SDSB 79-29
Technical Report
August 1979
Vehicle Efficiency - Road vs Dynamometer
by
Bruce Grugett
NOTICE
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 technical information and
to inform the public of technical developments 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
-------
-2-
I. Introduction
The accuracy of results obtained from vehicle fuel consumption mea-
surements performed on a dynamometer in predicting fuel consumption on
the road depends upon the ability of the dynamometer to simulate the
road experience of the vehicle. A previous study indicated that
vehicles consume less fuel (higher mpg) in a test run on the current EPA
dynamometer than when the same test is performed on the road. This
implies that in current dynamometer testing either fuel consumption is
not measured accurately, too little work is extracted from the vehicle,
or the vehicle operates more efficiently than on the road.
II. Background
In a program begun last summer and recently completed, a series of
steady-state tests were run on a 1976 Mercury Montego on a twin-roll
dynamometer at the EPA and on the Transportation Research Center test
track in Ohio. During these tests measured fuel consumption was man-
ually recorded while wheel torque and vehicle speed were recorded on
magnetic tape.
This is the first time that EPA has conducted an experimental pro-
gram in which vehicle torque and speed measurements were obtained during
a test in which fuel consumption was measured. These additional data
are sufficient to calculate the work done by the vehicle which can then
be divided by the fuel consumed to give a measure of vehicle efficiency.
III. Measurements
The data discussed in this report were all obtained from steady
state measurements on the test track of the Transportation Research
Center of Ohio or from dynamometer measurements at the EPA MVEL lab-
oratory. The data consisted of two general categories, measurements of
the fuel consumed by the vehicle and measurements of the energy expended
by the vehicle during the period of the fuel consumption measurement.
A. Fuel Consumed
All fuel consumption data were obtained by a Fluidyne 1240 flow
meter. For both the road and the dynamometer, the measured fuel con-
sumption was corrected to fuel volume at the common reference tempera-
ture of 60°F. This corrected fuel consumption was used throughout the
report when calculating vehicle efficiency on both the road and dyna-
mometer.
B. Energy Expended
The work done by the vehicle can be obtained by integrating the
instantaneous power, given by the product of the wheel torque and vel-
ocity, over the time of the test. The torque was measured with torque
transducers mounted on the wheels, so the torque values include the
torque exerted by the wheels on the tires. Since the tires are "down-
stream" of the torque sensors, tire losses are included in the data.
Losses "upstream" of the torque sensor, for example engine and drive
train losses are not included.
-------
-3-
The energy expended is the true integral of the power delivered to
the rear wheels. The instantaneous power is given by the product of the
torque and angular velocity.
= T±W± (1)
Where:
P. = the instantaneous power at the i sample period.
T. = the instantaneous torque at the i sample period.
W. = the instantaneous angular velocity at the i sample period.
The energy expended in any period is the product of the power
transmitted times the timed interval. In this test program the time
periods between the data. samples were always one second. Therefore, the
total energy expended during the test is, to a very close approximation,
simply the sum of the observed powers. That is:
E = £ P.At.
=AtIT.W. (2)
Where:
E = the total energy expended during the test.
At = the interval between data samples
= one second
Equation (2) was used to calculate the energy expended over each
road and dynamometer test. In all calculations the true angular vel-
ocities of the drive wheels, as measured by pulse encoders on the
wheels, were used.
IV. Discussion
In order to better compare the fuel measurements and energy calcu-
lations from the different tests, a specific fuel consumption and specific
energy expended were calculated by dividing the total energy and the total
fuel consumed by the distance travelled during the -test. That is:
e = E/d
and
f = F/d
Where:
E = Total energy expended.
F = Total fuel consumed.
-------
e = the specific energy expended (Joule/M)
f = the specific fuel consumption (cc/km)
d = the test distance
All of the data, for both the track and the dynamometer tests are
presented 'inathe appendix .and are plotted in Figure 1. There are two
very significant observations which can be made from this figure; the
linearity and repeatability of the dynamometer data, and the large
difference between road and the dynamometer data.
A. Dynamometer Data
The dynamometer data appear to represent a linear relationship
between the fuel consumed and the energy extracted. A linear regression
of these data resulted in a correlation coefficient of .97; that is 94
percent (R =97 = .94) of the variation of-the data may be explained by
the linear relation.
The linearity is surprising since the data vary over a +16,percent
range in fuel consumption and a factor of two in expended energy. These
data include tests at three different speeds, 40, 50, and 55 mph, and
three different dynamometer adjustments. The important point is that
the actual energy expended was measured during each test. The vehicle
fuel consumption appears to vary linearly with the energy expended over
a wide range of both parameters.
B. Road versus Dynamometer Results
The second significant observation is the large difference between
the vehicle fuel consumption on the road and on the dynamometer. While
this is apparent from Figure 1, it is even more evident when the differ-
ence in the engine efficiency on the road versus that on the dynamometer
is considered.
The efficiency of any system is defined to be the ratio of the
useful energy extracted from the system to the energy put into the
system. In this case the useful energy extracted from a vehicle is the
mechanical work done by the wheels while the input energy is the chem-
ical energy in the fuel. The efficiency can also be expressed as a
percentage if the amount of chemical potential energy contained in the
fuel is known. D.E. Foreiger reports in his paper "Gasoline Factors
Affecting Fuel Economy," (2) that the energy obtained from burning comm-
ercially available gasoline falls in a narrow range around 115,000
BTU/gallon, or equivalently 32023 J/cc.
The vehicle efficiencies were calculated for each of the 50 mi/hr
steady-state tests on the road and on the dynamometer. On the road, the
vehicle produced an average of 3,612 joules per cubic centimeter (J/cc)
of fuel used. This gives a vehicle efficiency on the road of 11.3 per-
cent. On the dynamometer, however, the average efficiency was 5430 J/cc
or 17.0 percent. These results are shown in the following table.
-------
-5-
180*00
;i. 48 A 00
Road Data
u 132 >00 1 $ ## #
*>!<** 2
§ * >!<>!<$ >!<
§ i ### #2##>K# *
^ • 2>K3* *
S * 2 *
_i 116,00 + * * *2 * •*
p >l< %
E . v. Dynamometer Data
•f 2*** * *
*2 #>!<
2>K*2)!< 3
100*00 -f X<
i. _ . .i _.. .. i i i i _.. .1 i ._ .1. i i.
i » • i r i ~' t r r I T )
400,00 . 600*00 800*00
500,00 700*00. 900.00
WORK EXTRACTED J/M
FIGURE 1.
-------
-6-
Vehicle Efficiency
J/cc Percent
Road
Dyno.
3612
5430
11.3
17.0
Difference 1822 6.3
These differences appear quite large. Compared to the results on
the dynamometer, the vehicle consumed 34% more fuel on the road for the
same amount of work. A statistical "t-test" confirmed this observation,
showing that there is greater than 99% confidence that the dynamometer
efficiency was greater than the road. Since the differences are so
large, and the statistical tests confirmed that these are not random
fluctuations, the possibility of systematic experimental errors was
investigated. Errors in either the fuel consumption measurements or
errors in the torque or speed measurements could result in the observed
differences between the road and the dynamometer efficiencies.
It is unlikely that the difference in efficiency is due to an error
in measuring fuel consumption. The flow meter was calibrated before the
test program and the calibration was confirmed during the program. In
any case the same instrument was used for both the road and dynamometer
tests. Even if the calibration was somewhat in error, the same error
would be present in both the road dynamometer measurements, therefore
this could not explain the difference between the road and dynamometer
results.
The dynamometer fuel consumption data were approximately confirmed
by carbon balance measurements, therefore if any error exists in the
flow meter data it would have to be the road data which is in error.
There is, however, no evidience of error in these data. All of the
steady-state data are in approximate agreement with any data variability
much less than the difference between the road and the dynamometer
results.
In the case of the torque measurements, some thermal drift problems
were reported in the course of the track measurements. Therefore, as a
check on both the torque and speed measurements, road coastdown times
were used as an independent estimate of the energy required by the
vehicle per unit distance traveled. In this case, the average force
exerted by the vehicle at 50 mi/hr was calculated from the coastdown
conducted immediately after each steady-state test. This force was
assumed to be the average work done by the vehicle per unit distance
travelled during the preceeding 50 mi/hr steady-state test. This
coastdown derived force was then divided by the fuel consumed per unit
distance to give an estimate of vehicle efficiency. The efficiencies
computed in this manner were 13.1 percent for the road and 19.2 percent
for the dynamometer. Although the magnitudes of these numbers are
slightly different than those obtained directly using the torque and
speed data, they are quite similar and they indicate the same difference
-------
-7-
in the vehicle efficiency on the road versus that on the dynamometer.
It should be noted that the vehicle speed measurements used to cal-
culate the work done and the fuel consumed per unit distance were slight-
ly different for the dynamometer and the road. The road velocity was
obtained from a fifth wheel, while the velocity during dynamometer
operation was obtained from the dynamometer rear roll. A recent EPA
study (3) concludes that the dynamometer rear roll velocity overesti-
mates the road velocity of the vehicle which would occur for the same
vehicle drive wheel conditions. The net effect of correcting the data
of this report would be to shift all of the dynamometer data of Figure 1
both upward and to the right. The total relationship between the road
and dynamometer data would change very little, since; in any case the
effect is only 1 to 2%. The effect of an error in velocity is minimal
with the methodology used to analyze the data of this report since the
energy extracted from the vehicle was directly measured. In addition,
when computing the efficiencies, any systematic errors in the vehicle
velocity or simulated velocity would be self compensating and no error
would be introduced in the efficiency.
No significant systematic errors could be identified in the measure-
ments, therefore, attempts were made to identify differences in the
vehicle state or condition between the road and dynamometer tests.
Engine efficiency can vary with loading. On the dynamometer, 50
mi/hr steady-state tests were run with power absorber settings of 10.4,
11.4, and 12.4 horsepower. The data in the table below confirms the
fact that in the normal operating range, engine efficiency, as pre-
viously calculated, increases with load.
Power Absorber Percent
Setting (HP) J/cc Efficiency
10.4 5148 16.1
11.4 5346 16.7
12.4 5459 17.0
It could be hypothesized that the higher efficiencies observed on the
dynamometer resulted from greater vehicle loading on the dynamometer
than on the road. However, the actual power at the rear wheels was mea-
sured and no increase in loading was observed. In addition, increased
load would result in increased fuel consumption and exactly the opposite
was observed.
Throughout the efficiency calculations the energy available at the
rear wheels was used as the measure of useful work. Any changes in
energy dissipation between the engine and the drive wheels would affect
the fuel consumption for the same resultant energy at the wheels, there-
by affecting the calculated efficiency.
It could be hypothesized, for example, that the track to dynamometer
differences could be a result of severe drive wheel brake drag during
-------
-8-
the track measurements. However, this would have been observed in the
vehicle coastdowns. As discussed earlier, the coastdown measurements
were in reasonable agreement with the torque wheel data.
Part of the lower efficiency on the road can be accounted for by
the fact that during tests on the road, electrical power for the test
equipment used was taken from the vehicle, while during dynamometer
testing it was not. The inverter used on the road to change the 12-volt
DC level into an AC supply for the test equipment is capable of out-
putting 200 watts. At a 50 mph steady-state, this is an additional
unaccounted load of about 10 J/m. Considering Figure 1, this would have
the effect of shifting the road data to the right by 10 J/m. This is in
the right direction, but inadequate to explain more than a small frac-
tion of the observed differences. Even if very poor efficiencies are
assumed for the vehicle alternator and the inverter and the data were
shifted 50 J/m this cannot explain the observed differences.
In a similar manner, the power steering of the vehicle would cont-
ribute to the greater fuel consumption on the road. Again, however,
this effect is not sufficient to explain an appreciable part of the
observed differences, especially since the test track was a very large
(7.5 mi) oval.
The differential probably dissipated a different amount of energy
on the road than on the dynamometer. This would occur since the road
measurements were obtained at lower ambient temperature .and because very
little cooling air is supplied to the differential when the vehicle is
operated on the dynamometer. In general, it seems likely that the
differential operates at higher temperatures on the dynamometer than on
the road. This would reduce the lubricant viscosity and could have a
significant effect on the energy dissipation of the differential and
hence affect the efficiency. 'Earlier dynamometer experiments have indi-
cated that on the dynamometer the differential consumes approximately
1000 watts at 50 mi/hr (4). This is only 50 J/m therefore, even if the
differential losses doubled or tripled on the road this would not explain
the differences observed between the road and dynamometer results. In
addition differential effects would have been observed during the
vehicle coastdowns.
It seems impossible to explain the observed fuel consumption
differences by any mechanism which affects the load imposed on the
engine. The most plausible explanation is that the .differences were
caused by a direct temperature effect on the engine. The road tests
were conducted with the ambient temperature in the range of 52-67°F,
while the dynamometer tests were conducted in the range of 71-76°F. If
the vehicle engine operated under much richer fuel-air conditions at the
lower temperature, then this might explain the large observed difference.
The enrichment might occur either intentionally, or by malfunction, such
as a sticking automatic choke.
V. Conclusions
Two important conclusions can be reached from the reported data.
-------
-9-
The fuel consumed during these measurements was linerly related to the
measured energy extracted from the vehicle. Second, a significant
difference was observed between the fuel consumed by the vehicle on the
road and that consumed by the vehicle on the dynamometer when the same
energy was extracted from the vehicle.
A. Linearity of the Fuel Consumption Data
The observation that more fuel is burned as more work in extracted
from the vehicle is inherently logical. The surprising observation is
that, in this experiment, strong linearity was observed over a wide
range of vehicle work. This implies that it should be possible to
easily, and accurately predict the fuel economy effects of many para-
meters. For example, if the energy required to operate an air condi-
tioner or other assessory is known, then the resulting fuel consumption
effect could be easily and accurately predicted. Similarly, the fuel
consumption effects of changes in driving cycles could be predicted
from the energy demand of the driving cycle, if vehicle performances are
known for similar cycles.
B. Track versus Dynamometer
It is concluded that, under the conditions of this experiment
significantly different fuel consumption was measured on the track
versus the consumption measured on the dynamometer. It has been re-
ported (5) that temperature does affect fuel economy, but reported
temerpature effects are not large enough to account for the efficiency
differences which were observed, Differences in vehicle assessory
loading and temperature effects on the drive train can account for some
of the track to dynamometer difference, but not the majority of the
effect. It must, therefore, be concluded that either the road fuel
consumption data are in error, or that a very significant temperature
realted effect occurred during the track measurements. No evidence of
error in the road fuel consumption data could be detected. If the ob-
served effect was temperature induced it might be very vehicle specific
such as extreme choke action and therefore be difficult to confirm.
VI. Recommendations
The conclusions of this report are somewhat surprising; therefore,
the first recommendation is that these results be confirmed. Since in
general, ambient temperature may be a direct or indirect cause of vehi-
cle fuel economy differences between track and dynamometer measurements,
temperatures of vehicle components should be monitored in future programs.
These programs should also attempt to quantify the energy demand and
fuel consumption effects of various accessories and drivetrain compon-
ents.
In this test program all parameters, except track fuel consumption,
were confirmed by some redundant measurement. For example the wheel
torque data were confirmed by the coastdown results. In any future
program it is recommended that some confirmatory measurement of the
track fuel consumption be made. Even the simple recording of odometer
readings and tank fills would provide some confirmatory data.
-------
-10-
References
1. J. Dillard, Murrell, Previews of Dyno vs. Track Fuel Economy
Findings, EPA memo to John P. DeKany, August 3, 1977.
2. D.E. Foringer, "Gasoline Factors Affecting Fuel Economy," SAE
Paper No. 650427.
3. J. Yurko, "A Track to Twin Roll Dynamometer Comparision of Several
Different Methods of Vehicle Velocity Simulation," EPA Technical
Report to be released.
4. R. W. Burgeson, "Tire Test Variability," EPA Technical Report,
March 1978.
5. A.C.S. Hayden, "The Effect of Technology on Automobile Fuel Economy
under Canadian Conditions," SAE 780935, 1978.
-------
-11-
Appendix
Dynamometer Efficiency
Fuel Consumed Energy Extracted
cc/km J/M
122 638
106 495
128 739
123 684
108 511
130 760
116 607
103 456
122 669
125 735
109 568
135 853
129 733
112 541
123 657
107 487
132 754
103 504
124 746
118 663
126 739
111 582
114 606
101 457
121 700
117 622
103 462
124 713
105 453
127 710
115 573
103 430
122 664
124 674
106 493
132 779
120 637
105 460
128 720
126 679
108 483
135 778
118 629
103 471
123 721
119 ' 630
-------
-12-
Dynamometer Efficiency (cont.)
Fuel Consumed Energy Extracted
cc/km J/M
104 462
127 720
116 618
103 453
124 704
122 647
107 476
131 765
117 702
103 503
125 810
118 610
103 453
128 706
124 708
109 529
133 807
126 700
136 804
115 589
103 434
122 662
121 663
107 495
129 756
115 635
103 495
124 725
171 586
177 673
164 59J
178 646
* US. GOVERNMENT PRINTING OFFICE 1980- 651-112/0145
------- |