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

-------