The Engineering
Resource For
Advancing Mobility 400 COMMONWEALTH DRIVE WARRENDALE, PA 15096
1
« % W m
f'
8 9
'g, #
% jf **|
*# &
f 11 If I
<% 1
840496
Development of Adjustment Factors for
the EPA City and Highway MPG Values
Karl H. Hellman
and J. Dillard Murrell
U.S. Environmental Protection Agency
International Congress
& Exposition
Detroit, Michigan
February 27-March 2,1984
-------
The appearance of the code at the bottom of the first page of this paper indicates
SAE's consent that copies of the paper may be made for persona! or interncl use, or
for the personal or internal use of specific clients. This consent is given on the con-
dition. however, that the copier pay the stated per article copy fee through the
Copyright Clearance Center, Inc., Operations Center, P.O. Box 765, Schenectady,
N.Y. 12301, for copying beyond that permitted by Sections 107 or 108 of the U.S.
Copyright Law. This consent does not extend to other kinds of copying such as
copying for general distribution, for advertising or promotional purposes, for
creating new collective works, or for resale.
Papers published prior to 1978 may also be copied al a per pSp#tt.fes;£f' $2-50
under the above stated conditions.
SAE routinely stocks printed papers for a period of three years following date of
publication. Direct your orders to SAE Order Department.
To obtain quantity reprint rates, permission to reprint a technical paper or pet*
mission to use copyrighted SAE publications in other works, contact the SAE
Publications Division.
6SN 0148-7191
Copyright © 1984 Society of Automotive Engineers, Inc.
This paper is subject to revision. Statements and opinions ad-
vanced in papets ot discussion aie the author's and aie his
responsibility, not SAE's; however, the paper has been edited
by SAL for uniform styling and format. Discussion will be
printed with the paper if it is published in SAE Transactions.
I or permission to publish this paper in full or in part, contact
the SAE Publications Division.
Persons wishing to submit papers to be considered for pre',
sentation oi publication through SAE should send the mmn.
script or a 300 word abstract of a proposed manuwnpt to
Secretary, Engineering Activity Board, SAE.
Printed in I S \
-------
840496
Development of Adjustment Factors for
the EPA City and Highway MPG Values
Karl H. Hellman
and J. Dillard Murrell
U.S. Environmental Protection Agency
ABSTRACT
This paper describes the development of ad-
justment factors applicable to the EPA City
and Highway MPG values. The paper discuss-
es the data bases used, and the analytical
methods employed to arrive at adjustment
factors of 0.90 for the EPA City MPG value
and 0.78 for the EPA Highway MPG value.
IT HAS BEEN WELL KNOWN FOR YEARS that the
EPA MPG values in advertising, on new car
labels, and in the Gas Mileage Guide over-
predict actual consumer fuel economy ex-
perience. This subject is discussed in
several earlier SAE papers (l)-(4).* To
improve the MPG information available to
consumers, EPA studied this shortfall and
proposed (5) to adjust the MPG values by
certain specific numerical adjustment
factors.
NATURE OF THE ADJUSTMENTS
The development of MPG adjustment factors
followed an EPA decision to adjust both the
City MPG and the Highway MPG by "uniform
constant adjustment factors". More sophis-
ticated adjustment methodologies had been
developed (1), but the use of uniform con-
stant adjustments was preferred because
these would preserve the overall MPG rank-
ing of vehicles, and the marginal improve-
ment of the more accurate adjustment meth-
odology was not considered worth the per-
ceived added complexity. Therefore, the
analytical task was to find values of cor-
rection factors for city and highway driv-
ing, fc and fh , so that MPG values of:
Adjusted City MPG = fc X EPA City MPG
and
Adjusted Hway MPG * fh X EPA Hway MPG
were the most representative of the aver-
age MPG achieved in actual use.
Bi-Modal Analysis
In a bf-moda? MPG label system, to develop
the adjustment for the EPA city MPG value
it was necessary to use data from vef'des
that were operated in the city. Tne -lgn-
way MPG adjustment was similarly concentra-
ted on data from highway-driven cars This
limited the on-road MPG data bases t: :nose
that contained information on the oeg^ee of
urbanization of the driving.
The data bases used contained "T"
portant parameters: city fraction . :• • snd
average miles per day (AMPD). Tne ca-a-
meter CF is subjective, based or ^escs-ses
to a question of the type, "How
'Numbers in parenthesis refer tc -«'• e^ces
listed at the en$i of this paper
0148-7191/84/0227-0496$02.50
Copyright 1984 Society of Automotive Engineers, Inc.
-------
your driving was done in the city?", with a
numerical response of CF ¦ 100 if all driv-
ing was done in the city, and CF = 0 cor-
resoonding to all driving being non-city
driving. The respondents were not given a
definition of "city", so their responses
reflect their own perceptions of city
ear i v 1 ng.
The other parameter, AMPD. is a non-
subjective indicator of how vehicles are
used. A vehicle driven many miles per
day would tend to accumulate many of these
miles in highway operation.
Stratifying the data by CF and AMPD,
two vehicle subsets are isolated: those
••city driven for shorter distances" and
those "highway driven for longer distan-
ces". From these two data sets, the ad-
justment factors were to be derived.
Fuel Economy Data Base
Characteristics of the fuel economy data
base are summarized in Tables 1 and 2.
The codes RAC, RMC, etc. refer to the
drive ("F"ront, "R"ear). the transmission
("A"utomatic. "M"anual), and fuel system
("C"arbureted, fuel "I"njected. nD"iesei);
a RAC vehicle is one with Rear wheel drive,
an Automatic transmission, and a Carbure-
ted engine.
Not all of the MPG data base was used.
To focus on newer model year vehicles, MPG
data used was restricted to 1979 and newer
models. Also, as shown in Table 1, not all
of the data sources had both CF and AMPD
data. After screening for model year and
the inclusion of both CF and AMPD, a total
of some 43,000 data points remained.
Table 1
Qualitative Summary of In-use MPG Data
GM»
EPA E.F.
Chrysler"
Ford*
DOE
Mult imanufacturer:
yes
yes
no
(Chry only)
no
(Ford only)
yes
Technology coverage:
excel lent
1imited
(no Diesels)
poor
(m i n i ma 1 FI,
no Diesels)
poor
(minimal FI,
no Diesels)
fair
(m i nima 1 F %
few Dieseijj
Vehicles consumer
owned and driven:
yes
yes
company cars
driven by
execut i ves
company cars
driven by
execut i ves
some
Geographical coverage: excellent fair poor poor*" ooor
(7 locales) (83% Mich. ) (74'/, Mich. ) (jnunOwn)
Seasonal coverage: fair fair fair fair csor
(biased to (biased to (spring 81, (biased to i-*-":-
summr/fal1) sprng/fall) winter 82) summer)
City fraction data: yes sort of yes yet""
Miles/day data: yes yes 1981- no ya»
1982- yes
• model years 1979 and later.
«¦ driving location known only by pnone area code: all other sources (except
DOE) known by ZIP code. There are 878 ZIPs out only 107 area codes.
««• systematic errors in 1979-80 Ford data as received; now corrected. Ford
says 1981 data do not nave this problem
-------
by data source:
Table 2
Quantitative Summary of In-Use MPG Data
Total No. Data Points: 66,518
Ford GM DOE Chrys EPA
37280 15280 &228 2630 2100
by model year:
1975 1976 1977 1978 1979 1980 1981 1982
2246 2160 3424 16207 13077 14303 . 13687 1414
by technology:
RAC
RAI
RAD
RMC
RMI
RMD
49494
1 175
596
6216
349
32
FAC
FAI
FAD
FMC
FMI
FMD
3543
262
1 1 1
4017
436
287
by manufacturer:
Ford
GM
Chrys
AMC
VW+AP
Datsn
41 184
14439
6591
871
836
625
Tyota
Honda
Mazda
Volvo
MerBz
Peugo
587
497
360
1 16
100
58
(balance of 254 divided among 6 others:
BMW, Subru. Fiat. Renau, Saab. Alfa)
by month:
Jan
Feb
Mar
Apr
May
Jun
....
....
1190
5 i 43
2473
1872
3764
4535
Jul
Aug
Sep
Oct
Nov
Dec
7141
2181
523
2626
2167
1 165
2312
"Wi nter" :
7934
"Summer":
13470
"Annua 1
": 8016
Cal if
Texas
NYork
Ohio
Penns
11 1 in
5660
2303
1566
2333
1262
1610
Flori
Michi'
NJers
Georg
NCaro
Indi
1040
30292
695
658
530
797
(balance of 10,218 divided among 38
other states, plus D.C. and P.R.
and 3 Canadian Provinces)
-------
4
The CF/AMPD Data Base
The analysis required determinatIon of
"typical" values of CF and AMPD, to des-
cribe typical city and highway driving for
consumer owned and driven vehicles. This
determination of typical consumer CF and
AMPD values used about 9,000 records of
consumer data from the 43,000-record fuel
economy data base (Ford and Chrysler non-
consumer data were omitted), and some
10,000 records of J.D. Power data, which
had valid consumer CF and AMPD data, al-
though its perceived MPS data is considered
unacceptable for MPS shortfall analysis.
This 19,000-record CF/AMPD data base
was then divided into ninths, based on the
tripartitionlng of the number of data
points for each of the two parameters.
This subdivision of the data base is shown
in Table 3.
The typical values of CF and AMPD were
selected as the average values in two cor-
ners of the matrix: high CF, low AMPD; and
low CF. high AMPD. These average values
are given in Table 4.
The vehicles in each subset were sub-
divided Into one of the twelve technology
groups: 1) Front or Rear Drive, 2) Auto-
matic or Manual Transmission and 3) Carb-
ureted, Fuel Injected or Diesel. For
each of the technology combinations, re-
gression equations were developed relating
their shortfall parameter, GPMR, to five
variables: CF, AMPD, EPA 55/45 MPG, temper-
ature, and odometer. To standardize for
influencing variables, the temperature and
odometer terms were reduced to constants by
using 55 degF and 4000 miles for these vari
ables, and the average value of the EPA
55/45 MPS of each class was substituted,
resulting in class equations containing
three terms: a constant, the CF term, and
the AMPD term.
These class equations were aggregated
by weighting each by its expected percent-
age in the future fleet (1).
The uniform adjustment factors (which
equal 1/GPMR) are plotted as functions of
AMPD, with CF as a parameter, 1n Figure 1.
The CF effect on the highway adjustment fac
tor is too small to show up on the graph.
Table 3
Cutpoints and Data Count for Tripartitioned
CF/AMPD Data Base
AMPD
26 or
39 or
CF
less
27 to 38
more
75
or more
3249
2120
902
31
to 74
1707
2360
2204
30
or less
1314
1791
3165
Table 4
Characteristics of C1ty-Driven
and Highway-Driven Cars
Driving Mode
CF
AMPD
C1ty-Dr1ven
89
17
Highway-Driven
14
62
The analysis of the 43,000-record
fuel economy data base then proceeded as
follows: The data base was divided into
nine subsets using the numerical cutpoints
of CF and AMPD from Table 3.
To retain as much data as possible in
each of the two modal subsets, the "city
4/9" of the data base was used as the
"city" subset and the "highway 4/9" as the
"highway" subset.
These vehicles in each subset repre-
sented different data sources, different
times of the year, and a technological ml*
different from that of the fleet for whicrt
the adjustments would be used. These non-
uniformities were addressed as follows.
DETERMINATION OF ACTUAL AMPD AND CF VALUES
Figure 1 shows how the value of the adjust-
ment factors are influenced by CF and
AMPD. In addition to the data bases that
were used to determine the factors, other
sources of data were Investigated that con-
tained information on why and where driving
is done. The references found to be most
useful were reports (6)-(7) from the 1977
Nationwide Personal Transportation Study
(NPTS).
City Driving
Ref. 6 presents city travel stratified in
several ways. Converting the miles per
year data into miles per day, the average
AMPD for overall driving is 27.9 miles.
The data in reference (6) are stratified by
degree of urbanizat1 on: inside or outside
of Standard Metropolitan Statistical Area
(SMSA). Investigating "city" travel char-
acteristics leads to the choice of "inside
SMSA. within central city". The AMPD for
this type of travel is 27.4.
The value of 27.4 for AMPD character-
istic of city driving represents driving on
all types of roads within a centra 1 city O'
a SMSA. This consists of driving on city
streets - the typical image of "city driv-
ing" - and also driving on freeways in the
central city of a SMSA. Because this free-
way driving is not considered really "city"
in nature, the VM-1 Table for 198! (8) was
used to determine what portion of travel in
urban areas was done on urban interstates;
this value is 0.200.
Therefore the AMPD of 27.4 was correc-
ted by a factor of 0.800 to adjust to
the AMPD most characteristic of city onV-
ing, which 1s 21.9 miles per day. Prom
Figure 1, at 21.9 AMPD and CF ¦ ioo :ie
value for fc is 0.90.
Another way to define city orwi'g is
to extract from each of the surveys tr.at
data which can be said to be "pure cny":
those survey responses indicating cc val-
ues of 100. (Table 5)
-------
5
H
g 0.9
to
3
<
o
9? 0.7
0.6-
5 10
CF=75
CF=89
CF=100
100
AVERAGE MILES PER DAY
Rgure 1 - Adjustment Factors
Table 5
City- Driv.n»
Adjustment Facxo
Data Source
Member Data
Points
Emission Factors
GM 1980
GM 1981
Chrysler 1981
Chrysler 1982
Ford 1979
Ford 1980
Ford 1981
A1 1
280
534
492
117
258
943
679
497
3782
0. 86
0. 87
0. 87
0.87
0. 85
0.89
0.84
0.87
0.87
Highway Driving
The NPTS data do not cover highway driving
In a manner comparable to that of city dri-
ving, which relates to a specifiable loca-
tion. Determination of a "typical" AMPD
for highway driving involves consideration
of the more general parameters of trip
type and trip length. The type of driving
that most would agree corresponds to "high-
way driving" is the vacation trip. In data
on vacation trip length from Reference (7),
vacation driving accounts for 0.1% of trips
and 0.6% of VMT, so it definitely 1s at an
extreme end of the distribution. The over-
all average trip length for vacations is 95
miles, and for certain segments of the pop-
ulation it is higher. Table 6 lists the
average vacation trip length for the high-
est trip length groups, for eight different
ways of stratifying the data.
-------
6
Table 6
Trip Length Data. Vacation Trips
(Average Vacation Trip Length * 95 miles)
Highest Average Stratum
Stratification Trip Length
Ey place of resi- In SMSA, not in centra1!
dence city 121 miles
By SMSA population 250.000 to 500,000
150 mi 1es
By annual household $15,000 to $25,000
i ncome 130 miles
By number of vehi- Three vehicles
cles owned 105 miles
By driver's age
21-25 years 120 miles
By driver's occu- Service work 165 miles
pat ion
By day of week trip Monday 300 miles
began
By hour of day trip 6 AM to 9 AM 141 miles
began
The average of the high-stratum vacation
trip lengths is 154 miles. Assuming con-
servatively that one-way trips of this
length occur in one day, we see from Figure
1 the value for fh at 154 miles per day is
0.79. (The value for fh at the average
vacation AMPD of 95 is 0.77.) If vacation
trips are the definition of highway driving
the highway factor, fh, woula be in the
range of 0.77 to 0.79.
Table 7
Highway Factors Correspond!ng to the
"No City Driving Equals Highway Driving"
Def i n i ton
Data Source
Number Data
Points
Highway
Factor
Emission Factors
56
0 78
GM 1980
GM 1981
383
390
0. 73
0.77
Chrysler 1981
Chrysler 1982
1 1
136
C.75
0. 74
Ford 1979
Ford 1980
Ford 1981
283
12
135
0.83
0. 82
0. 82
A1 1
1407
0. 79
Another possibility for the definition
of highway driving could be those survey re-
sponses for which city fraction was zero.
Using this "no city driving equals highway
driving" definition, the values for fn are
as shown in Table 7.
In the EPA Emission Factors survey, the
respondents were asked to describe their
driving by choices - one of which was "all
highway". So for this survey there is no
need for the assumption that "zero CF"
equals "highway".
DISCUSSION
The choice of adjustment factors depends on
what definitions of city and highway driv-
ing are used. Selection of a definition of
city and highway driving carries with it a
selection of miles per day and city frac-
tion that correspond to that definition.
In some of the approaches used, travel
characteristics entered into Figure 1 yield
adjustment factors.
In other approaches, data yielded ad-
justment factors directly; from these fac-
tors. travel characteristics that corres-
pond to tnese uniform adjustment factors
can be deduced from Figure 1.
In comparing the results of the two
approaches, consistency considerations al-
low the evaluation of definitions of city
and highway driving that yield results com-
patible with both the directly-derived ad-
justment factors from the in-use surveys
and the travel characteristics from the
data in references (6) and (7).
Tables 8 and S summarize tne results
for city and highway adjustment factors.
In the tables, values in parenthesis are
those implied using Figure 1. and values
not in parenthesis are those that come
directly from the data.
With respect to city driving, it ap-
pears that the survey results for implied
values of AMPD of 17. 23, and 27 AMPD are
consistent with the SMSA central city def-
inition of 21.9 AMPD.
It must be pointed out that the actual
data sets yielded lower uniform adjustment
factors (0.86 and 0.87) in Table 8 because
the data sets contain distributions of
vehicle technology different from what is
projected for the future. Because the ad-
justments are to be used for future vehi-
c'es. tne values given by Figure 1, which
are suitably adjusted for future fleet
characteristics, are the ones used.
With respect to highway driving, all
highway uniform adjustment factors are
clustered in a narrow range, 0.75 to 0.79,
(see Table 9) because the highway adjust-
ment is not as sensitive to AMPD as is the
ci ty adjustment.
Implied or actual AMPD values of 90 to
'18 represent a consistent estimate of
hignway driving. The range of adjustments
nere is 0.77 to 0.79.
-------
7
Table 8
Summary of City Adjustment Results
City
Definition of Adjustment City Average Miles
City Driving Factor Fraction
Non-Emission Factors; q S7 100
Emission Factors: 0.86 TOO
SMSA Central City (0.90)
"Typical" AMPD (0.90)
Per Day
(17)
"AU City" (0.92) 100 27
(14)
"AH City" (0.91 ) 100 23
100 22
89 17
Table 9
Summary of Highway Adjustment Results
Definition of
Highway Driving
H1ghway
Adjustment
Factor
Average Miles
Per Day
High-stratum
vacation trip
Non-Emission Factors:
"Zero City"
Emission Factors:
"All Highway"
Average vacation
trip
"Typ i cal " AMPD
(0.79)
0.79
(0.79)
0.78
(0.77)
(0.77)
(0.75)
154
(120)
1 18
(100)
90
95
82
CONCLUSIONS
1. Si-modal adjustment factors for EPA
City and Highway MPG values have been
shown to be sensitive to the definition
of city and highway driving.
2. Survey results and travel character 1s-
tlcs data agree on the nature of city
and highway driving. City driving iS
characterized by AMPD values within 5
of 22 miles per day; highway driving
is characterized by AMPD values within
15 of 105 miles per day.
The best uniform city adjustment factor
is 0.90 and the best uniform nigrn«ay
adjustment factor is 0.78.
-------
REFERENCES
(1) "Why Vehicles Don't Achieve the EPA MPG
on the Road and How That Shortfall Can
be Accounted For", SAE Paper 820791,
K.H. Hellman and J.D. Murrell, June
, 1982.
(2) "Fuel Economy Ratings vs. Road Experi-
ence - An Analysis of Ford's 1982 Lease
Fleet", SAE Paper 831034, D.L. Kulp and
J.C. McKenna, June 1983.
(3) "In-Use Fuel Economy of 1981 Passenger
Cars", SAE Paper 820790, R.W. Schnei-
der, W.S. Freas, and T.P. McMahon,
June 1982.
(4) "Comparison of EPA and On-Road Fuel
Economy - Analysis Approaches, Trends,
and Impacts", SAE Paper 820788, B.D.
McNutt, R. Dulla, R. Crawford, H.T.
McAdams, and N. Morse, June 1982.
(5) "Fuel Economy of Motor Venicles: Revis-
ions To Improve Fuel Economy Labeling
and the Fuel Economy Data Base", Fede-
ral Register, 48 FR 26698-26717, June
9, 1983.
(6) "Household Vehicle Utilization", Report
No. 5 of the 1977 Nationwide Personal
Transportation Study, U.S. Department
of Transportation, FHwA/PL/81/011,
Apri1 1981.
(7) "Purposes of Vehicle Trips and Travel"
Report No. 3 of the 1977 Nationwide
Personal Transportation Study, U.S.
Department of Transportation, FHwA/PL/
81/001, December 1980.
(8) Annual Vehicle Miles of Travel and
Related Data - 1981, by Highway Cate-
gory and Vehicle Type", Table VM-1,
U.S. Department of Transportation,
September 1982 (Preliminary).
------- |