improvement of 1.5% City and 0.2% Highway. Since some of the cars
required no maintenance, all of this average improvement came from those
which did, hence that fraction of the cars which actually received main-
tenance had higher percentage improvements.
Austin and Hellman, "Passenger Car Fuel Economy - Trends and Influencing
Factors", SAE paper 730790, September 1973.
56Bernard and Pratt, Op. Cit. (45)
57White, Op_._ Clt. (46).
-------
85
58
MPG effects of specific malfunctions as reported by Panzer"
59
and by Toulmin are listed in the next table, and indicate that carburetion
by White,
bhat
and spark system problems have the most significant MPG impacts.
Percent Effect on Fuel Economy
of Indicated Malfunction
Malfunction
One spark plug misfiring
Air/Fuel ratio too rich
Ignition timing retarded
Idle A/F rich
Plugged PCV
Choke rich
Idle RPM high
Distributor vacuum low
Idle A/F lean
Ignition timing advanced
EGR disabled
Air Pump disabled
Choke heater disconnected
Idle RPM low
Panzer
(1975's)
City Highway
-13
-11
-6 (8°)
-7
-4
-3
-3
-1
*
* (8°)
X
X
X
+3
-15
-12
-4
*
-3
*
*
ft
*
*
X
X
X
*
White57(75-77's)
Cit
X
X
X
-2
X
-2
X
*
V
+2
+1
+ 1
*
X
y Highway
X
X
X
+1
X
-1
X
-1
X
(5°) +1
+1
+1
+2
X
Toulmin (77-80's)
City Highway
X
X
X
-1
X
*
-4
X
X
+2
+4
X
X
X
X
X
X
+1
X
+1
-2
X
X
(5°) +1
+4
X
X
X
* = insignificant effect (<0.5%)
X = not evaluated
58
Panzer, "Fuel Economy Improvements Through Emissions Inspection/Main-
tenance", SAE paper 760003, 1976.
59
Toulmin, "Light Duty Vehicle Driveability Investigation", EPA Report
EPA-460/3-78-012, December 1978.
-------
86
The frequency of occurrence of malfunctions, by engine system, is given
in the next table. In terms of emissions performance, those vehicles failing
emissions tests show generally higher incidences of engine malfunctions.
Percent of Vehicles with Indicated Malfunction
(1975-1976 Models, EPA RM Program)
Air Induction system
Carburetor/Fuel System
Ignition System
Exhaust Gas
Recirculation
Air Pump
PCV System
Evap. System
Miscellaneous
At least one malfunction 50.4
Passing
Vehicles
(>.',
40.8
12.0
4.0
0
0.6
0
0
Failing
Vehicles
6.3
84.0
36.0
23.4
1.1
0.8
2.3
1.7
All
Vehicles
6.33
66.00
26.33
15.33
0.67
0.67
1.33
1.00
Most Frequent
Item3
Air cleaner element
Idle mixture caps
Timing out of spec.
Time delay solenoid
Disabled
PCV Filter
b
Early fuel evap.
91.4
74.33
based on one test site,
not all sites
several items equally frequent
If tuneups are performed on all of a group of cars, whether necessary or
not, mixed results can occur. In a study conducted by EPA , three vehicles
which had been worked on by private garage mechanics and then tuned to
specification by EPA technicians showed the following changes in cold-start
EPA City fuel economy:
Plungis, "A Study of Fuel Economy Changes resulting from Tampering with
Emission Controls", Report 74-21, Test and Evaluation Branch, ECTD, EPA,
January 1974.
-------
87
Measured MPG (1972 City Test Procedure):
1973 Compact
1973 Large Wagon
1974 Midsize (3 mos. old)
As
Received
17.3
7.7
10.5
Tuned to Spec
by EPA
19.3
9.1
10.3
Adjusted by
Garage
18.4
9.0
10.2
The two older cars benefited from the maintenance (by 11.5% EPA, 6.4%
Garage; and 18.2% EPA, 9.3% Garage, respectively), but the newer car's
fuel economy was reduced (1.9% EPA, 2.9% Garage).
In a tuneup study done for DOT, Claffey also reports mixed results for
a group of 22 1970-74 model cars, all of which were given plugs/points/
condenser replacements, plus other work deemed necessary on an individual
basis. The directional changes in MPG are shown in the next table.
Quantitative (percentage) changes were not readily discernible from the
data. Note that the three cars which had never been tuned up did not
always improve in MPG by being tuned.
Mixed Results of Tuneups Given Whether or Not Needed
Effect of Tuneup on MPG:
Stop-and-go Steady Cruise
Number of Cars and Odometer Values
10 Urban Cars 12 Rural Cars
l(12k/7k)
l(29k/7k)
l(98k/25k)
7(39k/12k)
l(6k/6k)
l(23k/23k)
X(Y/Z) = No. Cars (Avg. odometer/Miles since last tuneup)
MPG Better
MPG Better
(No Data)
(No Data)
No Change
MPG Worse
MPG Worse
MPG Worse
MPG Better
No Change
MPG Better
No Change
No Change
MPG Better
No Change
MPG Worse
2(27k/17k)
2(34k/8k)
l(60k/10k)
4(25k/17k)
l(17k/17k)
__ —
Claffey, "Passenger Car Fuel Conservation", DOT Report FHWA-PL-77009,
January 1977.
-------
88
Tuneups may not, in general, have the same effect on all engines. From the
EPA 1975-1976 RM data, tuneup effects estimates (derived from before-
and-after vehicle slip factors) seem to vary with engine size, as follows:
Engine Displacement:
(Cubic Inches)
£225
226-229
300-360
>360
Change in Relative MPG;
Highway
-0.7%
+0.6%
+2.4%
+3.5%
-0.3%
+1.2%
+0.4%
+0.9%
It is not clear whether the smaller engines did not improve in MFC,
because they responded poorly to maintenance or because they were in
less need of maintenance.
A Canadian Survey shows that smaller engines are tuned up more frequently
than larger ones:
Fraction of Cars Receiving Tuneups:
Mileage Interval
Between Tuneups
4-cylinder
Models
6-cylinder
Models
8-cylinder
Models
0 - 3000
3001 - 6000
6001 - 12,000
12,000
19%
48%
30%
3%
18%
45%
31%
6%
10%
36%
44%
10%
Approximate Average
5600
6000
7200
-------
89
The apparent engine size dependency in the RM data could be influenced
by similar "miles-since-last-tuneup" factors.
Many of the foregoing findings emphasize the need for a determination
(through careful inspection) of whether engines in fact need to be tuned
up. By making this kind of determination, the inspection and maintenance
(I/M) programs currently being developed by EPA and the States will
maximize the fuel economy benefits of proper engine maintenance.
f\ 9
In an EPA report the available data on the fuel economy impact of I/M
programs was evalulated in some detail. One basic conclusion was that
the fuel economy benefit of an I/M program depends to a great extent on
the capability of the mechanics who perform the tuneups. Without emission-
oriented mechanic training, the fleetwide fuel economy benefit is estimated
to be negligible; but with emission-oriented mechanic training, we can
expect fleetwide benefits of about 1 per cent for current types of cars
and possibly more for 1981 and later models.
"Effects of Inspection and Maintenance Programs on Fuel Economy",
Report IMS-001/FE-1, Inspection and Maintenance Staff, ECTD, EPA,
March 1979.
-------
90
b. Engine Response to Fuel Properties - One special class of ve-
hicle condition effects requires separate mention: that of relationships
between fuel properties and engine operation, specifically with regard to
fuel economy. Of many fuel properties which have been investigated by
EPA ' , four have the potential for affecting fuel economy to any measure-
able extent:
° Fuel density
0 Octane rating
0 Volatility
0 Additives
Fuel density has a straightforward fuel economy influence, in that the
denser the fuel, the higher the heat content per gallon consumed; in
general this means more available work per gallon, and hence more miles
per gallon for a constant amount of work per mile. The SAE test pro-
cedures' correction factors reflect MPG effects as follows:
MPG(SG) 1
MPG(.?3?) 1 + 0.8(0.737-SG)
where MPG(SG) = MPG with fuel of specific gravity = SG (dimensionless);
MPG(.737)= Reference MPG, with fuel of specific gravity = 0.737
or. MPG (API)
MPG(60.S) 1 + 0.0032(API-60.5)
where MPG(API) = MPG with fuel of gravity66 = API(degrees);
MPG(60.5) = Reference MPG, with fuel of API gravity = 60.5
0
£ O
Bascunana and Stahman, "Impact of Gasoline Characteristics on Fuel
Economy and Its Measurement", Report 76-10, Technology Assessment and
Evaluation Branch, ECTD, EPA, December 1976.
64
Harvey, "Representativeness of Emissions Certification Gasoline", Draft
Report, Technology Assessment and Evaluation Branch, ECTD, EPA, July 1978.
Society of Automotive Engineers, "Fuel Economy Measurement—Road Test
Procedure—J1082b" , January 1979.
66API gravity and specific gravity are related by: API° = — ' • - - 131.5
-------
91
Reference 63 reports a relationship involving fuel density directly:
MPG(P) 47,400 + 10,960P
MPG(P ) ~ 47,400 + 103960P0
where MPG(p) = MPG with fuel of density67 = p(Ib/gallon);
MPG(p o) = Reference MPG, with fuel of density = p
This relation is about the same as the SAE J1082 factors, and indicates
that fuel economy changes by about 10% for each Ib/gal change in fuel
density.
Average density of EPA unleaded test fuel in 1977-78 was 6.16 Ib/gallon
from numerous EPA measurements, to 6.17 Ib/gallon for EPA fuel analyzed
by DOE's Bartlesville Energy Technology Center. By comparison, densities
of commerical fuels, as reported by the Motor Vehicle Manufacturers
Association and DOE for recent years, are:
In-Use Fuel Gravity and Density
Degrees API:
MVMA DOE Average Ib/gallon
Unleaded, summer, 1976
1977
1978
Unleaded, winter,
1975-76
1978-79
The slightly higher summer grade densities would tend to give road MPG
overages from 0.2% to 0.4%, while the winter grade densities would
correspond to road shortfalls from zero to 0.5%. Hence, no significant
net fuel economy effect can be associated with differences between
average densities of EPA unleaded fuel and in-use unleaded fuels.
59.2
58.5
58.9
—
—
59.7
59.3
58.8
61.8
60.3
6.18
6.20
6.20
6.11
6.16
Fuel density and specific gravity are related by: p = SG x 8.3.
£ Q
Harvey, Op. Cit. and Johnson, "Composition and Octane number of U.S.
Motor Gasolines Sampled in the DuPont 1978-79 Winter Road Octane Survey",
DOE Report BETC-0012-1, September 1979.
Average API gravity = 60.2°
-------
92
There is some variation in in-use fuel density which can cause road MPG
differences for individual drivers, or all drivers in certain locales.
The figure below illustrates the distributions of API gravities and
densities for gasolines from a 1972 nationwide sampling . The extremes
of these distributions correspond to fuel economy variances of + 3%.
FIGURE 22. Distribution of API Gravity, 1972
70
J_
6.5 64 6.3 6.2 6.1 6.0
Density, Lb./Gallon
5.9
5.8
The DOE surveys mentioned above reveal density differences from one
region of the country to another: while the U.S. average API gravity
for 1978-79 winter unleaded gasoline was 60.3, regional extremes and
their respective road MPG differences were:
"API Ib/gallon
Seattle 56.4 6.28
Wichita/Oklahoma City/Tulsa 64.3 6.03
MPG Effect vs. EPA Fuel
+1.2%
-1.2%
70.
Phillips Petroleum Co., unpublished.
-------
93
Fuel Octane Rating itself has no direct effect on fuel economy ' ,
that is, supplying higher or lower octane fuel to an engine will not
result in an MPG change as long as the engine runs acceptably on either
fuel . Parameters closely related to fuel octane which do affect fuel
economy, and emissions, are compression ratio and spark timing. Fuel
economy tends to increase with increased compression ratio and increased
spark advance; unfortunately, so does engine knock tendency. To keep
engine knock at acceptable levels while increasing compression ratio
and/or spark advance, fuel octane must be increased or the "mechanical
octane" of the engine must be increased. To control both engine knock
and emissions, the compression ratio, spark advance, fuel octane require-
ments, and other engine design factors must all be optimized.
Up to model year 1974, auto manufacturers included compression ratio
reductions and spark retard in their approaches to emission control, and
72
octane requirements accordingly were eased for these vehicles . For
post-1974 models, improved emission control technology has allowed
optimization for more fuel-efficient compression ratios and spark
advance calibrations. Most recently, turbocharging has been introduced
to take fuel economy advantage of reduced engine displacements.
All of these developments have tended to cause late model vehicles'
octane appetites to increase. These vehicle octane requirements are
typically reported via a graphical "percent satisfaction" relationship,
73
an example of which appears in the next figure
The term "percent satisfied" that appears on the figure needs to be
explained. The tests used to generate these types of graphs are conducted
Cars equipped with knock sensor systems are an exception, as will be
discussed later.
72
Octanes of commercial fuels were not correspondingly reduced, however;
instead, average octane remained relatively constant while customer
satisfaction improved. See: Courtney and Newhall, "A Primer on Current
Automotive Fuels'1, Automotive Engineering, December 1979.
73
Ethyl Corporation, "Automotive Developments '79—A Survey for the Oil
Industry", 1979.
-------
94
FIGURE 23. New-Car Fuel Octane Requirement
98
I
g 92
I
e 90
1974 Cars
1975 Cars
20 40 60
Percent Satisfied
80
100
by trained technicians who are more sensitive to knock than is the
motoring public. Reference 72 states:
"'Trace Knock' is the knock intensity detectable by a trained
technician. Octane requirement measured by a trained rater is
about five or six RON's above the average an average untrained
observer would determine."
A factor which tends to offset this, however, is Octane Requirement
Increase (ORI). It is well-documented that the octane requirement
of an automotive engine increases during service. Reference 64 reports
that octane requirements increase 5 to 12 numbers (RON) in 10,000 to
30,000 miles, after which they remain constant.
74
Coordinating Research Council(CRC), "Influence of Leaded and Unleaded
Fuels on ORI in 1971 Model Cars, Phase I: 1970-71 CRC Road Rating
Program", Report No. 451, September 1972.
CRC, "Octane Requirement Increase in 1973 Model Cars, Phase II: 1973
CRC Road Rating Program", Report No. 476, February 1975.
Niles and McConnell, "Establishment of ORI Characteristics as a Function
of Selected Fuels and Engine Families", SAE Paper 750451, 1975.
Ahlquist, £t a±, "Some Observations of Factors Affecting ORI", SAE
Paper 750932, 1975.
78
Benson, "Some Factors which Affect Octane Requirement Increase", SAE
Paper 70933, 1975.
-------
95
The octane levels for fuel of recent years are shown in the next table.
Summer 1975
1976
1977
1978
Winter 1975-76
1978-79
Octane Ratings of In-Use
Unleaded Gasoline
Research Octane (RON)
MVMA DOE Chevron
72
Motor Octane (MOM)
MVMA DOE
—
92.1
93.1
92.4
—
—
92.1
92.4
92.5
93.0
92.3
92.5
92.0
92.2
92.7
92.8
—
—
84.1
83.4
83.1
—
—
84.1
83.8
83.9
—
83.8
In the worst case, it could be assumed that all of the in-use spark maladjust-
ment that was observed in the EPA Restorative Maintenance (RM) data relates
to dissatisfaction (at the levels in the figure above) with the octane of
the fuel. The resulting computation implies a fleetwide fuel economy
shortfall of 0.6% for this factor. While not an absolute upper bound, we
feel that this is a conservative estimate, i.e., greater than the actual
value that would be calculated if all of the necessary data and all of the
reasons for in-use spark maladjustment were known precisely.
Knock Sensors - Some vehicles are being introduced equipped with knock sensor
systems. In general, these systems incorporate a sensor which detects engine
knock and, based on the signal that indicates that incipient knock is present,
adjusts the spark timing to eliminate the knock.
Whether or not a shortfall in MPG occurs with knock sensor equipped
vehicles depends on several factors. First, the details of the control
logic must be known. Systems could be designed to always operate at
or just below the knock threshold, but others could be designed just to
retard from a fixed (vacuum/mechanical or electronic) timing calibration.
Secondly, the fuel octane that the base calibrations were determined for
must be known. Calibrations could be tailored for EPA's high octane
test fuel or they could be tailored for typical in-use fuel properties.
Thirdly, the octane of the fuel that was actually used for the official
EPA test should be known. Although the vast majority of the tests are
-------
96
run on high octane EPA test fuel, some vehicles with knock sensors have
been tested on lower octane fuel in response to a manufacturer request
for special test procedure. Fourth, the actual in-use spark timing
history of the vehicle, compared to the spark timing that occurred on
the EPA test should be known to allow the overall effect to be evaluated.
For example: a vehicle might have a system that results in a negative
vehicle slip when evaluated on a dynamometer with in-use fuel, but might
also have a positive road slip (due to more advance under many different
operating conditions) that could cancel or offset the vehicle slip.
The above discussion should not be taken to indicate that knock sensor
equipped vehicles are not and will not be a source of shortfall. As
long as EPA continues to use high octane fuel for testing, the potential
for a knock sensor shortfall will exist. EPA is investigating the whole
issue of the representativeness of its current test fuels, looking at
octane as well as other fuel properties. It is interesting to note,
however, that even if EPA did use a low octane fuel for testing, any
shortfall due to overly aggressive spark timing calibrations may not be
entirely eliminated, because the presence or absence of any degree of
knock during a test is not now a criterion that determines the acceptability
of an emission or fuel economy test.
Fuel Volatility is an important property due to its influence on such
performance parameters as vapor lock, driveability, startability, car-
buretor icing, and oil dilution.
Some volatility characteristics of in-use and EPA unleaded summer gasolines
64
are compared in the next table:
Distillation Temperature, °F (ASTM D86)
1976 1977 1978 1977-78 EPA
MVMA DOE MVMA DOE MVMA DOE Avg. Range
IBP
10%
50%
90%
EP
93
126
221
336
414
89
121
220
332
411
88
125
221
331
414
89
121
221
333
410
87
123
220
336
413
89
120
220
333
409
88
126
221
310
399
82-92
121-134
217-228
304-318
389-406
-------
97
The initial boiling point (IBP) and 10% and 50% evaporation points for EPA
fuel are about the same as for the 1976-78 commercial fuels, but the 90%
point and end point (EP) indicate higher volatility for EPA fuel. Since
this upper end of the fuel distillation curve is believed to be important
for driveability of vehicles with today's quick-release chokes, such
vehicles could tend to be vulnerable to problems with overly lean air/fuel
ratios when operating on less volatile fuels. The in-use "fix" for such
problems is usually enrichment of the idle mixture or choke setting.
Again referring to Section IV.B.S.a., fuel economy effects of idle and
choke enrichments have been reported to be:
EPA City Test EPA Highway Test
Rich idle mixture -1% to -7% zero to +1%
Rich choke zero to -3% -1% to +1%
The average EPA 55/45 MPG penalty for these effects, if they always
occur in combination and if they interact linearly (the worst case
assumptions), is 2.4%. As was done for octane considerations above, it
could be inferred that those vehicles found in the 1975-76 RM program
with carburetor or air induction system maladjustments (72% of all
vehicles) were in that condition due to EPA/in-use fuel volatility
differences and were incurring said 2.4% penalty in-use. Making these
worst-case assumptions, the maximum fleet MPG penalty due to the volatility
gap between EPA fuel and in-use fuels would be 1.7%.
Additives - In-use fuels use a wide variety of additives, most of which are
present in such small quantities that fuel economy is not affected. When
additives are present in significant amounts, however, miles per gallon
of the blended fuel can vary if the heating value of the additive(s) is
significantly different from that of gasoline. For example, early results
from our own studies on 10% alcohol blends ("gasohol") have shown volu-
metric MPG penalties on the order of 3%. Naturally, gasoline consumption
is reduced by about 7% with this blend, but individual motorists using
gasohol can expect to pump slightly more total fuel.
-------
6. Summary Findings: Vehicle Slip
The overall average shortfalls attributable to vehicle slip items are
summarized below:
Effects of Vehicle Slip Influences
on Fuel Economy
Relative Fleet MPG
Influence MPG Shortfall
Production Slip:
Administrative Variance 0.999 -0.1%
Hardware Variance 0.985 -1.5%
Vehicle Condition (Test)
Engine Tune 0.990 -1.0%
Engine/Fuel Response 0.977 -2.3%
Brake Drag 0.998 -0.2%
Wheel Alignment 0.999 -0.1%
If all of these factors act independently, a total shortfall of about
5% can be associated with Vehicle Slip. Recalling that the three-year
average Vehicle Slip shortfall from DOE-furnished data (Section IV.A)
is 4.7%, the influences included in our Vehicle Slip category—and our
estimates of their respective magnitudes—must be concluded to be reason-
ably accurate.
Note that influences of brake drag and wheel misalignment are included.
These items are discussed later under Road Slip, but part of their
effect—that associated with the two vehicle wheels which drive the
dynamometer rolls—is indeed measured in dynamometer tests. Hence, a
proper accounting of effects which show up on the dynamometer must
include these factors.
The two subcategories under Production Slip have been found to have
different MPG influences as a function of model year, as follows:
-------
99
Model Year Trends, Production Slip Items
(Average Percent Deviation from EPA Label/Guide MPG)
Model Year:
1974 1975 1976 1977 1978 1979
Administrative — +1.3% -1.1% -1.6% +0.9%
Variance
Hardware Variance +0.6% -7.5% -8.0% +3.3% +1.4% +1.2%
The significant hardware shortfalls in model years 1975 and 1976 are
believed related to the instantaneous injection of innovative emission
control technology into the fleet in those model years. Since these
shortfalls have not reappeared in 1977-78-79—although much of that
technology is still in use—we regard the 1975-76 slips as a temporary
response to the rapidity with which that technology was introduced,
rather than to shortcomings inherent in the techology itself.
As discussed in Section B.3, there seems to be a pattern of higher
Vehicle Slip shortfall for higher-MPG cars after break-in. This "MPG
tilt" does not appear to be characteristic of very low-mileage cars.
Since vehicle-condition influences take hold with increased vehicle
odometer mileage accumulation, the appearance and growth of MPG tilt
must be due to vehicle condition. This area definitely warrants
further study.
-------
100
(This page intentionally blank)
-------
101
FUEL ECONOMY INFLUENCES (Cont'd.)
Page
Road Slip 102
The Travel Environment • 103
Ambient Temperature 103
Barometric Pressure/Altitude •
Wind and Aerodynamics
Road Gradient U7
Road Surface and Condition 12°
123
Road Curvature ........
125
Summary - Travel Environment Effects ....
Travel Characteristics 126
Vehicle Speed 126
Influence of Traffic Volume 136
Trip Length . 133
1 OO
Trip Average Speed 1JO
Warmup Effects . . . 142
Average Miles Per Day 148
Dependence of AMPD on Population and Other Factors
Acceleration Intensity • •
Quantitative Studies I56
Driver-habit Studies ........ 163
Effectiveness of Fuel Economy Meters
172
Summary - Travel Characteristics Effects • • • • • •••* • • •
-------
102
C. Road Slip
The preceding discussion of vehicle slip considered only the fuel economy
behavior of production cars when operated according to the specific procedures
and conditions of the EPA tests. In this section, we will consider the
wider range of conditions to which vehicles are exposed in actual use, and
observed or estimated fuel economy sensitivity to these conditions. We
have grouped these^ factors under four basic headings:
° The Travel Environment - Conditions in which travel occurs, such
as weather and road surface, over which a driver has minimal
control—other than basic route selection, or simply the decision
not to travel at all;
° Travel Characteristics -.Details of the travel itself, such as
average speed, stopping frequency, and trip length, over which a
driver has partial control but is constrained by traffic flow and
the peculiarities of the trip route (speed limits, intersections,
lights, etc.); driving technique is also an important travel
characteristic, over which a driver has virtually total control;
Vehicle Condition (Road) - The "state" of the vehicle as configured
for traveling, including its mechanical condition and load; most
of .these .factors are matters of choice; and
Simulation Variance - Specifics of the EPA test which could lead
to either random variation, or directional offsets from real-
world fuel economy; and details related to mileage measurement,
fuel measurement, and miles-per-gallon calculation.
From earlier sections of this report, average in-use MPG shortfalls were
shown to have a definite dependence upon EPA MPG level for most of the
model years studied. In assessing fuel economy sensitivity to the in-use
factors, we have considered vehicle vintage and engine/ emission control
technology effects as they might relate to model year 'differences, and have
evaluated vehicle size effects as possible explainers of tilts in shortfall
with EPA MPG level.
-------
103
It must be emphasized that most of the fuel economy sensitivity measure-
ments reported here were derived from classical, "good engineering practice"
methods of testing, namely, measurement of the fuel economy effect of one
single variable at a time while going to great lengths to hold all other
influences constant. Unfortunately, when attempting to assess the combined
impact of a number of separate factors, their inter-dependence is not fully
known, and the resulting combined fuel economy effect is somewhat of an
estimate. Given an individual car with a known in-use MPG capability under
conditions which dp_ match those of the EPA tests, no one really knows
precisely what fuel economy it will get when driven aggressively around a
snow-covered gravel curve on a 3% grade in 22°F weather, with an out-of-
tune engine, misaligned wheels, and underinflated tires, towing a trailer.
Most would agree it would not be very good, but the precise factor that
should be applied to account for these combined influences is not immedi-
ately apparent, even if each individual MPG effect were known for that type
of car.
1. The Travel Environment
Included in this category are three weather conditions: ambient tempera-
ture, barometric pressure (altitude), and wind (aerodynamics); and three
road characteristics: gradient, surface type and condition, and curvature.
a. Ambient Temperature: The table lists temperature sensitivities
for a number of operating conditions, including steady cruises and cyclic
driving. For "cold-start" cycles, the MPG effect shown does not include
the effect of warmup, but only the sensitivity of MPG to ambient temperature
after some 15 miles of driving (7.5 miles and 10.2 miles, respectively, in
the case of the EPA City and Highway cycles).
While a temperature sensitivity of 1.0% to 1.5% per 10°F is generally accepted,
it is clear from these data that that sensitivity is characteristic of pre-1975
cars, basically large ones using relatively unsophisticated emission control
technology and calibrations. There is abundant evidence that later-model
conventional cars are more temperature-sensitive than pre-1975's and also
that smaller cars suffer higher percentage losses than do larger ones. To
-------
104
Effect of Temperature on Fuel Economy
(Percent MPG Change per 10°F Temperature Change)
Steady Cruise, Warmed-up
Claffey
79
Chrysler
80
20 mph 30 mph 40 mph
2.3 2.0 1.9
2.5 2.2 1.9
50 mph 60 mph 70 mph
1.8 1.6 1.5
1.5 1.2 0.9
B. Manufacturer and SAE Cycles, Warmed-up
Q-I
Scheffler/Niepoth and
O o
Tobin/Horowltz 1.1
(GM Urban)
80
Chrysler 2.6
(Chrysler ColdStart Urban)
SAE
84
1.4
AMC83
(SAE Suburban)
EPA City Cycle, Warmed-up
1.1-1.6
(SAE Urban, SAE Suburban,
SAE Interstate)
Q C
EPA Analysis of Mobil Oil Data:
(SAE Urban)
2000 Ib, 3.1
3000 Ib, 2.4
4000 Ib, 1.6
5000 Ib, 0.9
Eccleston, et al
86
. .Pre-1975, 0.7
1975 Models, 1.2
1973 Diesel, 2.0
Eccles ton/Hum
88
. . . City, 1.4
Highway, 0.5
D.
EPA Dyno/Track
EPA City Cycle
R7
City 2 0
Highway, 2.1
, after Cold Start
90
EPA Chicago
91
Marshall Gasoline Cars . .
Diesel Cars . .
0. 7
2.0
0.3
Eccleston, et al
86
Pre-1975, 1.8
1975 Models, 2.0
1973 Diesel, 2.4
PROCO Prototype, 1.6
Hayden
94
EPA CO Hot Spot
92
NYC Cycle, 3.9
EPA FTP, 0.8
1976-8 Conventional, 4.3
1976-8 Lean Burn, 2.7
1978 Stratified Chg, 2.7
1978 Diesel, 2.4
1978 Turbocharged, 1.7
95
EPA Chicago
93
Ostrouchov
.... 2000 Ib,
3000 Ib,
4000 Ib,
5000 Ib,
1.7
6.4
4.5
2.6
2.4
Ostrouchov
?nnn r 1975 Conventional,
zuuu - j 19?8 Lean Burn
2500 Ib
1978 Diesel,
6.0
1.8
1.8
Eccleston/Hurn 2.1
3000- i1975 Conventional, 3.4
3500 Ib I 1978 3-Way Catalyst 2.3
4500 - |1975 Conventional, 2.7
5000 Ib ' 1978 Diesel, 1.3
-------
105
Temperature Effects, Cont'd
E. EPA Combined Cycles, City Cold/Hot, Highway Warmed-up
94
Hayden 1976-8 Conventional, 3.2
1976-8 Lean Burn, 2.0
1978 Diesel, 2.0
1978 Turbocharged, 1.3
F. In-Use Driving
96
EPA Analysis of GM Data 1.2
79
Claffey, "Running Costs of Motor Vehicles as Affected by Road Design and
Traffic", NCHRP Report 111, 1971.
80
Chrysler Corp. data, submitted to: SAE Passenger Car and Light Truck
Fuel Economy Measurement Committee (unpublished).
ft 1
Scheffler and Niepoth, "Customer Fuel Economy Estimated from Engineering
Tests, SAE paper 650861, November 1965.
Q O
Tobin and Horowitz, "The Influence of Urban Trip Characteristics on Vehicle
Warm-up — Implication for Urban Automotive Fuel Consumption", SAE paper
790656, June 1979.
Q Q
American Motors data, submitted to: SAE Passenger Car and Light Truck
Fuel Economy Measurement Committee (unpublished).
84
Society of Automotive Engineers, Op. Cit. (65).
Q C
Mobil Oil Co. data, submitted to: SAE Passenger Car and Light Truck
Fuel Economy Measurement Committee (unpublished).
Q r
Eccleston et al, "Ambient Temperature and Vehicle Emissions'1, EPA Report
460/3-74-028, October 1974.
87
Data from EPA Dyno/Track Project, Phase II (unpublished).
Q Q
Eccleston and Hum, "Ambient Temperature and Trip Length — Influence
on Automotive Fuel Economy and Emissions", SAE paper 780613, June 1978.
89
Gulf Research and Development Company, "Passenger Car Fuel Economy in
Short Trip Operations", DOE Report HCP/W4248, July 1978.
90
Effects of Low Ambient Temperature on the Exhaust Emissions and Fuel
Economy of 84 Automobiles in Chicago", Report 78-3, Technology Assessment
Evaluation Branch, ECTD, EPA, October 1978.
-------
106
amplify on the latter effect, a 2.5% penalty on a 15-MPG large car yields a
loss of 0.4 MPG, but a 5.0% penalty on a 30-MPG small car results in a loss
of 1.5 MPG.
The data of the Canadian researchers (Hayden, and Ostrouchov) indicate that
vehicles powered by engines representative of future technologies appear to
be about one-half as sensitive to temperature as late-model conventionally-
powered cars.
These findings have bearing on the larger in-use shortfalls observed for
post-1974 models, and also may become important in the future, as vehicle
sizes shrink and new technology usage becomes more widespread.
The next table shows an estimate of the cumulative nationwide effect of
temperature on fuel economy. Using distributions of vehicle-miles traveled
as a function of temperature, and fuel economy responses of both "small"
and "large" cars to these temperatures, we can account for in-use MPG
shortfalls of 4 to 8% relative to the EPA test temperatures.
It will be noted that this computation assumes a continued rise in fuel
economy at temperatures up past 100°F, which is appropriate only for non-
air conditioned vehicles. The fuel economy penalties due to air conditioner
operation will be discussed later.
narshall, "Potential for Improving Short-Trip Fuel Economy by Fuel
Formulation", SAE paper 790655, June 1979.
92
Service and Kranig, "CO Hot Spot Preliminary Investigation1', Report 77-13,
Technology Assessment and Evaluation Branch, ECTD EPA, December 1977.
93
Ostrouchov, "Effect of Cold Weather on Motor Vehicle Emissions and Fuel
Economy", SAE paper 780084, February 1978.
94
Hayden, "The Effects of Technology on Automobile Fuel Economy under Canadian
Conditions", SAE paper 780935, November 1978.
Q C
Ostrouchov, "Effect of Cold Weather on Motor Vehicle Emissions and Fuel
Consumption — II", SAE paper 790229, February 1979.
General Motors 1975 Customer Survey (unpublished)
-------
107
Temperature Distribution and Estimated MPG Effect
Temperature, °F
<10
10-20
20-30
30-40
40-50
50-60
60-70
70-80
80-90
90-100
>100
% of
VMT3
0.5
2.0
6.4
12.2
14.6
16.1
18.3
18.6
9.2
1.8
0.3
MPG Relative
Small Cars
.692
.752
.792
.832
.872
.912
.952
.992
1.032
1.072
1.120
to EPA MPG
Large Ca:
.854
.882
.901
.920
.939
.958
.977
.996
1.015
1.034
1.058
VMT-Weighted
Average:
57°F
.920
(Shortfall
=8.0%)
.961
(Shortfall
=3.9%)
from Ref. 97; EPA test temperature = 68-86°F;
66% of VMT below EPA temperature
28% in EPA range
6% above EPA temperature
'Reference temp = 77°F (EPA MPG = 1.00)
Small-car Sensitivity = +4.0% per 10°F
Large-car Sensitivity = +1.9% per 10"F
97
U.S. Department of Transportation, "The Sensitivity of Projected
Aggregate Fuel Consumption to the Conditions of Individual Fuel Economy
Tests—Part A", Draft, May 1974.
-------
108
b. Barometric Pressure/Altitude: This effect is not without some
controversy. The most frequently quoted reference on altitude effects,
Ref. 79, indicates a large MPG penalty for operation at high altitude. The
reference also states (but this is rarely mentioned) that the penalty
applies to uphill driving on a 10% grade. The same reference also states
that for level road operation there is no MPG effect at least up to 2000
feet.
In its correction factors, the SAE fuel economy road test procedure
specifies a reduction of low-pressure (high altitude) fuel economy test
data to correct it to standard conditions, reflecting a fuel economy
increase caused by the high altitude condition; the magnitude of the
correction increases for higher speeds. More recent data, as shown in the
next table, do confirm the MPG improvement of high-altitude operation at
low speeds, but reveal small losses at higher speeds.
In the most tightly controlled of these projects, Ref. 99, a group of cars
was tested at low altitude, then trucked to a high altitude lab. At that
point, the vehicles had drifted out of tuneup specifications due to the
altitude; tested in this condition, they showed a 1.0% City MPG improvement
over the low altitude baseline. Upon being retuned to specification under
the high altitude conditions, they saw an additional 2.4% MPG improvement.
As shown in the next table, following, the estimated nationwide effect of
altitude on fuel economy is quite small, due to the low fraction of vehicle
miles traveled at high altitude, the low absolute magnitudes of the MPG
sensitivities, and the offsetting tendency of the positive and negative
signs assumed for city and highway sensitivities, respectively.
98
Liljedahl and Terry, "A Study of Exhaust Emissions from 1966 through 1976
Denver, Chicago, Houston, and Phoenix", EPA Report EPA-460/3-77-005, August
1977.
on
Edwards, Liljedahl ^t al, "1970 Passenger Car High Altitude Emission
Baseline", SAE paper 790959, October 1979.
-------
109
Effect of Altitude/Barometric Pressure on Fuel Economy
Steady Cruise, warmed up
Claffey79 . .
(one 1964 car)
.2000 ft
2500 ft
3000 ft
3500 ft
4000 ft
No change
1.8% loss
3.3% loss
7.8% loss
18.9% loss
0.6%
,6%
,0%
.1%
loss
loss
loss
loss
21.2% loss
Note: The Claffey data is for a 107. grade;
all other data is for level road conditions.
98
Liljedahl and Terry
(30 models, Denver cars
vs. Chicago and Houston
cars, 1976 models)
SAE Cycles, warmed up
30 mph, 4.0% gain at 5500 ft
60 mph, 1.0% loss at 5500 ft
AMC
(Ford Suburban Cycle)
SAE
65
32.0 in. Hg 1.6% loss
29.8 in. Hg (Detroit) Base
28.0 in. Hg 1.9% gain
26.0 in. Hg 3.0% gain
24.9 in. Hg (Denver) 3.8% gain
22.0 in. Hg
20.0 in. Hg
5.4% gain
6.2% gain
Urban, no effect
Suburban, 0.72% gain per in. Hg decrease
55 mph Interstate, 0.84% gain per in. Hg decrease
70 mph Interstate, 1.44% gain per in. Hg decrease
EPA Cycles
Liljedahl and Terry EPA cold/hot City, 3.7% gain at
(20 models, Denver cars vs. 5500 ft
Phoenix and Chicago cars, 1970
models)
Edwards, Liljedahl ^t aT . .
(25 cars tested at St. Louis,
then at Denver, 1970 models)
Liljedahl and Terry ...
(Denver cars vs. Chicago,
Houston, and Phoenix cars,
1976 models)
EPA cold/hot City, 3.4% gain at
5500 ft
EPA cold/hot City, 0.9% gain
at 5500 ft (31 models);
EPA Highway (warmed-up), 0.9% loss
at 5500 ft (27 models)
-------
110
Altitude Distribution and Estimated MPG Effect
Altitude, Feet
<500
500-1000
1000-2000
2000-5000
>5000
% of VMT
Urban Rural
56
29
11
4
1
32
26
23
16
4
MPG Relative to EPA MPG
City Highway
0.996
1.000
1.004
1.016
1.031
1.001
1.000
0.999
0.995
0.990
VMT-Weighted
Average:
750'
1300'
0.999
0.999
Combined City-Highway
Factor = 0.999
(shortfall = 0.1%)
From Ref. 97; EPA test altitude = 897.5 feet;
56% Urban/32% Rural VMT below EPA altitude
29% Urban/26% Rural VMT - EPA altitude
16% Urban/43% Rural VMT above EPA altitude
Reference altitude = 897.5 feet;
Urban Driving Sensitivity = +0.0006% per ft.
Rural Driving Sensitivity = -0.0002% per ft.
-------
Ill
c. Wind and Aerodynamics: Most of the technical literature
expresses fuel economy effects in terms of response to a change in
aerodynamic drag horsepower, occasioned by a change in effective frontal
area, drag coefficient, or both. It will be noted in the next table that
aerodynamic losses depend strongly on vehicle speed. The figures below
illustrate why: At low cruise speeds, chassis losses " are the primary
source of energy consumption; at higher speeds, aerodynamic drag becomes
the dominant energy loss.
FIGURE 24. Vehicle Road Load
Horsepower Versus Vehicle Speed
FIGURE 25. Distribution of Road Load
Horsepower Versus Vehicle Speed
Subcompact
Compact"
Standard"
20 30 40 50
Vehicle Speed. MPH
Curves from Ref. 100
(4 Passenger Compact Car)
20 30 40 50
Vehicle Speed, MPH
60
70
Tenniswood and Graetzel, "Minimum Road Load for Electric Cars", SAE
paper 670177, 1967.
Tire rolling resistance and rotational friction losses in the drive
train.
-------
112
Effect of Wind/Aerodynamics on Fuel Economy
A. Steady Cruise
102
Cornell , 18 mph Headwind 70 mph, 16.5% loss
18 mph Crosswind 2.2% loss
18 mph Tailwind 19.4% gain
103
Huebner/Gasser , 10% change in aero drag 70 mph, 4.3%
104
D01 , 10% change in aero drag 50 mph, 3.5%
Hurter e_t _al_ , 10% change in aero drag:
70 mph, 4.2%
at 50 raph: Compact cars, 1.3-1.6%
Large cars, 0.8-1.1%
at 70 mph: Compact cars, 2.0-3.0%
Large cars, 1.7-1.9%
Pierce , 107. change in aero drag 20 mph, 0.5-1.1%
30 mph, 0.7-1.4%
40 mph, 1.0-1.7%
50 mph, 1.2-2.7%
60 mph, 1.8-3.0%
70 mph, 2.7-3.8%
B. Cyclic Driving
Huebner/Gasser , 10% change in aero drag "Negligible"
(Chrysler Urban Cycle)
Arabs , 10% change in aero drag 225 CID engine, 0.9%
(EPA Urban Cycle) 318 CID engine, 0.7%
400 CID engine, 0.6%
104
DOT , 10% change in aero drag EPA City Cycle, 0.6%
EPA Highway Cycle, 2.3%
Hurter, £t £l , 10% change in aero drag Compact cars, 0.4-0.5%
(EPA City Cycle) Large cars, 0.3%
108
Ford , 10% change in aero drag 4-cyl. Cars, 2.2-6.3%
(EPA Combined City/Highway) 6-cyl. Cars, 2.9-8.1%
8-cyl. Cars, 2.2-5.3%
109
Nedley , 24% change in C A 1980 x-car, 7.2%
(EPA Combined City/Highway)
Change in Gal./100 Miles: Per HP Cnang£ p r s Ft chanee in r A
in 50 mph Road Load er bq' ht' Chan8e ln CDA
110
Marks Marks/Neipoth
Ford (Compact car) (Nominal car)
SAE Urban Cycle _ Oi027 Ot024
EPA City - 0.054 0.058
GM Surburban _ 0.049 0 048
EPA 55/45 o.l - o!o95
EPA Highway _ 0.188 0.140
SAE Interstate _ 0 42
-------
113
Aerodynamic losses also generally depend on vehicle size. As pointed out
112
by West et al , the relationship between frontal area (which, along with
drag coefficient, governs aerodynamics) and weight (which governs chassis
losses) varies with vehicle size:
Relation between Frontal Area, Weight, and Vehicle Size
Frontal Area (Sq. Ft.)
Per Pound of Vehicle Wt.
1975 Cadillac Sedan 0.0044
1975 Cadillac Seville 0.0047
1975 Honda Civic 0.0074
1975 VW Beetle 0.0089
102
Cornell, "Passenger Car Fuel Economy Characteristics on Modern Super-
highways", SAE paper 650862, November 1965.
103
Huebner and Gasser, "General Factors Affecting Vehicle Fuel Consumption",
SAE paper 730518, May 1973.
104
U.S. Department of Transportation, "Analysis of 1973 Automobiles and
Integration of Automobile Components Relevant to Fuel Consumption",
Draft, September 1974.
Hurter, et al, "A Study of Technological Improvements in Automobile
Fuel Consumption", DOT Report DOT-TSC-OST-74-40, December 1974.
Pierce, "The Fuel Consumption of Automobiles", Scientific American,
January 1975.
Arabs, "Passenger Car Design Influences on Fuel Consumption and Emissions",
AIAA paper 739113, August 1973.
108
Data from Ford Motor Co. (unpublished).
109
Nedley, "An Effective Aerodynamic Program in the Design of a New Car",
SAE Paper 790724, June 1979.
Marks, "Which Way to Achieve Better Fuel Economy?", Seminar at California
Institute of Technology, December 1973.
Marks and Niepoth, "Car Design for Economy and Emissions", SAE paper
750954, November 1975.
112
West, e_t aJL, A Technical Report of the 1975 Union 76 Fuel Economy
Tests", SAE Paper 750670, August 1975.
-------
114
113
These figures are in excellent agreement with data published by EPA :
Aerodynamic Fraction
of Total Road Load HP
Sq. Ft. Per Pound 40 MPH 70 MPH
"Luxury car" 0.0042 29% 52%
"Standard car" 0.0049 31% 55%
"Compact" 0.0059 36% 58%
"Subcompact" 0.0070 40% 62%
These aerodynamic load fractions are spotted on the previous figure, and
illustrate the magnitude of the vehicle size effect on the aerodynamic
load contribution.
The EPA test does simulate the aerodynamic loading experienced by a
vehicle, but only to the extent of aero drag encountered in still
(windless) air. The instantaneous effect of winds upon in-use vehicles
relative to the EPA test will, of course, be a function of wind speed,
and wind direction relative to the vehicle. For purposes of estimating
wind effects on overall nationwide fuel economy, we must expect a uniform
360° distribution of wind direction; this does not mean that all wind
97
effects cancel out, however. DOT has analyzed the aerodynamic effects
of wind direction at a constant vehicle speed of 55 mph and concluded—
as shown in the next figure—that losses at unfavorable wind angles
significantly outweigh gains at favorable angles. The relative areas
under these curves correspond to ratios of energy loss to energy gain of
2.2 to 1 (at 10 mph wind) to 2.4 to 1 (at 20 mph wind). So a vehicle
exposed to winds from all angles still suffers a net energy penalty due
to aerodynamic losses.
DOT also estimated the excess fuel consumption corresponding to these
net wind losses at 55 mph, which fuel penalty (we will assume) applies
to highway driving conditions. For city driving, the net wind penalty
is less due to lower speeds. For the EPA City cycle, of which some 25%
of miles traveled are affected similar to DOT's calculated 55 mph effect,
I I O
U.S. Environmental Protection Agency, "Factors Affecting Automotive Fuel
Economy", September 1975.
-------
115
FIGURE 26. Effect of Wind on Vehicle Aerodynamic Drag at 55 MPH
+ 100
+ 75 -
+ 50 —
c No
8, Change
(Drag Coefficient Sensitivity =1.5%
per Degree of Yaw Angle)
-25 -
-50 —
-75
60 90 120
Wind Direction, Degrees from Vehicle Path
there would be a net wind penalty about 20% as large as the highway
penalty. Further, from the data presented earlier, smaller cars are
influenced more strongly than larger ones by aerodynamic effects.
Using all of these factors and DOT's distribution of wind speeds,
estimated wind-related shortfalls are 2-3%, depending on car size, as
given in the following table.
It should be pointed out that the sensitivity of the calculation to
actual road conditions is not well established in the area of in-use C .
Most calculations are based on C yaw angle sensitivity, typically from
wind tunnel tests. Whether the C sensitivity to yaw angle will remain
the same for future cars, and how well the C and C /yaw angle data
obtained from wind tunnels actually simulate in-use conditions such as
flow field structure and the effects of nearby vehicles, is not known.
-------
116
Wind Speed Distribution and Estimated MPG Effect (360°)
MPG Relative to EPA MPG
Wind Speed,
MPH
£3
4-7
8-12
13-18
19-24
>25
% of
VMTa
16
28
30
18
6
2
City:
Small Car
1.000
.996
.986
.979
.971
.962
Large Car
1.000
.997
.989
.984
.978
.971
Highway:
Small Large
1.000
.981
.938
.903
.870
.828
1.000
.985
.952
.925
.900
.868
Combined City/Highway
Small Larae
1.000
.992
.968
.950
.931
.908
1.000
.993
.975
.961
.948
.930
VMT-Weighted
Average:
9 mph
.973 .979
(Shortfall (Shortfall
= 2.7%) = 2.1%)
From Ref. 97; EPA test wind speed = 0 mph;
»84% of VMT at wind speeds greater than EPA
Reference wind speed = 0 mph;
City Sensitivity, Small Car,
Large Car,
-0.13% per mph wind
-0.10% per mph wind
Highway Sensitivity, Small Car, -0.62% per mph wind
Large Car, -0.48% per mph wind
-------
d. Road Gradient: There is an underabundance of test data on the
effect of grades on fuel economy, and all of it is for steady cruise
conditions. As depicted in the following figure, it is clear that grade
has a dramatic MPG influence.
117
FIGURE 27. Effect of Road Gradient on Fuel Economy
Klein / Head " s ( 1 938 Ford)
O Winfrey"6 (1964 Chev.)
D Claffey (F'v« 1 964-68 Cars)
30 40
Steady Cruise Speed, MPH
Claffey, "Passenger Car Fuel Conservation", DOT Report FHwA-PL-77009,
January 1977.
Klein and Head, "The Effect of Surface Type, Alignment, and Traffic
Congestion on Vehicular Fuel Consumption", Oregon State Highway Dept.
Technical Bulletin No. 17, April 1944.
Winfrey, "Economic Analysis for Highways", International Textbook
Co., 1969.
-------
118
The data are fairly consistent with regard to upgrades, because climbing
an incline is very much a straightforward "brute force" mechanical
matter. The data for downgrades are quite scattered, since specific
engine and vehicle characteristics such as idle speed, compression
ratio, aerodynamics, and axle ratio come into play when descending an
incline. As in the case of wind, it must be assumed that over a period
of time, travel up and down grades will balance out. But again in this
case, the positive and negative MPG effects do not usually cancel. As
shown in the next figure, one study suggests that for small gradients
(less than 3.5%), fuel saved when descending overcompensates for excess
fuel consumed when ascending, giving a small net benefit in overall gas
mileage. For higher gradient values, the situation reverses, and there
is a net MPG penalty which grows worse with increasing grade.
FIGURE 28. Effect of Road Gradient on Fuel Economy
(Miles Traveled Up=Miles Traveled Down)
Claffey"
20MPH D50MPH
A30MPH T60MPH
• 40MPH O70MPH
-------
119
The estimated nationwide effect of road gradient is a 2% MPG shortfall,
when the distribution of grades is combined with the MPG factors derived
from these data, as given in the following table.
Gradient Distribution and Estimated MPG Effect
(Miles Traveled Up = Miles Traveled Down)
MPG Relative to EPA MPG
Grade, %
<0.5
0.5-1
1-2
2-3
3-4
4-5
5-6
>6
% of
VMTa
35
20
15
10
8
6
4
2
City
1.000
1.000
1.000
1.000
.990
.920
.844
.730
Highway
1.000
1.000
1.000
1.000
.980
.927
.878
.807
Combined
City-Highway
1.000
1.000
1.000
1.000
.985
.923
.859
.762
VMT-Weighted
Average:
1.6% grade
.981
(Shortfall
=1.9%)
from Ref. 97; EPA test excludes grade simulation;
All of VMT occurs at grades greater than EPA conditions
79
Reference grade = 0%;
Sensitivities above 3% grade per Claffey'7; no net MPG effect
assumed below 3% grade.
-------
120
e. Road Surface and Condition: The limited number of data sources
reviewed are in significant disagreement on the magnitudes of fuel
economy penalties of less-than-ideal road surfaces; this may be due to
the wide variances in actual characteristics that can exist for a given
"classification" of road type. The table below summarizes these data.
Effect of Road Surface and Condition on Fuel Economy
(All values shown are loss in steady-cruise MPG relative
to dry, well-maintained concrete or asphalt)
20 mph 30 mph 40 irroh
A. Dry Surfaces Claffey79'114 Claf79 K&H115 Claf K&H
Concrete w/cracks, settling 1% 2%
Asphalt, broken & patched 5% 17% 2-3% 25% 0-4%
Compacted Gravel 12-16% 21-22% 5-15% 36? 5-17%
Loose Gravel 5-21% 19-23%
Earth 9-21% 10-21%
Loose Sand 22% 29% 42%
B. Wet Surfaces (K&H)
Asphalt, good condition 1-4% 0-5%
Concrete 3% 3%
Compacted Gravel 11-20% 11-22%
Earth: Soft several inches 43%
Axle-deep mire 65%
C. Snow-Covered (Claffey117)
Hard-packed base 19-23% 14-17% 10-12%
New snow cover, 1/2" 26% 22% 17%
3M" 30% 24% 19%
1" 32% 26% 22%
1 1/2" 34% 31% 29% •
2" 38% 35% 32%
50 mph
Claf K&H
3%
33% 0-5%
41% 5-18%
1 1 A
(Winfrey °17%)
21-27%
10-24%
50%
0-5%
2%
12-24%
6-9%
11%
157,
19%
25%
31%
-------
121
There is also the possibility that vintage of the test vehicles has some
bearing on the data: Reference 115 employed pre-1940 autos, while References
79, 114 and 117 used models from the late 1960's and early 1970's in their
tests.
118
Recent research , which considers not only frictional factors but the behavior
of the vehicle suspension system, indicates that road roughness causes power
consumption penalties which imply increases in aerodynamic drag, along with
rolling resistance. This concept could explain why, in the previous table,
percentage MPG losses appear to increase with vehicle speed for those condi-
tions (broken pavement, gravel, etc.) involving surface roughness or macro-
friction, but percentage losses decrease with speed on new snow, which is more
of a purely microfrictional phenomenon. This is not to suggest that all of
the macrofrictional penalty is an aerodynamic effect: although some rough-
ness-induced losses due to vehicle pitch and yaw, and increased wheel-well
turbulence, would in fact be aerodynamic, other losses involving the sus-
pension system behave as though a drag coefficient increase had occurred.
The scatter in the data on MPG effects leaves wide margins for assessing
penalty points for various road surfaces; in addition, we are not in pos-
session of good data on the relative distribution of vehicle-miles traveled
among various surface classifications. Hence our evaluation of the nationwide
MPG shortfall due to road surfaces involves some estimation. Using known
distributions of roadway miles from DOT, we have made our own assumptions for
relative traffic flow among those road categories to estimate the VMT distri-
bution. These figures, and assumed MPG penalties, are given in the next
table. These estimates and assumptions yield a 4% shortfall.
Claffey, "Passenger Car Fuel Consumption as Affected by Ice and Snow",
Report sponsored by Task Force on Application of Economic Analysis to
Transportation Problems, 1971.
118
Velinsky and White, "Increased Vehicle Energy Dissipation Due to Changes
in Road Roughness with Emphasis on Rolling Losses, SAE paper 790653, June 1979.
-------
122
Estimated Road Surface Effect on Fuel Economy
of
Road Surface
Unsurfaced
Gravel, Slag, etc.
Low— Load Asphalt
Concrete and Hi-
Roadway
Miles3
18.2
31.1
27.3
23.4
Total
1.8
9.7
30.2
58.3
% of Vehicle Miles
Dry Wet Snow>
1.2 0.4 0.1
6.7 2.4 0.6
20.9 7.4 1.8
10.4 14.3 3.6
MPG Relative to EPA MPCT
Pry Wet Snowy
.80 .70 .65
.85 .82 .80
.96 .95 .90
1.00 .97 .93
Load Asphalt
VMT-Weighted Average MPG Factor: 0.956 (Shortfall = 4.4%)
a 119
Based on DOT data
EPA test simulates a dry, good-condition paved road; ^60% of VMT occurs on
poorer surfaces. The VMT distributions are EPA estimates; "wet" and "snowy"
VMT fractions are based on precipitation-data from NOAA , and on degree-days
1 ?1
data from ASHRAE .
Q
MPG penalties estimated from data presented earlier.
119
U.S. Department of Transportation, "Highway Statistics 1977", Report FHwA-HP-HS-77,
1979.
120
U.S. National Oceanic and Atmospheric Administration, Comparative Climatic Data,
annual.
121
American Society of Heating, Refrigerating and Air-Conditioning Engineers, ASHRAE
Handbook. 1978.
-------
123
f- Road Curvature: Once more, available MPG sensitivity data on this
influence are scarce, inconsistent, and limited to steady cruise operation, as
summarized in the figures. Klein and Head presented the important result
that total central angle per mile uniquely determines the MPG penalty for a
given vehicle speed, whether made up of many short, sharper curves or fewer,
longer gradual curves.
FIGURE 29. Effect of Road Curvature on Fuel Economy
to
10
20
# 10 -
o
g
I
30 40
Speed, MPH
50
60
60
-------
124
In the EPA's Dynamometer vs. Track Project, coastdown tests were run on both
the straightaways and the curves of the test track to determine the relative
road load values. The results differed with "car size", as listed in the
table. The smaller cars were not equipped with power steering, and the larger
cars were. The table also shows general fuel economy sensitivites to changes
in 50 mph road load HP, and the resulting estimated MPG penalties for the
curvature of this particular test track. This curvature is equivalent to
traveling in an 0.9 mile diameter circle, and the MPG penalty is about 1%.
Effect of Road Curvature on Road Load and Fuel Economy
(97° Central Angle per Mile)
Small Cars Large Cars
(Without Power (With Power
Steering) Steering)
Increase in 50 mph RLHPS 2.2% 4.6%
Change in MPG per % A RLHPb
City -0.16% -0.16%
Highway -0.33% -0.29%
MPG penalty, City -0.36% -0.75%
MPG penalty, Highway -0.73% -1.33%
Q
EPA Dyno/Track project coastdown tests
Vi
Ibid, and Southwest Research
A minimum estimate of the nationwide effect of curvature would use average
trip lengths and average speeds from the travel characteristics computations
in Appendix E, arriving at round trip average distances of 17.4 and 21.1
1 22
Martin and Springer, "Influence on Fuel Economy and Exhaust Emissions of
Inertia, Road Load, Driving Cycles, and N/V Ratio for Three Gasoline Auto-
mobiles", Final Report, Task No. 9, EPA Contract 68-03-2196, June 1977.
-------
125
miles, and average speeds of 31.7 and 33.2 mph for small cars and large
cars, respectively. These values—together with the central angle per
mile MPG sensitivity data from Klein and Head and the EPA tests—result
in overall curvature MPG penalties of 0.07% for small cars and 0.12% for
large cars. This is a minimum estimate, since greater circuity of
round-trip travel will increase these penalties; however, we have no
circuity data with which to make a more representative estimate.
g. Summary - Travel Environment Effects: The table below summarizes
the estimated fuel economy effects of the travel environment factors
analyzed:
Relative Fuel Economy Associated with Travel Environment
Effects (EPA 55/45 MPG = 1.000)
Total U.S. (VMT-Weighted) Range for Individual Cars
Factor Small Cars Large Cars Best Worst
Temperature 0.920 0.961 1.06 <0.69
Altitude 0.999 0.999 1.04 0.99
Wind 0.973 0.979 1.00 <0.83
Grade 0.981 0.981 >1.0 <0.60
Road Surface 0.958 0.958 1.00 0.35
Road Curvature <0.999 <0.999 1.00 0.75
Cumulative Effect 0.840 0.882
(Shortfall (Shortfall
= 16.0%) = 11.8%)
The cumulative effects shown are the products of the individual factors'
effects. The "best" and "worst" values shown are not the possible extremes,
nor even the extremes seen for individual test cars, but merely reminders of
the order-of-magnitude effects when each factor takes on highly favorable or
highly unfavorable values. It is important to note that the "best" conditions
correspond to little or no improvement over the EPA values, but there is ample
opportunity for a significant shortfall when any one or more of these factors
is highly unfavorable. Put another way, the Travel Environment simulated by
the EPA test is one that gives close to the best fuel economy of all possible
Travel Environments.
-------
126
2. Travel Characteristics
Travel characteristics which influence fuel economy in a quasi-instantaneous
manner (that is, for a given trip or segment of a trip) include vehicle
speed, stopping frequency, and acceleration intensity. Overlaid upon such
effects are influences related to recent vehicle history: The extent to which
the vehicle is warmed-up, which depends on distance traveled since being
started, and the degree to which it had cooled down from previous travel.
These factors are usually interrelated: short trips generally involve higher
stopping frequency and lower average speeds, and are influenced signifi-
cantly by warmup effects; longer trips are generally faster, smoother,
and less influenced by warmup.
a. Vehicle Speed: The dependence of fuel economy on vehicle
steady cruise speed has long been recognized. Vehicle and engine size
have also been noted for their dramatic effect on fuel economy over a
wide range of speeds, as in the figure.
Because of this vehicle/engine dependence, the relative effect of speed
alone can be seen more clearly through the use of normalized fuel economy,
as shown in the next double figure, with MPG at 40 mph used as the
FIGURE 30. Fuel Economy Versus Cruise Speed for Three Vehicles
40
O 30
20
10
I
Subcompact.
4-cyl. Engine.
Manual Transmission
Intermediate,
V-8 Engine, Automatic
Transmission
01-0-
Luxury Sedan,
V-8 Engine, Automatic
Transmission
I
Source: Huebner &Gasser"
I
20
30
40
Speed, MPH
50
60
70
-------
127
normalization base. The left-hand figure shows relative MPG:speed curves
for data previously published by the DOT and the MVMA. The right-hand
curves are from the EPA dyno/track project, and agree quite well with
the general curve shape of the left-hand figure; the dyno/track curves
also confirm that cruise speed affects fuel economy in about the same
relative way whether measured on the dynamometer or on a test track.
FIGURE 31. Effect of Speed on Fuel Economy—Steady Cruise
1.20
8 i.oo
0.80
§ 0.60
UJ
1
Li-
0.40
(Road Tests)
rO = Claffey7'( 1964-68 Models)
• = Claffeym(l970-74 Models)
D = MVMA/DOT'"(Pre-l973 Models)
I
I
I
20 40
Speed, MPH
60
70
1.20
i.oo
I
I
£
r
0.80
0.60 -
0.40
Dynamometer
Tests
Source:
EPA Dynon"rack
O • = Phase I (1975 Models)124
A A = Phase II (1976 Models. Unpublished)
I
20
40
Speed, MPH
60
80
In actual driving situations, particularly in urban traffic, far less than
125
half of travel time is spent cruising at steady speeds ' .
123
Motor Vehicle Manufacturers Association, "Motor Vehicles and Energy",
January 1974 (includes data from U.S. DOT Federal Highway Administration).
124
Austin, "Passenger Car Fuel Economy — Dynamometer vs. Track vs.
Road", Report 76-1, Technology Assessment and Evaluation Branch, ECTD,
EPA, August 1975.
125
Scott Research Laboratories, "Vehicle Operations Survey", Final Report
under Coordinating Research Council/EPA Project No. CAPE-10-68(1-70),
December 1971.
-------
128
Average speed in real-world traffic is determined by many, many factors;
a thorough treatment of the subject would be burdensome here, since our
primary concern is the fuel economy effect of average speed, not what
causes average speed to be what it is. The relation between average
speed, attempted cruise speed, and stopping requirements is of interest,
however. Using nominal vehicle acceleration and deceleration characteristics
1 7 f\
derived from traffic survey data , the top left figure illustrates how
stopping frequency affects average speed for specific attempted cruise
speeds.
FIGURE 32. Relationship Between Average Speed and Stopping Frequency
70 C
T
Source:
Bernard &McAdams'"
15 sec. idle assumed per stop
' attempted cruise speed
4 6
No. Stops Per Mile
8 0
No. Stops Per Mile
70
• = Johnson et al
O = EPA Employees (Michigan)
T = EPA Employees (Oregon)
• = EPA Test Cycles
A = SAE/Mfr. Test Cycles
10
No. Stops Per Mile
-------
129
Speed has been related to posted speed limits and other factors according to:
V = 20.5 + 0.433(PSL) - 0.407(NRE) + 0.1 IS (OSC) + O.OOOB(MSD)
[from Wortman, 1965, in Ref . 127]
where: V = mean spot speed, mph
PSL = posted speed limit, mph
NRE = number of roadside establishments per mile
OSC = percent out-of-state cars in traffic stream
MSD = minimum sighting distance, feet
If it is assumed that minimum free-flow speed, occurring when NRE, OSC, and
MSD are all zero, is a conservative estimate of attempted cruise speed, the
Wortman formula gives:
V ** . , = 20.5 + 0.433(PSL)
attempted
from which we see that attempted cruise speed exceeds posted limits, up to
about 40 mph. Applying this to the data in the top-left figure, it is inferred
that average speed in stop-and-go traffic is relatively independent of posted
limits, and dependent mainly on stop frequency, as in the top-right figure.
The lower figure depicts average speed/stop frequency relationships as observed
in actual traffic and in a recent survey of EPA employee home-to-work
129
commuting trips . Similarity of these data to the top right figure is obvious.
1 *? f\
Bernard and McAdams, "Automobile Exhaust Emission Modal Analysis Model",
EPA Report EPA-460/3-74-005, January 1974.
127
Voorhies, ej^ al, "Vehicle Operation, Fuel Consumption, and Emissions as
Related to Highway Design and Operation", Interim Report prepared for FHWA,
October 1977.
128
Johnson, et al, "Measurement of Motor Vehicle Operation Pertinent to Fuel
Economy", SAE Paper 750003, February 1975.
129
Unpublished.
-------
130
The lower graph in Figure 32 also shows specific average speed and stop
frequency values for the EPA tests, auto manufacturers' proving-ground
tests (some of which constitute the SAE road tests), and for specific road
types, from expressways to shopping center parking lots.
The fuel consumption effects of slowdowns and stops have been investigated
79 114
by Claffey ' as a function of attemptei
this data appears in the following figure.
79 114
by Claffey ' as a function of attempted cruise speed. An example of
FIGURE 33. Effect of Stops and Speed Variations on Fuel Consumption
o
u
3.0
I!
E
I 2.0
1.5
1.0
I I
Stop Cycles:
(No. Per Mile)
Source: Claffey7''114
(Generalized Representation of Data)
J_
I
Sp««d Chang* Cycto: —
On. Per Mile. A - «> MPH
On. Per Mile. A -30MPH
One Per Mil.. A -20MPH
One Per Mile. A - IOMPH
I
10 20 30 40 50
Attempted Cruise Speed, MPH
60
70
When this type of behavior is combined with speed/stop frequency data, the
pattern in the next figure results: for cyclic driving, relative fuel
economy is essentially a function of average speed only, virtually independent
of the particular stop frequency and attempted cruise speed that produced
that average speed.
-------
131
FIGURE 34. Relative Fuel Economy vs. Average Speed
1.00
0.75
o.so
0.25
1
Sources:
Claffey7'—Consumption of Stop-Cycles
Bernard/McAdams1 J6—Acceleration and
Deceleration Characteristics
I
T
0.2 Stop/Mile
• I Stop/Mile
+ 2 Stops/Mile
• 3 Stops/Mile
* S Stops/Mile
OS Stops/Mile
A 10 Stops/Mile
I
10
20 30
Average Speed. MPH
T
40
50
This relationship has been studied extensively by Evans e_t al
expressed in terms of fuel consumption:
130
and can be
= 3 + k/i
(for speeds < 40 mph)
where G is fuel consumption per unit distance, V is average speed, and j and k
are constants. The constant j is proportional to vehicle weight (according to
j = 9x10 W, where W = pounds, for the nine 1973-76 test cars in Ref. 130), and
the constant k is proportional to the engine's idle fuel flow rate (k
where I = gallons/hour).
1.251,
Evans, e^
-------
132
i n
EPA has found I to be dependent upon engine cubic inch displacement
(I = 0.27 + .0017(C), where C = cubic inches). Thus a vehicle weighing
"W" pounds, with an engine of "C" cu. in. displacement, traveling at "V"
miles per hour (average), will consume "G" gallons per mile, in cyclic
driving, according to:
G - 9xWM + •- [v<40
This equation is only valid for the types of vehicle/engine combinations
from which it was derived. Changes such as fuel shutoff at idle, or
significant changes in CID-to-weight ratio, would necessitate modification
of the equation. In addition, this equation is applicable only up to
average speeds of approximately 40 mph, after which cruise conditions
are approached and higher speed operation results in decreasing fuel
economy. The only cyclic-driving variables other than average speed
which have any significant influence on fuel consumption (for warmed-up
132
vehicles) are related to acceleration and deceleration intensities
Traffic network studies similarly show average speed to be
most significant determinant of vehicle fuel consumption.
The next figure compares steady cruise fuel economy and generalized
cyclic-driving fuel economy, for cars from several recent model years.
Earth, "Idle Fuel Consumption in Passenger Cars", Report 75-29,
Technology Assessment and Evaluation Branch, ECTD, EPA, July 1975.
1 QO
Evans, et^ al, "Multivariate Analysis of Traffic Factors Related to Fuel
Consumption in Urban Driving", General Motors Research Publication GMR - 1710,
May 1976. [See also: Evans, e_t^ a_l in Transportation Science, Vol. 10, No. 2,
May 1976, at 205.]
1 O O
Honeywell Traffic Management Center, "Fuel Consumption Study - Urban Traffic
Control System (UTCS) Software Support Project", Report FHwA-RD-76-81, February
133
Con
1976.
-------
133
The EPA curves are from EPA Emission Factors production car dynomometer
data, with the cyclic curves generated by Ref 126's Modal Analysis Model
and driving cycles synthesized after the methods in Ref. 134. The
agreement of the cyclic-driving simulations with the Evans road test
data is quite good up to 30 mph average speed.
FIGURE 35. Fuel Economy vs. Speed, Cruise and Cyclic Driving (Generalized)
25
20
Il5
UJ
1
10
Sources:
..-.EPA(Hud,k'»)
^— EPA (Unpublished) JDlti
^•» Evans
(Four 1973-74 Moddi. Four 1975. Oti« 1976)
10
20
30
Speed. MPH
40
SO
60
1 O /
Smith and Weston, "A Technique for Generating Representative Chassis
Dynamometer Test Cycles", APCA Paper 72-165, June 1972.
Hudak, "Effects of Driving Cycle Average Speed and Acceleration on
Emissions and Fuel Economy: A Modeling Study", Characterization and
Analysis Branch, ECTD, EPA, May 1979.
-------
134
In the next figure, fuel economy for specific test cycles is shown. As in the
previous figure, the fuel economy:average speed equation's behavior is shape-
wise consistent with MPG results of these driving cycles, for road tests,
dynamometer tests, and computer simulations.
Model year differences implied by this and the previous figures may or may not
be accurate; the vehicle mixes of the various data sources were intended to be
representative, but they are not exactly the same.
FIGURE 36. Fuel Economy vs. Average Speed, Specific Test Cycles
25
20
O
10
SAE
Cycles
Sources:
^^"^"^" Evans . Road Tests (From Previous Figure)
• 1975 fHudak135, Computer Simulations
SAE Committee Road Tests Round Robin 11 ft f 2 (Unpublished)
West et a/" 2, Track Tests EPA City, cold/hot, other cycles warmed up
Liljedahl & Terry", Dyno Tests, EPA City cold/hot, other cycles warmed up
I I I I |
A 1973
O 1975
• 1976
10
20 30 40
Average Speed, MPH
50
60
The distributions of vehicle speeds from actual traffic surveys are
compared with the EPA cycles' speed distributions in the next figure.
The EPA City cycle expends a higher percentage of time and mileage at
speeds below 30 mph and lower percentages above 30 mph, than the survey
data. The EPA Highway cycle is similarly biased toward lower speeds.
-------
135
FIGURE 37. Comparison: Speed Characteristics, EPA Cycles vs. Traffic Surveys
40
Speed. MPH
If speed differentials were independent of other travel characteristics,
the urban speed distributions would yield fuel economy better than the
EPA City cycle, and the highway distributions would yield fuel economy
poorer than the EPA Highway cycle, principally due to the MPG reduction
effect of high-speed driving.
-------
136
b. Influence of Traffic Volume: The speed characteristics of a vehicle
in a traffic stream are obviously influenced by that stream. One relevant
expression for that influence is:
= Vf (j -
[from Greenshields, 1930, in Ref . 127]
where: V = average vehicle speed
V- = free-flow speed
d = vehicle density, vehicles per lane-mile
d. = jam density,
250 vehicles per lane-mile
Note that "V" approaches zero as "d" approaches jam density.
79
Claffey has estimated the combined effects of traffic density and stopping
frequency for two urban road types, as shown in the next double figure. These
estimates indicate that fuel economy is influenced more by stopping frequency
than by traffic volume. That is, if traffic volume changed without affecting
1.00
FIGURE 38. Effect of Traffic Volume on Fuel Economy
1.00 i
Central Business District
(6 Lanes, 2 of which are
parking lanes)
Attempted Speed = 25
Stops Per Mile
-------
137
stop frequency, MPG would be affected only to a second-order degree. When
changes in traffic volume are evaluated as to total effect on the stream,
including stop frequencies, however, the impact on average speed and fuel
consumption can be dramatic. The next figure illustrates this. When vehicle
flow decreases, stream average speed goes up, and the remaining vehicles
operate at better fuel efficiency. Hence a 10% drop in traffic volume can pay
off in a fuel consumption reduction much greater than 10%, The negative
effects of traffic flow increases are compounded in the same way.
FIGURE 39. Effect of Traffic Volume Changes on Fuel Economy and Flow Speed
40
(J 20
1
LL.
fe
I
u
»
I 20
40
Decrease
in Fuel
Consumptioi
Source: Evans etal"4
Decrease
Increase
in Fuel
Consumption
10 IS
Initial Traffic Speed. MPH
Benefits
Resulting From
10% Reduction
in
Traffic Volume
Disbenefits
Resulting From
10% Increase
in
Traffic Volume _
20
25
1 o (L
Evans et al, "A Simplified Approach to Calculations of Fuel Consumption
in Urban Traffic", GM Research Report GMR-2181, August 1976.
-------
138
This would suggest that a change of traffic controls, in an attempt to increase
vehicle average speed and hence fuel economy, could show no net benefit if it
attracts more vehicles into the stream. A reduction in traffic volume, on the
other hand, guarantees fuel-saving results. It must be pointed out that the
above discussion and statements only apply to traffic flow with average speeds
less than about 40 miles per hour. The results cannot be extrapolated to
significantly higher speeds.
c. Trip Length: Two effects closely related to trip length are of
prime interest to fuel economy: average speed, and vehicle warmup.
(1) Trip Average Speed - Report No. 8 of the DOT Nationwide Personal
Transportation Study (NPTS) relates elapsed times and trip lengths for
home-to-work commuting, and has been a principal reference for analyses of
138 139
trip length and average speed in the literature '
There are three problems with literal use of the NPTS data. The first,
139
as pointed out by Joksch and Reidy , is that the data are based on a
questionnaire, and the questions may have elicited overestimated travel
times and correspondingly low travel speeds. In particular, total
reported trip times are believed to represent person time, from residence
door to workplace door, rather than vehicle operating time.
Joksch and Reidy observed that the difference in travel time between
trips of various lengths gives an indication of instantaneous speed.
The next figure shows this kind of interpretation of the NPTS data.
137
Svercl and Asin, "Nationwide Personal Transportation Study, Home-to-
Work Trips and Travel, Report No. 8", DOT/FHwA, August 1973.
1 38
Austin and Hellman, "Passenger Car Fuel Economy as Influenced by Trip
Length", SAE paper 750004, February 1975.
1 on
Joksch and Reidy, "Categorization and Characterization of American
Driving Conditions (Phase I)", DOT Report DOT-TSC-NHTSA-78-41, November
1978.
-------
139
FIGURE 40. Home-to-Work Commuting Time and Implied Average Speeds
50
Slope = 1.71 Min./Mile (35 MPH)
Slope = 2.86 Min./Mile (21C
Slope = 12 Min./Mile (5 MPH)
' 1 = Interpretation of NPTS Data
— after Joksch and Reidy'"
IS 20 25 30
Home-to-Work Trip Length, Miles
Trips between 1/2 mile and two miles in length are associated with a 21-
mph slope, trips between two and ten miles in length a 35-mph slope, and
longer trips a high-speed slope (There is some uncertainty about the
latter slope due to the way the NPTS data is presented, but Joksch and
Reidy argue plausibly for a 63-mph interpretation).
Applying these data to actual trip geometries, and assuming a trip of length
"X", the average commuter spends the first (and last) 1/4 mile entering and
exiting a 21 mph traffic stream, the next 3/4 mile (and the 3/4 mile segment
from X-l to X-l/4 miles) at 21 mph average, the next 4 miles (and the 4-mile
segment from X-5 to X-l miles) at 35 mph, and the middle 5 to X-5 miles at 63
mph.
The cumulative average speed for a given trip can be calculated from these
building blocks, and of course agrees with the NPTS data if one accepts the
data's suggestion that it takes 1/2 mile and six minutes total to enter and
exit the stream: this is shown in the lower curve in the next figure. The
-------
140
maximum cumulative average speed that can be calculated with this model
(assuming that entrance into, and exit from, the 21 mph stream are instan-
taneous) is represented by the top curve. This interpretation predicts home-
to-work average speeds six to eight mph faster than literal interpretation of
the NPTS data.
FIGURE 41. Average Speed vs. Trip Length in 1969: A Function of Data Interpretation
557MPHAvg .,
@ 100 Miles -
Cumulative Average Speed
Instantaneous Speed
5<7MPHAvg.
® 100 Miles
Total Distance ft Time
to Enter and Exit
Traffic Stream:
Zero/zero
= Literal Interpretation of NPTS Data
= Calculated speeds, Jokseh and Reidy Model
© = EPA "SS/45 Trip" (8.5 Miles. 16 8 MPH).
!/i Mi.16 Min (NPTS)
IS 20 25 30
Home-To-Work Trip Length, Miles
The second problem with the data is that it represents only work-related
commuting trips. The same NPTS report indicates that work-related travel
occurs predominantly in hours of the day when traffic density is high, the
work travel being in fact a prime contributor to that traffic density. Ap-
pendix D is a computation of the 24-hour average speeds for both work and non-
work travel, considering the different distributions of those respective types
of trips among periods of varying traffic density. From that analysis, it
would be estimated that non-work trips are 4.2% faster than work trips, and
all travel combined is 2.8% faster than work trips alone.
-------
141
The third problem is that the NPTS data applies to travel in 1969. More recent
surveys of the characteristics of relatively short trips show average trip
speeds to be significantly higher than [literal interpretation of] the NPTS
data. The next figure compares the two NPTS interpretations with the results
of two such surveys; the literal NPTS interpretation clearly underestimates
average trip speeds relative to the surveys, while the maximum interpretation
is quite a good fit to the data.
FIGURE 42. Data from Mid-1970's Surveys of Average Speed vs. Trip Length
50 -
40
30
20
10
= Home-to-Work Trips for
72 EPA Employees in Michigan
and Oregon. 1979 (Unpublished)
= Travel for Six Drivers.
12,000 Miles12'
I
10
Trip Length, Miles
IS
20
-------
142
140-143
(2) Warmup Effects - It has been well documented that vehicles
achieve poorer fuel economy when started "cold" than when fully warmed up.
The table below gives relative instantaneous fuel economy (MPG 4- fully warm
MPG) at specific operating times after start, for various pre-start soak
143
times, in the first part of the EPA city cycle :
Relative Instantaneous Fuel Economy
Minutes after Start:
Soak time, hours 1.05 2.10 3.15 4.20 8.42
1/2 or less
1
2
4
8 or more
Miles Traveled:
1.000
0.966
0.862
0.726
0.592
0.22
1.000
0.961
0.895
0.846
0.739
0.67
1.000
0.986
0.974
0.894
0.772
0.82
1.000
0.969
0.952
0.902
0.845
1.67
1.000
1.024
0.977
0.960
0.929
3.59
In similar tests, relative instantaneous EPA City MPG values were measured
for both dynamometer op<
of eight hours or more:
144
for both dynamometer operation and operation on a test track , after soaks
Relative Instantaneous Fuel Economy
Minutes after Start:
Test Site 2.33 5.67 8.42 17.33 22.87
Laboratory (Dyno) 0.658 0.884 0.934 0.979 1.022
Test Track 0.657 0.872 0.929 0.972 0.987
Miles Traveled: 0.68 2.65 3.59 6.34 7.50
140Scheffler and Niepoth, Op. Cit. (81)
U.S. Environmental Protection Agency, "A Report on Automotive Fuel
Economy", October 1973.
Matula, "Emissions and Fuel-Economy Test Methods and Procedures",
Consultant report to the National Academy of Sciences, September 1974.
Srubar, et al, "Soak Time Effects on Car Emissions and Fuel Economy",
SAE paper 7~8008li, March 1978.
14AEPA, Op. Cit. (87)
-------
143
This indicates that the fully-warmed up (FWU) condition has essentially
been reached by the end of the 7.5-mile EPA City Cycle.
These findings are comparable to cold start data from road course tests
using the SAE Urban driving cycle; an example of that data is shown in
the next two figures plotted as a function of ambient temperature. The
first graph shows low fuel economy for the first run of this 2-mile
cycle, and progressively improved fuel economy in succeeding runs. The
second graph gives the fuel economy values for the first 1/2 mile and
the next 1-1/2 miles of the first cycle.
145
FIGURE 43. Fuel Economy Data, SAE Urban Cycle
145
12
II -
10
I '
3
1975 Large Car
• Cycle I
• Cycle 2
A Cycle 3
T Cycle 4
10
20
30 40
Ambient Temperature °F
SO
60
70
145.
Mobil Oil Co., Op. Git. (85)
-------
144
I 7
x.
I
S 6
10
FIGURE 44. Fuel Economy Data. SAE Urban Cycle
T
1975 Large Car
Cycle I (Less First !/i Mile)
\
I
I
20
30 40
Ambient Temperature °f
SO
60
It is clear from all of these data that the excess fuel used for vehicle
warmup is consumed in the early stages of cold-start driving. The table
following lists the warmup fuel, i.e., that in excess of FWU fuel, for
these tests.
Excess Fuel Consumed for Warmups
Average Excess fuel,
% of FWU fuel:
All cars, Track
All cars, Dynamometer
Total Excess gallons:
4-cylinder cars
6-cylinder cars
8-cylinder cars
Four
SAE Urban
Cycles
(8 miles)
15.4%
___
.059
.088
.091
EPA City
Cycle
(7.5 miles)
12.0%
10.5%
.039
.056
.082
EPA Highway
Cycle
(10.2 miles)
10.6%
8.4%
.034
.053
.062
-------
145
The rate at which this excess fuel is consumed is plotted in the next figure:
within five minutes of start up, 75% of warmup fuel has been consumed; within
10 minutes, 90%.
•a 50 -
FIGURE 45. Rate of Consumption of Warmup Fuel
= SAE Urban Cycle145
= EPA City Cycle143 (Estim.)
= EPA City Cycle
= EPA Highway Cycle
10 IS 20
Time from Start, Minutes
Although vehicles achieve fully-warmed up MPG capability relatively soon after
start up, the cumulative fuel economy of a cold start trip is affected through-
out the trip by the initial usage of warmup fuel. A formula by Tobin and
146
Horowitz relating cumulative cold start and fully warm trip fuel consumptions
is equivalent to:
Cold Start Trip I4PG
Fully Warm t-lPG
1 + T
,-0.8
Where
T = trip length, in miles.
146
Op. Cit. (82)
I70°F]
-------
146
Using this formula, a trip length of 16 miles is required for cold start trip
MPG to reach 90% of fully-warm MPG; 40 miles for 95%, 130 miles for 98%, and
312 miles for 99%.
The triple figure illustrates how cumulative cold start trip fuel economy
varies with vehicle vintage, ambient temperature, and car size. The top graph
shows that cars built in the 1970's achieve relative warmup trip MFC's as high
as, or higher than, cars built prior to 1968. There is no significant dif-
ference in relative warmup trip MPG between pre-and post-1975 models.
The middle graph shows the depression of cold start MPG which results from
operation at low ambient temperatures. The solid curve is the Tobin and
Q-I
Horowitz formula, and the dashed curve from Scheffler and Niepoth , which
data was the basis for the formula. The circles are from EPA analysis of
oc
late-model car data from Mobil Oil , and reveal a significantly greater
temperature sensitivity for the later models (as was discussed earlier in this
report). At temperatures in the 70's, the later models' relative warmup
economy is quite like that of the older cars; however, the later models at
25°F achieve FWU fractions no better, and perhaps worse, than the older cars
at 10°F.
The bottom figure, also based on the Mobil data, shows that the warmup curves
vary with car size. This does not indicate that smaller cars warm up more
slowly than large ones, however, since the 75°F curves are nearly superimposed;
the difference in the curves at lower temperatures reflects higher sensitivity
to temperature for the small cars, also discussed earlier.
-------
147
£
_>-
3
t 40 -
o
g
v
t!
£
20
100
§
uj
•a
80
60
40
20
100
80
1
i
i «°
u.
"5
I 40
U
£
FIGURE 46. Cumulative (Trip) Fuel Economy
(a) Model year differences
75" F
O Pre-Emission
Controlled Vehicles-Scheffler and Niepoth'w _
IDA 1973 Vehicles
SAE Task Force Data (Unpublished)
• 1975-77 Vehicles-Mobil"15
_L
6 8 10
Trip Length, Miles
12
14
16
(b) Temperature /model year differences
T
T
T
7S°F
70° F
10° f
Scheffler and Niepoth \ Pre-Control
Tobin and Horowitz'*' ' Vehicles
1975-77 Vehicles
8
10
12
14
16
Trip Length, Miles
(c) Temperature / car size differences
T
T
1975-77 Vehicles
—— Large Cars
Small Cars
8
_L
10
12
14
16
Trip Length, Miles
-------
148
d. Average Miles per Day: A Useful Parameter: More often then not,
fuel economy surveys do not include detailed data on travel specifics such
as average speed and trip length. An easily obtainable parameter, which
does reflect the interaction of these characteristics, is average miles
traveled per day (AMPD). Correlation between AMPD and fuel economy has
147
been known for some time: the figure, from a 1964 study , shows the MPG
influence of AMPD for a small sample of cars from model years 1958 - 1963.
FIGURE 47. Fuel Consumption vs. Miles Per Day in User Operation
(1958-43 Models)
= Data
= Curve Fit:
MPG = 0.26 + 0.20 In (AMPD)
MPG
80 100
Average Miles/Day
Warren, "Some Factors Influencing Motorcar Fuel Consumption in Service",
SAE paper, 1964-
-------
149
For gasoline-powered cars, these data show that:
0 Fuel economy increases with AMPD; the hint of an MPG downturn above
120 miles/day is believed to reflect high-speed driving penalties
for long trips;
0 Fuel economy for the same vehicle can vary over a range of nearly 2-to-l
in response to AMPD;
0 At comparable AMPD, there is a clear distinction between vehicles' itPG
capabilities as a function of engine (and vehicle) size; these distinc-
tions can be masked by differential vehicle usage, however: four-cylinder
small cars can be driven inefficiently enough to subchieve MPG levels
lower than those of an efficiently-driven 8-cylinder large car;
0 Fuel economy response to AMPD can be approximated by a logarithmic relation,
shown by the heavy solid lines on the figure. The normalization point
for this curve fit was selected at 41.1 miles/day (15,000 miles/year).
The data also show the Diesel car to be relatively insensitive to AMPD, not
an unexpected result since Diesel engines are known to be highly efficient
at part load and less so at high RPM. This finding is also in line with
lower EPA Highway-to-City MPG ratios observed for late-model Diesel
Certification cars (1975-79 average ratio = 1.30) compared to gasoline
engine cars (1975-79 average ratio = 1.40).
Analysis of GM customer survey data shows the AMPD response of 1975 model
gasoline cars to be much the same as Warren's gasoline cars; the next figure
149
shows data for four GM 1975 model types. The apparent wider scatter in
this recent data comes from the cars being driven in different regions of
the U.S. and in different seasons of the year.
Murrell, "Light Duty Automotive Fuel Economy .,, Trends Through 1979",
SAE paper 790225, February 1979.
149
Janz, "Analysis of GM Fuel Economy Surveys", EPA Report EPA-460/3-76-029,
October 1976.
-------
150
FIGURE 48.
In-Use Fuel
Economy
vs.
Average Miles
Per Day
o
2
28
24
20
16
12
20
18
\ T
•
I
1975 Chevrolet
140 CO, 3.000 Lbs.
(4Cyl.,22Cars)
I
I
I
I
u
u
14
12
10
18
16
14
12
T
T
T
T
T
T
T
I
1975 Chevrolet
250 CID, 4,000 Lbs.
(6Cyl..32Cars)
I
I
10 20 30 40 SO 60 70 80 90
Miles Per Day
fe
2
Z 10
I I
I i I I L
_L
_L
197S Chevrolet
350 CID, 4,500 Lbs.
(8Cyl..77Cars)
J L_
10 20 30 40 50 60 70 80 90 100
Miles Per Day
100
10 20 30 40 SO 60 70 80 90 100
Miles Per Day
14
3 "
fc I0
2
Z 8
III
Cadillac
500 CID, 5,500 Lbs.
(8Cvl..83Cars)
I
I
10 20 30
40 SO 60
Miles Per Day
70 80 90
100
-------
151
The curves superimposed on each of the four data sets are not separate curve
fits for those models, but the curve fit for the entire survey data base,
MPG
MPG
= 0.535 + 0.125 In(AMPD)
41. 1
A travel characteristics analysis made for this report (see Appendix E)
relates normalized MPG to trip frequency and length; when plotted against
AMPD, the figure below is obtained. This model is more than just shapewise
similar to the vehicle data: it is (at constant trip frequency) in excellent
agreement with a logarithmic MPG:AMPD relationship, whose log term coefficient
lies between those of the curve fit equations for the Warren and GM data.
FIGURE 49. Relationships Between Fuel Economy and Travel Characteristics
(From Appendix E)
I 10
——— Relative MPG =
0.43 -I- 0.14 In. (AMPD)
10
40
Average Miles/Day
I
5,000
10,000
15.000
Average Mites /Year
20,000
I
25,000
-------
152
As an illustration of the value of knowing the AMPD characteristics of cars in
a data base, consider the following. Unpublished data furnished by Mobil Oil
Co. on the road fuel economy of a number of 1975 and 1976 company cars, each
driven by several employees over a period of weeks, was analyzed for in-use
fuel economy influences. Ranking the cars in order of average MPG slip (road-
to-EPA MPG ratio), the top 20% and bottom 20% were compared. The "best-slip
cars" were mostly 1975 models, relatively large, all V-8 powered (average CID
= 341), with average EPA MPG less than 15; the "worst-slip cars" were mostly
1976's, smaller, predominantly 4- and 6-cylinder (average CTD = 188), with
average EPA MPG above 20. Based on this alone, it could be concluded that
higher shortfalls are characteristic of small, high-MPG cars, and/or there is
a year-to-year growth in road shortfall.
When the usage patterns of these two groups of cars are compared, however, the
following appears:
"Best Slip" "Worst Slip"
Cars Cars
Average EPA MPG 13.8 23.3
All Possessions (up to 14 drivers each car)
Average miles/day 37.6 28.3
Average road MPG 13.2 17.1
Average slip, R/E 0.96 0.74
Best-MPG Possessions (all cars)
Average miles/day 50.8 35.9
Average road MPG 16.3 19.6
Average slip, R/E 1.18 0.85
Worst-MPG Possessions
Average miles/day 30.0 9.9
Average road MPG 11.3 14.7
Average slip, R/E 0.82 0.63
-------
153
The best-slip cars (notice we have labeled the cars) are the ones driven at
higher AMPD's, and the worst-slip cars are the ones subjected to the least
fuel efficient travel.
Relative success of the various drivers of the Mobil cars sheds yet more
light on the effect of AMPD on fuel economy; ranking the drivers in order of
MPG slip, and comparing the top 20% and bottom 20%, we find:
"best MPG" "Worst MPG"
Drivers Drivers
Average EPA MPG 16.1 17.2
Average miles/day 44.6 30.2
Average road MPG 14.9 13.4
Average slip, R/E 0.92 0.78
Although working with a group of vehicles with a 6% EPA MPG disadvantage, the
"Best MPG" drivers achieved 11% higher MPG on the road — by driving nearly
50% higher AMPD.
The "Best MPG" drivers, of course, are not necessarily the most fuel conserva-
tive. This is a good example of the pitfalls inherent in making value judg-
ments such as "Best" or "Worst" on the basis of fuel economy. When fuel
consumption is considered, the value judgements sometimes reverse as the
following table indicates:
"Best" (MPG Basis) "Worst" (MPG Basis)
Drivers Drivers
Miles Per Day 44.6 30.2
Miles Per Gallon 14.9 13.4
Gallons Consumed Per Day 3.0 2.3
Value Judgement, Based on
Fuel Consumption "Worst" "Best"
-------
154
This shows that emphasizing fuel economy (MPG) as a figure of merit can lead
to conclusions opposite those based on use of fuel consumption as the figure
of merit. One problem occurs, however, when attempting to evaluate fuel
efficiency on a consumption basis: fuel consumption is not a measure that
many people are familar with. While many people have an idea of how many
"MPG's" they get with their car, we would predict that significantly fewer
people know how many gallons per mile they consume. This discourages analysis
or discussion in fuel consumption terms, even though reduced fuel consumption
is probably a more appropriate measure, and higher fuel economy is only one
approach toward reducing fuel consumption.
AMPD has been found to vary with population density and other factors, from
the 1975 GM customer survey data (although the statistical significance of the
observations is not good). The following table lists these findings. AMPD is
notably lower for more densely-populated areas, and—for all population
densities—AMPD is lower for the smaller cars. Similarly, MPG shortfalls are
worse for higher population densities and for higher-MPG levels.
-------
155
Dependence of AMPD on Population
Density and Other Factors, MY 1975 GM Cars
Population (Owner ZIP Code)
Percent of Cars in Sample
Percent of VMT (NPTS#7)15°
Average miles/day
Standard Deviation of AMPD
Average slip, Road/EPA
AMPD vs. Vehicle Weight:
3000 Ib
4000 Ib
5000 Ib
AMPD vs. Engine CID:
150 cu. in
250 cu. in.
350 cu. in.
Road MPG:
EPA = 10
15
20
27.5
25.000
48.0
66.3
46.6
61.3
.901
40.9
44.8
48.7
42.3
44.5
46.8
9.2
13.3
17.2
22.5
25,000-
999,999 > 1 million
47.6
27.3
39.1
41.2
.855
31.3
36.6
41.9
34.3
36.8
39.3
8.6
12.5
16.2
21.2
4.<4
6.4
29.6
18.0
.798
28.1
29.1
30.0
23.1
26.2
29.4
8.0
11.7
15.2
20.3
U.S. Average:
Sample-
Weighted
42.3
—
.874
35.9
40.2
44.5
37.9
40.2
42.4
8.9
12.8
16.6
21.8
VMT-
Weighted
43.5
—
.881
37.5
41.6
45.6
38.9
41.2
43.6
9.0
13.0
16.8
22.0
Goley, et_ al, "Nationwide Personal Transportation Study, Household Travel
in the United States, Report No. 7', DOT/FHwA, April 1972.
-------
156
e. Acceleration Intensity: As mentioned earlier, once vehicle average
speed is determined by all the factors that influence it, the only other
factors that affect cyclic driving fuel economy are related to acceleration
and deceleration intensity. These two factors are highly correlated ,
reflecting a sort of "aggressiveness index" which is common to speed changes
in general, whether accelerating or decelerating. While decelerations are of
critical interest for emissions, it is the accelerations in a driving cycle
that impact most heavily on fuel consumption, and it is upon those that we
will concentrate.
Two different aspects of acceleration level are of interest here. There is
the matter of the usual quantitative engineering research, wherein acceleration
rates and their MPG influences have both been carefully measured. Perhaps
more striking, however, are reports of the significant effect on fuel economy
of driver acceleration habits and modification of those habits.
O-I
(1) Quantitative Studies - Scheffler and Niepoth reported in 1965 that
halving the 4.1 mph/sec acceleration rates of a city driving cycle improved
MPG by 7.7%, while doubling the base rates caused a 5.8% loss.
More recently, EPA has conducted and sponsored tests and computer simulations
of the fuel economy effects of altered speed change rates in the EPA test
cycles.
The test projects involved modified versions of the EPA Highway cycle, wherein
acceleration and deceleration rates were either increased by some 50% (an
"increased-noise" modification) or completely eliminated (a "smoothed" modifica-
tion, with a constant cruise at 50 raph between the cycle's initial acceleration
and final deceleration). The next figure shows the standard Highway cycle
speed vs. time schedule and an example of a noise-accentuated version of the
cycle.
The results of the tests are summarized in the table following. All of the
test cars were 1976 models.
151Evans, et al, Op. Git. (132)
-------
FIGURE 50. Example of Driving Cycle Modification in EPA Dyno/Track Project
15;
10
-------
158
Effect of Modified Acceleration/Deceleration Rates
on EPA Highway Cycle Fuel Economy: Test Data
Standard "Noisy1 "Smooth"
Cycle Cycle Cycle
Average Speed, mph 48.2 48.6 48.3
Average Speed Change
Rates, mph/sec
-Acceleration 0.33 0.48 Zero
- Deceleration -0.34 -0.48 Zero
Fuel Economy Effect ,
SwRI, 3 cars, dynamometer -4.3% +4.4%
EPA, 6 cars, dynamometer0 -3.0% +4.5%
EPA, CVCC, dynamometer -0.5% +2.0%
EPA, 6 cars, test track -4.4% +6.5%
Excluding initial acceleration and final deceleration
Reference 122.
ft
Dynamometer vs. Track Project, Phase II (unpublished)
Fuel economy sensitivity to accelerations did not vary with vehicle
size in either of these test projects, for those cars using conventional
engines. In the EPA dynamometer tests, the CVCC stratified charge car showed
noticeably less acceleration sensitivity than the other cars. In the SwRI
tests, the application of significant (^ 30%) increases in inertia weight and
road load horsepower resulted in only slight reductions in fuel economy sensi-
tivity to acceleration rate.
In the computer simulation projects, accentuated-noise versions of three
driving cycles (New York City, EPA City and EPA Highway) were employed. Model year
effects were also investigated, and two completely different simulation programs
were used. Although the simulations were computer models, all key inputs came
from test data from real vehicles or engines. The results are shown in the
next table.
-------
159
Effect of Modified Acceleration/Deceleration Rates
on Cyclic Driving Fuel Economy: Computer Studies
EPA City EPA Highway
NYC Cycle Cycle Cycle
Average Speed, mph
Base Cycles 7.5 19.6 48.2
"1.5x" perturbed cycle 9.4 21.0 48.3
"2.Ox" perturbed cycle 13.9 22.9 48.4
Average Acceleration rate, mph/sec
Base Cycles 1.38 1.13 0.44
"1.5x" perturbed cycle 2.97 1.73 0.65
"2.Ox" perturbed cycle 4.69 2.65 0.96
Average Deceleration rate, mph/sec
Base Cycles -1.36 -1.29 -0.50
"1.5x" perturbed cycle -2.96 -1.92 -0.72
"2.Ox" perturbed cycle -4.75 -2.72 -1.14
Fuel Economy Effect: l^5x 2.Ox l.Sx 2.Ox 1.5x 2.Ox
1973-74 Models3 -24.3% -49.3% -16.0% -36.5% -7.5% -16.0%
1975 Models3 -34.5% -62.2% -20.7% -44.4% -5.9% -16.1%
1975-77 Modelsb -19.9% -27.1% -16.8% -34/8% -6.6% -15.6%
1975 CVCCb -7.0% +4.2% -8.2% -17.0% -4.1% -10.2%
Analysis Simulations, Ref. 135; 180 1973-74 cars, 225 1975 cars
Modified DOT VEHSIM Simulations, Ref. 152: three cars
For the two lower-speed cycles, the 1975 models appear more highly
MPG-sensitive to acceleration thau the 1973-74 KGaeis. The stratified
charge car (as was observed in the EPA tests) shows significantly less
acceleration sensitivity than those with conventional rioverplants.
152
Thacker and Smalley, "Emission Modeling and Sensitivity Study" EPA
Report EPA 460/3-79--005, May 1979. '
-------
160
The average acceleration and deceleration rates for the various EPA cycles,
and also the SAE road test cycles, are compared with traffic survey data
in the next figure. The average rates for the NYC and EPA cycles are roughly
half those of the survey data, while average rates for the SAE cycles and the
"most-perturbed" EPA urban cycles are at least la above the survey average
rates. For perspective, the T'+2o" acceleration level corresponds closely
to the envelope of a wide-open throttle acceleration from zero to 50 mph in
13 seconds.
FIGURE 51. Comparison: Average Acceleration Rates (Time-Weighted):
EPA Cycles, SAE Cycles, and Traffic Survey Data
u
I
!<
I I
8 8
S y
OL <
c I
•s £
s S
< "5
t s
? Q
I
_ f Average Accel
• v w Traffic Survey
» • Avg. Decel.
f ... = Scott Survey
Standard EPA Cycles
O= Modified EPA Cycles
• = SAE Cycles
m= 13.3 Second Acceleration
from 0 to SO MPH100
10
20 30
Average Speed, MPH
153Scott Research Laboratories, Op. Cit. (125)
-------
161
It can be misleading to compare only the average speed change rates of these
cycles, because large differences exist among the cycles as to proportion of
time spent accelerating. A more consistent measure of cycle acceleration
intensity effects is the integral:
a-v-dt
a
where "a" is instantaneous acceleration rate, "v" is instantaneous speed, and
"dt " represents each time increment spent accelerating. The cycles can be
cl
compared on the basis of this integral, or its approximation:
a-v-1 /t,
a t
where a and v are cycle average values and t /t is fraction of cycle time
154 H t
spent accelerating . The resultant parameter has the units of power per
unit mass; for familiarity it is expressed as horsepower for a 3500-lb vehicle,
in the following table:
Cycle
NYC,
SAE Urban
EPA City,
base
1.5x
2x
base
1.5x
2x
SAE Suburban
EPA Hwy,
Survey
base
1.5x
2x
a
1.38
2.97
4.69
4.17
1.13
1.73
2.65
2.33
0.44
0.65
0.96
1.78
15.6 0.12 3.32
1.13
1.73
2.65
19.6
21.0
22.9
0.40
0.41
0.42
3.77
6.33
10.83
41.1 0.10 4.07
0.44
0.65
0.96
48.2
48.3
48.4
0.44
0.48
0.51
3.97
6.41
10.07
26.0 0.29 5.70
154
Use of the time fraction normalizes all the data, since total time
values range from minutes for the test cycles to hundreds of hours for
the survey data.
-------
162
These data are shown in the following figure, with the average acceleration
153
power for the traffic survey distributed over the speed range. The fact
that the base cycles' average power requirements lie on the same smooth curve
explains in part why fuel economy results (shown earlier) from these base
cycles also lie on a smooth curve.
FIGURE 52. Comparison: Test Cycles' Average Acceleration Power vs. Traffic Survey Data
is
10
8
o_
«
« 5
NYC
(2x)
10 20 30
Average Speed, MPH
O EPA-H
(2x)
50
It is clear that actual traffic, at low average speeds, involves significantly
higher acceleration power loadings than the base versions of the EPA and SAE
test cycles. The next figure illustrates the fuel economy penalties associated
with the harsher acceleration modified test cycles.
Fuel being wasted by higher acceleration driving can be saved through
more conservative driving habits, as discussed in the next section.
-------
163
FIGURE 53. Fuel Economy Penalties for Acceleration in Excess of EPA & SAE Test Cycles
It
Si
u
U 3
T3 C.
| «
II
m C
1l
1.00
10
20 30
Average Speed. MPH
40
50
(2) Driver-habit Studies - The Southern California Auto Club measured
fuel economy effects of subjectively-selected ("easy", "moderate", and ''heavy")
acceleration styles. A total of twenty 1969-73 cars were tested, including
six weighing less than 3000 Ib., eight between 3000 and 4000, and six over
4000 Ib. The tests were run on an auto raceway, and consisted of 0-40 mph
accelerations, followed by 40 mph cruises, for a total distance of 1/4 mile.
Duplicate runs were made at each acceleration intensity. The results are as
follows:
Effect of Acceleration Level
on Fuel Economy
"Easy" Acceleration
"Moderate"
"Heavy"
Small
Cars
+7.8%
-15.3%
Medium
Cars
+12.5%
.Base Case.
-11.1%
Large
Cars
+16.1%
-9.4%
This would suggest that smaller cars have less to gain and more to lose as
acceleration intensity is varied, but since actual rates were not measured, it
is not known whether the tests really imposed similar rates on all the cars.
Bintz and Banowetz. "Fuel Economy Testing", Automobile Club of Southern
California Report, September 1973.
-------
164
A more recent study , jointly sponsored by the SAE and California State
University at Chico, similarly evaluated the effect of subjective acceleration
variations on fuel economy. Three late-model cars were driven by 11 to 15
drivers each, over a four-mile course having 13 stops plus 14 right-angle
turns. The results are shown in the figure, with MPG plotted against average
speed. Since the test course and its stop and slowdown constraints were
fixed, the higher average speeds correspond to harder-acceleration driving.
F
50
S2 40
E
1 30
JJ
Stop & Go Cy
5 g
CURE 54. Fuel Economy Under Variable Acceleration Conditions
O
-
0 0
1
Source: Donoho"*
O O = 1 978 Ford Fiesta
A = 1 977 Merc. Monarch
• = 1 978 Chev. Monte Carlo_
O
0 0
° 0 0
• • f
1
00
A° °
. A .
1
5 20 25 30
Average Speed. MPH
As discussed earlier, fuel economy improves with higher average traffic flow
speeds (up to some 40 mph) when normal accelerations are used. The Chico
State data show that this potential MPG improvement is, at best, nullified
when the average speed increase is brought about by hard accelerations. In
the case of the smaller 4-cylinder car, the result of acceleration-related
speed increases caused significant fuel economy losses.
156
'Donoho, "EPA MPG ~ How Realistic?" SAE paper 780366, December 1977.
-------
165
Claffey conducted tests over a 3.5-mile course with 14 stops and 19
turns, using one 1968 car and 55 drivers. The drivers were divided into
four groups:
(a) 26 males, ages 16 to 70, who each drove the course once, using
"normal" driving techniques (unprompted), then repeated the
drive using a dash-mounted vacuum gauge to promote reductions
in acceleration rates;
(b) 20 females, ages 16 to 80, who each drove the course twice as
did group (a);
(c) 8 drivers (5 female, 3 male), ages 30 to 60, who each drove
the course twice: acceleration reductions in this group's
repeat drives were accomplished without vacuum gauge assistance;
this group initially (when driving normally) had a lower
average MPG than groups (a) and (b);
(d) One driver (male, age 53) who made multiple trips driving
normally, then more multiple trips at two progressive levels
of fuel conservation technique (without gauge assistance);
in the first (unprompted) round, this driver's average MPG was
lower than any of the averages for groups (a), (b) and (c).
In the "normal driving" trips, acceleration rates up to 6.2 mph /sec and
deceleration rates to -9.6 mph/sec were observed. Manifold vacuum
readings below 10 in. Hg (indicating hard acceleration) occurred for 35%
of the drivers (19 out of 55). Overall average "normal driving" speed
was 21.4 mph. In the gauge-assisted repeat trips for groups (a) and (b),
manifold vacuums were kept from going below 10 in. Hg.
In the gauge-unassisted repeat trips ("full fuel conservation" driving) for
groups (c) and (d), acceleration and deceleration rates were limited to +3.5
mph/sec and -4.5 mph/sec respectively. Overall average speed for these
conditions was 18.1 mph.
1570p. Git. (114)
-------
166
For groups (a) and (b), the effect of vacuum gauge assistance depended upon
whether "normal" driving was harsh or conservative. For those who were already
conservative drivers (58% of the males and 80% of the females), gauge-assisted
driving produced mixed results, ranging from small MPG improvements to small
MPG degradation: for some of these drivers, the gauge was more of a distraction
than an aid. For all of the initially hard-accelerating drivers (42% of the
males and 20% of the females), MPG was improved by gauge-assisted acceleration
reduction. The results are summarized in the next table.
Fuel Economy Effect of Limiting Accelerations with
Vacuum Gauge Assistance
Group (a) Group 'b)
26 Males 20 Females
A. "Easy" Drivers: Those with no low vacuums when driving normally
1. Those whose MPG improved with gauge assistance
Number 5 5
MPG, normal driving 10.47 10.60
MPG, limited acceleration 10.54 10.76
Change in MPG +0.7% +1.5%
2. Those whose MPG did not change with gauge assistance
Number 3 3
MPG, normal driving 10.25 10.38
MPG, limited acceleration 10.25 10.38
3. Those whose MPG worsened with gauge assistance
Number 7 8
MPG, normal driving 10.59 10.63
MPG, limited acceleration 10.33 10.33
Change in MPG -2.5% -2.8%
Total effect for all "easy': drivers combined:
Number 15 16
(58% of males) (80% of females)
MPG, normal driving 10.48 10.57
MPG, limited acceleration 10.38 10.47
Change in MPG -1.0% -0.9%
-------
167
Fuel Economy Effect of Limiting Accelerations with
Vacuum Gauge Assistance (Cont'd)
Group (a) Group (b)
26 Males 20 Females
B. "Hard" Drivers: Those with low vacuums when driving normally
Number 11 4
(42% of males) (20% of females)
MPG, normal (for them) 9.87 10.02
driving
MPG, limited acceleration 10.25 10.52
Change in MPG +3.9% +5.0%
For group (c), all of the initially "hard" drivers, and four of the
five initially "easy" drivers, achieved improved fuel economy by accelera-
tion reductions without gauge assistance, as shown in the next table.
Fuel Economy Effect of Limiting Accelerations
without Vacuum Gauge Assistance, Group(c)
Males Females
A. "Easy" Drivers;
o
Number 2 3
MPG, normal driving 10.46 9.93
MPG, limited acceleration 10.82 10.52
Change in MPG +3.4% +5.9%
B. "Hard" Drivers:
Number 1 2
MPG, normal driving 8.88 9.90
MPG, limited acceleration 9.29 10.49
Change in MPG +4.6% +6.0%
aTwo improved (+18.4% and +6.4%); one worsened (-6.0%)
The test results for the last driver, "group" (d), again confirmed the fuel
economy improvement potential of conservative speed change rates, and also
yielded data on fuel economy repeatability, given in the following table:
-------
168
Fuel Economy Effects of Limiting Accelerations,
Driver (d)
Normal Partial Fuel Full Fuel
Driving Conservation Conservation
Number of trips 15 20 11
Fuel Economy, MPG:
Average
Range
C.O.V.
9.84
9.51-10.28
2.17%
10.02
9.78-10.33
1.70%
10.50
10.28-10.76
1.26%
With progressive attention to minimizing acceleration and deceleration rates,
this driver improved his average fuel economy by 6.7% and improved in con-
sistency as well. Under the conditions most highly conducive to repeatability
(same driver, same car, same route, controlled driving technique), there
remained a 1% coefficient of variation, and a range of 0.5 MPG. This residual
variability includes the driver's trip-to-trip variability, the car's test-to-
test variability, and the variability of the fuel measurement hardware.
158
A driving-technique experiment was reported by Evans for 34 trips made by
nine different drivers over a fixed 16.8-mile route in suburban Detroit. All
trips used the same 1974 vehicle, and the drivers were given various driving
instructions. The driving techniques used were in response to these qualita-
tive instructions. Two instructions involved the use of a vacuum gauge "fuel
economy meter" with a three-color dial: a green (high vacuum) region intended
to indicate good fuel economy, and orange and red regions indicating lower
vacuums, higher power, and reduced fuel economy. The instructions used were:
1. "Drive normally with the traffic";
2. "Minimize trip time";
3. "Use vigorous acceleration and deceleration";
4. "Minimize fuel consumption";
5. "Maintain fuel economy meter in green region";
6. "Maintain fuel economy meter in green or orange region";
7. "Drive li'ke a hypothetical very cautious driver".
Evans, "Driver Behavior Effects on Fuel Consumption in Urban Traffic",
General Motors Research Report GMR-2769, June 1978.
-------
169
In response to instruction 4, the drivers divided into two subgroups which
interpreted the instruction differently. One group responded by reducing
accelerations and speeds, while the other group attempted to minimize stops
at traffic signals, even occasionally accelerating to "make a light".
In response to instruction 5, it was found impractical to keep the meter in the
green zone only; the required acceleration rates were unrealistically low with
respect to the prevailing traffic. Allowing some orange-zone meter readings
(instruction 6) permitted more realistic driving, while still imposing some
limits on acceleration rates.
The results of the experiment are given in the table, with the instructions
(and sub-interpretations of instruction 4) listed in order of decreasing fuel
economy.
Effect of Driving Technique on Fuel Economy and Trip Speed
Fuel Economy,
4b.
7 .
4a.
6.
1.
5.
2.
3.
Driving Technique
Minimize stops
Drive very cautiously
Reduce accels . and speeds
Keep meter in green/orange
Drive normally
Keep meter in green
Minimize trip time
Use vigorous acceleration
Avg.
13.52
12.51
12.44
12.31
11.64
11.42
10.59
9.92
Chan
+16.
+7.
+6.
+5.
>IPG
ge
n
4%
8%
7%
(base)
-2.
-9.
-14.
0%
0%
%
Average Speed, mph
Avg.
25
23
24
24
25
18
29
28
.93
.31
.06
.75
.11
.96
.06
.11
Change
+3
-7
-4
-1
.3%
.2%
.2%
.4%
(base)
-24
+15
+11
.5%
.7%
.9%
Trip Time, min.
Av;
38
43
41
40
40
53
34
35
.8
.2
.8
.7
.1
.1
.6
.8
Change
gain 1
lose 3
lose 2
lose 1
(base)
lose 13
gain 6
gain 4
-------
170
With two notable exceptions, those instructions which increased average speed
decreased fuel economy, and those which decreased speed increased fuel economy.
In the highest-MPG case, "playing the lights", both average speed and fuel
economy increased, in consonance with earlier observations that less stops (and
corresponding higher average speeds) are good for fuel economy. Tn the case of
instruction 5, average speed slowed, but fuel economy also decreased. Possible
causes may include failure to achieve fuel-efficient cruise speeds, and aero-
dynamic disturbance from the traffic passing the test car.
(3) Effectiveness of Fuel Economy Meters - The evaluation of devices such
as fuel economy meters is not within the purposes of this report; however, it
is clear from the preceding section that instantaneous reading MPO meters
are of questionable value in improving vehicle fuel economy. It is
pertinent here to quote conclusions from three sources vis-a-vis instanta-
neous meters:
[From Claffey114]:
0 "Nearly 70 percent of all drivers customarily drive without opening the
throttle plate excessively anyway (maintain engine vacuum above 10 inches
of mercury). These drivers would not be helped to save fuel by a vacuum
gauge.
° Fifty percent of these drivers would actually consume more fuel if they
were distracted in their driving by trying to adjust their driving habits
to a vacuum gauge. This is especially true of women and older persons.
0 Another 30 percent of all drivers, those who customarily jab the throttle
pedal on accelerations, could benefit from a vacuum gauge as it would
alert them to their fuel-wasting driving habits.
0 The gauge distracts driver's attention and is unsafe. During test opera-
tions, there were, in 225 vehicle-miles of travel, three near-miss acci-
dents directly caused by the driver's attention being distracted by the
vacuum gauge.
On the whole, a vacuum gauge mounted on the dashboard of all cars would
not only have a negative effect on overall fuel conservation but would
also increase the danger of highway accidents. It is not recommended."
-------
171
1 SR
[From Evans ]:
"Banowetz and Bintz (1977) [our Ref. 159] compared the fuel economies of
70 vehicles equipped with fuel economy meters to those of 70 control
vehicles not so equipped. The fuel economy meter displayed the (es-
sentially) instantaneous miles per gallon. The drivers of all 140
vehicles were motivated to save fuel. The vehicles were used for 12 weeks
of normal driving. The meter-equipped vehicles had fuel consumption, on
average, 3% less than the non-equipped vehicles, though the authors report
that they did not find this difference to be statistically significant."
[From DOT, in a report to the Congress and the President, commissioned by
the Energy Policy and Conservation Act, Section 512(a)]:
0 This study did not establish that use of mpg meters in new cars would save
enough fuel to measurably reduce the nation's fuel consumption and/or to
offset their own cost within a reasonable period of time, where the reason-
able period for cost offset was taken to be 3 years (i.e., first ownership).
0 It has not been established that use of mpg meters will save significant
amounts of fuel in average vehicles driven over average operating condi-
tions. Moreover, mpg meters have little potential for promoting fuel
savings under congested traffic conditions.
0 It would require about three years for a new large car or about six years
for a new small car to pay for a factory installed meter if a 5% fuel
economy increase could be obtained. For the least expensive commercially
available mpg meters, costing about $130 installed in a used car, fuel
economy increases of about 12% would be needed to cover the installed
cost of the mpg meter within three years. These percentage increases in
fuel economy are hypothetical examples.
0 Means for encouraging consumers to purchase automobiles equipped with mpg
meters include advertising, driver education, tax benefits, and subsidies
to manufacturers of meters. These measures are only likely to_ be effective
when it is shown that mpg meters are effective, economical, convenient and
safe to use.
There should be no requirement to install mpg meters in new cars.
0 The Federal government should take no action to promote the installation
and use of mpg meters in used cars at this time.
159
Banowetz and Bintz, "Field Evaluation of Miles-Per-Gallon Meters", DOT
Report DOT-TSC-OST-77-64, November 1977.
U.S. Department of Transportation, "Effectiveness of Miles-Per-Gallon-
Meters as a Means to Conserve Gasoline in Automobiles", July 1976.
-------
172
We emphasize that these conclusions apply to instantaneous-reading
devices. Other types of "fuel economy meters" which give cumulative
readings, rather than instantaneous readings, are in use; these other
types may or may not share the disadvantages of the instantaneous-
reading types.
f. Summary - Travel Characteristics Effects: The various aspects
of vehicle travel are closely interrelated, and cannot be treated separately.
Appendix E contains the analysis leading to our estimates of travel
characteristics of new cars of recent model years, i.e., the type of
vehicles addressed by this report.
Compared to EPA 55/45 conditions, that analysis shows that such vehicles
travel longer average trip distances, at higher average speeds, with
higher average acceleration rates, and a higher fraction of trips beginning
with a cold engine, as summarized below:
Comparison of Estimated Travel Characteristics
to EPA 55/45 Conditions
Estimated Actual MPG
EPA 55/45 Actual vs. EPA
Average trip length, miles 8.5 9.7 0.8% Higher
Fraction of trips started cold 26.9% 35.9% 0.7% Lower
Average trip speed, mph 26.7 32.7 10.5% Higher
Average acceleration HP 3.6 3.9 11.8% Lower
(3500 Ib car)
Thus, while specific travel details, on average, appear to be different from
the EPA tests, the overall fuel economy shortfall that can be associated
with travel characteristics in the aggregate is, due to the opposing effects,
less than might be expected.
The detailed analysis, which considers regional, seasonal, urban vs.
rural, and car size effects on vehicle travel, indicates small net
shortfalls relative to EPA 55/45 conditions, as given in the next table.
-------
173
Relative Fuel Economy Associated with
Travel Characteristics Effects
(EPA 55/45 MPG = 1.000)
Total U.S. (VMT-weighted)
Small Cars Large Cars
Spring
Summer
Fall
Winter
Annual
0.951
0.980
0.957
0.941
0.958
(Shortfall
= 4.2%)
0.977
1.004
0.982
0.968
0.983
(Shortfall
= 1.7%)
The seasonal and car-size differences in relative MPG seen in the table
above are due solely to differential vehicle travel patterns; the nominal 4%
Summer-to-Winter MPG loss, for example, reflects only vehicle usage, not
Winter losses related to temperature or road conditions.
-------
174
(This page intentionally blank)
-------
175
FUEL ECONOMY INFLUENCES (Cont'd.)
Ttoad Slip (Cont'd.) Page
Vehicle Condition (Road) ..... 176
Wheel Condition 176
Tire Size . 177
Tire Pressure 179
Lubricants -181
Availability of Lubricants 188
Total Energy Balance . • • • *88
Vehicle Weight Load 190
Simulation Variance 191
Low-Speed Dynamometer Loading 191
Tire/Dynamometer Interaction • 193
Weight Class Distributions • • .197
Manual Transmissions .....> 199
Power Accessories . . . ... 202
Open Windows vs. Air Conditioner Operation 206
Vehicle Cooling 207
Metric Slip 207
The EPA Carbon Balance Method .......... 207
In-Use Fuel Economy Determinations ........ 210
Fleet Car MPG , 210
Consumer MPG . . . . . . . . . . .... . . . . . 212
Variations Common to Both In-Use Methods 212
Summary -Metric Slip 213
Sunmary Findings: Road Slip 214
Model Year Differences 215
Fuel Economy Effects in Combination .216
*Math«»attoil .Implications » . . . . . . . . . .... ; 216
Engine M«p^ Considerations . , .,...,.,,... t 217
, Actual Examples 220
-------
176
3. Vehicle Condition (Road)
This section considers fuel economy influences whose MPG effects are seen
primarily, or exclusively, on the road.
a. Wheel Condition - The mechanical condition of the vehicle wheels can
affect fuel economy through excessive brake drag and wheel misalignment.
Normal levels of drag In disk brakes are sufficient to cause fuel economy
losses up to about 2.5%; improved brake system designs can eliminate more
than half of this loss . Additional data on abnormal disk brake drag, drum
brake systems, and parking brake effects, and on the in-use distributions of
these factors, would be required to evaluate precisely the magnitude of the
in-use shortfall due to excessive brake drag, but we believe a figure on the
order of 0.5% to 1% to be a conservative estimate.
162
Wheel alignment can have very significant MPG effects. Yurko reports
increases in tire rolling resistance of up to 25% per degree of slip angle,
for tests of wheels not connected to a vehicle front-end suspension system.
In vehicle tests at approximately 50 mph, the same report indicates rolling
resistance increases up to 75% for front-wheel misalignments exceeding manu-
facturer's recommended maximum toe-out by 1/2 inch, and up to 23% for a 1/2
inch excessive toe-in. The Motor Vehicle Manufacturers Association reports
an 0.3 MPG fuel economy loss for 1/4 inch improper front wheel toe-in align-
ment; the operating condition and base level MPG are not specified. Un-
published data submitted to EPA by Honda Motors reports fuel economy penalties
on the EPA City and Highway cycles of 3% and 2% respectively for front-wheel
misalignments of 2 millimeters, and even measurable effects (1% MPG loss on
the Highway cycle) of unbalanced tightening of wheel lug nuts.
161Porter, "Design for Fuel Economy - The New GM Front Drive Cars", SAE paper
790721, June 1979.
162Yurko, "The effect of Wheel Alignment on Rolling Resistance - A Literature
Search and Analysis", Report 78-12, Standards Development and Support Branch,
ECTD, EPA, July 1978.
Motor Vehicle Manufacturers Association, "Automobile Fuel Economy", September
1973.
-------
177
Reference 162 points out from a DOT survey of 125,000 vehicles in five states
that 19% of cars fail front-end alignment inspections, and concludes that
perhaps 10% of all in-use vehicles are operating with a 4% MPG penalty due
to front-wheel misalignment. This would correspond to a fleetwide MPG
shortfall of 0.4% relative to the EPA tests.
b. Tire Size - One tire effect which shows up in both dynamometer
tests and on the road is the effect of tire size. Since tires of different
sizes have different rolling radii, a tire size switch affects the
vehicle's N/V ratio164. A study of the rolling radii of various size
tires in the 1977 Certification fleet gives the following N/V changes
for various changes in tire size:
Average Change in N/V Ratio:
Tire Size Change iJ/V Change
One-letter shift in tire size (e.g. CR78-14 to DR78-14) -1.9%
One-inch shift in wheel size (e.g. F78-14 to F78-15) -2.6%
Both of the preceding (e.g. FR78-14 to GR78-15) -4.6%
Bias-belt to Radial type (e.g. B78-13 to BR78-13) +1.2%
Bias to Radial + 1 letter (e.g. C78-14 to DR78-14) -0.7%
Bias to Radial + 1 inch (e.g. H78-14 to HR78-15) -0.9%
Bias to Radial + both (e.g. F78-14 to GR78-15) -3.0%
Using average N/V sensitivity factors from Section IV.A.3. and the N/V
changes which accompany tire size shifts, and assuming a 4% advantage
for radial tires over bias-belted tires, the 55/45 fuel economy effects
of the tire size changes are as shown in the next figure. Only changes
which improve MPG are shown; of course, tire size changes against the
direction of the arrows will decrease fuel economy.
1 AA
See Section IV.A.3.
-------
178
FIGURE 55. Typical Fuel Economy Effects of Tire Size Changes*
Increase
Increase One |nd,
One Letter
( + 0.8%)
Radials
n
Bias-Belted
^S.
M
Both
Increase J^T* (+1-9%)
One Letter f^-^ ^
Switch and
... . Incre
Switch and Switch and
Increase lncrease
Switch One Letter °™'n<* 1
Only ( + 4.3%) ( + ^%)
'
3.5
XX
' *
k.
S~^
*
aseBoth
5.3%)
Bias-Belted. Switch to Radial
*lf tire make is switched. MPG change can be higher or lower.
These calculated effects may be conservative estimates of the influence
of tire size on fuel economy. In a study of over 30 types and sizes
of tires, the measured effects of wheel size changes over part of the
EPA City cycle were as follows:
Measured MPG Effect of Change
from 14" Wheels to 15" Wheels, First
505 Seconds of EPA City Test
Cold Start
Hot Start
Radial Tires
+3.7%
+3.6%
Bias-Belted Tires
+3.7%
+4.7%
Consumers have the option of making significantly fuel-efficient tire
size changes, within mechanical limits such as wheel well clearance.
Torres and Burgeson, "Comparison of Hot to Cold Tire Fuel Economy"
Report 78-16, Standards Development and Support Branch, ECTD, EPA,
December 1978.
-------
179
c. Tire Pressure - This tire effect is exclusively a road slip item,
since fixed, uniform tire pressures are used in dynamometer testing.
An extensive survey^^of the cold inflation pressures of nearly 9000
in-use tires reveals widespread underinflation, as much as 15 psi below
manufacturers' recommendations. The figure shows a typical pressure
distribution.
i
M
12
10
8
6
4
2
0
FIGURE 56. Example of In-Use Tire Pressure Distribution
~~! I I I 1 T
1 1 T
I
I
I
I
I
•17 -15 -13 -II -9 -7 -5 -3-1 I 3
Difference from Recommended Pressure, psi
Unpublished data from employee parking lot tire pressure surveys by B.F.
Goodrich and Uniroyal show much the same results.
The MPG effects of tire pressure can be estimated using a tire energy
sensitivity of -2.8% per psi and MPG sensitivity to tire energy of
166
Viergutz, et_ aJ^, "Automobile In-Use Tire Inflation Survey", SAE Paper
780256, February 1978.
167
Thompson, "Fuel Economy Effects of Tires", Report SDSB 79-13, Standards
Development and Support Branch, ECTD, EPA, February 1978.
-------
180
-0.20 City and -0.19 Highway
-0.55% per psi.
168
this gives a 55/45 MPG sensitivity of
Using the surveys' tire pressure distributions and the above MPG sensi-
tivity value, the overall MPG effects are as shown in the table:
Tire Pressure Surveys and Estimated MPG Effects
Underinflated Tires:
Other Tires:
Source
Viergutz
Goodrich
Uniroyal
Number
Surveyed
8900
6100
VLOOO
MPG
Loss
3
3
3
.1%
.7%
.4%
Fraction of
All Tires
70.
56.
46.
7%
8%
4%
MPG
Gain
1.
1.
1.
9%
3%
8%
Fraction of
All Tires
29.
43.
53.
3%
2%
5%
Notes
Chicago ,
5 sites,
3 sites,
radials
sumr/wint
sumr only
Weighted Average
3.4%
63.9%
1.7%
36.1%
If the underinflated tires were merely brought up to recommended pressure,
their 3.4% shortfall would be eliminated, and fuel economy for the overall
fleet would improve by 2.3%. An alternate calculation using Viergutz'
conclusions for overall average underinflations, 1 to 2 psi (Summer) and
5 to 8 psi (Winter), and a 52%/48% split between summer and winter VMT
yields fleet shortfalls of 1.6% to 2.7%.
If underinflated tires were inflated to match the pressure distributions
of those tires which are at or above recommended levels, their fuel economy
would improve by 5.3%, and that of the overall fleet by 3.4%.
168
Thompson and Torres, "Variations in Tire Rolling Resistance", Report
LDTP 77-5, Standards Development and Support Branch, ECTD, EPA,
October 1977.
-------
181
d. Lubricants - Lubricants can be significant to fuel economy
slippage in two ways: (1) Their effects show up in dynamometer tests,
and (2) It is possible for lubricant-related MPG shortfalls to be induced
by owner/drivers who replace original-fill improved oils with less fuel
efficient oils or who drive in a way which uses the friction reduction
of improved oils for increased performance instead of improved fuel
economy. In addition, there are questions regarding availability of
improved oils, and the overall energy efficiency (including process
energy penalties) of their use.
In addition to our own review of data on the MPG effects of improved
lubricants under a variety of conditions, we will excerpt heavily from a
comprehensive report on the broader aspects of these new oils, issued
by the Coordinating Research Council under U.S. Army and DOT-NHTSA
sponsorship. This report resulted from in-depth surveys of thirty
companies and/or agencies actively involved in improved lubricants
research in mid-1978.
There are two functional classes of lubricant improvements: viscosity
reduction, whose benefits derive from lower viscous drag and pumping
losses; and friction modification, which reduces rubbing friction in the
oil film between engine surfaces. In product terms, synthetic oils may
be considered a third category, exhibiting functional reductions in
viscosity or friction or both.
Marshall, "Su.rvev of Lubricant Influence on Light-Duty Vehicle Fuel
Economy", Coordinating Research Council Report 502, December 1978.
-------
182
Low-viscosity oils are generally "lighter", more refined oils. The CRC report
indicates a fuel economy change on the order of 1% per viscosity grade change:
Oil Comparison MPG Improvement
10W-30 vs. 10W-40 1%
5W-30 vs. 10W-30 1%
5W-40 vs. 10W-40 2%
5W-20 vs. 10W-40 2.5%-3.0%
A GM report equates a 67% reduction in viscosity to a 30% reduction in overall
mechanical friction, which in turn corresponds to a 5% to 6% gain in steady-state
(30-55 mph) vehicle fuel economy. Amoco reports a 90% reduction in viscosity
to produce a 6% fuel economy gain.
Limitations to viscosity reduction include concerns for engine wear rate, engine
cleanliness, increased oil consumption, and refiners' capacity to produce enough
acceptable light stocks. For these reasons, many (including some auto manu-
facturers) prefer the friction modification route for improved oils.
Friction modifiers include colloidally-suspended solids, such as graphite or
molybdenum compounds, and oil-soluble additives, such as oleates, sperm oil,
tallow, etc. The CRC report arrives at average fuel economy improvements from 1%
to 3.5% for steady state and cyclic driving vehicle tests, and 4.9% for field
tests, of friction-improved oils. The field improvements are higher, according
to this report, because actual driving is more severe than the EPA and SAE test
procedures, and the friction-modified oils are more effective in severe-duty
situations. A taxicab fleet, for example, showed an 8.2% improvement due to FM
oils.
Coodwin and Haviland, "Fuel Economy Improvements in EPA and Road Tests with
Engine Oil and Rear Axle Lubricant Viscosity Reduction", SAE Paper 780596, June
1978.
Passut and Kollman, "Laboratory Techniques for Evaluation of Engine Oil
Effects on Fuel Economy", SAE Paper 780601, June 1978.
-------
183
Thus far, friction modifiers have been applied only to multigrade base
oils (although they should improve monogrades as well) for marketing
reasons: the industry makes monogrades for those users who seek cost
savings via the price of the oil, and believes that those consumers who
don't buy multigrades—for cost reasons—will similarly avoid costlier
friction-improved monogrades. In 1976 and 1977, 40% of automotive
•i i • j -i 172
engine oils were multigrade oils
Of course, viscosity improvement and friction reduction packages can be combined,
with nearly additive effects, since their functions are relatively independent.
The CRC report cites one source as measuring a 1.6% MPG improvement with vis-
cosity reduction, a 1.3% improvement with friction improvement, and a 2.5%
combined effect. Research on combined VI and FM packages is underway, but
proceeding cautiously.
Synthetic oil can be used as a total product, or as a blending additive. One
major all-synthetic 5W-20 oil is available, with a measured average 4% fuel
economy improvement, according to the CRC. The primary advantages of the syn-
thetics are good low-temperature performance and longer drain intervals; the
primary disadvantage is cost.
The following table summarizes the measured fuel economy benefits of improved
173-184 „, ,
oils, from a number of sources . The referei
comparisons is SAE 10W-40, unless noted otherwise.
oils, from a number of sources . The reference oil for all of these
172
Oil and Gas Journal, August 1978.
-------
184
Fuel Economy Effects of Improved Lubricants
(Percent Change from SAE 10W-40)
A. Engine Dynamometer Tests
173
Exxon .friction modifier, at 1250 RPM P.1% to 12.5%
174
Amoco .friction modifier, single cylinder zero to 2.2%
six cylinder -0.7% to +2.9%
eight cylinder 1.7% to 2.6%
synthetic 5W-20 , eight cylinder 2.3%
Atlantic Richfield .friction modifer,
steady speeds, 0.5% graphite 0.5%
1.0% graphite 3.0%
multimode, 0.5% graphite 2.4%
1.0% graphite 5.1%
vis. reduction, steady speeds 0.7% to 1.3%
CRC (Marshall, Op. Git.): average for all oils, all sources,
Engine Tests 2.7%
B. Vehicle Tests, Steady Speed
Amoco, chassis dyno, 50 mph, 15 cars, friction modifier -0.5% to +3.2%
Lubrizol, track tests, 1/2 @ 35 mph, 1/2 @ 55 mph, 8 cars, 8000 miles each
vis. reduction 1.4%
10% synthetic 1.8%
Full synthetic 5W-20 2.3%
[required 1500 miles to reach full effect]
e dyno cycle, 2 spe(
(SAE 5W-30 reference oil) -1.1% to +11.2%
General Motors ,14.5 minute dyno cycle, 2 speeds, friction modifier
C. Vehicle Tests, Cyclic Driving
Atlantic Richfield, modified 45-min. A11A cycle run 8 times (10-55 mph, 38 mph avg.)
1%-graphite friction modifier, dyno tests 1.9% to 7.2%
10-mile road route, cold start 6.0% to 7.1%
52-mile road route, hot start 3.5% to 4.2%
[required 600 miles to reach full effect, 900 miles
to lose effect after drain]
1 7 Q
General Motors , GM City-Surburban cycles, 3 cars
Hot Start, low-vis commercial lubes in engine,
transmission, rear axle 1%
Hot Start, low-vis experimental lubes in engine,
transmission, rear axle 3.8%
Cold Start, low-vis commercial lubes in engine
and rear axle 4.8% to 11.0%
-------
185
C. Cyclic Driving, cont'd.
179
Univ. of Michigan , SAE J1082 track tests, 7 cars
Best friction modifier, 3 cars, Urban Cycle 11%
Suburban Cycle 4%
Interstate Cycle 4%
CRC, average, all oils, all sources, SAE J1082 Tests 2-3%
Lubrizol, hot start EPA dyno tests, 3 cars
Friction modifier, EPA City -1.27, to +2.27,
EPA Highway 0.3% to 2.3%
Vis. reduction, EPA City -0.7% to +1.2%
EPA Highway 1.0% to 2.8%
Amoco, std. EPA dyno tests, 2 cars.
Friction modifiers, EPA 55/45 -0.5% to +2.2%
[required 1500-1800 miles to reach full effect]
Exxon, std. EPA dyno tests, 6 cars,
Friction modifier, EPA City 0.1% to 10.2%
EPA Highway 1.9% to 14.9%
EPA 55/45 (average) 5.5%
178
General Motors , improved lubes in engine and rear axle
Low-vis, lubes in engine and axle, EPA City, 0.5% to 2.9%, 1.7% avg.
(3 cars) EPA Highway, -0.3% to +2.0%, 0.7% avg.
EPA 55/45, 0.4% to 2.3%, 1.3% avg.
Synthetic in engine, low-vis, in axle, EPA City 0.4%
(1 car) EPA Highway 2.4%
EPA 55/45 1.1%
CRC, average, all oil-soluble FM's, EPA City 2.3%
EPA Highway 1-6%
CRC, average, 5W-20 synthetic, 20 domestic cars,
EPA City 3.2%
EPA Highway 1-6%
8 import cars, 5W-20 synthetic vs. SAE 40,
Hot EPA City 4.6%
Cold EPA City 6.8%
CRC, average, all oils, all sources, EPA 55/45. . , . . , 2.0%
-------
186
D. Field Tests
Exxon, friction modifier, 19 cars, 6000 miles
Range zero to 7.1%
Average 4.6%
[required 1500 miles to reach full effect]
1 01
Climax Molybdenum , molybdenum disulfide friction modifier
7 cars, 102,000 miles 3.4%
10 cars, 395,000 miles 3.7%
23 school buses, 214,000 miles 4.6%
Atlantic Richfield 182, graphite friction modifier 4.9%
CRC, average, all oils, all sources, Field Tests 3.4%
E. Rear Axle Lubricants Alone
183
Ford , lab tests and computer simulations, synthetic vs. SAE 90W,
EPA tests, 70°F
Short trip winter driving
184
Edwin Cooper , friction modified lube vs. SAE 80W-90,
EPA Highway test with 50% increase in road load HP 2.0 to 2.2%
Field tests 3.2%
General Motors 178, low-vis lube (SAE 75W) vs. SAE 90W, 2 cars,
30 mph , 0.3% to 0.6%
55 mph 0.6% to 0.8%
-------
187
17 3
Waddey, et. al. , "Improved Fuel Economy via Engine Oils", SAE Paper 780599,
June. 1978.
174
Passut and Kollman, Op. Cit.
Broman, et al, "Testing of Friction Modified Crankcase Oils for Improved Fuel
Economy", SAE Paper 780597, June 1978.
Riester and Chamberlin, "A Test Track Comparison of Fuel-Economy Engine Oils",
SAE Paper 790213, February 1979.
Caracciolo and McMillan, "Effect of Engine Oil Viscosity on Low-Temperature
Cranking, Starting, and Fuel Economy", SAE Paper 790728, June 1979.
178
Goodwin and Haviland, Op. Cit.
179
Bennington, et. al. , "Stable Colloid Additives for Engine Oils — Potential
Improvement in Fuel Economy", SAE Paper 750677, June 1975.
1 80
Davison and Haviland, "Lubricant Viscosity Effects on Passenger Car Fuel
Economy", SAE Paper 750675, June 1975.
•I Q I
Risdon and Gresty, "An Historical Review of Reductions in Fuel Consumption of
U.S. and European Engines with MoS2", SAE Paper 750674, June 1975.
182
DeJovine, et al, "Consumer Fleet Testing of Friction Modified Motor Oils for
Fuel Economy Benefits", NPRA paper, March 1978.
Willermet and Dixon, "Fuel Economy - Contribution of the Rear Axle Lubricant",
SAE Paper 770835, September 1977.
184Papay, "Fuel Saving Gear Oils", SAE Paper 790745, June 1979.
-------
188
Availability of the options for improving oils is varied. For viscosity reduc-
tion, 5W-7.5W multigrade oils must be blended from "100-SUS" (Saybolt Universal
Seconds) paraffinic base distillates. The CRC report quotes figures from Sun
Oil indicating that currently 26% of total refinery lube output goes to 100-SUS
base oil; accounting for non-automotive use and assuming the refined 100-SUS
base fraction can increase to 49% in 1980 and 61% in 1985, projected supply and
demand are as follows, in millions of barrels:
Maximum Supply Forecasted Demand
1980 19.1 22.9
1985 19.1 28.6
Avoidance of this shortfall would require such steps as limiting use of 5W-7.5W
series oils to automotive purposes, importing refined 100-SUS base distillates,
expanding domestic refining capacity and/or relying more on FM additives and/or
viscosity reduction through blending with synthetics.
No availability problems are foreseen for friction modifier additives, and
synthetics are described by the CRC report as reasonably available—at least for
blending—for the next decade. Of the two types of synthetic base stocks most
likely to see increased use, the polyalphaolefins are made almost exclusively in
the U.S., with current capacity double the current demand, and the organic
esters are readily available, some widely used in non-lubricant applications and
obtainable from natural sources (coconut and palm kernel oils).
Total Energy Balance is concluded by the CRC report to be clearly in favor of
the improved oils. Mineral oils, when highly refined, would appear to reduce
lubricant yields, but refiners have other product markets for the broader boiling
range cuts which are rejected. FM additives should be no problem because of the
small quantities involved. For synthetics, the energy balance has been estimated
as follows, for a 6000 mile drain interval and a 4% fuel economy improvement
from a 15 MPG base:
-------
189
Fuel saved with +4% AMPG, 1,785,000 BTU
(15 gallons)
Net process energy increase -107,000 BTU
Net energy saved 1,678,000 BTU
[94% of gross (vehicle) savings]
The detailed calculation is given in Appendix C.
Currently, EPA has not knowingly given approval for the use of specific "slippery
lubes" in emission certification or fuel economy data vehicles, since they are
not yet extensively available as replacement oils, are not uniformly classified
as to fuel economy potential, and are generally more expensive. However, manu-
facturers are free to choose a range of oils for use in EPA test vehicles, as
long as the oil chosen meets owner's manual warranty requirements; it is possible
that oils that are optimum, from the fuel efficiency standpoint, have been
identified within the broad range of warranty specifications, and may be used
in some test cars. Whether or not these special lubricants do enter the certi-
fication process, in significant quantity, a positive road MPG slip will still
occur if indeed average road MPG improvements exceed the improvements measured
in the EPA tests. For respective improvements of 2.0% engine/1.0% axle, EPA,
and 3.4% engine/1.5% axle, road, this positive slip will be 0.019 percent AMPG
per percent use in the EPA fleet; for example, if 50 percent of EPA cars were to
use advanced lubes, the improvements for the total fleet would be 1.48% in EPA
MPG and 2.44% in road MPG, an 0.95% positive road slip. Of those cars factory-
filled with improved lubes, 55% could switch to conventional engine oil at the
first oil change before the fleet road MPG improvement ceased to exceed the EPA-
measured improvement.
-------
190
e. Vehicle Weight Load - When consumers use vehicles for carrying
additional (non-passenger) weight loads, road shortfalls are created. For
example, the carrying of an average of 50 Ibs of tools, sporting equipment,
vocational paraphernalia, etc., causes an average MPG loss of about 0.5%;
if 50% of all vehicle miles are traveled with 50 extra pounds, the net fleet
MPG penalty is 0.3%.
Trailer towing is an obvious example of a significant increase in weight
load. Usage and size characte:
trailer types are shown below:
•1 Q C I Q £
load. Usage and size characteristics ' , of the two most predominant
Travel Trailers Camping Trailers
Weight >2600 Ibs. 1000-2600 Ibs.
Avg. Annual sales,
1970-1976 180,000 82,000
Sales as percentage
of new-car and 1.5% 0.7%
pickup truck sales
Estimated annual miles 3000-4000 1200-2000
The road fuel economy penalty for towing a 3000-lb. travel trailer has been
•I QT>
investigated by Claffey for a large pre-1970 car under steady cruise
conditions; the penalty was found to be strongly dependent on speed, as
shown below:
GPM increase
MPG decrease
20 mph
.001
1.4%
30
.008
11.3%
40
.024
30.0%
50
.039
38.7%
60 mph
.066
48.6%
185
Automotive News, 1977 Market Data Book Issue, April 1977.
Personal Communication, numerous recreational vehicle dealers in
Southeast Michigan, 1979.
1 07
Claffey, "Running Costs of Motor Vehicles as Affected by Road Design
and Traffic", NCHRP Report 111, 1971.
-------
191
The fuel economy penalty for trailer towing in cyclic driving is estimated
using average weights of 1800 Ib. for camping trailers, 3000 Ib. for travel
trailers, and 4000 Ib. for the towing vehicles, and EPA cycle MPG sensiti-
vities of -0.39 City and -0.49 Highway. The resulting 55/45 MPG penalty es-
timates are 19.^% for towing camping trailers and 32.5% for towing travel
trailers. Annual MPG penalties of 2.4% occur for vehicles which tow cam-
ping trailers 10% of their VMT, and 8.8% for those which tow travel trailers
for 20% of their VMT.
All of these weight effects are estimated to combine into an annual average
MPG shortfall of 0.4% for the overall fleet.
4. Simulation Variance
The preceding road slip discussions dealt with conditions which are—or
are capable of being—different from the standardized conditions of the
EPA tests, and indeed different from generally accepted "standard test
conditions" used in all systematic research. The influences to be
discussed next are related not to conditions which the tests knowingly
depart from, but to effects that the test does attempt to simulate.
a. Low-Speed Dynamometer loading - Dynamometer power absorbers
are calibrated at 50 mph to match either the road load measured at that
speed for the specific vehicle, or a representative average 50 mph road
load calculated from an equation, based on vehicle specifications. At
speeds other than 50 mph, the load curve applied by the dynamometer may
or may not exactly match the vehicle's road load for those speeds.
-I Q Q
A comprehensive comparison of dynamometer and actual road power
loading over the 0-60 mph speed range was conducted by EPA on a total of
65 cars: 61 1975 models and four 1973-1976 models. Ten of the cars,
15%, were found to be overloaded by the dynamometer at speeds below 50
mph, and 55 cars, 85%, were underloaded by the dynamometer at low speeds.
-I Q Q
Thompson and Torres, "Comparison of Dynamometer Power Absorption
Characteristics and Vehicle Road Load Measurements", Report LDTP 77-3,
Standards Development and Support Branch, ECTD, EPA, July 1977.
-------
192
Average dyno overload at 20 mph for the ten cars was 6.2%, and average
20 mph underload for the 55 cars was 24.8%. The figure is an example of
road and dyno loadings for a dyno-underloaded car.
FIGURE 57. Typical Low-Speed Dynamometer Underloading
80
60
40
20 -
Road Load
Dynamometer Load
10
20
30
40
Speed, MPH
SO
The road load sensitivity coefficients in Section IV.A.3 are not applicable
to this effect, since they were developed based on 50 mph road load,
which by definition is the match point for the data of concern here.
Our estimate of the shortfall/overage effect of this item is based on
189
the calculation that road load-related energy constitutes 48.8% of
total EPA City cycle energy, and an assumed sensitivity of City MPG to
total cycle energy of -0.6. It is also assumed that low-speed dyno
misleading does not affect Highway cycle MPG.
With these assumptions, we estimate a road MPG shortfall of 3.3% for
dyno-underloaded cars (85% of all cars) and a road MPG overage of 0.9%
for dyno-overloaded cars (15%). Total fleet shortfall would be 2.7%.
-------
193
b. Tire/Dynamometer Interaction - The two tires in contact with
the dynamometer rolls are the link through which the dynamometer absorbs
power from the vehicle; in a perfect simulation, the interaction between
these two tires and the dyno rolls would match the power requirement of
the vehicle operating with four tires on the road.
Two aspects of tire/dyno interaction have been found to differ slightly
from road experience: the relative dynamometer response of tires of
different ply constructions, and differences in rotational speed between
the tires and the two separate dyno rolls.
Tire Construction affects fuel economy differently on the dynamometer
i actual
.193,194
190 191 192
than in actual driving; this has been documented by Ford ' GM ,
and EPA
190
Crum investigated the rolling resistance of H78-15 bias-belted and HR78-15
radial tires on dynamometers and on the road, with results as follows, in terms
of tire power consumption:
Rolling Resistance HP @ 30 mph
Bias-belted: Radial:
psig 45 psig 25 psig 45 psig
Twin-roll dyno
Road
(2
(4
tires)
tires)
3.
3.
8
9 "'
3.1
3.3
4.6 3.9
>T
3.8 *" 2.2
i aq
Thacker and Smalley, Op. Cit. (152)
190
Crum, "Road and Dynamometer Tire Power Dissipation", SAE Paper 750955,
October 1975.
191
Unpublished data furnished to EPA by Ford Motor Co., 1975.
192
Sterapel and Martens, "Fuel Economy Trends and Catalytic Devices", SAE
Paper 740594, August 1974.
193
EPA Dyno/Track Test Project, Phase II (Unpublished).
194
Burgeson, "Clayton Dynamometer-to-Road Tire Rolling Resistance Relationship",
Report LDTP 78-09, Standards Development and Support Branch, ECTD, EPA,
April 1978.
-------
194
Ideally, two tires at 45 psi cold inflation pressure on a twin-roll dynamometer
(the EPA conditions) should have the same power consumption as four tires at
recommended pressure on the road. Crum's data show that radial tires are
loaded about the same on the dyno and the road (3.9 vs. 3.8 HP), while bias-
belted tires are significantly underloaded on the dyno (3.1 vs. 3.9 on the
road). Reference 194 found similar results when comparing tire types over the
transient conditions of the EPA City cycle, as shown below:
Dyno Tire Energy/Road Tire Energy
Radials 0.925
Bias-belted 0.695
Bias 0.745
195
Using rolling resistance sensitivities of -0.20 City and -0.19 Highway ,
these data would indicate road shortfalls of 1% for radials, 6% for bias-
196
belted, and 5% for bias ply tires. Using new-car tire sales fractions of
69% radial, 18% bias-belted, and 13% bias ply, these figures translate to a
fleet road MPG shortfall of 2.5% due to the tire type malsimulation, for pre-
1979 vehicles.
Direct vehicle measurements more or less corroborate this analysis. The following
table lists radial vs. non-radial MPG comparisons for dyno tests and road/track
tests: These data (using pre-1979 test procedures) show a radial tire 55/45
MPG superiority of 3.7% on the road, but a radial tire inferiority of 1.5% on
the dynamometer.
195
Thnnipson and Torres, Op. Cit. (168).
196 Modern Tire Dealer, January 1979.
-------
195
Effect on MPG, Radial Tires
vs. Bias or Bias-belted Tires
Road/Track Tests
,191
Ford
197
2 cars
28 cars
Ford
192
GM, 3 cars
199
Hurter , 4 cars
193
EPA
6 cars
197
Bezbatchenko
197
4 cars
1 car
Firestone
Weighted Average
Dyno Tests
EPA, 6 cars
„ , 200 -
Honda , 1 car
191
Ford (computer model)
GM192, 3 cars
Weighted Average
EPA City
+3.7%
+4.2%
+2.0%
+3.8%
+3.9%
198
EPA Hwy
+4.2%
+2.6%
+3.6%
+3.4%
-0.2%
-6.0%
-2.5%
-1.1%
-7.8%
-4 . 2%
-1.2%
+4.6%
+2.3%
+6.4%
+8.3%
+4.5%
-5.4%
+4.0%
+3.0%
+2.8%
+7.0%
+8.6%
+4.1%
-1.4%
-1.2%
-2.0%
-5.4%
-1.4%
197
198
Ford Motor Co., "Fuel Economy Improvement with Radial Ply Tires", May 1974.
Ford City/Suburban Cycle
199
Hurter et al, "A Study of Technological Improvements in Automobile Fuel
Consumption", Report DOT-TSC-OST-74-40, December 1974. (computer model).
200
Unpublished data.
-------
196
Again using the 1978 new-car sales fraction of 69% for radials, a fleetwide
road MPG shortfall of 1.6% is indicated for 1978 models. The road shortfalls
are slightly different for prior model years, due to different sales splits for
the various tire types for those years:
Road MPG Shortfall due to
Malsimulation of Tire Type
1975 2.0%
1976 1.8%
1977 1.7%
1978 1.6%
For 1979 and later models, this particular shortfall has been reduced; the tire
type malsimulation has been corrected by EPA by means of an adjustment factor
for tire type in the road load setpoint equation.
An additional road MPG shortfall can occur, and is believed to be occurring,
due to owners switching from original equipment radial tires to aftermarket
201
non-radials. It has been estimated that 15% of new-car radials are replaced
with non-radials. This gives an estimate of 0.5% for the consumer-induced
shortfall. This shortfall will continue if consumer tire-switching continues
at a 15% rate. For newer tires which have even larger road MPG advantage over
non-radials ("P-metrics" and other higher pressure radials), the shortfall due
to switching will be larger than these figures by a factor of about three.
Tire Slip and differences in rotational speed between the two dynamometer
202 203
rolls have only recently been investigated ' . The loading applied to a
201
Thompson, Op. Git. (167).
202
Yurko, "Computer Simulation of Tire Slip on a Clayton Twin Roll Dyna-
mometer", Report 79-10, Standards Development and Support Branch, ECTD, EPA,
February 1979.
203
Yurko, "A Track to Twin Roll Dynamometer Comparison of Several Different
Methods of Vehicle Velocity Simulation", Standards Development and Support
Branch, ECTD, EPA, June 1979.
-------
197
vehicle on the twin roll dynamometer is transmitted mainly through the front
roll, while speed measurement is based on the rotation of the rear roll. If
the tires in the roll cradle slip relative to the front roll, the vehicle is
allowed to "accumulate mileage" for a less than realistic expenditure of energy,
giving it an MPG benefit which would not occur on the road. The energy error—
which relates to gallons consumed—has been estimated to be in excess of 4%,
and the speed error—which relates to miles traveled—is an additional 1%. When
the rolls are mechanically coupled, these errors are eliminated, and the re-
sulting decreased fuel economy is a better simulation of road fuel economy.
The MPG difference between the standard dyno configuration and the coupled-roll
configuration gives an estimate of the magnitude of the road shortfall
attributable to this effect, and is shown below. The first and third
vehicles represent unpublished EPA test data, and the second vehicle's
data is from Reference 203.
MPG Effect of Coupling Dynamometer Rolls
EPA City Cycle EPA Highway Cycle
Subcompact Car -0.54% -3.3%
Intermediate Car — -4.0% (50 mph cruise)
Large Car -3.3% -5.3%
If the car size difference implied by the numbers is real, overall 55/45 MPG
road shortfalls due to dynamometer tire slip would be 2% for small cars and 4%
for large cars. The shortfall for the overall fleet is estimated at 3.4%.
c. Weight Class Distributions - Vehicle weight simulation is based on
"laden vehicle weights", determined by adding 300 pounds to specified vehicle
curb weights. Due to the nature of the dynamometer equipment, however, it is
not possible to set test weights at the exact value corresponding to each
vehicle's laden weight; instead, a finite number of discrete test weights are
employed, and each vehicle is assigned to the test weight class nearest its
laden weight. If the true vehicle laden weights within each weight class
were uniformly distributed around the class weight, those vehicles which were
heavier than the class weight, and were therefore tested slightly under-loaded,
would be balanced off by vehicles below the class weight which were tested
slightly over-loaded.
-------
198
We have found that these distributions are not exactly uniform, however. As one
example, the figure shows laden weight distributions within each inertia weight
class for the 1977 Certification cars. It is clear that in most classes there
are more cars above the nominal class weight than below. Considering the
relative fuel economy effects of the respective underloadings and overloadings,
and the sales distributions among the weight classes, these distributions
result in a fleet fuel economy loss of 1.01 on the road. For the five lower
weight classes—2000 through 3000 pounds, the loss is 1.7%; for the five higher
classes, it is 0.5%. Initial steps have been taken in model year 1980 to
reduce the potential for weight class maldistribution, via modifications to the
weight classification system.
FIGURE 58. Distribution of Lt. Duty Vehicle Laden Weights Within Inertia Weight Classes (1977)
40
30
20
10
ill
1,900 2,100 2,300 2,100 2.300 2,500
Laden Weight Laden Weight
2,300 2,500 2.700 2,500 2.700 2,900 3,100 2.700 2.900 3,100 3,300
Laden Weight Laden Weight Laden Weight
IW = 2,000
30i—1 1-
20
2,250
2,500
2.750
3,000
10
1 I T
A
10 -
2,900 3,300 3.900 3.500 3,900 4,300 3,800 4,300
Laden Weight Laden Weight Laden Weight
4,300 4,800 5,300 5,300 5,800
Laden Weight Laden Weight
IW = 3,500 4,000
# = Laden Weight Corresponding to Inertia Weight
4.500
5,000
5,500
-------
199
d. Manual Transmissions - As was noted in the design parameter sensi-
tivity study in Section IV.A.3, manual transmissions' EPA MPG advantage over
automatics appears to have increased during the 1975-78 time period. The next
table shows this trend, from the sensitivity study.
1975
1976
1977
1978
Manual
Auto
EPA
City
1.000
1.011
1.031
1.034
MPG
MPG '
EPA
Hwy
1.133
1.129
1.152
1.129
Matched Cars
EPA
55/45
1.045
1.052
1.073
1.067
Chg. from
1975
+0.7%
+2.7%
+2.1%
The next figure illustrates the ratio of manuals' 55/45 MPG to that of auto-
matics, for each EPA weight class; the trend noted above can be seen clearly in
the lighter weight classes.
FIGURE 59. Comparison of Manual Transmission MPG to Automatic MPG, at Constant Weight
1.30
1.20 -
0.90
2.000 2,500 3,000 3.500
Weight Class. Pounds
4.000
4,500
-------
200
Data furnished by DOE on EPA-to-road MPG relationships for the two transmission
types were used to determine road manual-to-automatic factors. At a given EPA
MPG level, there can be significant differences in average vehicle weight
between automatics and manuals; hence weight normalization is required in the
use of the DOE data. This was accomplished by using the parameter "ton-miles
per gallon". The ratio of manual TMPC (MTMPG) to automatic TMPC (ATMPG) is a
measure of relative weight-normal fuel efficiency between the transmission
types, and the grand ratio of Road MTMPG/ATMPG to EPA MTMPG/ATMPG is a measure
of how well the EPA transmission comparison is matched on the road. The next
figure shows this grand ratio as a function of EPA MPG level.
FIGURE
!
g
o
1
1
3
1
60. M
I.OS
'I 1.00
UJ
1
£ 0.95
|
£ 0.90
K
T5
£ 0.85
0.80
odel Year Trends in Road and EPA Transmission MPG Relationship
1 1 1 1 Fleet Avg.
1975 1.005
fc^^ ^^J* 1976 0.943
^^*~**x>^ ^^f^"^ 1977 0.890^
^***t>^^ 1 975 1 978 0.8«4
NX ,976 ,'
^e * *
..•*" 1977 "* ^g^-*J
1978 ~
1 1 I 1
1
-------
201
Manual MPG vs. Automatic MPG
EPA 55/45
Road
1975
1976
1977
1978
4.5% better
5.2% better
7.3% better
6.7% better
5.0% better
0. 8% worse
4.5% worse
7.8% worse
Total Road Discrepancy
for Manuals
0.5% better
5.7% worse
11.0% worse
13.6% worse
This increasing road slip is not amenable to straightforward explanation. The
apparent improvement in EPA MPG with time has been shown to be the result of
increased manufacturer usage, in the EPA tests, of manual transmission shift
schedules tailored for good fuel economy and low emissions, but unrepresenta-
r\ n I rt f\ f
tive of what could be expected in consumer driving . The tightening of
shift point specifications by EPA beginning with the 1979 model year has decreased
or eliminated further usage of unrepresentative shift schedules, and restored
the test procedure to 1975 comparability; this should correct that portion of
the shortfall. The loss in manuals' advantage on the road could be due to
differential usage of manual vs. automatic vehicles, and/or increasingly fuel-
inefficient shifting by consumers. If this is the case, it is not a clear
issue of inexact simulation, but rather a question of whether vehicles driven
differently should be tested differently.
At any rate, the overall fleet shortfalls corresponding to the values above can
be determined by introducing the relative manual and automatic sales fractions.
The next figure gives the fleet average values for 1975-78, and also illustrates
manual transmissions' sales penetration as a function of vehicle test weight.
Below about 3000 pounds, more than half of passenger cars use manual trans-
missions.
f) f\ I
Hutchins, "The Effects of Manual Transmission Shift Points on Emissions and
Fuel Economy of a 1977 Chevrolet Chevette when Tested by the Hot LA-4 Proce-
dure", Report 77-15, Technology Assessment and Evaluation Branch, ECTD, EPA,
December 1977.
205
Hirabayashi, "Manual Transmission Shift Point Study", GM Environmental
Activities Publication A-3646, August 1978.
206
Rykowski, "Shift Schedules for Emissions and Fuel Economy Testing", Report
LDTP 77-6, Standards Development and Support Branch, ECTD, EPA, November 1977.
-------
202
FIGURE 61. Transmission Type Sales Fractions, by Vehicle Weight
100
U 75 -
a
Fleet Avg.
1975 19.6%
1976 17.1%
1977 16.8%
1978 174%
2,000
2,500
3.000 3.500
Weight Class (Pounds)
4.000
4.500
Using the fleet sales fractions and the ''Total Road Discrepancy" figures from
the preceding table, the road shortfalls for the overall fleet are estimated to
be 1.0% in 1976, 1.9% in 1977, and 2.5% in 1978.
e. Power Accessories - All accessories cause fuel economy losses due to
their weight, and power-consuming accessories cause additional fuel economy
losses due to hydraulic or electrical power consumption. The use of power
accessories in the U.S. car market is significant and growing with time ' ,
as in the next table:
207
Automotive News, 1976 and 1977 Market Data Book Issues.
208Forrest, et^ al, "Passenger Car Weight Trend Analysis (2 Vols.)", Report EPA-
460/3-73-006a, January 1974.
-------
203
Power Accessory Usage
Power Seats
Power Windows
Backlight Defogger
Air Conditioning
Power Brakes
Power Steering
1972
Domestic
—
—
—
70%
69%
86%
1976
Domestic
16.7%
22.7%
28.9%
74.9%
81.8%
90.8%
Import (Est.)
<5%
<5%
vL5%
^25%
vLO%
^30%
1977
Domestic
21.2%
27.2%
35.1%
81.9%
91.4%
95.8%
Import (Est
<5%
<5%
vL5%
^30%
VL5%
V35%
o.
The MPG penalty due to weight alone for air conditioning, power steering, power
209
brakes and power windows combined has been estimated at 4% for urban driving
and 3% at 70 mph.
The weight-related MPG penalty does not figure in EPA-to-road MPG shortfalls,
since accessory weights are accounted for in the EPA test vehicles. However,
those car buyers who opt for these items still absorb the added fuel consumption
due to their weight, whether or not these accessories are operated.
The fuel consumption effects of operation of power seats and the like are not
simulated in the EPA tests, but when averaged over a vehicle's life, this effect
should be very, very small. Power brakes are fully exercised in the stop-and-go
portions of the EPA tests, while the MPG effect of power steering is simulated
only to the extent of hydraulic pump losses exclusive of turning. We do not
have power steering system operational data with which to evaluate whether a
measurable shortfall exists in this area.
The effects of air conditioner operation are simulated in the EPA tests by means
of a 10% increase in 50 mph dynamometer road load power; vehicle air condi-
tioners are not operated during the tests. This simulation was originally
developed to represent the year-round average power increase for air condi-
tioning, not the MPG effect for that fraction of the time an air conditioner is
actually being used. With MPG sensitivities of -0.16 City and -0.33 Highway,
this 10% increase in 50 mph road load imposes a 2.2% penalty in 55/45 MPG on
209
Huebner and Gasser, "General Factors Affecting Vehicle Fuel Consumption",
SAE paper 730518, May 1973.
-------
204
air-conditioned models. With 70% of the total domestic and import fleet air
conditioned, the fleetwide penalty corresponding to this simulation is 1.6%.
The table below gives actual power steering and air conditioner fuel consumption
209—220
penalties from vehicle tests . Some power consumption characteristics of
these two accessories are also shown.
Fuel Economy Loss due to Power Accessory Operation
A. Steady Cruise
20mph J3£ 40 5_0 60 ]^ SOmph
Power Steering
Huebner & Gasser — — — — — 2.4%
Cornell 21° -- 2.4% 2.1% 2.0% 2.0% 2.0%
RLHP increase 211 9.5% 6.8% 5.3% 4.4% 3.8%
Air Conditioning
Cornell 8.1-10.4% 7.8-8.9% 6.6-8.1%
Huebner & Gasser — — —
212
Donoho
213
Coon, et al — — —
214
EPA Dyno/Track 14.4% 10.7% 7.5%
RLHP increase 63% 40% 29%
215
AC horsepower, Marks ("Climate control" system)
40°F 1.4 1.7 2.0
70°F 2.6 2.9 3.4
100°F 4.4 4.7 5.1
B. Cyclic Driving
01 £
Huebner & Gasser (Chrysler Urban)
5.5-7.2%
9.4%
11-15%
5.5%
23%
2.5
4.1
5.7
4.3-6.7% 3.3-6.1% 3.3-5.3%
6.0%
10-11%
5.3% 5.3% 4.5%
19%
3.3 4.6
4.9 5.9
6.6 7.6
AC + PS + Generator, 6 37
PS, 0.9%
AC. 13.0%
Air Conditioning: EPA City EPA Hwy EPA 55/45
,2]
al
,218
Eccleston, et al217, 15 1969-75 cars, 110°F 10.4%
EPA Dyno/Track 5 1976 cars 57°
71Q
Baker 30 1976-78 Calif, cars
770
Spindt & Hutchins 7 1972-79
-71°F
, 80°F
cars, 80°F ....
90°F ....
110°F ....
, . . . . 4.8%
. . . . 9.9%
. . . . 6.5%
. . . . 8.9%
. . . . 13.9%
8.0%
4.1%*
10 2%
17.6%
9 2%
ft n?
S A 7
Q La/
1 S 17
-------
205
The average in-use MPG penalty for air conditioner operation can be estimated in
two ways: (1) for a 70% vehicle installation factor and a fixed EPA 55/45
penalty of 6% for AC operation, the fleet MPG effect would be 0.4%, 0.9%, and
1.3% respectively for AC duty cycles of 10%, 20%, and 30%; (2) Applying temper-
ature adjustments to the AC penalty rather than using a fixed penalty, and
considering the distribution of VMT among various ambient temperatures (Section
IV.C.I.a.), the annual AC loss would be 4.0% for climate control systems, and
2.1% for standard systems. Again using a 70% vehicle AC installation factor and
estimating that 1/3 of those installations are of the climate control type, the
year-round penalty for the overall fleet would be 2.0%.
Thus, the EPA simulation's 1.6% MPG penalty compares well with even the worst
estimate (2%) of the real-world effect that it attempts to simulate: the
fleetwide, year round penalty of air conditioner usage.
210
Cornell, "Passenger Car Fuel Economy Characteristics on Modern Superhigh-
ways", SAE Paper 650862, Novermber 1965.
211
EPA-calculated composite from numerous sources, unpublished.
01 o
Donoho, "EPA MPG—How Realistic?", SAE Paper 780866, December 1977.
213
Coon, et al, "Technological Improvements to Antomobile Fuel Consumption",
Report DOT-TSC-OST-74-39, December 1974.
0 1 /
EPA Dyno/Track project, Phase II (unpublished).
215
Marks, "Which Way to Achieve Better Fuel Economy?", Seminar at California
Institute of Technology, December 1973.
91 fi
Scheffler and Niepoth, "Customer Fuel Economy Estimated from Engineering
Tests", SAE Paper 650861, November 1965.
217
Eccleston, et^ al_, "Ambient Temperature and Vehicle Emissions", EPA Report
460/3-74-028, October 1974.
218
Bernard, et al, "Automobile Exhaust Emission Surveillance Analysis of the
FY73 Program", EPA Report 460/3-75-007, July 1975.
219
Baker, unpublished data submitted to SAE Passenger Car and Light Truck Fuel
Economy Measurement Committee, September 1979.
220
Spindt and Hutchins, "The Effect of Ambient Temperature Variation on Emis-
sions and Fuel Economy - An Interim Report", SAE Paper 790228, February 1979.
-------
206
One more factor related to air conditioning deserves mention. An isolated
221
instance has been reported wherein a specific vehicle—at high speed—with
the air conditioner operating was supposedly found to be more fuel-efficient
than turning off the air conditioner and opening windows. The aftermarket air
222
conditioner industry wasted no time in taking out a large ad emphasizing the
AC-favorable aspects of that isolated observation, while omitting most of the
caveats. Two test projects that we know of ' included measurements of this
effect. The figure below shows the results. At steady cruise speeds up to 80
mph, the "windows-open" penalty remains significantly smaller than the air
conditioner penalty. The curves may indeed cross at 100 mph or so for the car
types tested, or at lower speeds for certain specific cars, but we must refute
any assertion that, as a rule, opening vehicle windows is more detrimental to
fuel economy than air conditioner operation, for any reasonable operating
condition. Also, as pointed out earlier, air conditioning systems impose a
full-time MPG penalty due to their weight, whether operated or not.
FIGURE 62. Fuel Economy Effects of Air Conditioning and of Open Windows
K2.S
-2.5
-5
-7.5
-10
-12.5 -
-15
20
A/C On,
Windows Closed
•. A = EPA Dyne/Track2" -
• = Donoho212
I
40 60
Vehicle Speed, MPH
80
100
221
222
The Washington Star, September 9, 1979, at F-10.
Automotive News, October 22, 1979.
-------
207
f. Vehicle Cooling - in the EPA tests, vehicle cooling is accomplished by
a fixed speed fan positioned in front of the vehicle grille. Tire cooling and
223
speed-dependent air flow, as seen on the road, are not simulated. A study
sponsored by EPA investigated the relative cooling effects of a number of fixed-
speed fan arrangements as compared with road operation, for two cars with large
V-8 engines (>400 CID): a 1975 Chrysler and a 1976 Pontiac.
The author concludes that one vehicle was consistently overcooled in the dyna-
mometer tests, while the other was consistently undercooled. Our own evaluation
of the data on road and dynamometer temperatures of various vehicle components
shows that, for both vehicles, road temperatures were usually lower than dyna-
mometer tests which used the standard EPA fan arrangement. If higher dyno test
temperatures are equated with higher fuel economy, this suggests the potential
for a simulation-related MPG shortfall for very large engines. However, the
small sample size and the absence of road fuel economy measurements in this
project prevent the drawing of specific conclusions on the magnitude or even the
direction of any such disparity for all vehicles. EPA is continuing to study
the question of vehicle cooling.
g. Metric Slip - Metric slip is the name given to any fuel economy
deviation caused by differences in the way fuel economy is measured. In contrast
to some of the other slip factors which are more amenable to direct estimates of
the average effect, metric slip is somewhat more amenable to a discussion of the
dispersions in the different approaches.
The two general methods are, of course, the method used by EPA to determine the
EPA fuel economy numbers and the method (or methods) used to determine In-Use or
Road Fuel Economy.
(1) The EPA Carbon Balance Method - The method used by EPA to calculate
fuel economy (the carbon balance method) is based on conservation of the mass of
carbon contained in the fuel, consumed by the vehicle, and ejected in the exhaust.
Knowing the amount of carbon in the exhaust and the amount of miles traveled on
the tests the fuel economy can be calculated.
223
Sharman, Temperature Comparison: On-Road versus Dynamometer Cooling",
Final Report, EPA Contract 68-03-2412, Task 2, June 1977.
-------
208
Generally the equation for fuel economy is of the form:
MPG =
D
where N, the numerator, is a function of the density (grams/gallon) of
the fuel and the carbon fraction (grams carbon/gram fuel) of the fuel;
D, the denominator, is a function of the emissions (grams per mile) of
the carbon-containing exhaust constituents and the carbon fraction
(grams carbon/grams exhaust constituent) or:
MPG =
(grams fuel \ / grams carbon \
gallon fuel ) X \ grams fuel )
^» / grams , \ / grams carbon \
( ' .,— exhaust constituent. 1 x [ r r rrr r— I
^ \ mile ^ f \ grams exhaust constituent. /
There are some underlying assumptions involved with the equation, and
some simplifications in its actual use, that can affect the fuel economy
calculated.
One assumption is that all of the fuel put into the gas tank is consumed
by the engine. This assumption is not totally correct. When liquid
fuel is pumped into the tank, gaseous vapors are displaced, and these
displaced vapors are not available to be burned by the engine. In addition,
vehicles lose gasoline vapors (evaporative emissions) due to diurnal
breathing and hot soak losses. These losses are not included in the
224
fuel economy calculations . When vehicles are in operation, running
losses may also be experienced, which are also not measured.
Another assumption is that all of the carbon in the fuel that is consumed
by the engine comes out of the exhaust tailpipe. This assumption neglects
exhaust system leaks.
These two assumptions are such that the fuel economy calculated by the
EPA carbon balance method tends to overestimate vehicle fuel economy.
?24
Other fuel measurement methods, such as volumetric and gravimetric
techniques, also virtually always overlook these fuel losses.
-------
209
Fuel economy dispersion can be influenced by the MPG equation also. For example,
at one time the "miles" value used to determine the gram/ mile parameter was
defined equal to 7.5 miles for the urban cycle (its nominal length); however,
considering test tolerances, not all tests are exactly 7.5 miles long. Be-
ginning with Model Year 1978, the miles traveled are calculated based on a count
of the revolutions of the dyno rolls on each test; the measurement has thereby
been improved. In another equation-related concern, the equation used currently
assumes that HC, CO, and CO- are the only carbon-containing exhaust constituents.
For vehicles powered by gasoline-fueled engines, this assumption means that
other carbon-containing materials which are not picked up by the Flame loniza-
tion Detector (FID) are not counted. This tends toward an overestimate of fuel
economy, since some materials such as aldehydes are not detected. For vehicles
powered by Diesel engines, there is carbon in particulate emissions, which is
also undetected.
Some values in the equation depend on certain properties of the exhaust con-
stituents, such as the carbon fraction. For CO and C0~, these fractions are
known precisely, but for HC the equation's carbon fraction is not necessarily
accurate for every exhaust emission control system and vehicle. Exact cal-
culations of MPG via vehicle-specific HC carbon fraction corrections are not
done at this time due to cost-effectiveness considerations.
The final concern has to do with fuel properties. The current approach is to
use a constant value for the product of density and carbon fraction. In the fuel
that EPA uses, there is some variability in both of these properties from batch
to batch; use of a constant value for this product can contribute to dispersions
in the MPG measurements.
The above fuel specification discussion only applies to discrepancies between
the carbon balance equation and the fuel used by EPA in its testing. This EPA
test fuel can differ in properties from the fuel actually used in the field.
This difference can also lead to a difference in calculated fuel economy.
Consideration could also be given to fuel density-ambient temperature relation-
ships.
-------
210
Summary - EPA's Carbon Balance Method - it appears that the carbon balance
method as currently used tends to overpredict fuel economy relative to direct
laboratory fuel measurements.
Calculating one single value for the MPG offset due to the way EPA uses the
carbon balance calculation is difficult for two reasons. First, in-use gasoline
varies substantially in its properties, and there is no one set of values to
compare to. Secondly, the amount of carbon that is missed by the current
procedure is not known precisely.
With the above caveats, some estimates of the offset due to the carbon balance
225
method can be made. Using survey data , one can compute that the in-use value
for g carbon/g fuel values ranges from 0.862 to 0.877 (EPA uses 0.866). The
value for the product of g carbon/g fuel and g gasoline/gallon in use ranges
from 2353 to 2424 (EPA uses 2421). bsing the average values of the in-use
fuels* carbon parameters, one calculates an offset of about 1%, with the EPA.
carbon balance calculations being high. If, however, one uses the values of the
parameters that maximize the offset and adds in an estimate of 1.0 g/mile for
the HC equivalent of all the carbon not accounted for, the offset could be as
high as 4%, again with the EPA method being high.
(2) In-Use Fuel Economy Determinations - Two subjects are relevant in this
area: the in-use fuel economy calculations performed to arrive at fleet car MPG
values, and in-use fuel economy calculations performed by consumers. Further,
there are some issues common to both of these in-use MPG determinations.
Fleet Car MPG - Fleet car MPG calculations are important because much of
the in-use data from which overall shortfall estimates are derived comes from
fleet cars.
There may be as many MPG calculation methods as there are fleets. It
can happen that data on miles driven and data on gallons consumed may not be
taken simultaneously on each vehicle in the fleet. Data on miles driven may be
taken from periodic odometer surveys and gallons consumed may be taken from
o o tr
U.S. Department of Energy, "Composition and Octane Number of U.S. Motor
Gasolines Sampled in the DuPont 1978-79 Winter Road Octane Survey", Report
BETC-0012-1, September 1979.
-------
211
credit card receipts or from a fleet's controlled fuel supply records. This can
lead to some inconsistencies in the MPG values for individual vehicles or the
fleet as a whole. It undoubtedly increases the dispersion in fleet MPG values
for nominally identical vehicles. Variance in either the figures for miles
traveled or gallons consumed could be partially responsible for the wide disper-
sion seen in the road MPG values given in Section III.C.2 for 123 fleet vehicles
with a 29 MPG EPA rating (high = _31 MPG, low = 3^ MPG).
Consider this example: a vehicle is driven for six months, delivering con-
sistent fuel economy near 15 MPG. If the vehicle is in fleet service, and
odometer readings (from the maintenance garage) do not reach the accounting
computer at the same time as fuel vouchers (say, from credit card purchases), an
end-of-month MPG calculation could match one month's miles with another month's
gallons, giving surprising results.
Effects of Led/Lagged Mileage and Fuel Use
Figures on MPG Calculations
Miles
Month Driven
1 1000
2 2000
3 1500
4 600
5 2100
6 1800
Gallons
Used
69.
135.
100.
39.
138.
118.
4
0
1
7
4
1
Calculated MPG:
Actual MPG: (Fuel lag 1 mo.) (Miles
Month Cumulative Mo. Cumulative Mo.
14.
14.
15.
15.
15.
15.
4
8
0
1
2
2
14.4
14.7
14.8
14.8
14.9
15.0
—
28.
11.
6.
52.
13.
8
1
0
9
0
_.
28
17
13
18
16
.8
.1
.5
.0
.6
—
7.4
20.0
37.8
4.3
17.8
lag 1 mo . )
Cumulative
—
7.4
12.8
16.4
12.3
13.6
Rather than report that "the car gets 4 to 53 MPG", an analyst armed only with
the data on the right side of the chart could use cumulative figures (13.6 to
16.6 MPG), average all of the monthly data (19.9 MPG arithmetic, 11.3 MPG
harmonic), or discard outliers below 8 MPG or above 22 MPG (60% of the data) and
average the remaining data (15.5 MPG arithmetic, 14.6 harmonic).
We have no data with which to evaluate the possible nationwide effects of such
metric anomalies; the examples above do illustrate the potential for significant
metric slips, possibly more likely to occur with fleet data than consumer data.
-------
212
Consumer Car MFC - Consumer-derived MPG values can also have some metric-
related dispersions. One possible error source is nonuniform sampling. We
surmise that most consumers do not keep detailed records of all their mileage
driven and fuel purchased, but rather that most consumers calculate MPG values
only when convenient, such as on a vacation trip where one or more tanks of fuel
are consumed in a relatively short time period. Such nonuniform sampling could
lead to MPG results that may not be completely representative of the overall MPG
performance of the vehicle.
Another consideration is that consumers might only compute tank-at-a-time MPG
values and not compute cumulative MPG over several tankfulls. Using tank-at-a-
time values leads to dispersion, and arithmetic averaging of these tank-at-a-
time MPG values can lead to overestimates of fuel economy compared to the
cumulative approach.
Variances Common to Both In-Use MPG Determinations - Both in-use sources of
MPG data have possible variances associated with them in the determination of
both miles driven and gallons consumed or purchased.
Variances in the miles-traveled parameter can come from a variety of causes, as
listed below :
Factors Affecting Odometer Mileage Accuracy
Take-off Pinion Design Limits
Tire Make, Tread, and Construction
Tire Inflation Pressure
Tire Wear
Tire Growth
Tire Size
Centrifugal Effects
Rear Axle Load
? 26
Society of Automotive Engineers, "Factors Affecting Odometer-Speedometer
Accuracy", SAE Information Report J862b, April 1969.
-------
213
Not all of the factors affect the odometer reading the same way, so an overall
average offset in mileage measurement cannot easily be determined. SAE Recom-
mended Practice J678 suggests that the odometer gearing ratio be such that
odometer accuracy will be within the limits of -1% to +3.75% at 45 mph. From this
factor alone, the allowable error favors fuel economy overestimation. In addition
to the odometer effects listed previously, miles indicated can differ from actual
miles driven due to wheel spin; this could be significant for vehicle operation in
slippery or icy conditions, and would also result in overestimation of miles
traveled and of MPG.
Determination of the "gallons" figure is subject to some variance also. The non-
repeatability of fill-up level, for example, will affect tank-at-a-time calcula-
tions, as discussed earlier. In addition, gallons consumed may not always equal
gallons purchased, due to fuel vapor displacement and gasoline spills during
refueling. When considering the accuracy of in-use MPG determinations, the
accuracy of gasoline dispensing pumps is also a factor.
(3) Summary-Metric Slip - Many slip factors in this report have been
assigned numerical values which typically are the difference between the averages,
or central tendencies, of two statistical distributions, one being "EPA" and the
other being "In-Use". For metric slip, assigning a specific value (or even a
direction) for in-use MPG slip is not feasible at this time. While certain
features of the EPA measurement procedure suggest that it overestimates fuel
economy, there are features of in-use measurement of miles traveled—and of in-use
record keeping—which make it likely that in-use MPG can be overestimated by a
margin at least as great as that of the EPA method.
-------
214
5. Summary Findings: Road Slip
Fuel economy influences which are felt mainly in operation of vehicles
on the road, or whose effects do not appear in dynamometer testing,
are listed in the table below:
Effects of Road Slip Influences
on Fuel Economy
Fleet MFC
Re^Lative MPG Shortfall
Travel Environment 0.868 -13.2%
Travel Characteristics 0.975 -2.5%
Vehicle Condition 0.954 -4.6%
Simulation Variance 0.898 -10.2%
The "relative MPG" figures in this tabulation were calculated by multiplying the
individual Road Slip influences' respective slip factors. Referring to Section
IV.A, the three-year average Road Slip shortfall from raw DOE data was approxi-
mately 10%. The average Road Slip inferred by the results of our studies exceeds
the DOE average by a factor of three.
This observation of course prompted a thorough re-examination of our analyses
with suspicions that we may have been unduly pessimistic in our assumptions.
Following that re-examination, we stand by the analyses for the individual MPG
influences. But the assumption of mutual independence among the many effects, and
the multiplication of their respective slip factors, "stacks them together" in a.
way which may not reflect their real-world interaction. This realization led to
the next section of the report, which does indeed confirm that the whole is not
necessarily as great as the sum of its parts, when it comes to multiple fuel
economy effects.
The Travel Environment category includes powerful, highly variable, and generally
unpredictable MPG influences; this is obvious enough not to warrant further
discussion. The relatively small overall shortfall attributed to the Travel
Characteristics class, however, conceals the fact that powerful MPG influences are
to be found here, too—particularly in the areas of vehicle speed and accelera-
-------
215
tion. These factors are no less variable, and no more predictable, than is the
weather, but they are subject to driver control.
As regards model year differences, only the Simulation Variance category lent
itself to year-by-year analysis. Two of the included factors and their shortfall
trends are listed below:
Model Year Trends, Simulation Variance Items
(Average Percent Deviation from EPA MPG)
Model Year:
1975 1976 1977 1978
Tire Type Malsimulation -2.0% -1.8% -1.7% -1.6%
Manual Transmissions +0.1% -1.0% -1.9% -2.5%
The tire type malsimulation item refers only to differences in the dynamometer
response of radial and non-radial tires, and does not relate to other dynamometer
loading factors.
While the time trends in these items' shortfalls are not of national fuel con-
sumption significance, it is important to note that these test shortcomings were
recognized and corrected in the EPA test procedures for 1979 and later model
years, driving their shortfall-producing potential toward zero. Another Simula-
tion Variance factor which we believe has contributed to MPG shortfalls on the
road is the system of assigning vehicle test weights to discrete classes. Be-
ginning with model year 1980, we think there will be a reduction in test weight-
related MPG discrepancies, due to the use of smaller weight increments between test
weight classes. The remaining Simulation Variance factors are receiving, and will
continue to receive, further study.
Finally, the Road Slip studies did reveal a number of ways in which the higher-MPu
cars can suffer worse MPG shortfalls. We find Travel Characteristics influences
such as trip length, average speed, cold-start fraction, and urban vs. rural
location for smaller cars to be typically in directions detrimental to fuel
economy. MPG sensitivities to low temperatures and wind effects'also appear to be
more detrimental for the smaller cars.
-------
216
D. Fuel
For most of this report, fuel economy effects of various influences have been
expressed as "slips": ratios of the MPG under the condition of interest to the
MPG under the standard conditions of the EPA tests. A simple way to estimate
the net effect of several influences acting in combination would be to multiply
the individual slip factors. However, as discussed below, other approaches are
preferable to this simplistic method.
1 . Mathematical Implications
Fuel economy, in MPG, can be expressed as a function of power (hp) , brake
specific fuel consumption (bsfc, in Ib/bhp-hr) , speed (mph) , and fuel density
(df, in Ib/gallon):
mph x d,,
MPG = *
hp x bsfc
At fixed speed and fuel density, the MPG slip for a change in hp is:
MPG. hp x bsfc
^ ^o J o
MPG hp . x bsfc.
a c^ J ^
MPG. hp x osfc
^ a o
or: =
+ A/zp J x
If the fuel economy slips for N different hp changes are measured separately,
and the combined effect of these changes estimated by taking the product of the
individual slips, we have:
MPG..
MPG **\ MPG .
O J=l\ 01
hp x bsfc hpo x bsfc hp x bsfa
0 x
(hr> + A/ZD )x bsfc-, , (hp + &hpn)x bsfc, n (hp + hhp )x bsfc,
'"•fs, c2 J npl o ^ J np^ o r n npn
o
-------
217
MPG
or:
N=n
MPG
= (hp x bsfc )n x fr (— , J, L \
O O 11 ^ ^p ^ Afcp Jx is/C, ./
But if all the Ahp's are applied at once, the MPG slip effect is;
MPG
N=m
hp x bsfc
*o J o
MPG
n
fe=2
x
'hpl
It is clear that the two methods are not at all mathematically equivalent, and
would produce equal results only by coincidence, or due to some special inter-
relationship of the variables which is not apparent.
2. Engine Map Considerations
It is well known that automotive engines generally become more efficient (i.e.,
bsfc decreases) as loads are increased. The figure, a portion of an engine
fuel consumption map, illustrates.
FIGURE 63. Portion of a Typical Engine Map (Early 1970's Domestic V-8)
I
30
20
10
I
1,200
.1,300
1.400
1.500
1.600
RPM
-------
213
Because of this engine map
portionately larger fuel economy
the next figure.
characteristic, small chanpes in load have a pro-
effect than large load changes, as shown in
FIGURE 64. Sensitivity of Fuel Economy to Change in Power Load (From Previous Engine Map)
-0.8
U
I -0.6
-0.2
(I400RPM, !25HPBase)
I
20
40 60
Percent Change in HP
80
100
Thus, simply multiplying the MPG slips for a number of relatively small load
changes — each measured at high MPG sensitivity — overestimates the effect of
simultaneous application of all those load changes. As a simple example, four
load increases of 5% each, at a sensitivity of -0.75% AMPG per %Ahp, would each
result in an MPG loss of 3.75%, or a slip factor of 0.9625. Using the product
of the slip factors, one would estimate the total MPG loss to be 14.2%, from:
MPG,
MPG
= (0.9625) = 0.858
A total load increase of 20%, however, with a sensitivity of -0.45% AMPG per
%Ahp, gives a loss of only 9.0% (slip factor = 0.910).
Analysis based on this engine map yields the following relationship for "derat-
ing" the calculated MPG slip of N separate load changes:
-------
219
MPG
N
MPG,.
Actual —-—•— = I Calculated
mG0 \ MPG
where cj> = 1-0.119 [ln(N) ] and the calculated slip is the product of the
individual slips. The equation is plotted in the next figure,
FIGURE 65. Relation for Derating Calculated MPG Effect of Multiple Influences
0.4
06 0.8 1.0 1.2
Calculated (Product) MPG Slip
1.6
General use of this equation should be made with caution, since:
this adjustment is derived from a simple point-to-point type of
analysis of one region of one specific engine map, and may not
accurately reflect MPG behavior for other engines, or the complex
transient conditions of stop-and-go driving;
the adjustment is based on engine load effects, and does not neces-
sarily apply to other MPG influences, such as ambient temperature.
-------
220
3. Actual Examples
In a study of engine modifications for improved fuel economy, vanBasshuysen
227
et al report the following changes in fuel consumption for six separate
factors, and their effect when combined. All changes are based on hot start
EPA City cycle tests.
Change in
Fuel Consumption
Change in Fuel Economy
Percent
AMPH
-1.8%
-1.5%
-2.2%
-2.8%
-3.2%
-3.2%
-12.7%
-14.7%
-13.8%
+1.8%
+1.5%
+2.2%
+2.9%
+3.3%
+3.3%
+14.5%
+14.9%
+16.0%
(25 MFC ba?
+0.5
+0.4
+0.6
+0.7
+0.8
+0.8
+3.6
+3.8
+4.0
Lightweight pistons
Softer valve springs
Cylinder head cooling
Hi-temp thermostat
Lighter engine oil
Decel. fuel cutoff
All, combined (actual)
(calculated, sum)
(calculated, product)
The combined slip is seen to be less than that calculated from the separate
slips, if the calculation is done multiplicatively. If the fuel economy
changes are added, either as absolute values or percentages, a better
estimate of the combined effect is obtained.
Applying the power equation developed in the preceding section, we find
that = 1-0.049 ln(N) produces the same results as the combined-effects
measurement, i.e.
MPG
MPG
6 = (1.160)*''"*" "'""' = 2.145, A = +14.5%
}1-.049 In(6)
from the fuel economy slip product;
MPG
l
MPG
e- = (0.862)1-049 ln(6) = 0.873, A- -12.7%
from the fuel consumption slip product.
977
vanBasshuysen et al, "Fuel Ecomomy Improvements by Reduction of Friction
Losses and Other Measures", Audi NSU Auto Union AG, November 1979.
-------
221
228
Data presented by Porter on nine vehicle improvements shows the following
fuel economy effects for the EPA 55/45 test:
Change in Fuel Economy
6 ' Change in
AMPG Percent Fuel Consumption
(Base=19.4 :MPG)
Reduced frontal area +0.2 +1.0% -1.0%
Reduced weight +3.4 +17.5% -14.9%
Modified cooling fan +0.3 +1.5% -1.5%
Wide ratio transmission +0.3 +1.5% -1.5%
Optimized torque converter +0.1 +0.5% -0.5%
Var. disp. transm. pump +0.8 +4.1% -4.0%
Reduced brake drag +0.3 +1.5% -1.5%
Reduced tire roll'g resistance +0.7 +3.6% -3.5%
Reduced aero drag +1.2 +6.2% -5.8%
All, combined (actual) +7.3 +37.6% -27.4%
(calculated, sum) +7.3 +37.6% -34.3%
(calculated, product) +8.4 +43.2% -30.2%
Again, use of the power equation, this time with = 1-0.050 ln(N),
adjusts the product-calculated improvements to the correct values:
2- = (1.432)1~'°50 lr"" = 1.377, A= +37. 71
MPG
o
from the fuel economy slip product;
= (0.698)1-'050 l»<9> - 0.726, A- -27.«
o
from the fuel consumption slip product.
0 9 Q
Porter, "Design for Fuel Economy - The New GM Front Drive Cars", SAE
Paper 790721, June 1979.
-------
222
Based on these two examples, both of which deal with fuel economy
improvements, the relation
n
S = I FTs . Y I* = 1-0.05 In N]
produces acceptably accurate estimates of the combined effects, S , of N
individually-measured slip factors, s.. The individual and combined
slip factors can be either fuel economy or fuel consumption slip factors.
Where some or all of the individual slips represent fuel economy degra-
dations rather than improvements, the equation above is not necessarily
accurate, but should provide a better combined-effects estimate than
the simple product of the slip factors.
-------
223
V. FOR THE FUEL DEM AND ANALYST
Page
The Past Revisited • • « . . 224
Average Fleet MPG by Year: EPA City, 55/45, and Road 225
The Future . . . 225
Forecasted Shortfalls • • 226
MPG Standards to Meet Selected Road MPG Values
Vehicle Age Effect
Relative Fuel Economy vs. Vehicle Age
-------
224
V. FOR THE FUEL DEMAND ANALYST
A. The Past Revisited
Since 1972, analyses of EPA fuel economy trends have been published
229
annually in the technical literature, in the form of EPA reports or SAE
230—235
papers by EPA authors . Whether taken at face value or "discounted",
the average EPA numbers have come to be widely used as a jumping-off
point for fuel demand computations. This report provides a basis for
assigning average Road MPG values to each model year, as a function of
their EPA averages. In Section III, relationships between road and EPA
MPG were developed for model years 1974 through 1979, based on raw data
for consumer-driven and fleet cars, known distributions of consumer and
fleet VMT, and odometer mileage effects. These data provide a straight-
forward means of labeling the 1974-79 cars with a Road MPG value.
229
U.S. Environmental Protection Agency, "Fuel Economy and Emission
Control", November 1972.
230
Austin and Hellman, "Passenger Car Fuel Economy - Trends and Influencing
Factors", SAE Paper 730790, September 1973.
231
Austin and Hellman, "Fuel Economy of the 1975 Models", SAE Paper
740970, October 1974.
232
Austin, £t al^ "Passenger Car Fuel Economy Trends Through 1976", SAE
Paper 750957, October 1975.
Tlurrell, et_ al, "Light-Duty Automotive Fuel Economy Trends Through
1977", SAE Paper 760795, October 1976.
Murrell, "Light-Duty Automotive Fuel Economy ... Trends Through
1978", SAE Paper 780036, February 1978.
235
Murrell, "Light-Duty Automotive Fuel Economy ... Trends Through
1979", SAE Paper 790225, February 1979.
-------
225
In the absence of EPA vs. Road MPG relations for pre-1974 cars, we must
note that (1) those cars are in most respects technologically similar to
the 1974's, but different from later models, and (2) their average EPA
fuel economy is very near that of the 1974's, but different from that of
later models. Accordingly, we have applied lonly] the 1974 relationship
to pre-1974 models. The following table lists the average EPA MPG
values and the corresponding Road MPG's, determined on the above basis:
Estimated Average .New-Car Fuel Economy at
4000 Miles, by Model Year
EPA Fuel Economy:
Road Difference:
Model
Year
pre-1968
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
72 FTP
City
12.22
11.89
11.98
11.89
11.67
11.56
11.47
11.51
—
—
—
—
—
75 FTP
City
12.88
12.59
12.60
12.59
12.27
12.15
12.01
12.03
13.69
15.23
15.99
16.97
17.60
55/45
14.90
14.69
14.74
14.85
14.37
14.48
14.15
14.21
15.79
17.46
18.31
19.57
20.11
Road
MPG
13.66
13.53
13.56
13.63
13.33
13.40
13.19
13.23
13.83
14.11
14.72
15.81
16.87
vs. 75 Citv
+6.1%
+7 . 5%
+7.6%
+8.3%
+8.6%
+10.3%
+9.8%
+10.0%
+1.0%
-7.4%
-7.9%
-6 . 8%
-4 . 1%
vs. 55/45
-8.3%
-7.9%
-8.0%
-8.2%
-7.2%
-7.5%
-6 . 8%
-6.9%
-12.4%
-19.2%
-19.6%
-19.2%
-16.1%
B. The Future
That was the easy part. Forecasts of the EPA and Road fuel economy averages
for future model years require additional consideration.
-------
226
First, it is not realistic to expect that the EPA average MPG for a
given model year will be exactly equal to the MPG standard for that
year. Several manufacturers are today above even the 1985 standard, and
few manufacturers will be below a given year's standard if they can help
it. The track record for the first three years in which the standards
have been in effect shows overall average EPA MPG leading the standard
by more than 1 MPG; we assume that this will continue, although by a
gradually-decreasing margin:
EPA MPG versus MPG Standards
Standard EPA MPG Difference
1978
1979
1980
1981
1982
1983
1984
1985
18.0
19.0
20.0
22.0
24.0
26.0
27.0
27.5
19.6
20.1
22.4
24.0
25.5
27.0
27.5
28.0
+1.6
+1.1
+2.4
+2.0
+1.5
+1.0
+0.5
+0.5
Actual
Estimated
Second, while the 1975-79 models may be considered representative of
some of the cars of the 1980's from the standpoint of technological
similarity, the 1974's are clearly a different breed of hardware. Just
as it was our judgement earlier that data from the 1974's is appropriate
for pre-1974 cars, 1974 data is simply not technologically appropriate
for post-1980 forecasting, whether or not the 1974 data might happen to
appear mathematically similar to data from the 1975-79 models.
Another technological consideration applies specifically to the 1975
and 1976 models. As noted in Section IV.B.4.b., significant shortfalls
occurred for these two (and only these two) model years at the basic
production hardware level. We note that 1975 marked the adoption of
-------
227
new emission control technology, and must allow for the possibility that
this change—not the technology itself but its sudden introduction—may
have had a negative effect which took two years to dissipate. The
phasing-in of technology improvements has since proceeded at a more
gradual, evolutionary pace, as evidenced by the lack of similar produc-
tion slippages for 1977-78-79; we expect that orderly influx of improvements
to continue in the 1980's, and there is no basis to project recurrence of
the 1975-76 kind of production slip anomaly. Hence when using the 1975-
76 EPA vs. Road relationships as (partial) predictors of the future, we
have removed the 7-8% production slips that were part and parcel of
these models' overall slips. In a similar vein, we have (as is only
fair) discounted the production slip overages observed in the 1977-79
models' data.
Thirdly, from a strictly numerical standpoint, not every model year's
data deserves equal weighting when combined with other years' data.
Given two data sets, one averaging 15 MPG and the other 25 MPG, the
first set is a better predictor in the neighborhood of—say, 18 MPG,
while the second set is more reliable when considering what might
happen at 27 MPG. In combining the 1975-79 data, we have chosen the
weighting parameter:
where: MPG = EPA MPG being forecasted;
X
MPG. = Average EPA MPG for year i; and
[ ] = absolute value.
Thus, if forecasting at 24 EPA MPG, road data from a 12 EPA MPG fleet
7
is weighted by (.50) = .250, while road data from a 23 EPA MPG fleet
2
is weighted (.96) = .920. Many analyses, including DOE's, throw out
data that varies from the value in question by more than 50%; our weighting
method does not go that far, but does recognize the need to soften the
influence of off-center data. The next table illustrates relative
weightings based on each year's average EPA MPG presented earlier.
-------
228
Weighting Factors for Use of Historical
Fuel Economy Data in Forecasting,
at Selected EPA MPG Levels
1975 data
1976
1977
1978
1979
EPA = 12
.468
.297
.225
.136
.105
11.
.974
.826
.732
.604
.552
20
.623
.762
.838
.957
.989
24
.433
.529
.582
.665
.702
27.5
.330
.403
.443
.506
.535
Finally, a portion of the road MPG shortfalls for the 1975-1978 models
has been attributed to certain inaccuracies in the EPA test procedures
for those models, such as imprecise simulation of the tire-to-dyno roll
interface and nonuniform distribution of vehicle weights within the EPA
weight classes (Section IV.C.). These problems have been reduced or
eliminated beginning with model year 1979, and the fuel economy shortfalls
due to these factors, are not projected to occur in post-1978 vehicles.
Hence the nominal 2.8% shortfall due to these factors has been removed
from the 1975-78 data for forecasting purposes, i.e., the projected
future MPG performance of vehicles similar to the 1975-78 models has
been purged of some road shortfalls where the cause of these types of
shortfalls in past models is known to be corrected.
Application of all of the above leads to the following MPG estimates:
Estimated MPG for 1980-85 Vehicles
with Technology Similar to 1975-79 Cars,
at 4000 Odometer Miles
MPG Standard EPA MPG Road MPG
1980
1981
1982
1983
1984
1985
20.0
22.0
24.0
26.0
27.0
27.5
22.4
24.0
25.5
27.0
27.5
28.0
18.1
19.0
19.8
20.6
20.8
21.1
-------
229
Note that these estimates apply to vehicles with a late-1970's mix of
technologies. The 1975-79 vehicle sample which generated the input data
consists mainly of gasoline engines with oxidation catalyst emission
control.
As noted earlier in this report, there is evidence that newer technologies
may reduce vehicle sensitivity to some real-world conditions. One
such technology for which road MPG has been analyzed in some detail is
*? "\ft
the Diesel; DOE reports an EPA-to-Road relationship for Diesel cars
as follows:
Road GPM = 1.1? x EPA GPM - ,001
This equation indicates a road shortfall of about 12% for Diesels,
compared to the 22% to 26% shortfalls in the preceding table for 1975-79
type vehicles.
The penetration of Diesels into the vehicle population can be estimated
from the trend of Diesel sales fractions ±n recent model years. In
examining these trends we note the Diesel fraction in the California
fleet to be a good one-year precursor of the Diesel fraction for the 49-
states fleet. With these trend data and the "Pearl equation" frequently
used in forecasting,
y =
a + be
-ox
a Diesel fraction of 14% is projected for 1985. The historical data and
the fitted Pearl curve are shown in the next figure; the "GM projections"
237
shown are from a trade journal in the public domain , and—although
not used to construct the Pearl curve—are in good agreement with it.
236
McNutt, Dulla and McAdams, "Comparison of EPA and In-Use Fuel Economy
of 1974-1978 Automobiles", Internal DOE Report, October 1979.
237
Automotive News, June 4, 1979.
-------
230
FIGURE 66. Estimated Diesel Fraction of New Car Sales vs. Year
Sales:
• = Calif. (+ I Year)
T = Federal (49 States)
• = GM Projections
1977
1978
1979
I960 1981
Model Year
1982
1983
1984
Of the many other technology alternates expected to see gradually increased
usage in the 1980's, the only other one for which there is any appre-
ciable road data is that of front-wheel drive cars.
Our analysis of GM's 1975 customer survey data shows total road slips as
follows for Eldorado and Toronado front drive cars and their most closely-
matched rear drive counterparts.
MPG Slip Factor, Road/EPA City
Front Drive Rear Drive
Cadillac
Oldsmobile
1.02
1.07
1.00
1.02
This indicates a road MPG advantage of 2% to 5% for the GM front drive
models. Road MPG for Ford's 1978 Fiesta front drive car has been reported
to be 2.3% above the overall Ford MPG regression curve while that of the
238
238
South and Raja, "In-Service Fuel Economy", SAE Paper 790227, February 1979,
-------
231
nearest rear drive competitors (Pinto/Bobcat and Mustang II) was 0.3%
below the curve, implying a net 2.6% road advantage for the front drive
model. These GM and Ford data together show an average front drive road
239
advantage of 3.2%. As to market penetration, GM is quoted as estimating
a front drive sales fraction of 50% for their car lines by the mid-1980's.
We have not analyzed any of the other new technologies in any detail,
for forecasting purposes; because of this, and because there is some
uncertainty as to the buildup rates for Diesels and front drives, we
have estimated improved technology impacts on the 1980-85 fleets in
terms of two scenarios which we believe constitute upper and lower
bounds, in fuel economy terms, to what will transpire in the next six
years. The lower bound, Scenario "A", assumes market penetrations by
1985 of 14% for Diesels and 25% for front-drive cars; the upper bound,
Scenario "B", assumes 20% and 50% respectively for Diesels and front
drives. The forecasts for these scenarios are given in the next table.
The pre-1985 market fraction buildups for these cases are also hypo-
thetical; the gasoline cars' and Diesel cars' respective EPA MPG values
are determined from these market penetrations and assumed overall EPA
MPG values presented earlier. The estimated Diesel fleet MPG's for each
year reflect consideration of sales shifts between the manufacturers of
Diesel cars.
By 1985, the effect of expanded use of these improved technologies
increases fleet road MPG over the "late 1970's technology" figures
presented earlier, by 0.7 MPG—for Scenario A, to 0.9 MPG—for Scenario B.
The road shortfalls implied by these figures are given in the table
following, using three reference values to define "shortfall": The EPA
55/45 value, the MPG Standard, and lastly-the EPA City MPG value.
Evaluation of the latter is important in view of the [interim] use of
only the City numbers in current Fuel Economy Labels and Gas Mileage
Guides.
OOQ
Ward's Engine Update, June 8, 1979.
-------
232
Forecasts of 4000-mile MPG
for two Improved-Technology Scenarios
% of vehicles:
FWD Diesel
EPA MPG:
Gasoline Diesel
Road MPG:
(all cars)
Scenario "A"; 1985 fleet 14% Diesel, gasoline cars 25% front wheel drive
1980
1981
1982
1983
1984
1985
5
8
13
17
20
21
6
9
12
13
14
14
22.1
23.6
25.2
26.7
27.2
27.7
28.6
28.5
28.5
29.0
29.5
30.0
18.4
19.5
20.3
21.2
21.5
21.8
Scenario "B"; 1985 fleet 20% Diesel, gasoline cars 50% front wheel drive
1980
1981
1982
1983
1984
1985
6
13
23
31
38
40
6
11
15
18
19
20
22.1
23.6
25.1
26.6
27.1
27.6
28.6
28.4
28.3
28.8
29.4
29.9
18.4
19.5
20.4
21.3
21.6
22.0
Road MPG Shortfalls for the Two Scenarios, at 4000 Miles
versus EPA 55/45 MPG: versus MPG Standard:
versus EPA City MPG:
1980
1981
1982
1983
1984
1985
4.
4.
5.
5.
6.
6.
0(18%)
5(18%)
2(20%
8(21%)
0(22%)
2(22%)
4
4
5
5
5
6
.0(18%)
.5(19%)
.1(20%)
.7(21%)
.9(21%).
.0(21%)
1.
2.
3.
4.
5.
5.
6(8%)
5(11%)
7(15%).
8(18%)
5(20%)
7(21%)
1.
2.
3.
4.
5.
5.
6(8%)
5(11%)
6(15%)
7(18%)
4(20%)
5(20%)
0.
1.
2.
2.
2.
2.
9(5%)
2(6%)
0(9%)
4(10%)
6(11%)
7(11%)
0.9(5%)
1.2(6%)
1.9(9%)
2.3(10%
2.5(10%)
2.5(10%)
-------
233
Since shortfalls are projected with respect to the City values, the
interim City-only system may not be acceptable as a permanent arrangement
for the EPA/DOE Fuel Economy Information program.
The historical and forecasted fleet road MPG values from all of the
analyses above are shown on the familiar "road vs. EPA" plot in the next
figure. The boundary curves for the historical data (see Section III)
are also shown. The forecasted band is seen to be within those boundaries,
and very near the line of a constant 20% shortfall.
FIGURE 67. Relations Between Road and EPA Fuel Economy, Historical and Forecasted
25
20
(D
CL.
-O
3
15
10
10
I
15
20 25
EPA S5/45 MPG
30
35
It is of interest to estimate the MPG Standards which correspond to
specific road fuel economy levels. Such estimates necessarily involve
some extrapolation; nevertheless, the next table presents such estimates
for the two 1980's scenarios, and also for the "1975-79 technology" case
described earlier.
-------
234
MPG Standards for Selected
Road MPG Values, at 4000 Miles
Road MPG
22
24
26
27.5
30
1975-79
Technology
30
35
41
45
51
Scenario
"A"
28
33
37
41
47
Scenario
"B"
28
32
36
39
44
C. Vehicle Age Effect
It has been emphasized that all of the foregoing applies to vehicles
at odometer mileages of 4000 miles. We have done that because, for
trend analysis and fuel demand forecasting, it must be recognized that
the fuel economy of a given model year's fleet is not a time-invariant
constant. The 1973 models' average road MPG in 1973 was one value, in 1978
another value, and in 1984 it will be yet another. Since vehicle fuel
economy is determined in large measure by how the vehicles are driven,
no analysis can produce a number guaranteed to occur. However, certain
aspects of vehicle behavior which affect fuel economy are known to vary
with odometer mileage, and do permit some estimate of time variations in
average MPG.
Section III.C.3. presented an equation relating odometer mileage to MPG:
4K
= 0.0186 In(ODO) + 0.846 [ODO = odometer mileage]
This relationship indicates an initially steep, and later more gradual,
rise in MPG with mileage accumulation under constant operating conditions.
But it is well known that vehicles are not used exactly the same way
throughout their lifetimes. The next figure illustrates patterns of
decreasing annual miles traveled, from three of many models in current
240-242
use in demand forecasting . This has a definite fuel economy
-------
235
FIGURE 68. Decreasing Vehicle Travel with Vehicle Age
25
SO 75 100
Odometer Mileage (Thousands)
r
£
f
6
ISO
effect, as discussed in Section IV,C,2,d. The applicable relationship is:
MPG
MPG
= 0.125 In(AMPD) + 0.535
[AMPD = Avg. miles/day]
41.1
or its equivalent,
MPG
14PG
= 0.125 ln(Al&?) - 0.202
[ AMPY = Avg. miles/year|
15K
(The fact that the equations fall apart below 5.0 miles driven per year,
or 73 feet per day, is of no practical significance.)
U.S. Environmental Protection Agency, "Mobile Source Emission Factors",
Report EPA-400/9-78-005, March 1978.
O / 1
Luchter, "The Methodology of Passenger Automobile Fuel Econgmy Rule-
making-Part 1: Technology", SAE Paper 790380, March 1979.
0/0
Cantwell et^ aJ^, "Projections of Motor Vehicle Fuel Demand and Emissions",
SAE Paper 780933, November 1978.
-------
236
When combined, the ODO and AMPY relationships provide an estimate of
relative fuel economy as a function of vehicle age, for a given annual
mileage trend model. The resulting time histories of relative MFC for
the three cases from the preceding figure are shown in the next figure.
In all cases, relative MPG rises initially, reaches its peak in one to
two years, and declines thereafter. This general behavior can be reason-
ably well represented by use of a function of the form
f(A) = p2A1e-<*A2
where A = vehicle age and A., and A~ are functions of A. The equation
for relative MPG using this function, as shown on the figure, is:
Relative 1-1PG = 0. 938 +
1/8=0.35]
FIGURE 69. Relative Fuel Economy vs. Vehicle Age
1.05
RMPG = FODO x FMPY;
FODO = -0186/n. (ODO) + .846
FMI>Y = .125 In. (AMPY) - .202
DOT
Curve Fit:
RMPG = 0.938 + TA;
.95
10
IS
Vehicle Age. Years
(DuPont)
25K
50K
I
75K
J
IOOK
I2SK
I
(DOT) I
2SK
50K
I
75K
I
IOOK
Odometer Miles
I25K
I
I50K
-------
237
VI. CONSUMER ADJUSTMENT OF EPA MPG
Page
Questionaire Approach 238
Field Trials 242
Adjustment Formula Approaches 243
Four-factor Equation: GM Data ....... 243
Four-Factor Equation: Ford Data 249
-------
238
VI. CONSUMER ADJUSTMENT OF EPA MPG
As recognized by the Conference Committee for NECPA 1978 (See Section I),
it would be desirable to devise a practicable approach with which individual
consumers could "adjust" the EPA MPG figures to reflect the particulars
of their own vehicle usage situations.
EPA and others have been exploring such adjustment methods for a number
of years. This section discusses the more promising of these approaches.
As a result of these studies, we are not optimistic as to the development
of an adjustment method that is acceptably accurate, and at the same time
acceptably simple for consumer use.
A. Questionnaire Approach
Ten factors, known to influence vehicle fuel economy and reasonably
well-quantified in the technical literature, were incorporated into a
questionnaire by means of which individual drivers can characterize
their own driving conditions. By totalling point values assigned to
each individual answer, a respondent arrives at a total score which, via
a look-up table, is used to adjust the EPA MPG value upward or downward
in accordance with the respondent's own "driving profile". Each individual
answer score is scaled logarithmically to the fuel economy adjustment
coefficient for that condition, and the total score is proportional to
the sum of all of the answers' logarithms: this has the same effect as
multiplication of the individual adjustment coefficients, and the total
score is thus related to the product of the individual factors, a "net
adjustment factor", as it were. For a given EPA fuel economy value and
total score, the look-up table gives the adjusted MPG corresponding to
this net adjustment factor.
The questionnaire and look-up table are shown below. It will be noted
that a total score of 26 corresponds to no adjustment of the EPA value.
Those questionnaire answers which describe the EPA combined City-Highway
test conditions do in fact produce a total score of 26, but other combinations
of answers which also yield this score also result in no net adjustment
of the EPA MPG value. ;
-------
239
The scoresheet below permits you to see how your own "driving
profile" affects your gas mileage. For example, you can calculate
the gas mileage you would get on a summer vacation trip and compare
that with your calculated fuel economy for winter home-to-work
commuting.
To fill out the scoresheet, pick the single most appropriate answer
for each question and enter the score for that answer in the space
provided. If two answers are equally appropriate, enter the average
of the two point values. After entering scores for all the questions,
total your score, find your car in the booklet, and note its listed
EPA combined City/Highway fuel economy: . Use the mileage
scoring table to determine the adjusted gas mileage value for your
driving profile.
In addition, you can use the scoresheet to help explain actual
measurements you make of your car's gas mileage. When you use the
scoresheet this way, answer the questions according to the way the
car was driven between the last two tankfills you used for your
measurement.
1. WHAT IS THE OUTSIDE TEMPERATURE FOR MOST OF YOUR DRIVING?
OVER 40°F (1) UNDER 40°F (0)
POINTS
2. WHICH OF THE FOLLOWING BEST DESCRIBES THE ROAD SURFACE YOU DO
MOST OF YOUR DRIVING ON?
SMOOTH BROKEN PAVEMENT OR LOOSE GRAVEL
PAVEMENT (4) PACKED DIRT/GRAVEL (2) OR DIRT (0)
POINTS
3. SELECT THE CATEGORY THAT BEST DESCRIBES THE TERRAIN IN
WHICH YOU DRIVE:
FLAT OR GENTLY
ROLLING (3) HILLY (2) MOUNTAINOUS(O)
POINTS
4. WHAT IS THE OVERALL AVERAGE SPEED FOR YOUR TRAVEL?
(REMEMBER, THE AVERAGE SPEED OF A TRIP IS ALWAYS LESS THAN
YOUR CRUISING SPEED)
UNDER 17-24 25-33 34-50 51-65 OVER 65
17 MPH (0) MPH (1) MPH (2) MPH (3) MPH (2) MPH (1)
POINTS
-------
240
5. WHICH CATEGORY BEST DESCRIBES THE NATURE OF THE TRAFFIC FLOW
YOU ENCOUNTER?
CRUISE OR
NONSTOP (A)
ESPECIALLY
FREE-FLOWING (3)
UNUSUALLY
CONGESTED (1)
ABOUT AVERAGE
FOR ROAD TYPE (2)
POINTS
6. SELECT THE TYPE OF ACCELERATIONS YOU NORMALLY USE WHEN DRIVING:
SMOOTH AND
GRADUAL (2)
ABOUT
AVERAGE (1)
RAPID (0)
POINTS
7. DO YOU OPERATE YOUR AIR CONDITIONER MOST OF THE TIME?
NO (3)
YES (2)
POINTS
8. FROM THE TOP ROW, SELECT YOUR AVERAGE TRIP DISTANCE; THEN FROM THE
VERTICAL COLUMN SELECT THE PERCENTAGE OF YOUR TRIPS WHICH ARE MADE
FROM A COLD START. (COLD START MEANS THE ENGINE HAS BEEN OFF FOR AT
LEAST FOUR HOURS.) PICK THE
9. DO YOU
ACCORDING
SINGLE MOST APPROPRIATE POINT VALUE.
0-5 5-10 OVER
MILES MILES 10 MILES
25% COLD STARTS OR LESS (3) (3) (3)
25-50% COLD STARTS (2) (3) (3)
OVER 50% COLD STARTS (0) (2) (3)
i
KEEP YOUR CAR MECHANICALLY MAINTAINED (AND ENGINE TUNED)
TO THE MANUFACTURER
'S SPECIFICATIONS?
POINTS
YES (3)
NO (2)
10. WHICH OF THE FOLLOWING CATEGORIES BEST DESCRIBES THE LOAD
YOU NORMALLY CARRY?
POINTS
1 or 2
OCCUPANTS (4)
3 or MORE
OCCUPANTS (3)
1,000-2,000 Ib
CARGO OR TRAILER (2)
2,000-3,000 Ib
CARGO OR TRAILER (1)
CARGO OR TRAILER
OVER 3,000 Ib (0)
POINTS
VEHICLE CODE NO. YOUR MEASURED GAS MILEAGE YOUR ZIP CODE
TOTAL POINTS
-------
241
Gas Mileage Scoring Table
Fuel Economy listed in Gas Mileage Guide (Combined City/Highway) :
Total
Points
17
18
19
20
21
22
23
24
25
26
27
28
29
30
10
4
4
5
6
7
7
8
8
9
10
11
12
13
15
11
4
5
6
7
7
8
8
9
10
11
12
13
15
16
12 13 14 15
Your ADJUSTED
5
5
6
7
8
8
9
10
11
12
13
15
16
18
5
6
7
8
8
9
10
11
12
13
14
16
18
20
6
6
7
9
9
10
11
12
13
14
16
17
19
21
6
7
8
9
10
11
11
12
14
15
17
18
20
22
16
Fuel
7
7
8
10
10
11
12
13
14
16
18
20
22
24
17 18
Economy
7
8
9
10
11
12
13
14
15
17
19
21
23
25
7
8
9
11
12
13
14
15
16
18
20
22
24
27
19
8
8
10
12
12
13
14
16
17
19
21
23
26
28
20
8
9
10
12
13
14
15
16
18
20
22
24
27
30
21
9
9
11
13
14
15
16
17
19
21
23
26
28
31
22
9
10
11
13
14
16
17
18
20
22
24
27
30
33
23
10
10
12
14
15
16
18
19
21
23
25
28
31
34
24
10
11
12
15
16
17
18
20
22
24
26
30
32
36
26
11
12
13
16
17
18
20
21
24
26
29
32
35
39
28
12
12
14
18
18
20
21
23
25
28
31
34
38
42
30
12
13
15
18
20
21
23
25
27
30
33
37
40
45
32
13
14
16
20
21
23
24
26
29
32
35
40
43
48
34
14
15
17
21
22
24
26
28
31
34
38
42
46
51
36
15
16
18
22
24
25
27
30
33
36
40
44
49
54
-------
242
The questionnaire was subjected to three field trials, whose results are
summarized in the table below. For each of the trials, the table shows
the ratio of in-use road MPG to EPA MPG, averaged over all respondents,
and the range and coefficient of variation (standard deviation divided
by the average) for this ratio. The right side of the table shows the
same types of figures for the ratio of adjusted MPG to road MPG.
Ideally — if the questionnaire adjustment was highly successful — the
latter ratio should average very near 1.00, the range of ratios should
be very narrow, e.g. 0.98-1.02, and the coefficient of variation should
be significantly reduced, which would indicate that the adjustments
produced nearly exact matches to each respondent's road MPG experience.
Results of Fuel Economy Questionnaire Trials
Road MPG/EPA MPG: Adjusted MPG/Road MPG:
Average Range C.O.V. Average Range C.O.V.
0.58-0.75 14%
0.70-1.30 21%
0.94-1.59 16%
In the EPA trial, a number of employees of the EPA's Ann Arbor, Michigan
laboratory worked with a preliminary version of the questionnaire; this
trial resulted in a general downward overcorrection and an increase in
scatter for the adjusted values. The questionnaire and look-up table
were modified somewhat as a result of the EPA trial, and the modified
version (which ia the one illustrated earlier), was used in trials by Ford
Motor Company employees in Dearborn, Michigan and GSA employees in
Washington, B.C. The Ford trial gave excellent average agreement between
adjusted MPG's and road MPG's, but the adjustment again increased the
dispersion on a driver-by-driver basis. In the GSA trial, the adjustment
overestimated road MPG; i.e., did not (on the average) bring the EPA
values down enough to match actual road MPG. As in the other trials, the
adjustment scheme in the GSA trial did nothing to reduce dispersion. It
must be concluded that this questionnaire is of only marginal value in
EPA
Ford
GSA
0.80
0.90
0.93
0.75-0.86
0.68-1.28
0.63-1.17
7%
18%
16%
0.71
1.00
1.12
-------
243
adjusting EPA MPG values on the average to match real experience, and of
no value at all in producing adjusted EPA MPG values which better match
individual drivers' actual fuel economy.
In our post-mortem on this adjustment scheme, we find, to be sure,
certain detailed shortcomings in this particular questionnaire; the
location of cutpoints between one answer/score combination and the next,
omission of additional factors such as odometer mileage, tire pressure,
and the like. But more importantly, we find that this general approach
to individual driver fuel economy adjustment has a fundamental deficiency;
there is in any questionnaire approach an inevitable trade-off between
accuracy and simplicity. There is little doubt that expansion of the
above questionnaire to cover 15, or 20, or 30 influences would improve
its driver-to-driver MPG accuracy, as would expansion of the number of
possible answers to the individual questions. Additional complication
in the latter vein would certainly reduce problems of gross digitization,
such as an average speed of 51 mph yielding a 10% MPG difference from 50
mph. However, even the existing questionnaire may be near, or possibly
beyond, that limit of complexity at which consumer acceptance and use
become extremely unlikely.
B. Adjustment Formula Approaches
EPA analysis of the 1975 GM customer survey data base produced the
following four-factor MPG equation:
Actual MPG .1268 (In AMPD) -.0118(ln POP) +.00123(T) -.00055(RH) +.6570
EPA City MPG
where AMPD = Average miles driven per day
POP = Population of owner's ZIP code
T = Temperature, °F
RH = Relative humidity, percent
In = Natural logarithm
The correlation coefficient is 0.620 and the standard error is 0.123.
-------
244
If AMPD, POP, T, and RH are known, an adjusted MPG ratio corresponding
to those four factors can be computed from the equation. For a given
vehicle, multiplication of the vehicle's EPA City MPG by this MPG ratio,
or factor, gives an adjusted MPG which should approximate the fuel economy
that particular driver should expect on the road.
If a consumer knows the EPA City figure for his car, the other four
input variables are readily available, and his adjusted MPG can be
arrived at graphically by means of the next chart. In the example
on the chart, a car with an EPA City rating of 26 MPG, in a location
with a population of 150,000, is driven 60 miles/day, at 30°F and 30%
relative humidity: the chart predicts a road fuel economy of 27 MPG
under these conditions.
This adjustment scheme was tested by playing it back through the GM data
base, using the four-factor equation and the specific conditions for
each car, to calculate individual adjusted MPG values. As with the
questionnaire approach, the averages, ranges and coefficients of variation
were determined for the aggregate road-to-EPA MPG ratio; again, the
adjustment would be considered successful if the latter ratio averaged
1.00 and showed a significantly smaller range and C.O.V. than the unadjusted
data. The results of this test are summarized below.
Results of Test of Regression Equation Adjustment
Average Range C.O.V.
Road MPG/EPA 55/45 MPG 0.8745 0.42-1.45 15.4%
Adjusted MPG/Road MPG 0.9999 0.71-1.40 9.7%
On the average, this adjustment performed quite well, but it had a
relatively minor impact on the dispersions. With a coefficient of
variation of 10% and individual variations as high as 40%, it is clear
that many individual drivers would have benefited little from this
method of adjustment.
-------
245
FIGURE 70. Construction Plot for Fuel Economy Adjustment
EPA City MPG
0 10 20 30 40 SO
20 3D 40
Predicted MPG
-------
246
This is depicted quite clearly in the next two figures. The first figure
shows the distribution of road MPG's for four "EPA MPG" strata of the GM
data. Each stratum includes all cars whose EPA MPG falls within a 5-MPG
band. For the 20-24.9 MPG stratum, for example, average EPA MPG is
22.8, and road MPG for these cars ranges from a minimum of 12.2 to a
maximum of 29.9; the peak of the distribution of road MPG's for this
stratum is at 19.9 MPG, and the average road MPG is 19.2. It is obvious
from this figure that most cars' road MPG's are below their EPA value,
although a small minority of cars did achieve or surpass their EPA
numbers. In this figure, the peaks of the road MPG distributions
coincide reasonably well with a simple adjustment of the EPA values
proposed by General Motors: 1/R = 1/E + .01
FIGURE 71. Distributions of Road Fuel Economy
(1975 GM Customer Survey)
30
20
IS
10
(29.9)1
EPA =
25-30
10
IS
20 25
EPA 55/45 MPG
30
35
-------
247
In the second figure, the regression-adjusted EPA values now coincide
with the peaks of the road MPG distributions, indicating good agreement
between average road MPG and the regression-adjusted EPA figures, but
the scatter in road MPG for a given adjusted EPA MPG is still significant,
FIGURE 72. Distributions of Road Fuel Economy with Respect to "Adjusted EPA MPG"
(I97S CM Customer Survey)
30
25
20
IS
10
10
Adj. EPA
= 25-30
IS 20 25
Regression-Adjusted EPA MPG, E1'
30
35
For further illustration, the data for a randomly selected group of ten
identical vehicles from the GM data base are shown in the following
table. The table shows actual fuel economies and actual-to-EPA City MPG
ratios, the four factors used in the regression equation adjustment, and
the resulting calculated fuel economies and calculated-to-EPA City MPG
ratios.
-------
248
Example of Fuel Economy Adjustment for Ten Identical GM Vehicles
(EPA MPG = 12.3 City, 19.0 Highway, 14.6 Combined)
Car
A
B
C
D
E
F
G
H
I
J
Actual
MPG
16.3
11.1
12.1
11.5
14.5
16.1
15.2
10.3
16.8
12.4
Act. MPG
EPA City MPG
1.33
.90
.98
.93
1.18
1.31
1.24
.84
1.37
1.01
Actual
i 11
Avg. Miles
Per Day
429.0
17.0
50.0
12.6
23.8
45.6
38.5
26.6
23.4
6.2
07(1 - - - Ql-anH
1R? flnpf f
Pop
(OOP)
3.47
3000
3.71
747.0
747.0
2.34
3.31
13.23
260.0
82.50
ge Ratio,
ard Devia
i F ^ £»nf" nf
Temp RH
(°F) (%)
46 75
72 71
72 71
34 88
34 88
70 78
68 75
39 68
72 73
66 77
Road MPG
Calc.
MPG
16.5
10.9
13.6
10.0
11.0
13.4
13.1
12.0
11.8
9.8
EPA City MPG
tion -----
\7a T ^ a t- -1 rm - - - - ~
Calc. MPG
EPA City MPG
1.34
.89
1.11
.81
.89
1.09
1.07
.97
.96
.79
Calculated
- - - - 0.99
0.16
. . 1 TV
In the aggregate, the adjustment resulted In a 12% underestimation of
average road MPG for this group of vehicles. Dispersions were not
appreciably reduced. For the first two individual cars, the adjusted
MFC's agree quite well with the actuals; for the other eight cars,
however, adjusted fuel economy differs from the actual by at least 1.5
MPG and as much as 5 MPG.
This example also exposes the possible error of assigning meteorological
and demographic data to individual cases based on survey dates and
vehicle owners' home ZIP codes: Vehicle A, averaging 429 miles per day
during the survey period, was very likely nowhere near home, and the
conditions existing in its home ZIP code at that time could have been
quite different from those where the car was being operated.
-------
249
From a fuel economy survey of 1978 model Ford Motor Co. employee lease
243
cars, South reports regression equations as follows:
MPG = .8232(MH) -1.231 +.9194x10 2 (MPD) -. 02528 (URBAN) +. 2921 (InAM)
MPG . , = .68?5(MH) -1. 118 +2.093xlO~2 (i&D) -.02076 (URBAN) +.32S6(lnAM)
GPMsummer= 1-149(1/MH} +-OH02 -. 392ZxlO~4 (1-1PD) +. W67xlO~S (URBAN) -. 1319xlO~2 (InAM)
GPMwinter= 1-294(1/m) +-J1SS9 -1.088xlO~4 (MPD) +. 1545xlO~3 (URBAN) -. 2042xlO~2 (InAM)
where MH = EPA 55/45 MPG
MPD = Average miles driven per day
URBAN = Percent urban driving, driver estimated
InAM = Natural log of average odometer miles
All correlation coefficients are 0.81 or higher. The first two equations
(fuel economy regressions) were used to adjust the raw data values. As
with the GM case, the adjustments reduced dispersions but did not eliminate
them, as indicated below:
Summer Data Winter Data
Unadjusted C.O.V. = 13.6% C.O.V. = 14.7%
Adjusted for MPD,
URBAN, AM C.O.V. = 11.3% C.O.V. = 10.6%
The figure shows road vs. EPA MPG for the raw data and the three-para-
meter adjusted data. The significant upward movement of the corrected
Winter regression line indicates that the Winter operational character-
istics were detrimental to fuel economy. The remaining Summer-to-Winter
MPG difference is due to factors unaccounted for, the most obvious of
which are temperature and road condition; this residual difference is
worse for higher-MPG cars (11% Summer-to-winter shortfall at 30 MPG vs.
6% Summer-Winter shortfall at 15 MPG).
243
South, "Further Results from the 1978 Fuel Economy Survey" SAE Paper
790931, October 1979.
-------
250
FIGURE 73. Effects of Usage Parameter Adjustment on In-Use MPG Regression Lines
IS
10
Source: South243
IS
20
25
30
EPA MPG
The effects of the adjustment on the summer data for one specific car
model (over 300 cars) are shown below.
Unadjusted
Adjusted
Average MPG
Maximum MPG
Minimum MPG
C.O.V.
17.32
24.93
12.05
11.5%
17.22
21.90
12.94
9.4%
Again, less than half of the C.O.V. was eliminated by the adjustment.
Adjustment of EPA MPG values by means of multi-factor equations such as
the above could prove valuable on an average basis; however, this
approach is no more acceptable for individual consumers than is the
questionnaire approach, principally because of the many fuel economy
influences which cannot be represented, and which therefore cannot enter
into the adjustment.
-------
251
VII. PUBLIC COMMENT
Page
Invitation , 252
List of Cotnmenters 252
Summary of Comments 253
Private Citizens 253
American Automobile Association (AAA) 253
American Motors Corporation .......... 253
American Oil Company 254
American Petroleum Institute ..... 254
Automobile Club of Southern California 254
Chrysler Corporation . . 255
Ford Motor Company . 255
General Motors Corporation . 256
Tosco Corporation 257
Comments of Other Federal Agencies . 257
DOE Comments (Letter) ............. 258
EPA Response 259
DOT Comments (Letter) ...... 260
EPA Response .................. 263
DOT Comments (Enclosure 1) 264
EPA Response 271
DOT Comments (Enclosure 2) . . . ... . . ... 272
EPA Response ............ . 275
-------
252
VII. PUBLIC COMMENT
A. Invitation
Notice of opportunity for public comment was published in the
Federal Register on July 12, 1979 (44 F.R. 40724). Eleven comments
were received as of August 27, 1979, as follows:
Private Citizens
W. Mark Day
M. P. Stombler
Consumer Groups
American Automobile Association
Automobile Club of Southern California
Industrial Concerns
American Motors Corporation
American Oil Company
American Petroleum Institute
Chrysler Corporation
Ford Motor Company
General Motors Corporation
Tosco Corporation
-------
253
B. Summary of Comments Received During Report Preparation
PRIVATE CITIZENS
° EPA estimates should be made more representative of actual
use, to compare present vehicle to potential replacement
vehicle;
0 Suggestions for changing EPA estimates and procedures:
decrease EPA estimates to represent typical engine mistuning
and wheel misalignment;
decrease temperature for cold-start;
use short trips with engine cool-down between trips;
refine altitude information given to public;
- advise people of temperature effects.
0 Get rid of the dynamometer;
° EPA should use a steady-state track test which the average
driver can relate to (e.g., 55 mph steady cruise);
° Driving his BMW at 70-90 mph, he gets 18 MPG at low altitude
and 25 MPG at high altitude.
0 Include effects of:
aerodynamics;
inertia;
altitude
AMERICAN AUTOMOBILE ASSOCIATION (AAA)
0 EPA test fuel is not representative of pump gasoline, and
differences in fuel properties can create road MPG shortfalls.
AMERICAN MOTORS CORPORATION
0 Test brake drag is 1/2 of actual;
0 Test tire pressure is greater than actual;
0 Tire loss is less for test (2 tires versus four);
0 4000 mile EPA test vehicles maintained better than older cars;
0 4000 mile EPA test vehicles are selected to represent mean of
production vehicles: more variability in actual vehicles'
0 Driver performance must be accounted for.
-------
254
AMERICAN OIL COMPANY
Road Fuel Economy Relationships:
Model Year No. Cars Equations
1974 1656 Road MFC = 1.04 (EPA l-IPG)'9642
1975 2485 Road MPG = 2.86 (EPA WG)
1976 2410 Road I® G = 2. 84 (EPA
1977 2375 Road MPG = 1.71 (EPA MPG)'
AMERICAN PETROLEUM INSTITUTE
0 Road MPG for Domestic makes in fleet use:
1968 to 1974, from 83% (1973) to 94% (1968) of 1967 MPG;
1975 to 1977, from 92% (1975) to 103% (1977) of 1967 MPG.
° Road MPG corrected for changes in vehicle design characteristics
(not corrected for odometer mileage effects):
1968 to 1974, from 88% (1973) to 99% (1969) of 1967 MPG;
1975 to 1977, from 97% (1975) to 106% (1977) of 1967 MPG.
0 Emission controls blamed for MPG losses (not credited for gains)
AUTOMOBILE CLUB OF SOUTHERN CALIFORNIA
0 Survey of 1.5 million club members:
- An increasing number of motorists are aware of the
EPA estimates;
A significant number of motorists (46%) compare their
own fuel economy with the estimates;
A majority (51%) indicate that their auto obtained
worse fuel economy than the 55/45 estimate.
0 Relationship of in-use fuel economy to EPA estimates:
mature engines give 7% better fuel economy than green
engines;
EPA ratings overestimate road MPG by 7% (1976) to
9% (1977);
carbon balance fuel measurement underest imates fuel
economy by 4% (1977) to 6% (1976).
0 Recommended:
whole number range (e.g., 13-18 MPG) for future EPA estimates;
intensive EPA/DOE data collection, computer analysis,
and mathematical modeling of influence factors.
-------
255
CHRYSLER CORPORATION
° Sees no reason to change EPA's testing procedures; supports
EPA's test. Since so many factors affect fuel economy, no
single test can be in all ways "representative";
0 EPA test is good for measuring relative fuel economy.
FORD MOTOR COMPANY
0 Ford's 1979 production cars are fairly represented by their
test prototypes;
0 Supports the EPA test procedure:
simplicity;
both emissions and fuel economy are measured at same time;
accurate relative MPG comparison;
- dyno test can be done consistently year-round (not
affected by weather, etc.).
0 In-use influences penalize smaller cars more than larger ones:
trip length and accumulated mileage tend to be less in
an urban environment, and smaller cars are found more
in that environment;
- small cars have smaller displacement-to-weight ratios,
which results in more carburetor enrichment;
payload effect is greater for smaller cars;
smaller tires on smaller cars have greater dyno effect.
0 Vehicle Condition
- older cars are driven less, and used less for long trips;
are also in worse state of tuning, tire inflation, and
wheel alignment;
power load from additional accessories is greater in
actual vehicles;
EPA test does not differentiate properly between bias
and radial tires.
0 Differences due to Environment
wind;
grades;
- temperature.
-------
256
0 Dyno/Road Differences
tire cooling is different on road; this causes higher
road rolling resistance;
road acceleration rates are greater than in the dyno test;
inertia simulation is hampered by tire slippage on dyno.
GENERAL MOTORS CORPORATION
0 The EPA test was never intended to represent actual fuel economy;
0 Estimated EPA fuel economy is useful to public for comparison,
as stated on label;
0 It is difficult to quantify the fuel economy effects of:
- road surface
state of repair of road
road grade, curvature, crown
- wind
- precipitation
altitude
humidity and temperature
accessory loads
vehicle maintenance
- vehicle load
0 Should consider other factors, also:
traffic conditions
driver behavior
cold start effects
assumptions in the EPA tests and data use
0 There is little or no difference between certification and
production vehicles.
0 Reasons not to change test:
the test has the precision needed to obtain valid
repeatable laboratory results;
if the objective of the EPA estimate is to predict
actual fuel economy, then tests must be developed to
measure fuel economy under all possible circumstances;
-------
257
it is better to test under known conditions [as in the
current EPA test] and use results only for comparison;
the purpose of the MPG standards is to reduce gasoline
consumption on a national basis; correlation between
fleet fuel consumption (EPA test) and actual fuel economy
is a constant: actual GPM = EPA GPM +• 0.01.
TOSCO CORPORATION
The EPA tests use high octane gasoline, while most in-use
vehicles use lower octane fuels. This factor could contribute
to the difference between actual fuel economy and EPA estimates,
C. Comments of Other Federal Agencies
The Departments of Energy and Transportation furnished information
and guidance during the preparation of this report. These agencies
were briefed on the study's findings at the time of completion of
the first draft, and submitted comments on the draft. Their comments
are reproduced in their entirety below, together with brief EPA
responses to the comments.
The Federal Trade Commission was also briefed and given copies of
the draft report; the FTC did not submit written comments.
-------
258
Department of Energy
Washington, D.C. 20585
'*«* 9 1380
Mr. Michael Walsh
Deputy Assistant Administrator
Mobile Source Air Pollution Control
ANR-455
U.S. Environmental Protection Agency
Washington ,xiD.p. 20460
Dear MrX'
^-*
I would like to thank you for the briefing your staff provided to
the Department of Energy (as well as DOT and FTC) on your draft
report Passenger Car Fuel Economy; EPA and Road - A Report to
the Congress. The report is both comprehensive and of excellent
technical quality. We are very impressed with the quality and
thoroughness of the analysis Dill Murrell and Karl Hellman have
done.
Both during the briefing and later we gave your staff technical
comments on the work and its presentation. The only major point
I want to restate here regards the treatment of 1979 model year
data. By the nature of the "fleet" versus "consumer" weighting
process used in the analysis, the EPA representation of 1979 MY
shortfall is almost completely dependent on one manufacturer's
(Ford) data. While we have no reason to doubt the validity of that
data for Ford cars, it may not be representative of the general
MY 1979 auto population. The GM MY 1979 "fleet" data is not only
different in absolute shortfall, but it shows a different trend.
We are not suggesting you change any conclusion you have reached
but rather that any conclusion on the 1979 MY trend must be made
with caution. This is all the more important since the auto
manufacturers seem to now be taking the tack that "while there
may have been a shortfall problem for 1974-1978, the 1979 (Ford)
data shows that things are getting better,"
While we hope this is true, DOE doesn't accept one year's data
from one manufacturer as being definitive of future years' trends.
Your report ought to be clear so as to avoid any unnecessary
misuse.
-------
259
One other point that does not appear to be addressed in the
EPA report is the lack of data and analysis on inuse light
truck fuel economy. While the report, almost by necessity,
is focused on passenger cars, the light truck mpg shortfall
problem may be even worse. The report should be specific with
regard to this point so that it is not assumed that this data
and analysis is equally applicable to trucks.
We' are anxious for you to release the report in its final form.
This will greatly improve understanding of the shortfall
issue (by a number of interested parties). It is also
necessary as part of the basis for the rulemaking on 1982
model year fuel economy labels, which we are anxious to
begin. I am concerned that any further delay may foreclose
action for the 1982 model year. DOE is prepared to offer
comments, conduct analysis and present the results of our
consumer survey work to support your rulemaking activity.
I would appreciate hearing from you as to what you schedule
is for this action and what we can do to assist EPA. We are
looking forward to working with you.
Sincerely,
Sydney Berwager
cc: Charles Gray - EPA
Dill Murrell - EPA
Marilyn Holmes - FTC
Barry McNutt - DOE
EPA Response
The predominance of Ford data in the 1979 data is pointed out in Section
III.C, Representativeness of the Data Sample.
The report title, "Passenger Car Fuel Economy ..." describes what is
addressed by the report; the Executive Summary indicates the need for
in-use data on light trucks and vans.
-------
260
U.S. DEPARTMENT OF TRANSPORTATION
NATIONAL HIGHWAY TRAFFIC SAFETY ADMINISTRATION
WASHINGTON. O.C. 20590
MAR 3 1 1980
IN REPLY REFER TO:
Mr. Michael P. Walsh
Deputy Assistant Administrator for
Mobile Source Air Pollution Control
Environmental Protection Agency
Washington, D.C. 20460
Dear Mr. Wa-lsh:
This transmits comments of the National Highway Traffic Safety
Administration (NHTSA), Department of Transportation (DOT), on the
Environmental Protection Agency (EPA) draft report, "Passenger Car Fuel
Economy: EPA and Road, A Report to the Congress." NHTSA review of this
report has been coordinated with the Office of the Assistant Secretary
for Policy and International Affairs, DOT.
Overall, the report presents an extremely broad treatment of the
multitude of factors which can affect fuel economy, together with
estimates of the effects of most of these individual influences. The
incorporation of some 260 references must surely make the report the
most comprehensive single document dealing with EPA versus on-road fuel
economy, and indicates a very thorough and conscientious effort by the
authors.
NHTSA's primary comment on the study concerns the relationship of the
Executive Summary to the remainder of the report. In our view, findings
and conclusions given in the Summary do not fully summarize the vast
amount of technical information and effort that has gone into subsequent
sections of the report. This shortcoming, we believe, also detracts
from focusing on the main objectives of the "404 study" which were to:
(1) Evaluate the reasons for deviation between EPA and consumer fuel
economy and; (2) provide a basis for consumers to better evaluate the
fuel efficiency they could expect to achieve as a function of individual
driving characteristics, optional equipment, etc. The orientation here
is to provide the consumer with better information on which to base a
vehicle purchase decision and to predict the fuel economy he can expect
to achieve in the real world. The study, as written, has achieved the
first objective, but was not able to meet the second objective.
-------
261
It is our recommendation that the Executive Summary be revised, and
offer the following, more specific suggestions:
(1) Reconcile the information under Conclusion One, pages 1-3 to
1-8, concerning EPA-to-on-road shortfall with information
given later in the report on the same topic. The Executive
Summary states that a shortfall exists between EPA and on-road
fuel economy, and that the shortfall is greater the higher the
EPA rating. However, elsewhere in the report information is
given to support a narrowing of this gap. For example on
pages 111-16, 17, it is indicated that since 1976, the
shortfall for high mpg cars has decreased. Pages IV-108 and
V-7-9 cite a tightening of test procedures and a projection of
significantly increased penetration of diesel power and front-
wheel drive technologies (features which show less dynamometer
to road slip than their conventional technology counterparts),
all of which should serve to decrease the shortfall gap.
Further, on pages IV-17 and IV-172 vehicle usage factors are
cited which could, in the past, have contributed to higher
shortfalls for higher mpg cars, but which may well be changing
as consumers begin to use small cars for other than primarily
local travel. Some recent NHTSA research (Enclosure 1) -
supports this premise of changing vehicle usage.
(2) In view of the comments under (1), above, and degree of
applicability to the study objective, reconsider the use of
the charts and tables on longe-range, projected fuel
consumption on pages 1-3 through 1-6. Such information
concerning the implications on long term fuel conservation,
Iras, we believe, only a secondary relationship relative to the
primary one of consumer information, yet this information
appears first in the discussion of Conclusion One.
(3) The final statement of page 1-6 concerning fuel savings due to
the standards should be qualified. Both the Congress and the
DOT recognized at the outset that EPA fuel economy would not
equal real-world fuel economy. Accordingly, NHTSA has
employed a specific discount factor (11%) in all consumption
projections in recognition of this historical shortfall.
(4) Some idea of the fuel economy slips due to individual factors,
as covered at length in the text, should be included in the
Executive Summary.
(5) Relative to the comments at the top of page 10 concerning
future work, we suggest you also cite our "On-the-Road Fuel
Economy Survey" as a source to fill the gap on light truck
in-use data. The survey will also provide much needed
nationally representative data on late model passenger car
-------
262
fuel economy. In this regard we believe more acknowledgement
should be made in the Executive Summary of the present lack of
nationally representative and statistically valid data for on
road, consumer fuel economy data. Potential biases due to
fleet data, lack of import representation, and odometer
differences are given in the report in Section III. To these
three reasons we would add a fourth: Existent data are
comprised of various special purpose tests and brief surveys,
as opposed to being based on a nationally valid, probability
sample of consumer owned vehicles.
In addition to the Executive Summary, we suggest that the first
paragraph under Section II-BACKGROUND be revised to reflect a third
purpose for fuel economy data, namely, "to assist in the promulgation,
enforcement, and evaluation of fuel economy standards." Along with
this change, we propose the following paragraph to be inserted as
paragraph number four (4) on page II-l: "Standards engineers and
personnel concerned with standards enforcement and program evaluation
need data to assist in: (1) the establishment of levels for future fuel
economy standards; (2) determination of compliance with existing
standards; and (3) to evaluate the real-world effects of those standards
as to their energy conservation effects."
This concludes our comments on Sections I and II, the Executive Summary
and Background. Enclosure 2 contains a few additional technical
comments on Sections III-V.
We appreciate the opportunity to review this draft report. Should you
require further information, -please don't hesitate to contact me
(426-1560).
Sincerely,
Barry Felrice
Associate Administrator
for Plans and Programs
2 Enclosures
-------
263
EPA Response
The Executive Summary has been rewritten.
Conclusions regarding high-MPG cars' road shortfalls, and nationwide
fuel consumption, have been modified to address only those model years
for which there is actual road fuel economy data.
Conclusions on fuel savings associated with the standards now reflect a
scenario in which assumed road MPG improvements parallel, but do not
equal, EPA MPG.
In the Background section, we have clarified the point that the historical
users of MPG figures were fuel demand forecasters and consumers, joined
later by those responsible for management of fuel economy standards.
-------
264
TSC r 1 J29.4 (S/7«)
UNITED STATES GOVERNMENT
Wl1
DEPARTMENT OF TRANSPORTATION
RESEARCH AND SPECIAL PROGRAMS ADMINISTRATION
TRANSPORTATION SYSTEMS CENTER
SUBJKT. Passenger Car Fleet Projection Parameters
.
MOM , John K. Pollardjfand K. H. Schaeffer
DATE: February 14, 1980
««ptvto
nuniion of: DTS-321
TO
Chief, Transportation Industry Analysis Branch, DTS-322
Contained in Tables 1-4 (attached) are estimates of passenger
car sales and curb weight by size class for the years 1979-
1990 as well as VMT and scrappage curves. Data sources, metho-
dology and caveats associated with Tables 1-4 are described
below.
1. Auto Sales By EPA Size Class
Retail sales of domestic and imported cars in calendar year 1979
were as shown below:
Total
% by
TOTAL
Class Size
Two Seaters
Minicompact
Subcompact
Compact
Mid-Size
Large Size
A. vwa^.
164,464
672,387
3,001,071
1,431,680
3,343,279
1,897,352
10,510,233
Class Size
1.6%
6.42
28.67.
13.6%
31.82
18. 1%
100.1%
Applying the 1-979 percentage distribution by size class to the
DRI "Trendlong 2004" passenger car sales forecast yields the
results shown In Table 1. The DRI forecast was chosen because,
In the opinion of TSC staff, it represents the best attempt to date
to deal with the issue of light trucks and their substitution
for passenger cars.
ENCLOSURE ONE
-------
265
2. Curb Weights
Proprietary submissions from the four domestic manufacturers were
used to compile the data shown in Table 2. Since no domestic mini-
compacts are planned for production in the years 1981-84, weights
for imports (e.g., Honda Civics) were used for those cells.
3. Survivability Data
Some analyses of survival data disaggregated by size class have
shown an inverse relationship between size and longevity. (Domestic
subcompacts have been an exception to this rule, but their early
retirements reflect problems peculiar to the Vega and the Pinto.
Imported subcompacts survive longer than the average of all domestic
cars.) Relatively shorter lives for larger cars are consistent with
the data on VMT by age which have shown that larger cars accumulate
more miles per year. Survival data typically show small (compact)
cars remaining in the fleet more than two years longer than full-
sized cars of the same make, while being driven one to two thousand
fewer miles per year. Several possible explanations for this
traditional pattern of vehicle use have been offered, among them:
(1) higher income families, who definitely travel more, prefer
larger cars; and (2) multi-car households who have tended to use their
larger cars for vacation and other long trips for reasons of comfort.
The sharp increase in the real price of gasoline in 1979 and the
now widespread fears of continuing availability problems may have
Inverted the rationale for the traditional patterns of vehicle use.
That is, individuals and families who want or need to drive a great
many miles per year may now prefer smaller cars. Similarly, in
the face of availability problems or gasoline rationing, the smaller
car could become the vehicle of choice for vacation trips.
In light of the above, it is the judgment of TSC staff that currently
available data on survival and VMT disaggregated by size class do
not provide a good indication of future patterns of use. We would
not go so far .as to assume that the pattern will be completely
Inverted. Rather, our best guess is that the factors which favor
higher utilization of small cars will roughly balance the factors
which have traditionally favored the higher utilization of larger
cars* Thus, we recommend the use of the same survival and VMT
curves for all vehicle classes.
-------
266
The survival data shown In Table 3 were calculated from R. L.
Folk's registration data for July 1, 1977 and 1978. Since the
Polk" figures cover only 15 model years, survival factors for
years 16 through 25 were estimated by extrapolating an assumed one
year survival rate of 0.7, which produced an estimate of total
car population greater than 15 years old consistent with Polk data.
4. Annual Mileage By Age
For the reasons described in Section 3, above, we recommend the
use of the same schedule of annual miles travelled by age for all
size classes.
The data shown in Table 4 are based upon the National Science
Foundation's 1978 National Transportation Survey. The 1095 respon-
dents to this survey completed questionnaires providing information
on the make, model, age, and miles travelled during the past year
on each of the 1766 vehicles in their households. From these data,
a table showing mean mileage by age for passenger cars was constructed.
Several curve fitting routines were tested on this table. Simple
linear regressions yielded the highest r2 value, .78. The equation
is:
VMTn - 12,012 - 427.39n
where VMTn - miles travelled in year n
n « year of life
Since the survey data covered only eleven vintages, values for years
12 through 25 were extrapolated.
The smoothed data were then input to the TSC Auto Fleet Fuel Consump-
tion Model ("FUEL3") along with the survival data from Section 3. The
resulting estimate of total fleet VMT for 1978 was .940 trillion
miles, some 19.7% less than the FHWA estimate of 1.17 trillion miles.
This was to be expected since the NTS data cover private households
only, whereas the total fleet contains a substantial fraction of
business-use vehicles which are much more heavily utilized. To
correct for this, the NTS derived mileages were multiplied by 1.246
and reinserted into FUEL3. The adjusted figures, shown in Table 4,
result in an estimate of 1978 total fleet VMT of 1.17 trillion.
Attachments
cc:
DTS-321/R. Ricci
-------
TABLE 1. AUTO SALES BY EPA SIZE CLASS
(annual sales In millions)
Year
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
Minlcompact
0.7
0.6
0.7
0.7
0.7
0.7
0.7
0.8
0.8
0.8
0.8
0.8
Subcompact '
3.2
2.8
3.1
3.3
3.3
3.3
3.4
3.6
3.7
3.6
3.6
3.7
Compact
1.4
1.3
1.4
1.5
1.5
1.5
1.6
1.6
1.7
1.6
1.6
1.7
Midsize
3.3
3.0
3.2
3.4
3.5
3.5
3.6
3.8
3.9
3.8
3.8
3.9
Large
1.9
1.7
1.8
1.9
2.0
2.0
2.1
2.1
2.2
2.2
2.2
2.2
Notes: 1 Includes two-seaters
2 Includes subcorapact wagons
3 Includes compact wagons
4 Includes midsize wagons
5 Includes large wagons
SOURCE: Data Resources, Inc., Trendlong forecast allocated to size classes according to 1979
actual percentages.
Cs
-------
N3
O
00
TABLE 2. ESTIMATED AVERAGE CURB WEIGHT (AS SOLD, INCLUDING OPTIONS) FOR DOMESTICALLY MANUFACTURED CARS
Mini-Compact
Subcompact
Compact
Mld-Slza
LairS*
1979
2600
2700
3000
3500
4000
1980
2600
2700
2800
3350
3900
1981
1850*
2500
2700
3250
3900
1982
1850*
2400
2700
3250
3800
1983
1800*
2400
2600
2950
3800
1984
1800*
2350
2600
2950
3450
1985
1750
2150
2500
2800
3350
1936
1750
2100
2400
2800
3300
1987
1750
2100
2400
2800
3300
1988
1750
2100
2350
2750
3250
1989
1700
2100
2350
2750
3250
1990 1991 1992 1993 1994 1995
1700
2000
2350
2700
3200
SOURCE: TSC Staff (baaed upon proprietary data submitted by domestic manufacturers extending to 1986 or 1987 and extrapolated to 1990).
Values for mini-compacts marked with asterisk are for imports, since no domestic minis are scheduled for production
in these yeara.
-------
269
TABLE 3. SURVIVAL RATES FOR PASSENGER CARS
Fraction of Original
Production Still
Registered
Year (All Size Classes)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
1.000
.992
.968
.951
.925
.884
.824
.750
.656
.550
.447
.356
.279
.219
.170
.119
.083
.058
.041
.029
.020
.014
.010
.007
.004
SOURCE: R. L. Polk & Co., registration data for July 1, 1977 and
1978; years 16 through 25 extrapolated at an assumed one
year survival rate of 0.7.
-------
270
TABLE 4. VMT BY AGE
Miles Travelled per
Vehicle, All Size
Classes
Year
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
(Thousands)
14.436
13.903
13.371
12.838
12.306
11.773
11.240
10.708
10.176
9.643
9.110
8.577
8.045
7.513
6.980
6.447
5.927
5.382
4.850
4.317
-------
271
EPA Response
Enclosure One (1) of the DOT comments was prepared prior to receipt
by DOT of EPA's first draft, and is not a direct comment on the report.
The information is useful, and is appreciated. EPA would suggest that
DOT reconsider the assumption, implied by Table 1, that the mix of car
size classes will remain constant through 1990.
-------
272
NHTSA COMMENTS ON SECTIONS III-V
DRAFT REPORT, "PASSENGER CAR FUEL ECONOMY: EPA AND ROAD"
A REPORT TO THE CONGRESS'
U.S. ENVIRONMENTAL PROTECTION AGENCY
January 1980
o Page IV-15. The analogies here are misleading even though the author admits
to "a bit of far-fetchedness." A chemical reaction and the use of db which
is on a logarithmic scale do not illustrate the point that all slips are not
additive. A simple analogy utilizing the factors in point (vehicle and
environmental factors) would be much more useful.
o Page IV-16. It would seem that the more frequent tune-up interval
attributed to smaller engines would be related to owner perceived reduction
in performance rather than less improvement per tune-up. Obviously, a
malfunction which affects one cylinder will have a proportionately greater
effect on a 4-cyl. engine than an 8-cyl. engine.
o Pages IV-76-78. The effect of grades of fuel economy is calculated by using
steady state fuel consumption values. It should be pointed out that steady
state fuel consumptions will likely underestimate the effect of grades since
transient throttle excursions will cause accelerator pump injections, short
periods of power enrichment and other nonsteady state phenomena which will
likely degrade fuel economy.
o Pages IV-112 & 113. The discussion on the pitfalls inherent in making value
judgments such as best or worse based on fuel economy misses the point. The
only message contained in the table at the bottom of IV-112 is that people
who drive more miles may consume more fuel even though they achieve greater
mpg. I believe we understand the intention but the message that is coming
forth from the illustration is not appropriate.
o Page IV-130. The DOT study used electronic flow meters to indicate fuel
efficiency and the conclusion regarding manifold vacuum gauges is therefore
inappropriate.
o Page IV-138. The conclusion is drawn that fuel savings from friction
reducing oils may be lost if drivers use the friction reduction for added
performance. The notion that drivers would begin "hot-rodding" their
vehicles after gaining one or two horsepower is stretching the imagination a
bit too far. The most important slip is the one ignored, a positive slip due
to the fact that the improvement on the dyno is far less than that achieved
in the real world.
o Page IV-146. The section on lubricants skirts the issue of current practices
by stating that EPA has not knowingly given approval for use of "slippery
lubes." How do we reconcile the Honda experience with low viscosity lubes
and the fact that the current Ford certification oil (which is not the oil
used as factory fill) exhibits fuel efficiency characteristics equivalent to
most publicized slippery lubes?
ENCLOSURE TWO
-------
273
o Page IV-154. The discussion of errors created by the energy and velocity
discrepancies is straight forward and the thought that coupling the rollers
will eliminate those errors is plausible. However, the conclusion that this
difference is a measure of the road shortfall is rather vague. Given that
the twin-roll dyno results in unnatural distortion and loading of tires in
the first place, it is difficult to reconcile that coupling of the two rolls
will immediately make the tire losses representative of the real world, it
would seem that the different response of radials vs. bias, front-wheel drive
vs. rear-wheel drive and all the other anomalies would have to be considered
prior to drawing any conclusions regarding the representativeness of the
coupled rollers.
o Page VI-160. The statement is made' that power steering operational data is
not available to evaluate whether a shortfall exists due to lack of turning
of the front wheels during tests. We suggest that any one of the following
will provide a good indication of this loading:
1. Program Summary Report Study on Reduction of Accessory Horsepower
Requirements, Air Research Manufacturing Company, June 15, 1977.
2. Automotive Accessory Drive System Study, 74-310772, Air Research.
3. C.W. Coon et al, Technological Improvements to Automobile Fuel
Consumption, Vol. IIA, DOT-TSC-OST-74-39. December 1974, (Southwest
Research Institute).
4. C. Marks, "Which Way to Achieve Better Fuel Economy," General Motors
Engineering Staff, Detroit, Michigan, December 3, 1973.
o Pages V-6, 7. An estimate of 14% diesel penetration in 1985 is made using
the "Pearl Equation." It is suggestd that the DOT submission with regard to
diesel penetration (9% in 1985) to the Three Agency Diesel Task Force is a
much more reliable estimate of this value. The claim is made that the G.M.
estimate supports this figure. The G.M. estimate is of their own
penetration; given that Ford and Chrysler as well as a number of the
importers plan little or no diesels it is difficult to reconcile 15% as a
number for the total car fleet.
o Page V-8. G.M. is quoted as projecting 50% front-wheel drive by the mid-
1980's (June 8, 1979, Ward's Engine Update). This estimate has long been
superseded by announcements which place this projection at 90-95% (e.g.,
Motor, Oct. 1979). Further, Chrysler has said 100% FWD by 85 and Ford is now
projecting 40-50 % by 85 and 100 % by 86-87.
It is therefore concluded that the estimates for FWD in the two scenarios are
too low. At least a third scenario should be added which places FWD
penetration at 90-100%.
-------
274
In view of the emphasis on "running changes" in recent Congressional hearings
and in the media, we suggest this factor be added to the report, along with
appropriate comments.
To the above we would like to add one general comment. Whereas many sources
of data and test results are cited in the report, there seems to be a rather
universal lack of interpretive comments concerning the sources as to their
strong, weak points etc. For example, in the conflicting data on tune-ups
(pages IV-48, by Claffey), is it not possible that some of the difference in
mpg could be due to inadequately performed tune-up work? Also concerning the
EPA 75-76 RM data, it may be that insufficient data are available in the
small engine category (i.e., say 100-150 CID) to draw definite conclusions.
In the interpretive comments, the -0.7% mpg for <225 CID engines may not be a
true indication, but may be due to random variatTon. Were all tests done by
the same laboratory with the same technicians, test/tune-up equipment, etc.?
All of these are potential sources of variation in fuel economy results.
-------
275
EPA Response
Many of the comments in Enclosure Two (2) are editorial; some have led
to corrections, some have not. All are acknowledged and are appre-
ciated.
Surveillance over the certification use of "slippery lubricants" is
hampered by the non-existence, at this time, of a precise system for
classifying oils as to vehicle fuel efficiency potential. In the "Honda
experience", EPA questioned the representativeness of the engine lube
oil used in some Honda test cars. Honda was required to rerun those cars
using oil that was considered representative. Thus, EPA took action
which was equivalent to disapproving use of an unrepresentative oil.
In the case of Ford, certification vehicles use an unexotic oil, [Brand
Name] 10W-30, which is commercially available and is not excessively
costly. The DOT comment suggests that there is something unusual about
use in certification of oils which are not used for factory fill. This
is not at all unusual: factory-fill oils are frequently "break-in oils",
while certification oils resemble, or at least should resemble, post-
break-in refill oils.
Data on the fuel economy effects of power steering under parametric test
conditions is not in short supply. What is needed is data on the frequency
and degree of turning experienced by vehicles in actual use.
It is acknowledged that EPA's crystal balls may be subject to clouding,
with regard to the prediction of future market shares of Diesels and
front wheel drives. The report now specifies, as it should have in
draft, that Section V is an illustration. It should be noted, though,
that "forecasted" road MPG, at 27.5 EPA MPG, varies by less than one MPG
from the most optimistic Diesel and front drive scenario to a scenario
in which there are virtually no Diesels or front wheel drives at all.
-------
276
(This page intentionally blank)
-------
277
APPENDICES
Page
Method for Averaging Fuel Economy Data 278
Examples of Averaging and Regression Analysis Methods 280
Energy Balance for a Synthetic Motor Oil . 282
Relation Between Home-to-work Trip Speed and
Trip Speed for Non-work Travel 283
Computations of Travel Characteristics and Effects . . . • 286
Factors Related to Annual VMT 286
Trip Length and Frequency • • • « « 288
Average Vehicle Speed and Regional VMT ... 291
Relative Fuel Economy 294
Cold Start Fraction 297
U.S. Average Road Fuel Economy, Passenger Cars, through 1978 . . . . . 300
Fleet Fuel Consumption Implications 302
-------
278
APPENDIX A
Method for Averaging Fuel Economy Data
Suppose a motorist takes the following trips:
200 miles, using 15.0 gallons;
100 miles, using 9.4 gallons;
140 miles, using 11.8 gallons.
The fuel economies of these trips are:
200 miles
15.0 gal.
100 miles
9.4 gal.
140 miles
13.33 MPG;
- 10.64 MPG;
- 11.86 MPG.
11.8 gal.
By merely averaging the trip MFC's, the motorist would calculate:
(13.33 + 10.64 + 11.86) + 3 - 11.94 MPG
But this is incorrect. The motorist traveled 440 miles and used 36.2
gallons, so the overall fuel economy was:
440 * 36.2 - 12.15 MPG
To calculate fuel economy correctly for multiple trips, the following
equation must be used:
Miles Total Miles Traveled (A-l)
Gallon Total Gallons Used
If the individual trip lengths and fuel economy values are known, but
the gallons used are not known, the proper equation is:
Miles. + Itiles. + ... + Milesu
.__ J _ 2 _ N
' Milee, Mies, Miles (A-2)
1
where: miles - length of trip "x";
MPG - gas mileage for trip "x"; and
A
N • number of trips.
If all of the trips are of the sane length (such as in a standardized test),
equation A-2 is equivalent to:
-------
279
Miles x N
... . . (A-3)
Miles
i \
PG,,)
i\ f
t \MPG f MPG r ' ' ' r MPG
where: miles = the standard test length, and
N = the number of tests.
Equation A-3 simplifies to:
which is the "harmonic average" of the test MPG's.
If only trip MPG values (but not trip lengths) are known, no averaging
technique assures accurate computation of cumulative >fPG; however, the
harmonic average always gives a more conservative (lower) estimate than
244
the arithmetic average
n f I
McNutt et al, "A Comparison of Fuel Economy Results from EPA Tests
and Actual In-Use Experience, 1974-1977 Model Year Cars (Appendix A)",
SAE paper 780037, February 1978.
-------
280
APPENDIX B
Examples of Averaging and
Regression Analysis Methods
A large (>1500 car) data base on in-use fuel economy has been exten-
sively analyzed by EPA. The following analysis results illustrate how
different conclusions can be drawn from the same data, depending on the
method of analysis. No one method or conclusion is "more correct" than
any of the others. Note that even the average fuel economy is subject
to some debate.
A. Average Values
Road MPG
EPA (55/45) MPG
Road GPM
EPA GPM
In(Road MPG)
In(EPA MPG)
In(Road GPM)
In(EPA GPM)
Road MPG - EPA MPG
Road GPM - EPA GPM
Road MPG/EPA MPG
Road GPM/EPA GPM
Mean Value
13.76
15.79
0.07678
0.06572
2.593
2.739
-2.593
-2.739
-2.030
0.01106
0.875
1.172
Standard
Deviation
3.44
3.43
0.0174
0.0113
0.234
0.191
0.234
0.191
2.28
0.0127
0.134
0.194
Implied Implied Road MPG
Avg. MPG Shortfall
13.763
15.793
13.02b
15.22b
13.37
15.48
13.37
15.48
12.9%
14.5%
13.6%
13.6%
12.9%C
14.4%d
12.5%C
14.7%d
Arithmetic Average
Harmonic Average
Using Arithmetic Average EPA MPG
Using Harmonic Average EPA MPG
-------
281
B. Regression Equations (R = Road MPG, E = EPA MPG)
R = 1.401 + .7827[E]
R = -24.41 + 13.93[ln(E)]
R = 28.94 - 231.1[1/E]
ln(R) = .1109 + .9062[ln(E)]
ln(R) = 1.800 + ,05027[E]
ln(R) = 3.593 - 15.22[1/(E)]
1/R = .007372 + 1.056[l/E]f
1/R = .2471 - .06216[ln(E)]
1/R = .1306 - .03496[E]
E - R = -1.401 + .2173[E]
E - R = -8.472 + 3.834[ln(E)]
E - R = 6.201 - 63.47[1/E]
R/E = .7827 + 1.401[1/E]
R/E = 1.095 - .08055[ln(E)]
R/E = .9437 - .004375[E]
1/R - 1/E = .01451 - .000218[E]
1/R - 1/E = .02118 - .003692[ln(E)]
1/R - 1/E = .007372 + .05619H/E]
Correlation
Coefficient
0.780
0.775
0.758
0.742
0.737
0.734
0.683
0.683
0.670
0.327
0.322
0.314
0.118
0.115
0.112
0.059
I 0.055
0.050
Standard Implied Road MPG Shortfal
Error E = 15 MPG E = 25 MPG
2.15
2.17
2.24
0.157
0.158
0.159
0.0127
0.0127
0.0129
2.15
2.16
2.16
0.133
0.134
0.134
0.0127
0.0127
0.0127
12.4%
11.2%
9.8%
13.3%
14.3%
12.2%
14.3%
15.4%
16.2%
12.4%
12.7%
13.1%
12.4%
12.3%
12.2%
14.4%
14.4%
14.3%
16.1%
18.3%
21.2%
17.4%
17 . 4%
20.9%
19.4%
14.9%
12.0%
16.1%
15.5%
14.6%
16.1%
16.4%
16.6%
18.5%
18.9%
19.4%
e"Fuel economy" regression
"Fuel consumption" regression
-------
282
APPENDIX C
245
Energy Balance for a Synthetic Motor Oil
BTU's Per
6,000-Mile
Drain Period
AUTOMOBILE FUEL ECONOMY EFFECT
Conventional SAE 10W-40 (? 15 mpg
6,000 miles/15 mpg = 400 gallons
400 gallons X 116,000 BTU/gallon 46,400,000
Synthetic @ 15.6 mpg (4% Improvement)
6,000 miles/15.6 mpg = 385 gallons
385 gallons X 116,000 BTU/gallon 44,615,000
Net Automobile BTU Savings 1,785,000
PROCESSING ENERGY EFFECT
Conventional SAE 10W-40
(5 quarts charge plus 3 quarts makeup)
Processing energy including atmospheric and vacuum
distillation, solvent extraction and dewaxing, and
additive preparation = 19,000 BTU/quart
19,000 BTU/quart x 8 quarts 152,000
Synthetic
(5 quarts charge plus 2 quarts makeup)
Processing energy including preparation of SHF
stock, ester, and additives = 37,000 BTU/quart
37,000 BTU/quart x 7 quarts 259,000
Net Processing Energy Increase 107,000
ENERGY SAVINGS
Automobile savings 1,785,000
Processing energy increase -107,000
Net savings 1,678,000
Marshall, "Survey of Lubricant Influence on Light-Duty Vehicle Fuel
Economy", Coordinating Research Council Report 502, December 1978.
-------
283
APPENDIX_D
Relation Between Home-to-Work Trip Speed and
Trip Speed for Non-Work Travel
246
A traffic survey on the Los Angeles road route corresponding to the
EPA urban driving schedule shows the following average traffic speeds as
a function of time of day:
Average Speed
7-9 a.m. and 3-5 p.m. (4 hours) 17.4 mph
9-11 a.m. and 1-3 p.m. (4 hours) 21.0 mph
Relative traffic densities can be estimated from time-of-day distribu-
247
tions of vehicle travel :
Fraction of Relative Traffic
Daily VMT Density
7-9 a.m. and 3-5 p.m. (4 hours) 31.30% 1.656
9-11 a.m. and 1-3 p.m. (4 hours) 18.90% 1.000
Other daylight hours (7 hours) 37.00% 1.119
9 p.m. - 6 a.m. (9 hours) 12.80% 0.301
248
An equation (the Greenshields formula ) relating average speed to traffic
density is:
(j \
1 - /d'} where: V = average speed;
0 '
Vf = free-flow speed;
d = traffic density;
d = jam density.
O / £
Scott Research Laboratories, "Vehicle Operations Survey", CRC/EPA
Project No. CAPE-10-68(1-70), December 1971
r\ i -j
Svercl and Asin, "Nationwide Personal Transportation Study, ,Home-to-Work
Trips and Travel, Report No. 8", DOT/FHwA, August 1973.
248
Voorhies, et^ al_, "Vehicle Operation, Fuel Consumption and Emissions as
Related to Highway Design and Operation", Interim Report for DOT/FHwA,
October 1977.
-------
284
This equation is equivalent to:
V = a + b(RTD) where: "a" and "b" are constants, and
RTD = relative traffic density.
Using the preceding data in this equation (we are here assuming that
nationwide relative traffic densities by time of day are applicable to
the Los Angeles route),
17.4 = a + b(1.656) and 21.0 = a + b(l.OOO)
Which yields a = 26.49 and b = -5.49. Applying this to the two other time
blocks, we have:
Average Speed
7-9 a.m. and 3-5 p.m. 17.4
9-11 and 1-3 p.m. 21.0
Other daylight hours 20.4
9 p.m. - 6 a.m. 24.8
249
It has been shown that average night driving speeds are lower than daytime
speeds; accounting for this, the traffic density-derived 24.8 mph night
average speed becomes 23.8 mph.
The NPTS report gives the following distributions of vehicle-miles traveled
by trip purpose, and time of day:
Work Travel Non-Work Travel
7-9 a.m. and 3-5 p.m. 44.64% 24.55%
9-11 a.m. and 1-3 p.m. 10.42% 23.19%
Other daylight hours 34.22% 38.40%
9 p.m. - 6 a.m. 10.72% 13.86%
100.00% 100.00%
2^9Claffey, "Passenger Car Fuel Conservation", DOT Report FHwA-PL-77009,
January 1977. (Urban test route, 75 drivers).
-------
285
Combining these distributions with the average traffic speed values for the
four time blocks (44.64% of work travel is conducted in traffic flowing
at 17.4 mph, etc.), the following VMT-weighted average speeds are obtained:
100.0
Work Travel: ^ = 44fg4 10m42 ^^ ^— = 19^27, mph
17.4 + 21.0 + 20.4 + 23.8
100.0
Non-Work Travel; V = = 20.08 mph
• «" 24.55 + 23.19 + 38.40 + 13.86 *
17.4 21.0 20.4 23.8
100.0
All Travel: V -,-, = = 19.80 mph
2^ all 31.30 18.90 + 37.00 12.80 V
17.4 21.0 20.4 23.8
Since the absolute values for these average speeds are unique to this
particular Los Angeles route, the relative values are more generally
applicable:
20.08
— J, • U 5 Ci
mph , 19.27
v work
traveI 19.30 .. )00
———^— — — ±.(J£iQ
-------
286
APPENDIX E
Computations of Travel Characteristics and Effects
A. Factors Related to Annual VMT
250
0 The Nationwide Personal Transportation Study shows that,
in 1969, personal passenger vehicles of all ages had annual
VMT (vehicle miles traveled) averages of 11,105 miles for
urban households and 15,387 for rural households, based
on a questionnaire survey of approximately 6000 households.
251
A 1975 survey by General Motors of approximately 2000 owners
of new GM cars showed annual VMT averages of 13,968 for urban
areas and 17,015 for rural areas. These VMT rates are more
applicable to a late-model, new-car analysis than the NPTS rates.
The GM survey also reveals differential mileage accumulation
O ^ /
rates as a function of car size (as do other studies ) .
The GM data are separable into rural and urban strata, and
show the following differential VMT rates as a function of
vehicle weight:
Urban: 170.5 Amiles/year per 100 Ib. Aweight
Rural: 142.4 Amiles/year per 100 Ib. Aweight
250
Goley, et_ a^,"Nationwide Personal Transportation Study, House-
hold Travel in the United States, Report No. 7", DOT/FHwA, December 1972.
251
Unpublished.
252
Scardino, et al, in "Impact of the FEA/EPA Fuel Economy Information
Program", FEA~Report No. ISBN: 0-89011-487-0, June 1976, showed that
intermediate and larger cars accumulated 641 miles per year (average)
more than compact and smaller cars;
253
Canada Department of the Environment, in "Canadian Automobile Driver
Survey", Report EPS 3-AP-73-10, October 1973, finds a VMT average dif-
ference of 570 miles per year for the same two broad divisions of car sizes;
254
Data (unpublished) furnished to EPA by Mobil Oil Co. shows average
differences in VMT rates of 1800 miles per year for [larger] cars with
300 CID engines compared to [smaller] cars with 100 CID engines.
-------
For 1976, the median year of the five model years considered
in this report, small cars' average weight is 2650 Ib. and
large cars' average weight is 4320 Ib'
255
287
Urban and rural areas differ as to proportioning of registrations
9 S 6
between small and large cars , small cars constituting 35.3%
of all urban registrations but only 28.6% of rural registrations.
257
The NPTS gives trip and VMT fractions by season and place of
residence as follows:
Spring Summer Fall Winter Total
Trips: Urban 26.2%
(incorporated)
Rural 28.3%
(unincorporated)
25.0% 25.6% 23.27, 100%
24.4% 23.8% 23.5% 100%
VMT: Urban
24.7%
28.8% 25.2%
21.3%
100%
Rural
26.9%
27.3% 23.0%
22.8%
100%
255
from EPA data, unpublished.
256
Shonka, "Transportation Energy Conservation Data Book, Edition 3",
Oak Ridge National Laboratory Report ORNL - 5493, February 1979.
257
Strate, "N.P.T.S., Seasonal Variations of Automobile Trips and Travel,
Report No. 3", DOT/FHwA, December 1972.
-------
288
B. Trip Length and Frequency
As discussed above, increases in total VMT from the NPTS values can
stem from increases in trip length, trip frequency, or both.
258
Growth in average trip length has been reported by Tobin and Horowitz
for various trip purpose categories as follows:
Annual Growth Rate
Work-Related trips 0.9% - 1.5%
Shopping 1.3% - 2.8%
Social/Recreational 0.5% - 0.7%
Non-Home based 0.7% - 2.5%
The VMT growth (relative to NPTS averages) discussed in the preceding
section is greater than even the maximum trip length growth rates above
would explain; evidently some of the VMT difference is due to vehicle
"newness" in terms of age. Noting that the newest cars, and highest VMT
259
values, in the NPTS data correspond to higher income brackets , we
can infer, using high income travel characteristics in NPTS Report 7
(Table A-6), that high new-car VMT results much more from increases in
trip frequency than trip length. Applying nominal trip length growth
rates from Tobin and Horowitz, and accounting for the balance of the VMT
difference using "new-car" trip frequency and trip length changes scaled
from that NPTS table, we estimate the following adjusted trip character-
istics for new vehicles, circa model year 1976.
258
Tobin and Horowitz, "The Influence of Urban Trip Characteristics on
Vehicle Warmup — Implication for Urban Automotive Fuel Consumption",
SAE Paper 790656, June 1979.
259
Strate, "N.P.T.S., Annual Miles of Automobile Travel, Report No. 2"
DOT/FHwA, April 1972.
-------
289
NPTS Trip Characteristics Adjusted for New-Car
VMT and Seven Years' Growth in Trip Length
(Model Year 1976 represented)
Urban Households:
Rural Households:
1.
2.
NPTS Data
Work-Related
Family Business
Educ. /Civic/Rel.
Social/Recr.
Other
All Purposes
Adjusted Data
Work-Related
Family Business
Educ. /Civic/Rel.
Social/Recr.
Other
Trip
Length
9.54
4.92
4.05
13.02
7.69
8.41
10.99
5.77
4.35
13.44
8.20
Trips
Year
483
406
118
301
13
1321
549
461
142
348
16
Miles
Year
4609
1999
478
3920
ino
11105
6034
2660
618
4677
131
Trip
Length
11.48
6.73
5.85
13.26
12.38
9.81
12.58
7.82
6.14
14.01
13.83
Trips
Year
561
493
157
336
21
1568
576
501
161
341
21
Miles
Year
6438
3316
918
4455
260
15387
7246
3918
989
4777
290
All Purposes
9.31
1516
14120
10.63
1600
17220
-------
290
The next three matrices present the corresponding trip characteristics,
broken out by car size and season of the year, derived from all of the
above data.
Average Vehicle Miles Traveled, New Cars, 1976
Urban Households: Rural Households:
Small Cars Large Cars Small Cars Large Cars
Spring
Summer
Fall
Winter
Total
Spring
Summer
Fall
Winter
Total
3025 3740 4167
3526 4362 4230
3086 3816 3563
2608 3225 3532
12,245 15,143 15,492
Average Number of Trips, New Cars, 1^76
Urban Households Rural Households
397 453
379 390
388 381
352 376
1516 1600
Average Trip Length, New Cars, 1976
._, —
4818
4890
4120
4084
17,912
Urban Households: Rural Households:
Spring
Summer
Fall
Winter
Small Cars Large Cars Small Cars
7.62 9.42 9.20
9.30 11.51 10.85
7.95 9.84 9.35
7.41 9.16 9.39
Large Cars
10.64
12.54
10.81
10.86
Overall
8.08
9.99
9.68
11.20
-------
291
C. Average Vehicle Speed and Regional VMT
Section IV.C.2.c(l) concluded that the "maximum speed" interpretation of
NPTS trip speed data gives the best agreement with recent GM and EPA trip
survey data. One additional factor: the Canadian survey cited previously
indicates that average speeds for commuting trips in Canada are some 7%
slower in Winter than in Summer. The GM and EPA trip surveys were conducted
in the months of March-June and March-April, respectively, and would not
fully reflect this influence.
We account for this, as a first-level approximation, by dividing the U.S.
into two zones, separated more or less by the 5000 degree-day/year contour
0 f\C\ O £ 1
on Figure APX-1 , and by Federal Region boundaries on Figure APX-2
FIGURE APX-I. Normal Number of Degree Days Per Year.
American Society of Heating, Refrigerating, and Air Conditioning
Engineers, ASHRAE Handbook, 1978.
261
Greene, et_ a!L, "Regional Transportation Energy Conservation Data Book",
Oak Ridge National Laboratory Report ORNL - 5435, September 1978.
-------
292
FIGURE APX-2. United States Federal Regions
REGION I
Principal characteristics of these two zones are as follows:
North Zone
Regions/(States) Included
Zone fraction of U.S. VMT
Urban VMT fraction in Zone
South Zone
1,11
VII,
,(PA,WV),V
VIII,(NV),X
53.5%
58.7%
III(except PA & WV) ,
IV,VI,IX(except NV)
46.5%
51.5%
Trip average speed values determined from the trip length data and from
zonal considerations are given in the next matrix.
-------
293
Average Trip Speeds, New Cars, 1976
Urban Households:
Small Cars Large Cars
Rural Households:
Small Cars Large Cars
Spring, North
South
Summer, North
South
Fall, North
South
Winter, North
South
All
The corresponding VMT
30.7
31.3
32.1
32.1
31.2
31.5
29.9
31.1
31.2
distributions
31.5
32.2
34.8
34.8
32.0
32.3
30.8
32.0
32.6
are shown in
Urban Households:
Spring, North
South
Summer, North
South
Fall, North
South
Winter, North
South
Small Cars
.0274
.0209
.0319
.0244
.0279
.0213
.0236
.0180
Large Cars
.0501
.0383
.0585
.0446
.0512
.0391
.0432
.0330
31.4
32.0
33.8
33.8
31.8
32.1
30.9
32.1
32.3
the next matrix:
32.8
33.5
36.1
36.1
33.4
33.8
31.5
32.8
33.8
Rural Households :
Small Cars
.0170
.0174
.0173
.0176
.0145
.0149
.0144
.0147
Large Cars
.0424
.0433
.0431
.0440
.0363
.0371
.0360
.0367
All
.1954
.3580
.1278
.3188
-------
294
The above analysis yields overall average speeds of 31.7 mph for small cars,
and 33.2 mph for large cars. These values are—if anything—conservative in
both absolute level and car size differential, compared to the findings of
? ft ?
an extensive GM traffic survey , as indicated in the figure below.
16
14
12
I '•
i:
4
2
0
FIGURE APX-3. Percent of Miles Spent in Speed Bands
I [ I I
Average = 36.9 MPH
Compact Cars
I
I
I
7.5 17
5 27.5 37.5 47.5 57.5 67.5 77.5
Speed, MPH
16
14
12
I 10
'o
£ 8
V
I 6
4
2
0
I I I I I I I
Average = 39.4 MPH
rui
Standard Cars
I
I
I
7.5 17.5 275 37.5 47.5 575 67.5 77.5
Speed, MPH
D. Relative Fuel Economy
Fuel economy performance as a function of trip length was modeled using
the preceding trip length:average speed relations and data given in the
text for acceleration intensity penalties (which vary with average speed) ,
and warmup fuel consumption (which varies with average speed and trip time)
o /• o
Johnson, et al, "Measurement of Motor Vehicle Operation Pertinent to
Fuel Economy", SAE Paper 750003, February 1975.
-------
295
The resultant data were normalized to an EPA 55/45 cycle fuel economy
reference value developed from the same model. The next figure shows norm-
alized trip MPG versus trip length for the base case (with no penalty for
excessive acceleration), and for fully-warmed up and cold start trips with
acceleration penalties reflecting real-world observations.
0.7
FIGURE APX-4. Relative Fuel Economy vs. Trip Length
I I I i I I
Trip Average Speed, MPH
10
12
14
Trip Length, Miles
-------
296
This model is the basis for the following matrix of normalized MFC, values.
Relative Fuel Economy (EPA 55/45 = 1.000)
Urban Households: Rural Households
Small Cars Large Cars Small Cars Large Cars
Spring, North, Cold Start
Hot
South, Cold
Hot
Summer, North, Cold
Hot
South, Cold
Hot
Fall, North, Cold
Hot
South, Cold
Hot
Winter, North, Cold
Hot
South, Cold
Hot
.883
.977
.892
.987
.922
1.003
.922
1.003
.894
.985
.899
.990
.871
.963
.889
.983
.915
.995
.924
1.005
.955
1.022
.955
1.022
.925
1.005
.929
1.010
.901
.983
.919
1.003
.912
.993
.921
1.003
.946
1.017
.946
1.017
.918
.999
.922
1.004
.905
.985
.923
1.005
.934
1.006
.943
1.016
.968
1.028
.968
1.028
.940
1.012
.944
1.017
.923
.998
.946
1.018
The proportioning of vehicle travel between "cold start" and hot start
trips depends on the pre-start "soak" time which defines a cold start;
this is a function of travel patterns (trip frequency and time between
trips), and ambient temperature.
-------
297
28.0%
30.0%
35.6%
44 . 1%
23.9%
26.7%
31.6%
43.1%
9 (~\ *\
A driving pattern survey which gathered extensive data on trip frequen-
cies and distributions throughout typical weekdays and weekend days in
six urban areas yields the following definition of cold start fraction
as a function of minimum soak time:
Fraction of Engine Starts Which Are Cold Starts
Minimum Soak Time
for Cold Start North South
8 hours
6 hours
4 hours
2 hours
To clarify: if 8 hours of soak are required for an engine start to be a
"cold start", only those trips which begin after soaks of at least 8
hours are cold starts; if soaks of 4 hours are sufficient to make a
start a "cold start", those additional trips preceded by soaks of 4 to 8
hours qualify as cold starts, and the cold start fraction is higher.
264
Engine cooldown data from a soak time effects study can be used to
estimate what soak time defines a cold start. The selection of a specific
engine temperature to define a start as "cold" is not exact; in this
study, 130°F was judged to be appropriate based on examination of relative
fuel economy vs. engine temperatures, after various soak intervals. (A
difference of 10°F in this assumption makes a difference of 4 to 6
percentage points in calculated cold start fraction.) The soak time to
o /; Q
Kearin, et al, "A Survey of Average Driving Patterns in Six Urban
Areas of the United States", System Development Corporation Report TM-
(L)-4119/007/00, January 1971. An analysis of the cold start impli-
cations of the SDC report appears in Vogt, "Hot Start/Cold Start Weighting
Factors as Determined from the Study 'A Survey of Average Driving Patterns
in Six Urban Areas of the United States'", Draft Report, Standards
Development and Support Branch, ECTD, EPA, April 1976.
264
Srubar, et^ a^, "Soak Time Effects on Car Emissions and Fuel Economy",
SAE paper 780083, February 1978.
-------
298
reach 130°F varies with engine size and ambient temperature, as follows
Hours of Soak Time to Reach 130°F Engine Temperature
(Average of Oil and Water Temp; Initial Temp = 190°F)
Ambient
Temp, °F
20
40
60
80
4-Cyl. Engine
(2.3 litre)
1.7
2.1
2.9
4.3
8-Cyl. Engine
(5.8 litre)
2.1
2.8
3.8
6.0
Average ambient temperatures, as a function of geographical region and
^)f\ ^
season of the year , are shown in the following matrix, together
with cold start soak times and cold start fractions computed using the
foregoing data.
Cold Start Characteristics
Spring,
Summer ,
Fall,
Winter,
North
South
North
South
North
South
North
South
Average
Temp.(F°)
45.2
62.0
68.4
77.6
49.5
64.5
25.0
47.6
STCS (hrs)/Cold
Small Cars
2.32/.420
2.98/.360
3.327.376
4.07/.312
2. 407. 415
3. 107. 353
1.787.453
2.367.399
Start Fraction
Large Cars
3. 057. 385
3.96/.315
4.46/.341
5.587.274
3.18/.380
4.177.307
2.227.427
3.12/.352
VMT-Weighted Cold Start Fraction:
,385
.359
.347
STCS = Soak time for cold start definition; defined here as time
to reach 130°F engine temperature.
265
Newspaper Enterprise Association, The World Almanac and Book of Facts
1980, 1979. ———
-------
299
Applying these weightings to the preceding relative MPG values, the cold/
hot weighted relative MPG's are shown in the next matrix:
Spring, North
South
Summer, North
South
Fall, North
South
Winter, North
South
All
CoId/Hot Weighted Relative MPG
(EPA 55/45 = 1.000)
Urban:
Small Cars Large Cars
Rural:
Small Cars Large Cars
.935
.951
.971
.976
.945
.956
.919
.943
.963
.978
.998
1.003
.973
.984
.946
.972
.957
.972
.989
.994
.964
.973
.947
.970
.977
.992
1.007
1.011
.983
.993
.967
.991
.950
.977
.971
.990
The overall 35.9% cold start fraction can be compared with the cold
f\ e r
start fraction of the EPA combined City-Highway test :
Cold Start Fraction, EPA 55/45 Test
(10,000 mile basis)
Urban trips, 5500 miles @ 7.47 miles/trip: 736 trips
Rural Trips, 4500 miles @ 10.24 miles/trip: 439 trips
Total: 1175 trips
Cold starts, 43% of urban trips:
Hot starts, 57% of urban trips and
all of rural trips:
Cold start fraction, 316/1175:
316 starts
859 starts
26.9%
266
A determination of cold start fraction for emissions purposes would
be conducted differently from this fuel economy-related analysis,
and would address only the urban test, which is the only test used for
emissions certification.
-------
300
APPENDIX F
U.S. Average Road Fuel Economy, M
— Passenger Cars —
Source: DOT/FHwA VM-201A and VM-1
Pre-Emission Control:
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
15.29
15.29
15.29
15.29
15.29
15.30
15.20
15.08
15.05
15.04
14.96
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
14.95
14.96
14.97
14.95
14.99
14.67
14.70
14.58
14.53
14.36
14.39
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
14.28
14.27
14.24
14.33
14.31
14.19
14.17
14.07
14.00
13.93
Emission Control;
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
13.79
13.63
13.57
13.57
13.49
13.10
13.43
13.53
13.72
13.94
14.06
-------
301
(This page intentionally blank)
-------
302
APPENDIX G
Fuel Economy Targets and
Fuel Consumption Implications
A. Targets
"Target" fleet MPG values can be expressed in terms of percentage improve-
ments, EPA figures, or on-road figures. The target for the voluntary
improvement period (Section II) was expressed as a 40% improvement, by
1980, over 1974 levels. The mandatory improvement targets (standards)
were specified in average 55/45 MPG terms, with an implied base level of
13.9 MPG for 1974. This 13.9 MPG figure was the best estimate of the
1974 models' average 55/45 MPG existing between October 1975 and January
f\ r -i
1978, the interval during which the standards were set . No targets
have (thus far) been expressed in terms of road MPG. The following
table lists the specified EPA 55/45 targets, all using the 13.9 MPG
value for 1974 that they were based on, and the road targets implied by
the specified targets, using the current best estimate for the 1974
models' road fuel economy, 13.23 MPG (See Section V.A.).
267
EPA estimates of 13.9 MPG for 1974 were published in Austin, et al,
"Passenger Car Fuel Economy Trends Through 1976", SAE Paper 750957^
October 1975, and again in Murrell, £t al, "Light-Duty Automotive Fuel
Economy Trends Through 1977, SAE Paper 760795, October 1976. (Updated
retrospective estimates of the 1974 models' fleet average fuel economy
have since been published, beginning with Murrell, "Light Duty Automotive
Fuel Economy...Trends through 1978, SAE Paper 780036, February 1978.)
-------
303
Target Fleet Fuel Economy Values,
EPA and Road
Voluntary Program
Model
Year
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
Specified
Target
1.067x1974
1.133x1974
1.200x1974
1.267x1974
1.333x1974
1.400x1974
1.733x1974
Implied EPA
55/45 Target
13.9 Base
14.83
15.75
16.68
17.61
18.53
19.46
24.09
Mandatory Program
Specified Target,
EPA 55/45
18.0
19.0
20.0
22.0
24.0
2JLO
27.0
Implied
Road Target
13.23 Base
14.12
14.99
15.88
17.13
18.08
19.04
20.94
22.84
24.75
25.70
27.5, 26.0 minimum 26.17, 24.75 minimum
XX = specified
YY = implied
B. Fuel Consumption
768
A simplified model was used to investigate the sensitivity of total
U.S. passenger car fleet fuel consumption to the road target values
above, and to the maximum and minimum bounds of in-use fuel economy from
Section III. The size of the fleet was held constant at 100 million
vehicles, so that all observed fuel consumption effects would be due
to fuel economy factors only, and not to growth in the vehicle population,
VMT/car/year, or total VMT. The model's assumptions for annual miles
traveled, and vehicle scrappage rates, as a function of vehicle age, are
given in the reference. The results of this sensitivity study are shown
in the following tables.
Ward and Thompson, "Prediction of U.S. Annual Fuel Consumption by
Passenger Automobiles", Report SDSB 79-12, Standards Development and
Support Branch, ECTD, EPA, 1979.
-------
304
INPUT DATA; Average Road Fuel Economy, MPG
—Individual Model Years —
Model
Year
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985 - 2000
Road
MPG
13.
13.
14.
14.
15.
16.
15.7 -
16.2 -
16.7 -
17.1 -
17.3 -
17.5 -
23
83
11
72
81
87
18.
20.
21.
22.
23.
23.
7
1
3
6
1
5
-------
OUTPUT DATA:
U.S. PASSENGER CAR FLEET FUEL CONSUMPTION
(FLEET SIZE = 100 MILLION)
ANNUAL FUEL CONSUMPTION
CUMULATIVE FUEL CONSUMPTION
D
O
in
3
Z
Z
O
O
CALENDAR
YEAR
1974
2000
(BILLION GAL/YR)
TARGET HIGH LOW
70.4 70.4
70.4
(MILLION BBL/DAY)
TARGET HIGH LOW
4.59 4.59 4.59
(bILLION GALLONS)
TARGET HIGH LOW
(BILLION BARRELS)
TARGET HIGh LOW
0
0
0
35.6
39.5 53.0
2.32 2.58 3.46
1224 1300 1515
0
0
29.1 30.9
0
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
70.0
69.2
68.0
66.3
64.0
61.6
58.9
55.8
52.7
49.5
46.6
44.0
41.8
40.1
38.7
37.7
36.9
36.4
36.1
35.8
35.7
35.6
35.6
35.6
35.6
70.1
69.6
68.9
67.7
66.0
63.8
61.1
58.2
55.2
52.2
49.5
47.1
45.1
43.5
42.3
41.3
40.7
40.2
39.9
39.7
39.6
39.5
39.5
39.5
39.5
70.1
69.6
68.9
67.7
66.0
64.4
63.2
61.6
60.1
58.9
57.7
56.5
55.6
54.9
54.4
53.9
53.6
53.4
53.3
53.2
53.1
53.1
53.0
53.0
53.0
4.57
4.51
4.43
4.32
4.18
4.02
3.84
3.64
3.44
3.23
3.04
2.87
2.73
2.61
2.52
2.46
2.41
2.38
2.35
2.34
2.33
2.32
2.32
2.32
2.32
4.57
4.54
4.49
4.42
4.31
4.16
3.99
3.80
3.60
3.41
3.23
3.07
2.94
2.84
2.76
2.70
2.65
2.62
2.60
2.59
2.58
2.58
2.58
2.58
2.58
4.57
4.54
4.49
4.42
4.31
4.20
4.12
4.02
3.92
3.84
3.76
3.69
3.63
3.58
3.55
3.52
3.50
3.49
3.47
3.47
3.46
3.46
3.46
3.46
3.46
70
139
207
273
337
399
458
514
566
616
663
707
748
788
827
865
902
938
974
1010
1046
1081
1117
1153
1188
70
140
209
276
342
406
467
526
581
633
683
730
775
818
861
902
943
983
1023
1062
1102
1142
1181
1221
1260
70
140
209
276
342
407
470
531
592
650
708
765
820
875
930
983
1037
1091
1144
1197
1250
1303
1356
1409
1462
1.7
3.3
4.9
6.5
8.0
9.5
10.9
12.2
13.5
14.7
15.8
16.8
17.8
18.8
19.7
20.6
21.5
22.3
23.2
24.0
24.9
25.7
26.6
27.4
28.3
1.7
3.3
5.0
6.6
8.2
9.7
11.1
12.5
13.8
15.1
16.3
17.4
18.4
19.5
20.5
21.5
22. 4
23.4
24.3
25.3
26.2
27.2
28.1
29.1
30.0
1.7
3.3
5.0
6.6
8.2
9.7
11.2
12.7
14.1
15.5
16.9
16.2
19.5
20.8
22.1
23-4
24.7
26.0
27.2
28.5
29.8
31.0
32.3
33.6
34.8
36.1
U)
o
Ul
-------
* UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON. DC 20460
SEP 2 9 1980
THE ADMINISTRATOR
Honorable Walter F. Mondale
President of the Senate
Washington, D.C. 20510
Dear Mr. President:
In accordance with section 404 of the National Energy Conservation
Policy Act I am transmitting to Congress the enclosed document entitled
"Passenger Car Fuel Economy: EPA and Road." This is a detailed report
on the degree to which fuel economy estimates, required to be used in
new car fuel economy labeling and in the annual fuel economy mileage
guide, provide a realistic estimate of average fuel economy likely to be
achieved by the driving public.
If there are any questions, or if additional information is needed,
please call me, or your staff may want ^call Gregory Dana at 755-0596.
yours
as M. Costle
Enclosure
I 2J2Z.S UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
\,^ ^ WASHINGTON. DC 20460
SEP 2 9 mo
THE ADMINISTRATOR
Honorable Thomas P. O'Neill, Jr.
Speaker of the House of Representatives
Washington, D.C. 20515
Dear Mr. Speaker:
In accordance with section 404 of the National Energy Conservation
Policy Act, I am transmitting to Congress the enclosed document entitled
"Passenger Car Fuel Economy: EPA and Road." This is a detailed report
on the degree to which fuel economy estimates, required to be used in
new car fuel economy labeling and in the annual fuel economy mileage
guide, provide a realistic estimate of average fuel economy likely to be
achieved, by the driving public.
If there are any questions, or if additional information is needed,
please call me, or your staff may want to c^ll Gregory Dana at 755-0596.
Enclosure
------- |