EPA-AA-SDSB 79-09
An Analysis of Test Procedure Changes Made During
1975-1979 With Respect to Measured Fuel Economy Effects
December 1978
Technical Reports do not necessarily represent final EPA decisions
or positions. They are intended to present technical analysis of
issues using data which are currently available. The purpose in
the release of such reports is to facilitate the exchange of tech-
nical information and to inform the public of technical develop-
ments which may form the basis for a final EPA decision, position
or regulatory action.
Standards Development and Support Branch
Emission Control Technology Division
Office of Mobile Source Air Pollution Control
Office of Air, Noise and Radiation
U.S. Environmental Protection Agency
-------
Introduction
This paper presents an analysis of each of the changes in EPA
test procedure since 1975 which have been identified as potential
areas where it may be argued that changes in EPA regulations have
resulted in a decrease in measured vehicle fuel economy. These
areas are:
A. Changes in the EPA prediction of the dynamometer power
absorption.
B. Changes in allowable practices for determining and
requesting alternate dynamometer power absorptions.
C. Changes in the dynamometer inertia simulation increments.
D. Changes in allowed alternate shift point schedules.
E. Changes in dynamometer calibrations.
F. Changes in vehicle selection.
G. Changes in laboratory humidity levels.
H. Use of the actual simulated distance traveled by the
vehicle versus the nominal test distance.
I. Changes of the value used for the density of CC^ in the
EPA calculations.
In many cases these changes were technical improvements which
increased the accuracy of the test procedure or provided for
improved fuel economy recognition of technical improvements. In
other instances these changes were considered necessary to prevent
or restrict abuses of various aspects of the EPA fuel economy
measurement procedures. In the first category, these changes
either increased measured fuel economy or had no net directional
effect. The second type of change generally reduced measured fuel
economy; however, these changes were introduced only as the need
arose. Consequently, they generally corrected abuses which did not
exist in 1975, since there was little pressure for fuel economy
improvements in that model year. The changes would not have
affected the 1975 results, but were necessary to insure that the
results of future model years were comparable to those of 1975.
In general, it is easy to demonstrate that under specific
circumstances a given vehicle might show a degradation in fuel
economy between the 1975 and the 1979 model year test procedure.
Therefore, if the true net fuel economy difference between the test
procedures of these years is to be evaluated the following ques-
tions must be asked:
-------
-2-
Did the change have an effect on all vehicles or what percent-
age of the total population of vehicles were affected?
Was the change systematic for all vehicles affected or what
was the net effects of this change?
Would the change have had an effect on the 1975 test results?
What was the net effect of all changes?
It is intended that this paper provide the background informa-
tion and the direction for indepth questioning of premises that
changes in EPA regulations since 1975 have reduced measured fuel
economies.
Analysis of the Changes in EPA Procedures
The following sections address each of the previously identi-
fied changes in EPA test procedures or test conditions. In those
instances where EPA Technical Support Reports, Analyses of Comments
in response to NPRMs, or other supporting material is available,
this material is provided as attachments or referenced.
A. Changes in the EPA Prediction of the Dynamometer Power
Absorption.
For the 1979 model year, EPA revised the equation used to
predict the dynamometer adjustment for LDV testing. The revision
replaced the existing table based on the vehicle inertia weight
with an equation primarily based on the vehicle reference area, but
also including the effect of vehicle protuberances and tire type.
This change was developed, at least partially, in response to
requests by the automotive industry for a change in EPA practices
to better represent the fuel economy advantages of radial tires
which were not directly reflected in the twin roll dynamometer
tests.
The potential effect of the change to the area based equation
can be seen from the attached plot. This graph indicates that the
dynamometer adjustments obtained by EPA were, on the average, in
good agreement with the existing weight-based table for vehicles
with bias or bias belted tires. For vehicles with radial ply
tires, the data indicated that the current table was approximately
8% higher than appropriate for a median vehicle. This result is
reflected in the frontal area equation which was subsequently
developed by regressing these data against frontal area data
supplied by the vehicle manufacturers.
-------
-3-
Approximately 80% of the 1979 OEM tires were radials; there-
fore, the majority of all light-duty vehicles currently tested
would be expected to have slightly lower predicted dynamometer
power absorption than would have occurred if this change had not
been made. This effect is estimated to be about 1.5 percent
increase in measured fuel economy or about 0.3 mpg for a 20 mpg
vehicle.
The previous analysis assumes that the vehicle weight-area
relationships remain the same as observed in 1975, the model year
most predominant in the EPA vehicle sample used to develop the
reference area based prediction equation. To some extent this has
changed, since some vehicle weight reductions may have occurred
without reduction in vehicle reference area. To the extent that
this has occurred, a vehicle tested with a dynamometer adjustment
based on the reference area formula might show reduced fuel economy
compared with test results using the previous weight based table.
However, since the area based formula more accurately predicts the
aerodynamic drag of the vehicle, which is the predominant force
simulated by the dynamometer, the use of this equation should more
accurately simulate the road experience of the vehicle. That is,
the previous weight-based table would have provided a larger
predicted fuel economy reward for changes in vehicle weight,
without changes in vehicle reference area, than would have occur-
red in consumer use of the vehicle.
EPA technical support reports have discussed the development
of the area-based equation in detail.(1,2) This material was also
presented to the technical community in the form of an SAE paper.(3)
The analysis of comments received in response to the proposed
change from the weight-based equation is also available in the
public docket.(4) When reviewing these comments, it should be
noted that they were received in response to an originally proposed
equation which was considerably more complex and which, in general,
predicted higher dynamometer power absorptions. As a result of
these comments the equation was revised to its present simpler
form.
B. Changes in Allowable Practices for Determining and
Requesting Alternate Dynamometer Power Absorptions.
EPA has always provided the option that a manufacturer may
request, for specific vehicles, dynamometer adjustments which are
different from the values predicted by EPA. A request for such
alternate dynamometer power absorptions must be supported by road
test data demonstrating the appropriateness of the request. In
1975, the regulations implied that manifold pressure measurements
were the required method of generating acceptable road data. Later
the manifold pressure approach was deleted, and subsequently the
-------
-4-
coastdown technique has become the prevalent method of generating
supporting data for alternate dynamometer power absorption re-
quests. An acceptable coastdown procedure has been provided to the
industry as an EPA Recommended Practice which has been distributed
as an attachment to Advisory Circular 55.
Advisory Circular 55 has been revised several times to clarify
the requirements of similarity between the road test vehicles, the
production vehicles represented and the exhaust emission certifi-
cation vehicles. Technical changes have also been made to the
Recommended Practice to improve precision and to make the standard
test conditions more representative of typical vehicle use. While
many of the revisions to AC-55 may appear to increase the resulting
dynamometer power absorption slightly, these changes were intro-
duced to eliminate abuses which became prevalent after 1975.
In general, compared with the practices of 1975, the entire
area of alternate dynamometer power absorption should result in
reduced loadings of the test vehicles. This has occurred because
very few alternate dynamometer adjustments were requested in 1975,
while the use of alternate dynamometer power absorptions is preva-
lent today. Since an alternate dynamometer power absorption is
generally only requested when it is beneficial to the manufacturer,
much of the current testing is conducted at vehicle loadings lower
than those which would have been used if the same vehicles had been
tested in 1975. EPA is currently investigating in detail the
average dynamometer adjustments used in the 1979 model year versus
the inertia weight table values of the Federal Register Code of
Regulation prior to the 1979 model year.(5)
Even for those vehicles for which alternate dynamometer power
absorptions were requested in 1975, the current test procedure is
likely to yield slightly lower values. This occurs because the
added experience and precision of the current procedure allows
correction to more ideal standard conditions. While these correc-
tions were not prohibited in 1975, the procedures actually in use
were generally not this sophisticated.
C. Changes in Dynamometer Inertia Simulation Increments.
Beginning with the 1979 model year, EPA reduced the increments
of simulated inertia by approximately a factor of two. This change
was made because it was becoming apparent that manufacturers were
increasingly tending to design vehicles which were at the upper
bound of the older, larger inertia weight increments.
This decrease in the inertia weight increments improves the
precision of the emissions and fuel economy measurements from any
given test vehicle. It will decrease the fuel economy results only
-------
—5—
if the vehicle would otherwise have been tested at a lower, less
appropriate inertia weight class.(6)
When the reduction of the inertia weight increments was
proposed, none of the manufacturers argued that this change would
have reduced the fuel economy results of the 1975 model year.
Comments were received that this change would reduce some of the
fuel economy benefits that had been anticipated from vehicle weight
reduction programs. These comments supported EPA's concern that
vehicle down-sizing programs were directed toward the EPA inertia
weight classes. To the extent that this was occurring, EPA mea-
surements would have over-estimated the fuel economy benefits of
vehicle down-sizing if the intervals of simulated inertia weights
were not reduced.
The full record of the comments received in response to the
rulemaking which decreased the inertia weight increments, and the
analysis of these comments by EPA can be found in the public
docket.(7)
D. Changes in Allowed Alternate Shift Point Schedules.
In 1975, the regulations provided that test vehicles would
normally be shifted at 15, 25 and 40 mph. In order to provide for
unusual vehicles the regulations also provided for the option of
shifting the vehicle at the shift points recommended by the manu-
facturer. During 1975 model year testing, most vehicles were
shifted at the default, 15-25-40 shift points.
On July 16, 1976, the default shift points were deleted from
the regulations and the vehicles were to be shifted according to
the manufacturer's recommendation to the ultimate purchaser. EPA
soon began to receive shift point requests which appeared to be
selected primarily to minimize emissions or to maximize fuel
economy. For example, different requested shift speeds for hot
versus cold vehicle operation and generally very early shift points
for the EPA Highway Cycle. The shift point recommendations pro-
vided by the manufacturers to the vehicle consumer were generally
not nearly as specific as the shift point requests recieved by EPA,
but rather were sufficiently vague to encompass the requests made
to EPA.
EPA investigated this problem and concluded that many of the
shift schedules requested by vehicle manufacturers were unrepresen-
tative of typical vehicle use. This investigation was docu-
mented in an EPA technical report. (8)
In order to insure more reasonable future shift schedules EPA
restricted acceptable shift schedules with Advisory Circular No.
72. This Advisory Circular provides that allowable
-------
-6-
may be the 15-25-40 mph schedule originally presented in the
regulations, a shift schedule based on a percentage of maximum
recommended engine RPM which was developed in the attached report,
or any other recommended shift schedule based on typical vehicle
use data.
At the present time virtually all manufacturers are using the
15-25-40 shift points schedule, as was the case in 1975. Conse-
quently, the changes in allowed shift schedules eliminated an abuse
which occurred primarily after 1975. Consequently, change in
allowable shift schedules can account for little or no syste-
matic fuel economy degradation between the 1975 and present model
years.
E. Changes in Dynamometer Calibrations.
In 1975, EPA used a manual method of adjusting the dynamometer
power absorption prior to each test. In this case a dynamometer
warm-up vehicle was used to drive the dynamometer at 50 mph while
the vehicle operator adjusted the water level in the hydrokinetic
power absorber, by means of a solenoid valve, until the desired
torque reading was obtained.
In May 1977, EPA changed to an electronic feedback dynamometer
control system which eliminated the need for manual adjustment of
the dynamometer prior to each test. This change reduced the
laboratory effort required for each test and reduced test variabil-
ity .
This change, if independently made, could have resulted in a
change in the loading experience of the test vehicles. However, at
the same time EPA altered the dynamometer calibration procedures,
both because this was required to support the automatic control
features and to ensure that the change to the automatic dynamometer
adjustment would not result in a net change in the dynamometer
experience of the test vehicle. The most significant change in the
calibration procedure was an allowance for the angular velocity
differences between the front and rear dynamometer rollers which
were previously observed with the dynamometer adjustment vehicle.
The changes in the operation modes and the dynamometer cali-
bration procedures have been described in detail in the EPA com-
munications. (9)
F. Changes in Vehicle Selections.
It has been suggested that changes in the selection of test
vehicles may have resulted in a decrease in the measured vehicle
fuel economy. However, when attempting to research this question,
few changes could be identified in the vehicle selection process.
-------
-7-
The emission test vehicles are selected on a structured basis
which provides for several EPA selections of "worst case" vehicles.
It is possible that EPA is currently more astute in the selection
of "worst case" vehicles; however, this would not reflect any
change in the procedure or the intent of the procedure.
In addition to the emission test vehicles which are selected
by EPA, the manufacturer can elect to test supplementary fuel
economy vehicles. In 1975 the fuel economy program was voluntary,
and the manufacturers were consequently given greater latitude in
the selection of these optional vehicles and in the use of the
resulting data than is presently allowed. Currently the optional
fuel economy vehicle selections are constrained so that hopefully
the data from these vehicles increases the accuracy with which the
EPA sample represents the total vehicle population. These con-
straints probably reduce the fuel economy advantage obtained by a
manufacturer for each optional test vehicle. However, in 1975 few
optional vehicles were tested compared with those tested in the
1979 model year. The net effect these optional vehicle selections
have had on the corporate average fuel economies between the 1975
and 1979 model years is difficult to accurately assess. Certainly
the two major factors, the increase in the number of optional
vehicles test data in the 1979 data base, and the EPA constraints
on the selection of these vehicles tend to be compensatory.
A vehicle selection process also occurs in the review of fuel
economy data and selection of vehicles for confirmatory testing.
EPA is currently conducting a greater number of confirmatory tests
than were conducted in 1975. If these results are consistently
lower than those reported by manufacturers, inclusion of these data
into the fuel economy calculations could reduce the computed
corporate average fuel economy. If this is affecting fuel economy
results, it is a more vigilant maintenance of unbiased results, not
a change in the EPA procedure.
G. Changes in Laboratory Humidity Levels.
In April 1976 EPA changed the average laboratory humidity from
approximately 55 grains of water per pound of air to 75 grains.
This change was made to reduce the magnitude of the effects of the
humidity correction factor in the calculation of the NOx emissions.
Reducing the correction factor effect was desirable to improve the
accuracy of the determination of vehicle NOx emissions.
This change would, in general, be expected to decrease the
measured fuel economy of a vehicle since the combustible portion of
the incoming fuel-air charge would be reduced and the vehicle would
tend toward enriched operation. This is, however, entirely depen-
dent on the calibration of the vehicle. For 1975 through 1978
-------
-8-
model vehicles, the theoretically anticipated enrichment effect
would probably result in some loss of fuel economy. However, for
1979 and later model year vehicles using fuel system feedback
technology, this enrichment condition would be sensed and compensa-
tion would be made in the fuel delivery. Therefore, no fuel
economy degradation would be anticipated for current or future
vehicles using sensor-feedback technologies. In fact, under some
conditions the increased humidity might benefit fuel economy and
exhaust emissions in the same manner as exhaust gas recirculation
can be beneficial. That is, a non-combustible diluent can reduce
engine pumping losses and allow optimization of the spark timing
without knock or excessive NOx production.
It should also be noted that the higher test humidity condi-
tions were chosen as standard conditions well before 1975. This is
evident since the NOx correction factor in the EPA exhaust emis-
sions calculations has used 75 grains of water per pound of air as
the standard condition from very early in the regulations. The
only change which was made was to make the actual test conditions
correspond to the theoretical standard conditions of the calcula-
tions. This change was made as soon as the Ann Arbor facility
could consistently and accurately maintain the higher humidity.
H. Use of the Actual Simulated Distance Traveled by the
Vehicle Versus the Nominal Test Distance.
In 1975 EPA calculated fuel economy based on the nominal, that
is theoretical, distance of the test cycles. This was done primar-
ily for convenience in the calculations even though the actual
simulated distance traveled by the vehicle might be slightly
different from this value.
Deviation from the nominal value can occur for two reasons,
normal allowable driver fluctuations about the test speed versus
time schedule and the inability of the vehicle to maintain the
speed called for by the test schedule. EPA investigated variations
in the simulated distance traveled in 1975 and concluded that most
of the distance variations occurred because of driver fluctuations.
This investigation was presented in an EPA technical report.(10)
The largest data base, which considered the results of 170 repeat
tests on a single vehicle, was contributed by Ford Motor Company.
These data indicated that the coefficient of variability of repeat-
ed fuel economy measurements would diminish by about 30% (a de-
crease from approximately 3.6% to 2.4%) when the actual distance
was used in the fuel economy calculations. These same data indi-
cated this change would result in an increase in the harmonic mean
fuel economy of 0.16%.
It was concluded that for most vehicles no significant effect
-------
-9-
on measured fuel economy occurred when the actual simulated dis-
tance was used in the fuel economy calculation. Consequently, the
change was made in the fuel economy calculation to reduce test
variability. For some underpowered vehicles which might not be
able to follow the EPA driving schedules, some systematic decrease
in vehicle fuel economy would result from this change. Few of
these vehicles, however, were present in the 1975 test fleet. If
such vehicles are becoming more prevalent it would be illogical to
credit these vehicles with greater fuel economies than actually
achieved because of an erroneous assumption that they traveled a
greater distance than was actually simulated in the fuel economy
test.
I. Change of the Value Used for the Density of CC^ in the
EPA Calculations.
On November 14, 1978 EPA changed the value for the density of
C02 used in the fuel economy calculations from 51.85 gm/ft3 to
51.81 gm/ft3. This change was made so that the value for the
density of CO^ would be the same in all EPA calculations. The
current value is considered to be more accurate, and it is believed
that the slight discrepancy from the previous value may have been
the result of round-off error in calculations of the original
density.
This change will have a direct systematic effect on all EPA
light-duty vehicle fuel economy calculations. It will increase the
measured fuel economy by slightly less than 0.08% or 0.015 mpg for
a 20 mpg vehicle. This effect is quite small, however, it is
significant because it systematically improves all fuel economy
measurements.
Conclusions
It is concluded that none of the identified changes in the EPA
test procedures should result in a net systematic decrease in the
measured corporate fuel economies. The one possible exception, the
change in laboratory humidity, might reduce the fuel economy of
some vehicles not using the advanced technologies introduced by
some manufacturers in 1979 and widely anticipated in the 1980's.
-------
10
References
1. G. D. Thompson, "Light-Duty Vehicle Road Load Determination" ,
EPA Technical Report LDTP-76-3, December 1976.
2. G. D. Thompson, "Prediction of Dynamometer Power Absorption to
Simulate Light-Duty Vehicle Road Load" , EPA Technical Report,
April 1977.
3. G. D. Thompson, "Prediction of Dynamometer Power Absorption to
Simulate Light-Duty Vehicle Road Load", Society of Automotive
Engineers, 780617.
4. Analysis of Comments received in response to the September 10,
1976 EPA NPRM regarding fuel economy and emissions testing.
5. J. D. Murrel, Discussion.
6. T. R. Norman, T. Rarick, "The Effect of Dynamometer Weight
Simulation on Fuel Economy Measurements" , EPA Technical Report,
February 1976.
7. Analysis of Comments op cit.
8. R. A. Rykowski, "Shift Schedules for Emissions and Fuel
Economy Testing" , EPA Technical Report, November 1977.
9. D. Paulsell, "Analysis of Ford Presentation Regarding Road
Load Problems", Memorandum to Michael Walsh, August 28, 1978.
10. T. Rarick, "Dynamometer Distance"*, EPA Technical Report, July
1975.
------- |