PB82-126111
PRELIMINARY CALCULATION OF FUEL ECONOMY ADJUSTMENT
FACTORS. (TECHNICAL REPORT. )
Environmental Protection Agency
Ann Arbor, MI
Jul 81
J
U.S. Department of Commerce
National Technical Information Service
NTT1S
-------
EPA/AA/CTAB/FE-81-17
Technical Support Reporc for Regulatory Action
Preliminary Calculation of Fuel Economy
Adjustment factors
July, 1981
Notice
Technical support reports for regulatory action do not necessarily
represent the final EPA decision on regulatory issues. They are
intended to present a technical analysis of issues and recom-
mendations resulting from the assumptions and constraints of that
analysis. Agency policy constraints or data received subsequent
to the date of release of this report may alter the recom-
mendations reached. Readers are cautioned to seek the latest
analysis from EPA before using the information contained herein.
Control Technology Assessment and Characterization Branch
Emission Control Technology Division
Office of Mobile Source Air Pollution Control
Office of Air, Noise and Radiation
U.S. Environmental Protection Agency
2565 Plymouth Road
Ann Arbor, Michigan 48105
-------
TECHNICAL REPORT DATA
(Please read Intinicrions on the reverse before completing)
1. REPORT MO.
EPA/AA/CTAB/FE-81-17
3. RECIPIENT'S ACCESSION-NO.
PM1 12611 1
•J. TITLE AND SUBTITLE
Preliminary Calculation of Fuel Economy Adjustment
Factors.
5. REPORT DATE
July 1981
-,A n
J I U
6. PERFORMING ORGANIZATION CODE
7. AUTHORISI
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
United States Environmental Protection Agency
Office of Mobile Source Air Pollution Control
Control Technology Assessment & Characterization Branch
Ann Arbor, MI 48105
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
:3. TYPE OF REPORT AND PERIOD COVERED
Technical
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT
This report is a dual purpose report: 1) it is a cover report for EPA
1'inal report 460/3-81-003, "Development of Adjustment Factors for On-Road Fuel
Economy" and 2) it is a preliminary determination of fuel economy adjustment
factors in support of on-going fuel economy rulemaking activity.
In is role as a cover for the EPA report, it extracts key findings, critiques
the findings and offers modifications to the findings where there is reason to
adopt alternate assumptions. In its rulemaking support role as an initial
estimate of MPG adjustment factors, it compares the EEA (Energy And Environmental
Analysis) results with those of other recent, significant publications in this area.
The resulting preliminary adjustment factors herein are thus an amalgam of the
factors derived by these four data sources.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Fuel Consumption
Computation
Motor Vehicles
Fuel Economy
Calculation
13. DISTRIBUTION STATEMENT
release unlimited
19. SECURITY CLASS (This Report/
unclassified
21. NO. OF PAGES
45
20. SECURITY CLASS (This page/
unclassified
22. PRICE
EPA Form 2220-1 (9-73)
-------
ABSTRACT
This is a dual purpose report:
'a) It Is a cover report for Contractor Final Report EPA-460/3-81-003,
"Development of Adjustment ?actors for On-Road ?uel Economy", March
1981, by Energy and Environmental Analysis, Inc., (EEA);
(b) It is a preliminary determination of fuel economy adjustment factors in
support of on-going fuel economy rulemaking activity.
In ita role as a cover for the EEA report, it extracts key findings In that
report and che assumptions associated with them, critiques those findings and
assumptions, and offers modifications to the findings where there is reason to
adopt alternate assumptions. This is not done "in a vacuum", however. In its
rulemaking support role as an initial estimate of MPG adjustment factors, it
compares the EEA results with those of other recent, significant publications
in this area. The resulting preliminary adjustment factors herein are thus an
amalgam of the factors derived by these four data sources. Data analysis is
continuing.
-------
-3-
I. BACKGROUND and PURPOSE
There are currently four major "collection agents" for data on the In-use fuel
economy of motor vehicles:
a) EPA's Emission Control Technology Division
b) DOE's Office of Policy Evaluation
c) General Motors Technical ("-enter
d) ford Motor Co. Fuel Economy Planning
EPA-ECTD, primarily through its Emission factors and Inspection and
Maintenance Programs, has been gathering in-use MPG data via owner
questionnaires for several years, and has accumulated over 10,000 data points
on owner-perceived MPG of cars and light trucks from model years 1975 through
1981. Recently, in-use data on a subpopulation of the questionnaire-surveyed
Emission factors owners has been collected using postcard-type fuel economy
diaries, which give measurements of miles traveled and gallons of fuel for
several successive fuel purchases. The vehicles surveyed by EPA-ECTD were
localized in the vicinities of the four to six commercial test labs contracted
by EPA to administer the Emission factors and I/M Programs.
DOE has become the ad_ hoc national center for in-use vehicle MPG data,
acquiring since 1977 a total of well over 50,000 data points on vehicles from
the 1975 through 1980 model years. This data base covers a large number and
variety of sources, including the aforementioned EPA and auto company surveys,
along with those of oil and chemical companies, non-auto/oil light industry
firms, utilities, commercial fleet operators, commercial collectors of auto
maintenance data, and a host of state agencies who operate fleets. In
addition to collecting, systematizing, and analyzing others' data, DOE has
sponsored in-use MPG surveys of its own.
General Motors has surveyed the in-use MPG of a total of more than 6,000
General Motors cars from the 1975, 1976, and 1978 model years, and recently
surveyed approximately 5000 model year 1980 in-use cars from thirteen
manufacturers. These GM surveys were nationwide and generally multi-seasonal,
and employed postcard diaries for logging in-use mileage and gallonage.
lord's in-use MPG surveys are restricted to the more recent model years, but
are respectably prolific nonetheless. Since 1978, rord has collected postcard
diary data on 29,000 cord vehicles from model years 1978 through 1980. These
vehicles were leased and operated by .7ord Motor Co. employees in many parts of
the country and over all parts of the year. Most of the data, however,
represents vehicle operation in the Detroit,. Michigan area.
-------
-4-
Some parts of these data bases were examined by EPA to illustrate approaches
to road adjustment factors in Technical Support Report for Regulatory Action
EPA/AA/CTAB/FE-81-6,"Light Duty Vehicle ?uel Economy Labeling", October 1980.
In the total universe of analyses that will lead to development of road
factors for the MPG labeling rulemaking, that report represented Round One.
The EEA report is considered Round Two. It is a thorough, detailed analysis
of the DOE data base (which included some of the ECTD data, some of the GM
data, and some of the ?ord data at the time); it perfected most of the
analytical approaches that will be used in the final rule support analysis,
and — within the limits of its data base and assumptions — began to zero in
on the quantitative ballpark for road adjustment factors.
The final rule support analysis, Round Three, will use all of the data from
all of the aforementioned sources, plus any data supplied to EPA in the rule
making process, and will of course be performed in-house. The data now in
hand are being cleaned and characterized. The data base has at this writing
been augmented with climatic, demographic, and topographic data to permit
careful examination of the distributions of fuel economy influences within and
among the data sources. The data are now being assigned EPA MPG values to
permit calculation of the "in-use shortfall". When this is completed, the
full-scale analysis and determination of road factors will commence.
Pending completion of these analyses, however, it is necessary to identify the
approximate range of values within which some forms of the final road
adjustment relationships are expected to fall, so that all interested parties
can respond from an informed standpoint to the Notice of Proposed Rulemaking
for fuel economy labeling soon to be released. This report, then, provides
that estimate of the probable range of road adjustment factors. It would be
considered Round Two and a_ Half. It makes that estimate by examining [not
data, but] reported results of analyses of data, from the four major in-use
data sources. In so doing, it reviews the EEA Report results along with
reports of analyses by the other three sources.
-------
-5-
II. DISCUSSION
A. Data Sources
Table 1 lists Che four reports which are the subject of this report. All are
recent publicatons.
Details of each of the data bases used for the four reports' analyses appear
in Table 2, and a narrative summary of the distinguishing features of the four
surveys is given in Table 3.
None of the four data bases is considered "ideal" in all respects. The 1980
GM Survey might be evaluated as nearly ideal based on tables 2 and 3, but two
aspects of that survey not addressed in these tables are worth mentioning:
(a) The overall response rate was quite low (11.62 of mailed diar^^s were
returned; after screening, 79% of these survived, leaving a net usable
fraction of 9.2Z); this leaves the possibility of non-response bias in MPG,
which was not evaluated — although the distribution of vehicle types in the
returned sample was quite representative of the model year 1980 vehicle
population. (b) While the GM cars in the survey were sampled from GM sales
records, covering all states, the other manufacturers' cars were sampled via
R.L. Polk registration data, which excluded 15 states. Most notable among the
excluded states are Massachusetts, Connecticut, New Jersey, and Pennsylvania,
which together account for most of the VMT in the Northeast.
B. Single Mode and Dual Mode Stratagies
At present, EPA is considering two alternative strategies for fuel economy
labeling: a labeling system based on one MPG number, and a system based on two
MPG numbers. The one-number system would use the EPA combined (55M5) MPG
value as the basis for road adjustment ; the two-number system would use the
EPA City and EPA Highway MPG values as the bases for adjustment.
The general case of the adjustment relationship for both strategies is an
equation of the form "Label MPG equals a function of EPA MPG". Based on
linear regression analysis in GPM ratio versus EPA MPG space, as discussed in
detail in the EEA report, the general equation form is "Label MPG equals EPA
MPG/'a + b EPA MPG)". For the final rule, the generalized case will be
analyzed in detail.
The next section, Section C, addresses the one-number 'single-mode) system.
Almost entirely, it addresses the simple case of no dependence of the
shortfall on EPA MPG level (b=o). This results in a .simplified equation form:
"Label MPG - K EPA MPG" (K - I/a).
The section after that, Section D, addreses the two-number (dual-mode) system;
it discusses some results in terms of terms of the "simple multiplicative
factor" form of the equation, and some results in terms of the MPG-dependent
form.
-------
-6-
Table 1
Data Sources
"EEA": Energy and Environmental Analysis, Inc.,
"Development of Adjustment Factors for
On-Road Fuel Economy", EPA-460/3-81-003,
March 1981.
'1980 Ford": N. E. South, "1978 to 1980 ?ord On-Road Fuel
Economy", SAE Paper 810383, February 1981.
'1980 GM": R.W. Schneider e£ al, "In-Use Fuel Economy of
1980 Passenger Cars", SAE Paper 810384,
February 1981.
'EPA Emiss.Fact": B. Bradley e£ al, "Fuel Economy of In-Use
Passenger Cars: Laboratory and Road", SAE
Paper 810780, June 1981.
-------
-7-
Table 2
In-Use F. E. Data Bases: EEA and Three Other Sources
Survey Characteristics
Model Years
Manufacturers
Geographic Spread
Seasonal Spread
Total No. Data Points
Average EPA Comb. MPG
Vehicle Characteristics
Car Size
Engine Size
Manual Transm.
Front Drive
Diesel
Air Conditioning
Avg. Odometer
Driving Characteristics
Consumer Use, Measured
Consumer Use, Perceived
Fleet Use, assumed meas
Avg. Miles per Day
Fraction City Driving
Overall Road MPG Factor
(vs. EPA Combined MPG)
EEA
1975-80
95% Domestic
Yes
Yes
25,239£/
19.6
60% mid-Lge
55% 8-Cyl
13%
5%
1%
7
8,000l/
. /
40%!'
40%d/
20%d/
7
7
0.90
1980
Ford
1980£/
All Ford
Mostly Detroit
Yes
8,689
21.0
7
?
20%
5%
None
?
3450
100%
—
—
52
54%
0.86
1980
Gen Mtrs
I980b/
76% Domestic
Yes
Mostly Fall
4,871
21.6
7
7
33%
23%
5%
7
3958
100%
—
—
45
53%
0.84
ay Ford also ran surveys in 1978-79, Ford
vehicles only.
b/ GM also ran surveys in 1975, 76, and 78, GM
vehicles only.
c_/ Includes the earlier Ford & GM data, and
also EPA Emiss. Fact, and I/M questioriaire
(perceived; MPG data.
d/ Estimated.
e/ High odometer and temp. should make it
higher; extensive A/C use, city driving
fraction, and population density should make
it lower; effects apparently cancel.
EPA
Emis.Fact
1975-80
81% Domestic
Mostly Houston
Mostly Summer
440
20.0
51% Mid-Lge
46% 8-Cyl.
24%
16%
None
92%
24,000
100%
33
67%
0.83£/
-------
-8-
Table 3
Comparison of Data Bases
EEA: Multi-manufaccurer
Most: data points, good geographic & seasonal spreads
Smaller cars under-represented
Imports underrepresented
"Fuel Efficient" technologies underrepresented
60% of data does not represent measured, consumer use
Overall road factor 4Z-7Z higher than other surveys
1980 Ford: Only Ford vehicles
Southeastern Michigan » locale for most driving
Front drives and Diesels underrepresented
All data measured, consumer use (Ford employees)
Highest miles per day
Lowest Avg. odometer
1980 GM: Multi-manufacturer
Reasonably high import percentage
Most data represents October & November
All data measured, consumer use (private owners)
Fuel efficient technologies well represented
Avg. miles per day and city fraction are typical
EPA Emis. Fact: Very small data base
Multi-manufacturer
Reasonable import percentage
All data measured (private owners)
Driving highly localized in geography & season
Manuals & front drives well-represented; no Diesels
Highest odometer average
Low miles per day, high city fraction
92% of vehicles were air-conditioned
-------
-9-
C. Single Mode Analysis
Table 4 summarizes Che four references' findings on single-mode road factors.
Even wichin a given reference, "the answer" is not a fixed road factor, but
depends upon how it is calculated. Using EEA as an e::ample, if one takes the
model year-specific average on-road MPG values (R) and average EPA MPG values
^E) from EEA's table 3-4, the model year sample weightings from their table
3-3, calculates a total sample R of 16.5 and a total sample E of 19.6 and
takes the ratio of these values, a road factor of 0.84 is obtained. But if
one enters an E value of 19.6 in the EEA MPG-dependent algorithm, this
produces an E/R ratio of 1.108, yielding a road factor (R/E) of 1/1.108 -
0.90. Thirdly, if the E/R ratio is taken for each vehicle in the data base
and averaged, a figure of 1.109 results; the road factor from this method is
then 1/1.109 - 0.90*.
Also shown in Table 4 is the fact that MPG dependence of the shortfall, at the
fleet level of aggregation, is not the same for the three sources which
evaluated it. for the EEA report, a 10 MPG increase in EPA fuel economy gives
a 3.1% increase in road GPM ratio; for Ford, it is a 1.2% increase, and for GM
it is a 4.9% decrease. This apparent inconsistency between sources can be due
to a number of factors, some of which could be or could have been adjusted
for, and some which cannot/could not.
One of these factors is the proportion of various unique-shortfall vehicle
technologies in the respective survey populations. Given the wide variance in
representation of various vehicle technology types among the four surveys
(Table 2), technology specific shortfalls and weightings should be taken into
account. Even if a road adjustment factor at the fully-aggregated level is
all that is sought, the development of that one factor from these sources'
results still requires disassembly of their findings into specific technology
strata and reassembly using appropriate weightings of those strata.
Remember: The ultimate use of road factors will be for post-1982 vehicle
fleets, not on fleets with the technology mixes of these surveys.
Table 5 gives the road factors for twelve technology strata, as developed by
the four analyses.
* The EEA report does not include this calculation. It was
furnished to EPA after publication.
-------
-10-
Survey .\vcrage£'
MPG Dependence:
- Method
Table 4
Single-Mode Road factors
1980
,7ord
1980
Gen Mtrs
EPA
Emis.Fact
ERA
.84-. 90
- Shortfall increases/decreases
with higher EPA MPG
.84-. 86
Increases:
f- .915
-------
-11-
Table 5
Vehicle Technologies & Road factors: Four Data Sources
Vehicle Technology
RWD/Auto/Carb
RWD/Manu/Carb
RWD/Auto/FI
RWD/Manu/FI
RWD/Auto/Dsl
RWD/Manu/Dsl
FWD/Auto/Oarb
fWD/Manu/Carb
FWD/Auco/FI
rWD/Manu/FI
FWD/Auto/Dsl
rWD/Manu/Dsl
1980
EEA Ford
201277. 865£/ 6918/.8261/
2154/.888 1360/. 867
2637. 891
(2 cars)
2137. 898
(40 cars)
1687.975
6457.982 4H/.966
(19 cars)
12W.991
(2 cars)
727.983
1980 EPA
Gen Mtrs Emis.Fact
25397 800£/ 3C1/.798
7297.887 597.818
(4 cars)
(7 cars)
2017. 879
(23 cars)
7317.855 (15 cars)
5277. 975£/ (28 cars)
(13 cars)
— (13 cars)
667.849
557.949
aj x/y denotes no. cars/road factor
b/ may include some rl; does include 3159 overdrives
c/ may include some FI.
-------
-12-
The "Acceptable Stratum Sample Size" Issue
Also shown in cable 5 are the sample sizes associated with the technology
strata. We have not shown road factors for strata with sample sizes less than
approximately 50*.
The "Which Vehicle Technology Weightings to Use" Issue
Table 6 shows projected sales penetrations for the twelve technology strata
for the 1981 model year, the latest model year for which we have reliable
predictions. At the very least, "new-fuel efficient" technologies should be
represented at these sales fractions in re-aggregating technology specific
shortfall observations.
Other Issues . Survey findings Weighted Equally, not by Sample Size
Pooling of FI and Diesel Road Factors
Retain or Discard Obsolescent Technologies
In Table 7, the aforementioned source/technology road factors and strata
weightings are brought together. ?or each stratum, one road factor was
determined by combining the sources' individual road factors, equally
weighted, consumption-combined (i.e. harmonically averaged). This approach
was used rather than sample weighting. The rationale is illustrated by the
following analogy. If four inspectors using identical copies of an NBS
standard yardstick measure N]_, N£, N*3, and N4 widgets, the average
widget length vould, by all means, properly be calculated as a weighted
average, using the individual inspectors' average measurements and their N^
weights. However, if they use four hand-crafted but uncalibrated yardsticks,
sample-weighting does not a_ priori lead to the best estimate of average widget
length. The measurements made by inspector //I should not dominate an
aggregate calculation just because he had a good day and measured more widgets
than the other inspectors. So it is with in-use fuel economy surveys. Until
there is a precise standard procedure (that is, a science) for conducting such
surveys, the efforts of quite competent surveyors will feature
distinguishable, individualistic traits (that is, art). Sample size, then, is
used only to determine a threshold of acceptability (50 cars for this
analysis), not as a weighting factor when aggregating across data sources.
To estimate MPG within a precision of +_ 1 MPG in the neighborhood of 20
MPG (+5% precision), with a coefficient of variation of 20%, at a
confidence level of 90%, a sample size of 45 is required; for 95%
confidence, a sample size of 64 is required (from P2 - Z2 COV2 (1/N)
and t-tables).
-------
-13-
Table 6
Vehicle Technologies: Sales Fraction "1981) and MPG Range
Drive
Rear
Rear
Rear
Rear
Rear
Rear
Front
Front
Front
Front
Front
Front
Transm.
Auto
Auto
Auto
Manu
Manu
Manu
Auto
Auto
Auto
Manu
Manu
Manu
Engine
Garb
FI
Diesel
Garb
Fl
Diesel
Garb
FI
Diesel
Garb
FI
Diesel
% Sales
43.6
3.2
3.4
9.5
2.4
0.1
18.8
2.0
0.6
13.4
1.8
1.2
EPA
Combined
MPG
12-32
10-28
25-29
12-45
13-32
30-31
18-34
17-29
23-25
24-42
23-33
41-47
Comments :
Transmission
>50Z Lockups
Mostly non-lock, non-OD
Mostly lockup
-50/50 OD/non-OD
Mostly OD
Mostly non-OD
Mostly non-OD
No OD
No OD
Mostly 'OD
Mostly OD
All OD
Note - Diesels:
Each of :h«; *. strata has narrow EPA MPG range,
and the 4 MP& ranges are mutually-exclusive
(non-overlapping)
Note - FWD Automatics: There are no lockups in these strata in 1981.
-------
-14-
Table 7
Sales Significance and Road Factors for
Technology Strata and Upward Aggregations of Them
RWD-Auto (sales)
(road factor)£/
RWD-Manu (sales)
(road factor)
FWD-Auto (sales)
(road factor)
FWD-Manu (sales)
(road factor)
All RWD (sales)
(road factor)
All FWD (sales)
(road factor)
All Auto (sales)
(road factor)
All Manu (sales)
(road factor)
All (sales)
(road factor)
Carb
.436
.82
.095
.86
.188
.92
.134
.97
.531
.83
.322
.94
.624
.85
.229
.92
.853
.87
FI
.032
.89
.024
— —
.020
— —
.018
.99
.056
.89
.038
.99
.052
.89
.042
.99
.094
.93
Diesel
.034
.89
.001
— —
.006
•85-/
.012
.97
.035
.89
.018
.92
.040
.88
.013
.97
.053
.90
All
.502
.33
.120
.86
.214
.91
.164
.98
.622
.84
.378
.94
.716
.85
.284
.93
1.000
.87
All Gasoline
(Carb + FI)
.468
.83
.119
.86
.208
.91
.152
.98
.587
.83
.360
.94
.676
.85
.271
.93
.947
.87
All FI
(FI + Dsl)
.066
.89
.025
— —
.026
•85-/
.030
.98
.091
.89
.056
.97
.092
.88
.055
.98
.147
.92
a/ Road factor • simple average of the
applicable surveys' road factors;
all source/strata with >50 cars
are counted.
b/ Obsolescent version of FWD-Auto technology?
-------
-15-
The combining of technology straca in Table 7 was done via sales-weighted har-
monic averaging.
Noting more similarity between the road factors of gasoline FI and Diesel cars
than between those of gasoline fl and carbureted gasoline cars, a pooled road
factor was calculated for both gasoline fuel injected and Diesel cars.
It should be noted that the road factor for front drive automatic Diesels
comes mainly from cars whose design r'-.r«..
-------
-16-
Table 8
Summary - Constant Road factors, One Mode
No. Straca
Road MPG
EPA Combined MPG
Road Factor(s)
ONE:
0.87
FOUR-
(a) ?WD .94 RWD .84
(b) Auto .85 Manu .93
(c) Garb .87 FI- .92
'a) RWD-Auto .83 RWD-Manu .86 FWD-Auto .91 FWD-Manu .93
(b) RWD-Carb .83 RWD-r"I .89 i?WD-Carb .94 fWD-FI .97
^c) Auco-Carb .85 Auto-fl .88 Manu-Carb .92 Manu-FI .98
EIGHT
RWD-Auto-Carb .82
RWD-Manu-Carb .86
FWD-Auto-Carb .92
?WD-Manu-Carb .97
RWD-Auto-FI .89
RWD-Manu-FI (?)
d/
FWD-Manu-FI .98
a/ There are chree cwo-scraca opcions
b/ FI means gasoline FI and Diesel pooled
£/ There are chree four—scrata options
d/ Obsolescent version of i?WD-Auto technology?
-------
-17-
Table 9
Apparent Effect of Combustion/Carburetion Technology on Road Factor*
(Strata >50 cars)
Gas FI vs. Carb.
Diesel vs. Garb.
Diesel vs. Gas FI
D =• DOE/EEA
E =» EPA Emis. Fact
F = Ford
G =• GM
RWD-Auto
D +3Z
G +8%
D +1%
RWD-Manu
FWD-Auto
G -IX
FWD-Manu
D -H2
D 0%
G -3%
D -1%
For example, in the GM survey, RWD-Auto Diesels had an 8 per cent higher
Road Factor than did RWD-Auto Carbureted vehicles.
-------
-18-
Table 10
Apparent Effect of Drive Train Technology on Road Factor
(Strata >50 cars)
Manu vs. Auto, RWD
Garb Cars
D +2% F +4%
E +2% G'+9%
FI Cars
Diesel Cars
Manu vs. Auto, FWD D +1% G +12%
FWD vs. RWD, Auto D +11% G +6%
FWD vs. RWD, Manu D +9% F +10%
G +9%
G +10%
G -3%
D = DOE/EEA
E = EPA Emis. Fact,
F = Ford
G =» GM
-------
-19-
Table 11
MPG Dependency of GPM Ratio, evaluated at the Technology Stracum Level
Range of MPG
coefficients
EEA GM£/ EmFacJi/
.0033-.0159 .0023-.0082 .0030-.0156
Average of MPG
coefficients
.009
.006
.006
Trend?!/
No
No
b/
not counting one negative slope known to be
influenced by spurious cars (Cadillacs) and
mixing of Garb and ?I cars
not counting one negative slope based on
15 cars (R2 = 0.017) in an MPG range that is
too narrow to really develop an MPG dependence
£'i.e. does the GPMR slope increase, or decrease,
consistently with MPG level?
-------
-20-
Histogram Tests of Single-Mode Road Adjustment Systems
Using Che adjustment factors derived above, the effectiveness of road
adjustment at various levels of technology aggregation was tested. The
Emission Factors Postcard data base (the only one fully accessible for
computer analysis at the time) was the test "subject".
The test consisted of multiplying each car's EPA combined MPG (Eo) by the
appropriate road factor, taking the difference between this road-adjusted
value (£') and actual road MPG, and histographing this difference.
Figure 1 is the histogram for road A MPG using unadjusted EPA MPG as the
reference. Figure 2 is the road A MPG histogram using the single-factor (no
strata) adjusted EPA MPG as the reference. The effect of the adjustment in
moving the distribution toward zero average A MPG is obvious.
Table 12 summarizes the effects of the various road adjustment schemes.
Conclusions that can be drawn from this test, in order of increasing
subtlety/complexity, are:
Road adjustment unquestionably produces label values that reflect
in-use average MPG better than the unadjusted EPA 55/45 values.
There is an average error of -3.6 MPG using raw EPA numbers, while
average error .Is reduced to less than 1 MPG with road-adjusted numbers.
Unadjusted EPA numbers, even with a 2 MPG discount, are too high for
two-thirds of the in-use cars; however—adjusted numbers, with a +2 MPG
window, accurately reflect the road MPG experience of two-thirds of the
population.
In this particular test, stratification produces no significant
improvement in the adjusted label values' road representativeness, on
the average.
The primary reason for the previous conclusion is that the four-source
road factors are not the best set of factors to try on only one of the
sources' data. The four-source no-strata adjustment factor is 0.87,
while the appropriate factor for the Emission Factors data base is
0.83. Obviously, the 0.87 factor is not stringent enough, as verified
by the consistently negative residual MPG errors remaining after road
adjustment.
The distribution inclusion/exclusion results of this test also show
little or no improvement for stratification compared to a single
fleetwide adjustment factor.
The primary reason for this is that this one influence, vehicle
technology, is responsible for only a fraction of the total in-use
variation; all of the other influences (driving habits, etc., etc.)
still contribute to the dispersions in the in-use data.
-------
-21-
Figure 1
Difference: EPA 55/45 MFC minus Road MFC
:DPOINT HIST* COUNT
:o
9
8
•?
6
.000
.000
.000
.000
.000
5.000
4
3
n
1
0
.
{ t
,
. ,
•
.
,
.
,
,
,
,
n
.
'.
.
.
,
0
.000
.000
.000
.000
.000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
.000
0.
0.
0.
0.
.
0.
,
0.
,
1 .
.
2.
1.
7.
8.
14.
14.
16.
1 1 .
10.
3.
4.
1.
1.
,
0.
0.
0.
0.
0.
0.
2
5
7
1
5
0
6
0
2
5
3
4
4
7
6
1
1
1
9
0
0
0
0
1
0
2
0
3
5
o
i.
1
7
31
36
64
63
72
50
47
16
18
5
5
4
0
0
0
0
0
0
+
•f
t
•f
+x
+
+ XX
+
+ XXX
+XXXXX
+ XX
+XXXXXXXXX
+XXXXXXX
mxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
+XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
+XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
+XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
+XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
+XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
+XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
+XXXXXXXXXXXXXXXX
mxxxxxxxxaxxxxxx
+XXXXX
+XXXXX
+ XXXX
+
+
*•
+ <
+
+
A
/%
/?//
yV° &/
/t>^f/
Aw
/ o -i6/
/Jl •»/
&\y
\ X) •/
ft
w
/
OTAL
440
-------
-22-
Figure 2
Difference: Adjusted EPA 55/45 MPG minus Road MPG
DPOINT HISTZ COUNT
5.000 0. 0 +
4.000 0. 0 +
3.000 0. 0 +
2.000 0. 0 +•
1 .000 0. 0 +
0.000 .2 1 + X
.0000 .7 3 + XXX
.0000 .5 2 + XX
.0000 1.1 5 +XXXXX
.0000 1 .6 7 +XXXXXXX
.0000 4.1 18 +XXXXXXXXXXXXXXXXXX
.0000 5.0 22 +xxxxxx*xxxxxxmxxxxxx
.0000 10.5 46 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX.XX
.0000 14.4 72 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
.0000 18.2 80 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
12.0 53 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
.0000 13.2 58 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
.0000 6.6 29 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXX
.0000 4.1 18 +XXXXXXXXXXXXXXXXXX
.0000 3.0 13 +XXXXXXXXXXXXX
.0000 .9 4 +XXXX
.0000 .7 3 +XXX
.0000 .7 3 +XXX
.0000 .7 3 +XXX
.0000 0. 0 +
D.OOO 0. 0 +
1.000 0. 0 +
2.000 0. 0 *
3.000 0. 0 +
4.000 0. 0 +
5.000 0. 0 +
]TAL 440
-------
Number of
Strata
1
1
2
2
2
4
4
4
8
Results of
Road Adjustment
Strata Adjusted
No adjustment
Fleet adjustment
FWD/RWD
Auto/Manual
Carb/FI
R-A/R-M/F-A/F-M
R-C/R-FI/F-C/F-FI
A-C/A-FI/M-C/M-FI
Each Technology
Histogram
Factors:
(One-Mode
Avg. Error
(MPG)
-3.6
-0.8
-0.6
-0.9
-0.9
-0.6
-0.5
-0.9
-0.5
-23-
Table 12
Tests of Four-Source
Test Subject - Emis.
Adjustment System)
Composite
Fac. Data Base
X of Population
Below Within 2 MPG of E1
67
24
20
25
25
22
20
25
19
31
66
68
65
66
68
68
66
69
Above
2
10
11
10
9
11
12
9
12
-------
-24-
D. Dual-Mode Analysis
The dual-mode analysis chat can be mined from che chree non-EEA sources is
narrower in scope Chan Che foregoing single-mode analysis:
The ?ord paper did noc separate "ciCy-driven" cars and "highway-
driven" cars and develop mode-specific road factors; it did com-
pare all driving modes' aggregate MPG results with the EPA City
value (R - 1.01 Ec, a 1% overage) and with the EPA Highway value
(A - O.d8 Eg, a 32% shortfall); it also showed the percent of
all of the sample cars' MPG experience that would be captured by
two-number ranges built from several combinacions of City and High-
way adjustment factors.
The Emission factors paper did show road factors for city-driven
and highway-driven cars for the sample fleet, but not for vehicle
technology straca; like the ford report, it also compared all of
the data to the EPA city value (R - 0.96 Ec, a 4Z shortfall) and
to the EPA Highway value (R - 0.68 EH, a 32% shortfall).
The GM paper compared overall fleet MPG to the EPA City value (R »
0.95 Ec, a 5% shortfall), but not to the Highway value. The GM
paper also gave technology-specific equations involving city
driving fraction and the EPA City value, from which road factors
for city-driven cars of the specific technologies can be derived.
Table 13 lists the two-mode road factor results from the cord, GM, and Emis-
sion Factors papers.
The EEA report presented an extensive two-mode analysis, including MPG de-
pendence and technology stratification, and using two versions of two-mode
analysis on the more highly-populated cechnology strata. The downside of
EEA's cwo-mode analysis rescs with its use of only perceived MPG data from
the 1978-79 J.D. Power survey.
Table 14 gives the equations for MPG-dependent City and Highway GPM ratios
from the EEA report, for conventional technology (carburetted, automatic
transmission) vehicles. The A and B sets of equations come from the two
alternate analysis techniques used by EEA to develop dual-mode factors. The
techniques are described in the report, and will not be discussed further
here except to point,out that one technique calculates road factors directly
by using only city-driven cars and highway—driven cars, and the other (more
complex) technique uses all of the data and the full range of city driving
fractions.
Dual-mode road factors for alternative technologies were determined in the
"constant" form (MPG independent); they appear in Table 15. The factors
shown as estimated (type B) do not exactly match those calculated by EEA for
the two technologies shown; they are EPA estimates based on EEA report data
but on different assumptions than those used by EEA. Two points should be
noted about these estimates:
-------
-25-
Table 13
Two-Mode Road Adjustment ?accors from Non-EEA Sources
A. Ford Motor Co.
City Adjustment
factor
Highway Adjustment
factor
Z Cars Included!/
Between the Adjusted
Numbers
0.79
0.77
0.74
0.69
0.81
0.82
0.84
0.88
80Z
85Z
90*
952
a/ Excluded cars are equally distributed high and low.
B. Emission Factors
Type of
Driving
All City
Mostly City
Both
All Highway
Mostly Highway
Both
% of
Sample
25%
32Z
57Z
2%
10%
12%
City Adj
factor
.88
.93
.91
—
—
Highway Adj,
cactor
.79
.73
.74
C.
General Motors
b/
City Adjustment factors, by Technology —
rtWD-Auto-Gas,
RWD-Manu-Gas,
RWD-Auto-Dsl,
.89
.92
.95
FWD-Auto-Gas,
FWD-Manu-Gas,
fWD-Auto-Dsl,
FUD-Manu-Dsl,
.95
.97
.93
1.03
W From the paper1 s table 12; all technologies evaluated at 100Z city
friction, 44.6 miles/day, 3958 odometer miles, and 54.6°?. These
factors are MPG-dependent.
-------
-26-
Table 14
Two-Mode Road Adjustment Algorithms from Che EEA Report
(Conventional Technology vehicles)
Using subsets of pure mode-driven cars (46% of the data)
City GPM ratio*:
Hwy GPM ratio:
B. Using all of the data
City GPM ratio:
Hwy GPM ratio:
RWD-Auto-Carb (1978 + 1979), 0.782 + .0181E
c
RWD-Manu-Carb (1978 + 1979), 0.717 + .0134E
RWD-Auto-Carb (1978 + 1979), 1.130 -I- .0097Et
RWD-Manu-Carb (1978 + 1979), 1.017 + .0097E,
H
RWD-Auco-Carb (1978), .845 + .0136E
c
(1979), .963 -1- .0091E
RWD-Manu-Carb (1978), .591 + .0208E
(1979), .557 + .0166E
RWD-Auto-Carb (1978), 1.127 -I- .0096E
(1979), 1.144 + .0088E
RWD-Manu-Carb (1978), 1.312 (no MPG dependence)
(1979), 1.343 (no MPG dependence)
road MPG adjustment factor is the inverse of GPM ratio.
-------
-27-
Table 15
Two-mode Road Adjustment factors from the EEA Report
(Alternative Technology Vehicles)
A. As presented in EEA report:
1-mode GPM ratios:
Technology
-------
-28-
for good reason at the time, EEA excluded certain 1978 models from
the 2-mode road factor calculation for ?WD-Manu-Carb vehicles; EPA
used all of the data.
due to sample size considerations, EEA chose not to recommend spe-
cific two-mode factors for ^WD-Diesels; although somewhat risky, we
did estimate them.
Before filling the matrix of two-mode road factors for all of the technology
strata, we must return to the matter of the MPG-dependent factors for con-
ventional technologies (Table 14). Since the alternate technology factors are'
MPG - ^ndependent, upward aggregation of all technologies' factors is best
done in a constant-factor scenario, as was done in the one-mode analysis.
The question for the conventional technologies is: what MPG values should be
selected to extract constant factors from the equations? The answer chosen
for this report is as follows. (a) given a fleet 55/45 MPG level, estimate
the fleet City MPG and Highway MPG using reasonable Highway to City, City to
55/45, and Highway to 55/45 MPG ratios; v'b) assume that the manual
transmission subfleet will have higher average MPG values than the automatic
transmission subfleet, as has been the case historically; (c) select
reasonable MPG ranges for the subfleets consistent with 'a) and (b) and use
them in the factor equations to determine constant City and Highway factors.
Table 16 summarizes this process and its results.
To make a conservative estimate* of che range of the two-mode GPM ratios, it
is assumed that a 20 MPG Automatic-RWD-Carb subfleet has with it a 25 MPG
Manual-RWD-Carb subfleet, and a 25 MPG Automatic-RWD-Carb subfleet has with it
a 30 MPG Manual-RWD-Carb subfleet. A consistent pairing of GPM ratios would
be RWD-Auto-Carb, City, 1.100 and RWD-Manu-Carb, City, 0.996. These
Lower-bound values are the ones that correspond to the alternate technology
GPM ratios in Table 15. Thus, for the EEA data base, an array of constant,
technology-specific GPM ratios would be (next page):
*An assumption of a 5 MPG difference between automatic and manual subfleets'
EPA average is conservative compared to actuality:
Auto MPG Manual MPG
EPA Emission "actors Survey 18.3 28.2
1979 Models 18.9 26.9
1980 Models 21.9 28.2
1981 Models 23.3 30.8
-------
-29-
City Hwy
RWD-Auto-Carb 1.10 1.36
RWD-Manu-Carb 1.00 1.30
FWD-Auto-Carb 0.92 1.15
FWD-gas Fl 0.88 1.22
etc.
etc.
etc.
Projecting the conventional technologies' GPM ratio growth of 0.06 for a 5
MPG increase in fleet MPG onto the alternate technologies as well, GPM
ratios appropriate to cars in a higher MPG space might be:
City Hwy
RWD-Auto-Carb 1.16 1.42
RWD-Manu-Carb 1.06 1.36
FWD-Auto-Carb 0.98 1.21
etc.
etc.
etc.
The technology-specific dual mode road MPG adjustment factors ^the inverse
of che above GPM ratios) can now be specified, and also collapsed into
four-, two-, and one-strata systems, sales-weighted, as was done in Tables 7
and 8 for the single-mode analysis. The resulting two-mode MPG adjustment
factors are given in Table 17.
Histogram Tests of Dual-Mode Road Adjustment Systems
As was done in the Single-Mode analysis, the effectiveness of two-mode road
adjustment was tested at various levels of technology aggregation, using the
Emission /actors postcard data base as the test subject. These tests used
adjustment factors in the middle of the ranges given in Table 17.
For the dual-mode system, there are two measures of the usefulness of
adjusted MPG values. One is the degree to which the two numbers accurately
represent their respective types of driving; in effect, the adjusted City
-------
-30-
Table 16
Estimation of Single Two-Mode Factors for Conventional Vehicles
A. Reference MPG levels
If EPA 55/45 MPG is: 20^ 25^ 30
then EPA City MPG is (approx): 17.4 21.8 26.1
and EPA Hwy MPG is (approx): 24.4 30.5 36.6
c/
B. Pooled road GPM ratios— at the indicated MPG levels
RWD-Auto-Carb,
RWD-Manu-Carb,
City
Hwy
City
Hwy
1.100
1.364
—
—
1.165
1.421
0.996
1.305
__
—
1.065
1.365
— Approximately equal to the Model Year 1979 U.S. fleet EPA average
and approximately equal to the four data sources' EPA average.
— Approximately equal to the Model Year 1981 U.S. fleet EPA average.
c/
— Calculated from the equations in Table 14 and averaged.
-------
-31-
Table 17
Summary - Constant Road Factors, Two Modes
/ B Road MPG for driving type i.\
\• *• EPA MPG< i
No. Strata Road c'aetors
ONE: C .93-.98, H .73-.77
TWO (a) FWD C 1.01-1.08, H .78-.82 RWD C .88-.93, H .71-.75
(b) Auto C .91-.96, H .73-.77 Manu C .98-1.04, H .73-.77
,'c) Garb C .94-.98, H .73-.77 FI* C .97-1.03, H .73-.76
FOUR (a) RWD-Aut C .87-.92, H .70-.73 RWD-Man C .94-1.00, H .73-.76
fWD-Aut C 1.02-1.09, H .82-.86 FWD-Man C 1.00-1.07, H .74-.77
(b) RWD-Crb C .87-.92, H .71-.74 RWD-FI C .93-.98, H .70-.73
FWD-Crb C 1.01-1.06, H .78-.82 FWD-FI C 1.07-1.14, H .78-.82
v'c) Aut-Crb C .90-.96, H .73-.77 Aut-FI C .96-1.02, H .72-.75
Man-Crb C .97-1.03, H .73-.76 Man-Fl C 1.00-1.06, H .75-.78
SEVEN RWD-Aut-Crb C .86-.91, H .70-.73 RWD-Man-Crb C.94-1.00,H .73-77
RWD-FI C .93-.98, H .70-.73 FWD-gasFI C 1.07-1.14, H .78-.82
FWD-Aut-Crb C 1.02-1.08, H .82-.87 FWD-Man-Crb C .99-1.05, H .73-.76
FWD-Dsl C 1.06-1.14, H .80-.84
* FI means gasoline FI and Diesel pooled
unless noted otherwise.
-------
-32-
number La tested as a mode MPG estimator against the in-use MPG of city-
driven cars, and the adjusted Highway number is tested against highway-
driving MPG. The other measure is the fraction of the total MPG population
captured between the two numbers (also of interest is the distribution of
the MPG tails outside the inclusion range).
The results of the first kind of test, mode MPG estimates, are typified by
figures 3 through 6. Figure 3 is the City histogram for unadjusted City
values, and Figure 4 the corresponding Highway histogram. figure 5 is the
City histogram for che first (unstratified) level of adjustment, and Figure
6 is the corresponding Highway histogram.
Table 18 summarizes the mode estimate histogram test for all levels of tech-
nology stratification. The need to adjust the EPA Highway numbers is
obvious. There is a slight hint that increased technology specificity helps
the adjusted City number to become increasingly representative; considering
that the adjustment factor system comes from J.D. Power data, it is en-
couraging that this test on a totally different data base shows any such
trend at all.
Figures 7 through 10 illustrate the range inclusion type of test. Figures 7
and 8 compare all MPG experience to the unadjusted City and Highway numbers,
respectively, and Figures 9 and 10 give the same type of comparison against
fleet-adjusted numbers. The included range is easily determined from such
pairs of histograms: those data which fall neither below the City value nor
above the Highway value are, by definition, included in the range between
the two numbers. Table 19 summarizes the results of all of the range
inclusion tests. When viewed in the range inclusion sense, it is again
clear that road adjustment of some kind is beneficial. Again, technology
specificity shows a weak tendency to improve representativeness.
-------
-33-
Figure 3
Difference: EPA City MPG minus Road MPG
(City-driven Cars)
MIDPOINT HISTZ COUNT
-13.000 0. 0 +
-12.000 .4 1 + X
-11.000 0. 0 +
-10.000 0. 0 +
-9.0000 .8 2 +XX
-8.0000 0. 0 +
-7.0000 1.6 4 + XXXX
-6.0000 .8 2 +XX
-5.0000 7.3 18 +XXXXXXXXXXXXXXXXXX
-4.0000 7.7 19 +XXXXXXXXXXXXXXXXXXX
-3.0000 14.2 35 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
-2.0000 19.1 47 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
-1.0000 17.9 44 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
0. 13.4 33 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
1.0000 6.9 17 +XXXXXXXXXXXXXXXXX
2.0000 4.5 11 +XXXXXXXXXXX
3.0000 2.8 7 +XXXXXXX
4.0000 .8 2 +XX
5.0000 .8 2 +XX
6.0000 .4 1 +X
7.0000 .4 1 +X
8.0000 0. 0 +
9.0000 0. 0 +
10.000 0. 0 +
TOTAL 246
-------
-34-
Figure 4
Difference: EPA Highway MPG minus Road MPG
(Highway-driven Cars)
MIDPOINT HISTZ COUNT
-22.000
-21.000
-20.000
-19.000
-18.000
-17.000
-16.000
-15.000
-14.000
-13.000
-12.000
-11.000
-10.000
-9.0000
-8.0000
-7.0000
-6.0000
-5.0000
-4.0000
-3.0000
-2.0000
-1.0000
0.
1.0000
2.0000
0.
0.
1.7
1.7
0.
1.7
1.7
1.7
3.4
1.7
1.7
6.8
3.4
10.2
11.9
10.2
15.3
6.8
11.9
5.1
1.7
0.
0.
1.7
0.
0 +
0 +
1 + X
1 + X
0 +
1 + X
1 +X
1 -fX
2 + XX
1 +X
1 +X
4 + XXXX
2 +XX
6 +XXXXXX
7 +XXXXXXX
6 +XXXXXX
9 +XXXXXXXXX
4 +XXXX
7 +XXXXXXX
3 +XXX
1 +X
0 +
0 +
1 +X
0 *
TOTAL
-------
-35-
Figure 5
Difference: Adjusted EPA City MPG minus Road MPG
(City-driven Cars)
MIDPOINT HI5T2 COUNT
-12.000 0. 0 +
-11.000 0. 0 *
-10.000 .4 1 +X
-9.0000 0. 0 +
-8.0000 .8 2 + XX
-7.0000 0. 0 +
-6.0000 1.2 3 + XXX
-5.0000 1.2 3 +XXX
-4.0000 8.1 20 +XXXXXXXXXXXXXXXXXXXX
-3.0000 7.3 18 +XXXXXXXXXXXXXXXXXX
-2.0000 16.3 40 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
-1.0000 20.7 51 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
0. 17.1 42 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
1.0000 12.2 30 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
2.0000 4.9 12 +XXXXXXXXXXXX
3.0000 5.3 13 +XXXXXXXXXXXXX
4.0000 2.0 5 +XXXXX
5.0000 '.8 2 +XX
6.0000 1.2 3 +XXX
7.0000 0. 0 *
8.0000 .4 1 +X
9.0000 0. 0 +
10.000 0. 0 *
11.000 0. 0 +
12.000 0. 0 +
TOTAL 246
-------
-36-
Figure 6
Difference: Adjusted EPA Highway MFC minus Road MPG
(Highway-driven Cars)
HIDPOINT HISTZ COUNT
-12.000
-11.000
-10.000
-9.0000
-8.0000
-7.0000
-4.0000
-5.0000
-4.0000
-3.0000
-2.0000
-1.0000
0.
1.0000
2.0000
3.0000
4.0000
5.0000
6.0000
7.0000
8.0000
9.0000
10.000
1 1 .000
12.000
0.
0.
0.
1.7
3.4
1.7
1.7
3.4
1.7
3.4
16.9
22.0
15.3
11.9
6.8
5.1
1.7
3.4
' 0.
0.
0.
0.
0.
0.
0.
0 +
0 +
0 +
1 + X
2 + XX
1 + X
1 +X
2 +XX
1 +X
2 + XX
10 +XXXXXXXXXX
13 mxxxxxxxxxxx
9 +XXXXXXXXX
7 +XXXXXXX
4 +XXXX
3 +XXX
1 +X
2 +XX
0 *
0 +
0 «•
0 *
0 +
0 +
0 +
TOTfll 59
-------
-37-
Table IS
Results of Histogram Tests of Two-Mode Adjustment Factors
Test Subject = Ends. Fac. Data Base
** Mode MPG Estimate **
Number of ~ Average
Strata Strata Adjusted MPG Error
1
1
2
2
2
4
4
4
7
No adjustment City
Hwy
Fleet adjustment City
Hwy
FWD/RWD City
Hwy
Auto/Manual City
Hwy
Carb/FI City
Hwy
R-A/R-M/F-A/F-M City
Hwy
R-C/R-FI/F-C/F-FI City
Hwy
A-C/A-FI/M-C/M-FI City
Hwy
Each technology City
Hwy
-1
-7
-0
-0
-0
^ ^
-0
-0
-0
-0
-0
-0
-0
-1
-0
-1
-0
-0
.6
.9
.7
.9
.2
.1
.6
.9
.8
.9
.1
.8
.1
.1
.1
.0
.1
.9
% of mode-specific experience
Below Within 2 MPG Above
33
97
19
17
14
24
18
17
19
17
13
24
13
27
18
17
14
20
62
3
71
73
75
63
74
73
71
73
79
64
76
61
74
73
77
68
5
0
10
10
11
14
8
10
9
10
8
12
11
12
8
10
9
12
-------
-38-
Figure 7
Difference: EPA City MPG minus Road MPG
(All Cars)
DPOINT HISTZ COUNT
3.000 0. 0 +
2.000 .2 1 + X
1 .000 0. 0 +
0.000 .2 1 + X
.0000 .5 2 + XX
.0000 .7 3 +XXX
.0000 1.4 6 +XXXXXX
.0000 .7 3 +XXX
.0000 5.2 23 +XXXXXXXXXXXXXXXXXXXXXXX
.0000 5.7 25 +XXXXX-XXXXXXXXXXXXXXXXXXXX
.0000 9.3 41 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
.0000 u.i 7i +xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxmxxxxxxxxxxxxxxxxxmxxxxxxxxxxxm
.0000 17.7 73 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXJUXXXXXXXXXX
14.1 62 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
.0000 10.0 44 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
.0000 6.8 30 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
.0000 4.5 20 +XXXXXXXXXXXXXXXXXXXX
.0000 2.5 11 +XXXXXXXXXXX
.0000 2.0 9 fXXXXXXXXX
.0000 .7 3 *XXX
.0000 .9 4 +XXXX
,0000 0. 0
.OCOO .7 3 +XXX
).000 0. 0
iTAL 440
-------
-39-
Fignre 8
Difference: EPA Highway MPG minus Road MFC
(All Cars)
MIDPOINT HISTZ COUNT
-23.000
-22.000
-21 .000
-20.000
-19.000
-18.000
-17,000
-le.OOO
-15.000
-14.000
-13.000
-12.000
-11 .000
-10.000
-9.0000
-8.0000
-7.0000
-6.COOO
-5.0000
-4.0000
-3.0000
-2.0000
-1 .0000
0.
1 .0000
2.0000
0.
.2
.2
.5
.9
.9
.5
1 .1
1 .4
3.0
3.2
4.8
8.2
7.5
12.0
13.4
T3.0
10.7
8.9
5.2
2.3
1 .4
.?
0.
1
• *.
0.
0
1
1
o
i.
4
4
A.
5
6
13
14
21
36
33
53
59
57
47
39
23
10
6
3
0
1
0
+ XX
+ XXXX
+ XXXX
+ XX
+XXXXX
+XXXXXX
mxxxxxxxxxxx
+XXXXXXXXXXXXXX
+XXXXXXXXXXXXXXXXXXXXX
+XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
'XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
+xxxxxxxxxxxxxxxmxxxxxxxx'xxxxxxxxxxxmxxxxxmxxxxx
+XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
uxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
+XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
+XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
*uxxxxmxxxxxxxxxxxxxx
+XXXXXXXXXX
+XXXXXX
+ XXX
TOTAL
440
-------
-40-
Figure 9
Difference: Adjusted EPA City MPG minus Road MPG
(All Cars)
DPOINT HISTZ COUNT
2.000 0. 0 +
1 .000 0. 0 +
0.000 .2 1 + X
.0000 0. 0 +
.0000 .7 3 + XXX
.0000 .2 1 *X
.0000 1.4 6 +XXXXXX
.0000 1.1 5 +XXXXX
.0000 5.2 23 +XXXXXXXXXXXXXXXXXXXXXXX
.0000 5.9 26 +XXXXXXXXXXXXXXXXXXXXXXXXXX
.0000 10.7 47 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
.0000 17.3 76 +XXXXXXXXXXXXXXXXXXXXX.XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
17.0 75 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
.0000 13.4 59 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
.0000 9.3 41 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
.0000 7.0 31 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX •
.0000 3.2 M +XXXXXXXXXXXXXX
.0000 3.2 14 fXXXXXXXXXXXXXX
.0000 2.0 9 +XXXXXXXXX
.0000 .5 2 +XX
.0000 .5 2 +XX
.00,00 .5 2 +XX
;.ooo .7 3 +xxx
;.000 o. o +
2.000 0. 0 +
JTAL 440
-------
-41-
Figure 10
Difference: Adjusted EPA Highway MPG minus Road MPG
(All Cars)
:HPOIN7 HISTZ COUNT
2.000 0. 0 +
1.000 .5 2 tXX
0.000 J 3 + XXX
'.0000 .7 3 +XXX
1.0000 1.1 5 +XXXXX
'.0000 2.3 10 +XXXXXXXXXX
,.0000 3.6 16 +XXXXXXXXXXXXXXXX
i.OOOO 8.4 37 tXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
.0000 8.2 36 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX.XXXX
.0000 12.7 56 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX.XXXXXXXXXXXXXXXXXXXXXXXXX
.0000 17.3 76 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
.0000 17.3 76 fXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX.XXXXXXXXXXXXXXXXXXXX
10.7 47 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
.0000 7.7 34 +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
.0000 3.6 16 +XXXXXXXXXXXXXXXX
.0000 2.0 9 +XXXXXXXXX
.0000 1.4 6 +XXXXXX
.0000 1 .4 6 +XXXXXX
.0000 .5 2 +XX
.0000 0. 0 +
.0000 0. 0 +
.0000 0. 0 +
0.000 0. 0 *
I.000 0. 0 *
2.000 0. 0 +
OTAL 440
-------
-42-
Table 19
Results of Histogram Tests of Two-Mode Adjustment Factors
Number of
Strata
1
1
2
2
2
4
4
4
7
** Range
Strata Adjusted
No adjustment
Fleet adjustment
FWD/RWD
Auto/Manual
Carb/FI
R-A/R-M/F-A/F-M
R-C/R-FI/F-C/F-FI
A-C/A-FI/M-C/M-FI
Each technology
Inclusion **
Z of all
Below Between
40
25
20
26
27
19 ••"• •-
*
19
25
19
MPG experience
C'-l and H'+l
60
66
70
65
65
69
71
66
69
Above
0
9
10
9
9
11
10
q
12
-------
III. CONCLUSIONS
A. The preferred general equation for analyzing and expressing the relation
between Road MPG (R) and EPA MPG (E) is an MPG-dependent fuel con-
sumption ratio; (1/R) T (1/E) - a + b E;
this is equivalent to: E/R - a + b E;
the general expression for road adjustment factors to multiply EPA MPG
numbers by is: £ - l/(a + b E);
a special case of the road adjustment factor, wherein there is no MPG
dependence and therefore b - zero, is: f • I/a;
a road adjusted EPA MPG number (E') is: E' • f x E.
B. All four data sources evaluated herein agree that the ETA- 55/45 MPG
value overestimates average in-use fuel economy in the range -from 102 to
20%. A road adjustment factor between 0.8 and 0.9 is therefore
indicated. (Road factors will be calculated to two decimal places in
susbsequent analyses.)
C. In single-mode analyses, there is a consensus among those data sources
which investigated vehicle technology stratification that:
Front wheel drive (FWD) cars have less MPG shortfall than rear
wheel drive (RFD) cars, for carburetted, automatic or manual
transmission power trains;
Diesel powered cars have less MPG shortfall than carburetted
gasoline cars, for RWD-automatic transmission configurations;
- Manual transmission cars have less MPG shortfall than automatics,
for carburetted FWD and RWD power trains;
Generally, specific vehicle technology strata show a worsening
MPG shortfall as MPG level increases.
D. Depending on the mix of technology strata in a data base, MPG dependence
of shortfall for the strata may be obscured when the data base is
analyzed at high levels of aggregation; however, such obscuration of
MPG dependence via high aggregation analysis does only that: it
obscures it. It does not make it cease to exist.
E. Using emission factors data, histograms show that 672 of in-use MPG ex-
perience falls more than 2 MPG below the EPA 55/45 MPG value; how-
ever—when single-mode road adjustment factors are applied to these
data, more than 652 of the population is accurately represented (within
a window of + 2 MPG) by the adjusted 55/45 value.
F. In dual-mode analysis, 332 of city driving MPG experience falls more
than 2 MPG below the EPA City value (702 of city-driven MPG falls below
-------
-44-
the City value); however, road-adjusted City values accurately
represent more than 702 of the city driving population, within a ^ 2 MPG
window;
For highway driving, 97% of in-use MPG experience falls more than 2 MPG
below the EPA Highway value (99.8Z of it falls below the Highway
value); with road adjustment, more than 60% of the highway driving MPG
experience (+ 2 MPG) is accurately represented.
G. At the fleet level, city adjustment factors of 0.9 to 1.0 are
indicated; highway factors of 0.7 to O.S are indicated. (Road factors
will be calculated to two decimal places in subsequent analyses).
-------
-45-
IV. ISSUES FOR FUTURE ANALYSES
Some of Che issues which surfaced in this analysis are listed below.
Section II described how this analysis dealt with them, but future analysis
will reconsider them carefully; they may be dealt with differently in these
future analyses.
A. Is minimum acceptable sample size for a given stratum to be fixed by
preordained precision and confidence level specifications and an
estimated measure of variance, or are precision and confidence level to
be outputs of a stratum analysis, given its sample size and actual
measure of variance?
B. When aggregating across data sources, should they be equally weighted or
sample-weighted, i.e., is the analogy of the widget inspectors
accurate?
C. When aggregating technology-specific road factors or road factor
algorithms, what sales-weighting values should be used?
D. Should data from obsolescent vehicle technologies be discarded?
E. What vehicle technologies can/should be pooled, and what criteria should
determine whether to pool or not to pool: sample size? engineering
argument? similarity of shortfall behavior?
F. Should pre-1979 models of manual transmission vehicles whose shortfall
changed significantly in 1979 be excluded from future road factor de-
velopment (che premise being that shift schedule optimization has been
stopped), or should they be included CChe premise being that the return
of shift schedule optimization cannot/will not be prevented)?
G. What method ("driving-mode extremes" versus "all data") should be used
Co develop cwo-mode adjustment faccors?
H. Will perceived MPG daca be accepced as chough it were measured, thrown
out because it isn't, or can it be "corrected" to make it comparable to
measured MPG data?
I. Is an analysis at the model-cype level an appropriace one Co employ?
------- |