EPA-AA-SDSB-82-02
Technical Report
TTI Track/Dynamometer Study
Final Report
By
Martin Reineman
and
Glenn Thompson
January 1983
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 Sources
Office of Air, Noise and Radiation
U. S. Environmental Protection Agency
-------
Abstract
Seven passenger cars and one light truck were operated
over the EPA urban and highway driving cycles to compare fuel
economy measurements obtained on a test track, with the fuel
economy results obtained on a chassis dynamometer. The test
program was designed to duplicate, as closely as possible, the
track force loading (as determined by standard EPA road
coastdown procedures) on the dynamometer. Experimental
parameters which were investigated included loading differences
between front- and rear-wheel drive vehicles, volumetric versus
carbon balance fuel measurement techniques, coupled versus
uncoupled roll dynamometer tests, and curved track versus
straight track coastdowns.
I. Summary
Analysis of the results from this program provides the
following primary conclusion:
Dynamometer fuel economy is higher than track fuel
economy. Paired comparisons of the average track result and
the test configuration most representative of the EPA
certification test (uncoupled dynamometer rolls and the carbon
balance method of measuring vehicle fuel economy) show that the
fuel economy measured on the dynamometer test is higher than
the £rack result. This difference is statistically significant
at the 95 percent confidence level. The average difference for
the Federal Test Procedure (FTP) test was 8.1 percent, and the
average difference for the Highway Fuel Economy Test (HFET) was
11.7 percent.
Several conclusions about the reasons for the track to
dynamometer difference can be deduced from analysis of the data
trends. The average values presented with each of these
conclusions are the best estimates of the magnitude of each
effect. It is important to note, however, that because of the
small sample size and the observed vehicle-to-vehicle
variability, there may be large standard deviations associated
with these average values. These conclusions are:
1. FTP carbon balance fuel economy measurements are
higher than the corresponding volumetric fuel economy results.
In 16 of 16 comparisons, the FTP fuel economy results based on
carbon balance measurements are higher than corresponding
volumetric fuel measurements. The average difference between
FTP results is approximately 2 percent. Using the HFET cycle
eight of 16 comparisons showed higher carbon balance fuel
economy. Here, the average carbon balance measurement is about
0.5 percent higher than the comparable volumetric measurement.
-------
-2-
2. Coupling the dynamometer rolls reduces the measured
fuel economy. In 29 of 32 comparisons, uncoupled roll fuel
economy results are higher than coupled roll results. FTP and
HFET dynamometer fuel economies determined • with the rolls
uncoupled are about 3 percent higher than similar
determinations with the rolls coupled.
3. The FTP track-to-dynamometer fuel economy
discrepancy is lower for front-wheel drive vehicles than for
rear-wheel drive vehicles. In all comparisons, the average
difference between dynamometer and track fuel economy for the
FTP cycle was higher for rear-wheel drive vehicles than
front-wheel drive vehicles. For the HFET cycle this trend
reverses and the discrepancy is less for rear-wheel drive
vehicles.
The following general conclusions were also observed:
1. A track-to-dynamometer fuel economy difference
exists even when the dynamometer force replicates the track
force as accurately as possible with current test methods. An
average difference of 3 percent was observed between track FTP
cycle fuel economy results and dynamometer FTP results obtained
with the dynamometer rolls coupled and using volumetric fuel
measurement.
2. Force loading, as determined by the road coastdown
procedure, did not fully explain this program's discrepancies
between track and dynamometer fuel economy.
3. The discrepancies between this program's dynamometer
fuel economy and official EPA-Certification results appear to
be the combined effects of prototype-to-production differences
and the differences between the test procedures of the two
programs.
II. Introduction
It is generally acknowledged that EPA light-duty vehicle
fuel economy estimates exceed real-world fuel economy results.
Several EPA studies have attempted to quantify the magnitude of
the difference, and more importantly, provide explanations for
the existence of the observed difference. However, the results
from these studies have been questioned because these studies
were not specifically designed for addressing this issue and
consequently the results were often inconclusive. This study
was designed and performed to examine one specific aspect of
the overall difference between road and dynamometer fuel
economy—the difference between EPA fuel economy results when
the test is conducted on a test track, versus the result when
the test is duplicated on a chassis dynamometer.
-------
-3-
The track test results were obtained by testing in a
narrow range of environmental conditions at the test facility
of the Texas Transportation Institute (TTI), at College
Station/ Texas. Dynamometer tests conducted at TTI included
"standard" Federal test procedures and "modified" test
procedures using a roll coupler and increased dynamometer
horsepower settings.
Ill. Experimental Procedure
Comparisons of track and dynamometer fuel economy tests
were obtained by running a sequence of urban (EPA-FTP) and
highway (EPA-HFET) fuel economy tests on a test track and
repeating those tests on a chassis dynamometer. Current EPA
test procedures were used to adjust the chassis dynamometer to
simulate the road experience of the vehicle. The following
sequence of events was employed for each test vehicle:
procurement/inspection, track coastdown tests, track fuel
economy tests, dynamometer horsepower determinations,
dynamometer fuel economy tests.
A. Procurement/Inspection
The test fleet was selected to represent a diverse group
of vehicles that included a range of engine sizes, transmission
types and estimated fuel economies. Both rear-wheel drive
vehicles and front-wheel drive vehicles were selected. The
majority of the test fleet were small vehicles, which are
representative of current and future U.S. vehicles.
The vehicle test fleet used for the track/dynamometer
comparisons is shown in Table 1. Most of the information in
Table 1 is self explanatory. The data in the column labeled
EPA Guide MPG are the fuel economy estimates published in the
1980, 1981, and 1982 Gas Mileage Guide. These are the results
of the EPA urban cycle (FTP) fuel .economy tests. The EPA
highway MPGs are the result of the EPA highway cycle tests
which are used in computing a manufacturer's corporate average
fuel economy (CAFE), as required by Department of Energy fuel
economy standards.
Vehicles were obtained from several sources. Some
vehicles were borrowed from private owners by offering leaner
vehicles and cash incentives. Others were leased from auto
dealers. All vehicles were visually inspected, tuned to
manufacturers specifications, checked for proper wheel
alignment, and test driven prior to instrumenting for fuel
-------
Table 1
Test Fleet Descriptives
1.
2.
3.
4.
5.
6.
7.
8.
Vehicle
1980 Oldsmobile
Cutlass Supreme
1980 Ford
Pinto Hatchback
1981 Ford
Custom F-100
1980 Chevrolet
Citation
1981 Ford
Escort
1981 Plymouth
Horizon
1981 AMC
Concord
1982 Honda
Engine
CID/Cyl.
260/8
140/4
300/6
173/6
98/4
104/4
258/6
81/4
Drive Inertia EPA Guide EPA Highway
Trans Wheels (lo) MPG (FTP)
A3
A3
A3
A3
M4
A3
A3
M5
R 4,000
R 3,000
R 4,250
F 3,000
F 2,500
F 2,750
R 3,500
F 2,250
18
21
18
20
30
26
19
41
MPG (HFET)
24
29
22
30
44
35
26
55
Civic
-------
-5-
economy measurement and coastdown time determinations. A
detailed description of each vehicle is presented in Tables A-l
through A-8 in the Appendix.
B. Track Coastdowns
Road force measurements were determined by using the test
procedures specified in EPA Advisory Circular (A/C) No. 55B,
"Determination and Use of Alternative Dynamometer Power
Absorption Values."[1] The test instrumentation used for
recording the required velocity/time data is presented in Table
B-l of the Appendix. In addition to the typical straight track
coastdowns described in A/C No. 55B, coastdown times were
measured over a curved section of the TTI track. Figure B-2 of
the Appendix describes the layout of the test track including
the portion designated for curved coastdown tests.
C. Track Fuel Economy Tests
Track fuel economy tests included the same sequence of
events required for a typical dynamometer test, including the
following procedures and test conditions: 1) a preparatory
warm-up cycle, 2) a 12-36 hour stabilization period at 68 to
86°F, 3) a true cold start, 4) constant specification test
fuel, and 5) ambient test temperatures of between 68 and 86°F.
Additional constraints for conducting track tests were winds
averaging less than 15 mph with gusts less than 20 mph, zero
precipitation, and a dry track.
Several minor modifications were made to the standard
dynamometer fuel economy (emissions) test procedures[2] to
reduce the test time. These changes included omitting
evaporative emissions measurements and deleting the requirement
for heating the test fuel in the vehicle from 60 to 84°F.
These variations from the typical EPA test procedure were
consistently made in all track and dynamometer testing. None
of the deviations from standard EPA measurement procedures were
expected to change the validity of the test results. Howell
Hydrocarbons' EEE Clear amd AMOCO Indolene were used as the
test fuels. The specifications of these fuels are summarized
in Table B-3 of the Appendix.
Two technicians were required to operate a test vehicle on
the track—one person controlled the vehicle speed while the
second person steered the vehicle. The equipment used for the
track fuel economy tests is also presented in Table B-l of the
Appendix. Following the required preparatory and stabilization
periods, a typical sequence of track fuel economy tests
consisted of a cold start urban fuel economy" test ~(FTPT
followed by a warm-up highway fuel economy test (HFET) and then
-------
-6-
a second HFET test. All track tests were conducted with the
tires adjusted to the manufacturer's minimum recommended
inflation pressure. The tire specifications ar.e summarized in
Tables A-l to A-8 of the Appendix.
D. Dynamometer Horsepower Determinations
Track coastdown data from straight and curved track
results were reduced and corrected, following the guidelines of
A/C No. 55B, to calculate values of total road force and
dynamometer 55-45 mph coastdown time. These data were then
used to set an equivalent force loading on the dynamometer.
The corresponding horsepower settings were developed from
straight and curved coastdown times using a Clayton twin-roll
dynamometer and the same dynamometer with the rolls coupled.
Thus for each test vehicle, two values of road force were used
to determine a total of four values of dynamometer force (or
power) loading.
Fu|l range dynamometer coastdowns (60-20 mph) were
conducted for each vehicle. These data were analyzed and
compared with the track coastdown data to determine the
differences between the vehicle loading on the track and the
loading curve obtained on the Clayton dynamometer when it was
adjusted to match the 50 mph road force value.
E., Dynamometer Fuel Economy Tests
All vehicles received at least three series of
FTP-HFET-HFET tests. To the extent possible, the dynamometer
tests were conducted at conditions identical to the track fuel
economy tests. The values for parameters such as: inertia
weight, axle loading, driver, volumetric fuel measurement
instrumentation, test fuel composition, and ambient temperature
conditions were all identical to, or adjusted " as close as
possible, to the test conditions which existed for the track
fuel economy measurements. For example, the mass of the
vehicle, and the axle loading on the dynamometer were the same
(or adjusted to be the same) as on the track. The driver and
the volumetric fuel measurement equipment were also the same
for track and dynamometer tests. A list of the test equipment
which were used for the measurement of vehicle fuel consumption
is presented in Table B-4 of the Appendix.
Table 2 is an overview of all test activities which were
performed during this program.
-------
Table 2
Test Plan
1.
2.
3.
4.
5.
6.
7.
8.
Vehicle
Cutlass
Pinto
P-100
Citation
Escort
Horizon
Concord
Civic
Trdck
Straight
7
7
7
7
7
7
7
7
Coastdowns [1]
(S) Curved (C)
7
7
7
7
7
7
7
7
Dynamometer Coastdowns [3J
Track FTP-HFET
Sequence [2]
3
3
3
9
3
3
3
3
Rolls
S
4
4
4
4
4
4
4
4
Uncoupled
C
4
4
4
4
4
4
4
4
Ho 11s
S
4
4
4
4
4
4
4
4
Coupled
C
4
4
4
4
4
4
4
4
Dynamometer FTP-HFKT
Rolls Uncoupled
S C
3
3
3
9
3
3 3
3
3 3
Sequence 14 J
Rolls Coupled
S C
3
J
3
9
3
3 3
3
J 3 V
11]Each track coastdown run consists of one coastdown in one direction immediately followed by a coastdowa iu tne opposite
direction. Each coastdown sequence thus requires reporting data from 8 pairs (runs) of tests.
[2] A track sequence consists of one FTP followed by two HFET tests.
13) Each dynamometer coastdown consists of eight measurements of coastdown times.
S = Straight track coastdown time.
C ° Curved track coastdown time.
[4] A dynamometer sequence consists of one FTP followed by two HFET tests. Volumetric and carbon balance fuel consumption
results are measured during each test.
-------
-8-
IV. Results
A. Track and Dynamometer Coastdown Results
Table 3 summarizes the track coastdown data and presents
the corresponding 50 mph force and horsepower results for each
vehicle. Total force is calculated by the equation:
F = FO + F2V2
where,
F = total road force at 50 mph
FO = constant force term
F2V2 = quadratic force term.
The force coefficients Fg and F2 are computed from:
FQ » MAQ
F2 = MA2
AQ and A2 are the vehicle acceleration coefficients
which are determined by fitting a quadratic acceleration versus
speed equation to the track or dynamometer speed versus time
data. The temperature at which track coastdowns were conducted
sometimes differed from the temperature of the track fuel
economy tests since those tests were conducted at different
times. Therefore, in accordance with A/C No. 55B, acceleration
coefficients were corrected to 68°F, 29 in. of Hg, and zero
wind speed.
The terms M and V in the above equations represent the
corresponding vehicle mass and velocity, respectively. The
vehicle mass includes a correction factor of 1.-035 to account
for the rotating inertia of the four wheels and the drive axle
components.
The average values for the track and dynamometer coastdown
force coefficients are included in Appendix E.
Table 4 summarizes the dynamometer coastdown results *
Since two separate sets of dynamometer coastdowns were
conducted, care must be taken in understanding these results.
The first set of dynamometer coastdowns were the 55 to 45 mph
.coastdowns which are used to determine the dynamometer
adjustment necessary to reproduce the measured road force at 50
mph. The dynamometer adjustments which resulted from these
-------
Table 3
Track Coastdown Results
Vehicle
1.
2.
3.
4.
5.
6.
7.
8.
Cutlass
Pinto
F-100
Citation
Escort
Horizon
Concord
Civic
Type[l]
S
C
S
C
S
C
S
C
S
C
S
C
S
C
S
C
Vehicle Mass
(lbm)[2]
4113/1702
4091/1686
3199/1341
3192/1338
4298/1707
4317/1726
3217/1989
3222/1978
2519/1444
2513/1433
2789/1690
2803/1686
3631/1493
3615/1488
2357/1334
2356/1332
55-45
Coastdown
Time
(sec)[31
15.81
15.34
11.78
11.60
12.61
12.24
13.90
13.79
13.87
13.32
13.52
13.41
14.39
14.11
12.14
11.98
Total
Road Force
@ 50 mph
(Ibf)
118.1
121.4
118.4
120.6
157.3
161.6
100.5
101.4
83.8
87.4
94.8
95.3
113.2
115.6
86.3
87.7
Total Road
Horsepower
6 50 mph
(hp)
15.7
16.2
15.8
16.1
21.0
21.5
13.4
13.5
11.2
11.7
12.6
12.7
15.1
15.4
11.5
11.7
1] Coastdown type: S = Straight track section/ C = Curved track
i section.
[2] Total mass/drive axle mass loading. Total loading includes 0.035 x
vehicle mass for driving and. non-driving rotating equivalences.
[3] Track coastdown times corrected to zero wind, 68°F and 29 in. Hg,
and equivalent dynamometer mass loading.
-------
Table 4
Dynamometer Coastdown Results
1.
2.
3.
4.
5.
6.
7.
8.
Vehicle
Cutlass
Pinto
F-100
Citation
Escort
Horizon
Concord
Civic [5]
Type[l]
D-U
D-C
D-U
D-C
D-U
D-C
D-U
D-C
D-U
D-C
b-u
D-C
D-U
D-C
D-U
D-C
Vehicle
Mass[2]
4071/1666
4071/1669
3056/1295
3056/1294
4325/1732
4326/1735
3056/2065
3056/2061
2544/1497
2544/150$
2798/1734
2798/1701
3563/1442
3563/1463
2290/1334
2290/1335
55-45
Coastdown
Time
(sec) [3]
16.23
16.34
12.06
11.32
13.05
13.10
13.28
13.29
14.15
12.98
13.21
13.11
13.55
13.50
12.07
12.07
Total Force
@ 50 mph
(Ibf ) [3J
114.2
114.9
110.5
117.8
151.9
149.9
105.2
105.0
82.0
89.5
97.2
98.2
120.6
120.8
85.5
86.3
Total
Horsepower
@ 50 mph
(hp) 13]
15.2
15.3
14.7
15.7
20.3
20.0
14.0
14.0
10.9
11.9
13.0
13.1
16.1
16.1
11.4
11.5
Dynamometer
Horsepower
(§50 mph
(AHP) 14]
10.5
10.9
9.7
10. a
15.1
lb.3
8.3
8.4
5.8
b.l
7.9
8.0
10.97
11.0
8.4
9.0
o
I
[1] Coastdown type: D-U = Dynamometer rolls uncoupled, D-C = Dynamometer rolls
coupled. D-U and D-C tests attempt to match straight track coastdown force.
[2] Total mass/drive axle mass loading. Total loading includes 0.018 x vehicle
mass for driving rotating equivalences.
[3] Coastdown time, total force @ 50 mph, and total horsepower @ 50 mph measured
during 60-20 mph coastdowns. Dynamometer times are not ambient corrected.
[4] Dynamometer AHP from PAU determination.
[5] Dynamometer horsepower inadvertantly set at equivalent AHP plus 10 percent for
air-conditoning simulation.
-------
-11-
coastdowns are given in the right-hand column of Table 4
(dynamometer horsepower at 50 mph (AHP)). These are the
dynamometer adjustments which were used when the fuel economy
measurements were obtained.
After completion of the fuel economy measurements, a
second series of coastdowns were conducted to characterize the
dynamometer performance throughout the speed range of the fuel
economy tests. These coastdowns were conducted over the usual
coastdown speed range, 60 to 20 mph. The data were analyzed to
yield force versus speed in the same manner as the road
coastdown data. However, no ambient corrections were necessary
since laboratory conditions were well controlled and wind and
air density effects are not present on dynamometers. A mass
correction factor of 1.018 was used to account for the effects
of rotating inertia. This is less than the road correction
factor since the non-drive wheels of the vehicle are not in
motion on the dynamometer. The force coefficients of this
analysis were then used to calculate the force acting on the
vehicle for all speeds below 60 mph. This force versus speed
information is plotted in Figures 1 through 8. Each figure
contains force versus speed curves based on 60-20 mph straight
track, coupled roll dynamometer, and uncoupled roll dynamometer
coastdown data.
The force coefficients were used to calculate the total
vehicle-dynamometer force at 50 mph, the power for the system
at 50 mph, and the 55-45 mph coastdown times for the vehicle on
the dynamometer. The coastdown time for the vehicle on the
dynamometer is a cross-check against the track coastdown time
of Table 3. These are all presented in Table 4.
B. Track 'a'nd Dynamometer Fuel Economy Results
Table 5 is a summary of the average fuel economy results
(track and dynamometer) for each test vehicle.
Track and dynamometer volumetric fuel economy measurments
are calculated using an average fuel flowmeter temperature for
a segment of the FTP (e.g., Bag 1) or the entire HFET, and the
gasoline density correction factors (€3) in SAE recommended
practice J1256. The average fuel flowmeter temperature is
based on the temperature at the start and end of an FTP segment
or an HFET. FTP composite fuel economy is then calculated
using the EPA cold/hot weighting factors to provide a correct
comparison with the weighted carbon balance fuel economy
measurements.
Individual track and dynamometer results _for the test
vehicles are presented in Tables D-l to D-48 of the Appendix.
-------
-12-
co T
CD
LUGO
O
o:
o
u_
CD
CD
o
o
c=>
a
o
o
TRflCK
DYNO - C
DYNO - UC
10.00
20.00 30.00 40.00 50.00 60.00
SPEED (MPH)
CUTLflSS - TRflCK/DYNflNOMETER COflSTDOUNS
FIGURE 1
-------
-13-
o
O
CD
CO
CM ••
GQ
LUOO
CJ
CC
O
<=>
CM
TRflCK
DYNO - C
DYNO - UC
0.00
10.00
20,00 30.00 40.00
SPEED (NPH)
50.00 60.00
PINTO - TRflCK/DYNflNOMETER COflSTDOUNS
FIGURE 2
-------
-14-
TRflCK
DYNO - C
DYNO - UC
20.00 30.00 40.00
SPEED (MPH)
50.00
60.00
F10Q - TRflCK/DYNflNONETER COflSTDOUNS
FIGURE 3
-------
-15-
o
o
CD T
O
CD
CM
O
O
O
O
CD
LUGO
O
CT
O
U_
CO
o ..
TRflCK
----- DYNO - C
DYNO - UC
0.00
10.00
20.00 30.00 40.00 50.00 60.00
SPEED (MPH)
CITflTION - TRRCK/DYNflMQMETER COflSTDOUNS
FIGURE 4
-------
-16-
DYNO - C
DYNO - UC
20.00 30.00 40.00
SPEED (MPH)
50.00
60.00
ESCORT - TRflCK/DYNflMONETER CORSTDOUNS
FIGURE 5
-------
-17-
TRflCK
DYNO - C
— DYNO - UC
10.00
20.00 30.00 40.00
SPEED (MPH)
50.00
60.00
HORIZON'- TRflCK/DYNflMOMETER COflSTDOUNS
FIGURE 6
-------
-18-
o
o
O
CO
O
o
O
O
LO-
GO
LUGO
<_>
o:
o
o
CO
o 4
CNJ
0.00
10.00
TRflCK
DYNO - C
DYNO - UC
20.00 30.00 40.00 50.00 60.00
SPEED (MPH)
CONCORD - TRflCK/DYNnnONETER COflSTDOUNS
FIGURE 7
-------
-19-
TRflCK
- DYNO - C
DYNO - UC
0.00
20.00 30.00 40.00
SPEED (MPH)
50.00 60.00
CIVIC - TRflCK/DYNRMOMETER COflSTDOUNS
FIGURE 8
-------
Table 5
Fuel Economy Results, mpg[l]
Coupled Roll
Dynamometer
Uncoupled Roll
Dynamometer
Cycle
FTP
HFET
Vehicle
Cutlass
Pinto
F-100
Citation
Escort
Horizon
Concord
Civic[2]
Cutlass
Pinto
F-100
Citation
Escort
Horizon
Concord
Civic[2] .
Track
13.68
17.04
14.62
18.83
23.33
20.77
18.24
33.33
19.19
23.53
18.47
26.70
34.48
30.18
28.08
43.52
Volumetric
13.98
18.18
15.16
19.02
24.11
21.60
18.60
34.03
20.23
25.75
20.51
28.79
39.93
33.04
28.63
42.88
Carbon
Balance
14.56
19.36
15.25
19.06
24.21
22.89
18.83
35.27
20.72
26.84
20.42
28.21
39.21
33.04
28.83
43.41
Carbon Official EPA
Volumetric
14.09
19.45
15.68
18.83
24.65
22.22
18.65
36.62
20.31
26.68
21.60
29.43
40.57
33.87
29.33
48.38
Balance
14.49
19.46
15.85
19.06
24.81
23.42
19.36
37.88
20.89
27.60
21.29
28.75
39.97
33.82
29.38
48. 8b
Results
18
21
18
20
30
26
19
41
24
29
22
30
44
35
26
55
I
to
o
I
"I'll Eacn mpgvalue is a 3-vehicle average, with the exception of the
Chevrolet Citation results which are average values based on nine
repeat tests. Offical EPA results are the corresponding FTP or HFET
values from offical EPA certification tests.
[2] Dynamometer horsepower inadvertantly set at equivalent AHP plus 10
percent for air-conditioning simulation.
-------
-21-
These tables include actual distance measurements and fuel
economy and fuel consumption calculations for the three
segments (Bag 1, Bag 2, and Bag 3) of the FTP, and composite
calculations for the FTP. Similar results are presented for
the warm-up HFET, and the official HFET, which immediately
follows the warm-up HFET. Tables C-l through C-8 of the
Appendix present the statistics of the composite FTP results
and the official HFET results (in miles/gallon) for all
vehicles and all test configurations. Included are
calculations of the mean, standard deviation, and coefficient
of variation for each vehicle/test configuration.
Table 6 summarizes the dynamometer and track fuel economy
results as the percent fuel economy deviation from the track
fuel economy. Thus, for the Oldsmobile Cutlass, the coupled
roll dynamometer volumetric result is 2.2 percent higher than
the corresponding track measured fuel economy. The data are
referenced to the track fuel economy result because it is most
representative of an actual driving experience, and therefore
the most logical method of comparison. The official EPA
results from emission certification and highway fuel economy
tests are also compared to the track test data. The data are
stratified according to drive axle type , in order to
characterize the higher tire losses expected for front-wheel
drive vehicles.
V. Analysis of Results
A. Findings Based on Comparisons within TTI Data
Tables 5 and 6 demonstrate that large vehicle-to-vehicle
variations were observed. Thus, the mean values discussed in
the following analysis may have large values of standard
deviation associated with them because of the combination of
inherent variability of fuel economy measurements and the small
sample sizes. Consequently, statistical confidence is obtained
only for the larger effects, such as the differences between
the track results and the carbon balance results obtained on
the dynamometer with the rolls uncoupled. The individual
vehicle results, however, show less variation. For example,
the overall coefficient of variation, based on samples of nine
track and 36 dynamometer tests with the Citation, is 1.7
percent. This value, approximately 2 percent, is believed to
be typical. Therefore most of the analysis for the individual
steps of this program, such as coupling the dynamometer rolls,
is presented in terms of the data trends. When a majority of
the test results show a directionally consistent shift, this is
judged to be of engineering importance, even if the amount of
the shift is similar in magnitude to the coefficient of
variation.
-------
Cycle
R-FTP[2]
F-FTP
R-HFET
F-HFET
Table 6
Percent Deviation from Track Fuel Economy Results[1]
Vehicle
Cutlass
Pinto
F-100
Concord
X
s
Citation
Escort
Horizon
X
s
Cutlass
Pinto
F-100
Concord
X
s
Citation
Escort
Horizon
X
s
Coupled
Dynamometer
Volumetric
2.2
6.7
3.7
2.0
3.7
2.2
1.0
3.3
4.3
2.9
1.7
5.4
9.4
11.0
2.0
7.0
4.1
7.8
15.8
9.5
11.0
4.2
Roll
Results
Carbon
Balance
6.7
15.0
4.5
3.2
7.4
5.3
1.2
3.7
10.2
5.0
4.6
8.0
14.1
10.6
2.7
8.9
4.8
7.8
13.7
9.5
10.3
3.0
Uncoupled
Dynamometer
Volumetric
3.0
14.1
7.3
2.3
6.7
5.4
0.0
5.7
7.0
4.2
3.7
5.8
13.4
16.9
4.5
10.2
6.0
10.2
17.7
12.2
13.4
3.9
Roll
Results
Carbon
Balance
6 .2
15.6
8.6
6.1
9.1
4.7
1.2
6.3
12.8
6.8
5.8
8.9
17.3
15.3
4.6
11.5
5.8
7.7
15.9
12.1
11.9
4.1
Official
EPA
Results
39.2
30.6
23.3
4.1
24.3
15.0
6.2
28.6
25.2
20.0
12.1
20.0
23.2
19.1
-7.4
13.7
14.2
12.4
27.6
16.0
18.7
7.9
i
M
to
[1]Positivenumbers indicate dynamometer fuel economy is higher than
track fuel economy.
[2] R = Rear-wheel drive and F = Front-wheel drive
-------
-23-
Four findings are apparent from an examination of the data
from these eight vehicles: 1) vehicle fuel .economy on the
dynamometer is greater than the vehicle fuel economy on the
track, even when the 50 mph dynamometer force is matched to the
track force as accurately as possible, 2) vehicle fuel economy
when the dynamometer rolls are coupled is less than
corresponding measurements made without the roll coupler, 3)
fuel economy measurements obtained by the carbon balance method
are greater than those obtained by volumetric methods, in
general, and 4) dynamometer fuel economy is 'higher than track
fuel economy even when dynamometer force versus speed loading
appears to exceed track loading conditions.
1. Dynamometer vs. Track Fuel Economy
Table 6 shows that dynamometer fuel economy measured with
a volumetric flowmeter is always greater than the track fuel
economy results when the 50 mph road force was matched to the
50 mph dynamometer force as accurately as possible.
A paired statistical t-test analysis using the data of
Table 5 proves that dynamometer fuel economy as determined in
the manner most representative of the EPA certification test
(uncoupled dynamometer rolls and the carbon balance measurement
method) is significantly higher than the corresponding track
fuel economy. This finding is statistically significant at the
95 percent confidence level and is especially noteworthy
because much care was taken to duplicate the total road force
on the chassis dynamometer. The average difference between
these test configurations was 8.1 percent for the FTP cycle and
11.7 percent for the HFET.
A difference exists between the fuel economy measured on
the track and on the dynamometer even when the dynamometer
replicates the track experience as accurately as possible with
current methods. With fuel economy measured with a volumetric
flow meter to provide a consistent comparison, the results
obtained on the dynamometer with the rolls coupled are still
considerably greater for all vehicles than the results obtained
on the track. The average difference was 3.3 percent for the
FTP cycle, and 8.7 percent for the HFET.
Rear-wheel drive vehicles showed a greater fuel economy
difference between the track and. dynamometer results for the
FTP cycle than occurred for front-wheel drive vehicles. This
was observed for all four possible comparisons—volumeteric and
carbon balance measurement methods with the dynamometer rolls
coupled or uncoupled. In the case of the HFET cycle, exactly
the reverse trend was observed. Front-wheel drive vehicles
showed a greater difference between the track and the
-------
-24-
dynamometer results than was observed from rear-wheel drive
vehicles. Again, this observation was consistent for all four
possible comparisons.
The data from the Honda Civic is not included in Table 6,
and was not used in the previous analysis, since in the case of
the Civic the dynamometer was not adjusted to match the
measured road force of the vehicle. For this vehicle, the
dynamometer adjustment was inadvertantly increased by 10
percent, the factor used to simulate air-conditioning usage in
the official EPA certification test procedure. Interestingly,
even with a 10 percent increase in the dynamometer actual
horsepower, all average values of the FTP fuel economy of the
Civic measured on the dynamometer still exceed the average fuel
economy measured on the track. In the case of the HFET
results, the average dynamometer measured fuel economy was
still greater than the average track fuel economy for those
tests in which the dynamometer rolls were not coupled. Only in
the case of the HFET results with the dynomometer rolls
coupled, did the track fuel economy results exceed those
obtained on the dynamometer.
2. Coupled vs. Uncoupled Roll Results
The effects of coupling the dynamometer rolls can also be
observed from an examination of the fuel economy data reported
in Table 5. In 29 of 32 paired comparisons, coupled roll fuel
economy is lower than uncoupled roll fuel economy. Volumetric
measurements on rear-wheel drive vehicles produced an average
fuel economy decrease of 2.7 and 2.9 percent for the FTP and
HFET, respectively, when the dynamometer rolls were coupled.
The corresponding carbon balance fuel economy decreases are 1.6
and 2.4 percent. A similar analysis for front-wheel drive
vehicles shows fuel economy decreases of 2.8 and 4.4 percent
for volumetric measurements conducted over the FTP and HFET,
respectively, and decreases of 2.9 and 4.3 percent for FTP and
HFET results, respectively, when carbon balance is used to
determine fuel economy. This study produced an overall 3.0
percent decrease in fuel economy with a roll coupler device.
This result is in good agreement with previously reported
results obtained at the EPA-MVEL in 1978, where a small number
of tests produced coupled roll effects of 1.8 and 3.8 percent
for the FTP and HFET, respectively.[3]
It should be recalled that throughout this study, coupled
roll dynamometer tests were based on an independent coupled
roll PAU determination. Thus, the coupled and uncoupled roll
total horsepower at 50 mph should be in good agreement. This
is demonstrated in Table 4, where the values of total
horsepower agree within the engineering precision of the
procedure currently used for determining PAU adjustment.
-------
-25-
Previous studies investigating dynamometer velocity and
acceleration simulation concluded that the roll coupler
provides a more accurate method of simulating.the road driving
experience of a vehicle by reducing the speed simulation/tire
slip errors. [4] Thus, it is not surprising that the coupled
roll fuel economy more closely agrees with the track fuel
economy results, as this test program demonstrates.
An analysis of the fuel economy data indicates that the
magnitude of the coupled roll difference is slightly higher
over the HFET cycle. Coupling the dynamometer rolls reduces or
eliminates tire-roll slip which is responsible for vehicle
speed simulation errors. Computer modeling indicates that the
accumulated speed error, the distance error, is greater over
the HFET cycle. This may be the reason that coupling the
dynamometer rolls produced a greater fuel economy effect on the
HFET cycle than on the FTP. Similar experimental results have
been observed by researchers at Ford Motor Company.[5]
3. Carbon Balance vs. Volumetric Measurements
Table 5 presents 32 paired comparisons of carbon balance
and volumetric fuel economy measurements. The data are
analyzed and partitioned according to FTP and HFET results
because there is no logical basis for expecting differences in
fuel measurement method as a function of roll configuration or
vehicle drive axle type. In all comparisons, the FTP fuel
economy results based on carbon balance measurements are higher
than corresponding volumetric fuel measurements, while eight of
16 comparisons using the HFET cycle showed higher carbon
balance fuel economy. This study finds an^average 2.5 percent
(standard deviation of 2.0) and 0.4 percent (standard deviation
of 2.0) fuel economy increase for the FTP and HFET results,
respectively, when fuel economy is calculated using the carbon
balance technique. These results corroborate the findings of
earlier EPA laboratory studies. For example, the average
carbon balance to volumetric difference of 2.5 and 0.4 percent
for the FTP and HFET, respectively, are similar to the average
FTP and HFET values of 1.6 and 0.2 percent previously reported
by Newell.[6]
While both fuel measurement methods contain sources of
possible error, the flow meter approach is direct and simpler.
Many known potential errors of the carbon balance method, such
as fuel lost by evaporation, carbon particles lost in the
exhaust system or in the sampling system, the assumptions of
values of density and H/C ratio, which may be too high, and the
inability of the flame ionization detector to measure
oxygenated hydrocarbons, all result in higher measured fuel
economy.
-------
-26-
Additional analysis of the individual ,fuel economy data in
Appendix D shows that Bag 1 results have the largest carbon
balance to volumetric difference. A bag-by-bag analysis shows
average fuel differences of 8.4, 1.4, and 3.9 for Bags 1, 2,
and 3 of the FTP, respectively. Previous experimental studies
have determined that a significant portion of the carbon
balance/volumetric differences in Bag 1 can be attributed to
fuel which must be replenished to the fuel float bowl at the
beginning of the cold-start portion of the FTP.[7]
Experimental results usually indicate that carbon balance
measurements are subject to greater variability than volumetric
measurements.[7,8] However, the data of Appendix D indicate
that the variability of carbon balance and volumetric
measurements were similar during the course of this program.
4. Track/Dynamometer Loading Comparison
Since the results of this study show that vehicle fuel
economy obtained from a dynamometer test is greater than that
obtained on a test track, the data were analyzed for possible
reasons for this difference.
The fuel economy differences were first examined for the
presence of any systematic factors which would result in lower
dynamometer loading. To assist in this examination, the plots
of total force versus speed (Figures 1-8) were examined.
Inspection of the force curves does not show a systematic
underloading effect on the dynamometer. The dynamometer force
curves are higher than the corresponding road force curves for
some vehicles and lower for others. This is true for both
front and rear-wheel drive vehicles.
Weight variations between track and dynamometer tests were
examined, but again, no systematic differences are apparent.
Effective vehicle mass on the dynamometer is higher for some
vehicles and lower for others than the corresponding mass
during the track tests. In general, the average total vehicle
mass and the drive axle mass were within 1 percent of the track
loading conditions.
An energy modeling analysis was conducted to combine the
effects of mass and force into a single parameter, specific
energy.[9] Table 7 is a summary of the energy analysis. The
actual energy consumed, eln, and the predicted energy
required, eReq, to operate a vehicle over the FTP or HFET is
presented on a BTU/mi basis for each vehicle. The energy
consumed for a particular cycle is based on the weighted
volumetric fuel measurement. The predicted energy requirements
were calculated from a computer energy model which used the
force-speed curves shown in Figures 1-8 and the vehicle mass to
-------
Table 7
Energy Analysis
Vehicle
1. Cutlass ein[2] =
eReq. I3)
2. Pinto
3. F-100
4. Citation
-
5. Escort
6. Horizon
7. Concord
-
8. Civic(4J
__.
Track
8,509
857
0.101
6,831
796
0.117
7,962
995
0.125
6,182
696
0.113
4,989
553
0.111
5,604
642
0.115
6,382
783
0.123
3,492
"i53
0.158
FTP
D-C [11
8,326
848
0.102
6,403
748
0.117
7,678
975
0.127
6,120
708
0.116
4,828
607
0.126
5,389
651
0.121
6,258
777
0.124
3,410
534
0.156
D-U [11
8,261 eln =
834 eReq. =
0.101 ej
-------
-28-
determine the total vehicle load during the test cycle. The
ratios of required to input energy are presented as an estimate
of the energy efficiency. Table 7 shows a general pattern of
higher dynamometer specific energy, relative to the track
condition for the front-wheel drive vehicles due to greater
tire energy losses on the dynamometer. For three of four
vehicles, the dynamometer required specific energy was greater
than the track specific energy. All rear-wheel drive vehicles
were predicted to require less specific energy on the
dynamometer relative to the road test condition. In either
case, however, the modeled dynamometer energy requirements were
generally within 1 percent of the modeled track energy
requirement. Thus, the specific energy analysis is not
sufficient to explain the observed differences between road and
dynamometer fuel economy results.
The energy modeling analysis was based on track coastdowns
obtained on the straight segments of the track. Consequently,
track curvature effects could introduce an effect which would
not be observed in the previous analysis. To test the track
curvature effect, all vehicles were also coasted down on curved
segments of the track. Table 3 shows a consistent small
increase in road force of about 2 percent due to road curvature
effects.
The effect of the track curvature on the vehicle fuel
economy was investigated by conducting fuel economy tests on
the Horizon and Civic with the dynamometer adjusted to simulate
the loadings measured during the curved track coastdowns.
These results are summarized in Tables C-6. and C-8 of the
Appendix. Although considerable data scatter is evident, fuel
economy results are about 1 to 2 percent lower using the curved
track loadings. This observed fuel economy effect is the upper
bound of the anticipated track effect because these dynamometer
loadings simulated operation of the vehicle on a continuous
curved surface, while the test track has only several curved
sections.
This analysis found no vehicle loading errors which could
account for the observed fuel economy differences. But, it
should be noted that the coastdown measurements are indirect
quasistatic approaches to force and energy measurements—not
direct measurements of the dynamic system which is acting
during road and dynamometer transient tests. It is possible
that a dynamic error in the dynamometer could yield a
systematic fuel economy effect. However, such an error would
have to be quite large to cause the observed effects and it is
unlikely that errors of this magnitude exist. It should be
noted that the Civic fuel economy measured on the dynamometer
-------
-29-
was higher than the track result even though the dynamometer
was inadvertently adjusted to simulate a 10 percent power
overload.
The energy efficiency data of Table 7 suggest that
vehicles operate more efficiently on the dynamometer. This
finding is consistent with previous EPA track-to-dynamometer
studies.[10] This energy efficiency discrepancy may be due to
differences in internal vehicle operating parameters during the
track and dynamometer tests. This program collected
measurements of four fluid temperatures during the road and
dynamometer tests of the Citation. An inspection of the
temperature data indicates that the peak temperatures of engine
coolant, engine oil, and transmission oil are approximately
20°F higher on the dynamometer than are the peak temperatures
achieved under track test conditions. Higher oil temperatures
suggest lower lubricant viscosity and less frictional energy
dissipation.
B. Findings Based on Comparisons with EPA-Certification
Data
The fuel economy results of this program are compared to
official EPA-Certification results in Tables 5 and 6. Table 5
compares fuel economy results on an average absolute mpg basis,
while Table 6 expresses the track-to-dynamometer difference as
a percentage.
The significance of the comparison to EPA-Certification
results is in the magnitude of the differences between the
track-to-official EPA numbers, and the carbon balance uncoupled
roll results versus official EPA results. Table 6 shows an
average of 24 and 14 percent higher fuel economy for official
results from rear-wheel drive FTP and HFET test
configurations. Similar results for the front-wheel drive FTP
and HFET tests are 20 and 19 percent, respectively. Table 6
also shows a large discrepancy between official EPA fuel
economy and this program's uncoupled roll, carbon balance fuel
economy results (the test configuration closest to official
Certification test procedures). An inspection of the data
shows an increase of 15 percent and 2 percent for the
rear-wheel drive vehicle FTP and HFET results, respectively.
The 15 percent value for the rear-wheel drive vehicles on the
FTP cycle is the difference between the 24.3 percent deviation
of the official EPA rear-wheel drive FTP results from the track
results, and the 9.1 percent deviation of the uncoupled roll,
carbon balance results from the track. The corresponding
increases for the front-wheel drive vehicle FTP and HFET
results are 13 percent and 7 percent, respectively.
-------
-30-
Historically, EPA has established the presence of a
production-to-prototype vehicle difference as a mechanism of
explaining the difference between EPA official fuel economy
results and results obtained from the EPA Emission Factors
program. The median discrepancy currently observed between the
Emission Factor and Certification programs is 8 and 6 percent
for the FTP and HFET, respectively.[11] The larger
discrepancies of this program appear to be a combination of
production-to-prototype vehicle differences and differences in
the procedures of this test program and those of the EPA
Certification program. For example, in this program all
vehicles were operated on the track and on the dynamometer at
their production weights plus the additional weight of the test
instrumentation and test personnel. Table 8 summarizes the
dynamometer adjustments used by TTI and EPA. Six of eight TTI
test vehicles exhibit higher (4 to 10 percent) inertia
loading. Overall, TTI inertia settings are 5 percent higher
than EPA Certification values. Values of TTI dynamometer AHP
range from 20 percent higher to 10 percent lower than
EPA-Certification horsepower settings, with the average load
being 6 percent higher.
The increased average loading of the vehicles in this
program is at least partially caused by the increased weight of
the test instrumentation and personnel. However,
inconsistancies in the required weights and test weight
simulation during the Certification coastdown and
emissions/fuel economy test contribute to the observed fuel
economy differences.
-------
-31-
Taole 8
Comparison of TTI and
EPA-Certification Dynamometer Settings
TTI
1.
2.
3.
4.
5.
6.
7.
8.
Cutlass
Pinto
F-100
Citation
Escort
Horizon
Concord
Civic
IW (lorn)
4,000
3,000
4,250
3,000
2,500
2,750
3,500
2,250
AHP (hp)
10.5
9.7
15.1
8.3
5.8
7.9
11.0
8.4111
EPA-Certif ication
IW (Ibm)
3,750
3,000
3,875
3,000
2,375
2,500
3,375
2,125
AHP (hp)
11.6
9.7
14.0
6.6
6.4
6.4
10.9[1
7.8
[1] Includes 10 percent increase in horsepower to account for
air-conditioning simulation.
-------
-32-
References
1. EPA OMSAPC Advisory Circular (A/C) No. 55B, U.S.
EPA, OANRf QMS, ECTD, SDSB, December 6, 1978.
2. Code of Federal Regulations 40 Protection of
Environment, Part 86, Control of Air Pollution from New Motor
Vehicles and New Motor Vehicle Engines.
3. EPA-MVEL Unpublished Data, U.S. EPA, OANR, QMS,
ECTD, SDSB, D. Paulsell, August 1978.
4. "Fuel Economy Test Procedure issue Paper," (Draft),
U.S. EPA, OANR, QMS, ECTD, SDSB, E. LeBaron, September 1981.
5. "A Quantitative Analysis of the Effects of Rolls
Coupling on Fuel Economy and Emission Levels," U.S. EPA, OANR,
QMS, ECTD, SDSB, T. Downey, K. Besek, I. Parekh, Society of
Automotive Engineers, SAE Paper No. 810827.
6. "Carbon Balance and Volumetric Measurements and Fuel
Consumption," U.S. EPA, OANR, QMS, ECTD, SDSB, T. Newell, SDSB
80-05, April 1980.
7. Personal Conversation With S. Bergen, General Motors
Corporation Proving Grounds, M. Reineman, November 1982.
8. EPA Memorandum, Comparison of Volumetric and Carbon
Balance Fuel Economy Measurement Repeatability, U.S. EPA, OANR,
QMS, ECTD, SDSB, from T. Penniga to R. Stahman, June 10, 1981.
9. "An Energy Demand Model for Light-Duty Vehicles,
with Concepts for Estimating Fuel Consumption," U.S. EPA, OANR,
QMS, ECTD, SDSB, T. Newell, SDSB-81-2, April 1981.
10. "Vehicle Efficiency - Road vs. Dynamometer," U.S.
EPA, OANR, QMS, ECTD, SDSB, B. Grugett, SDSB 79-29, August 1979.
11. EPA-MVEL Unpublished Data, U.S. EPA, OANR, QMS
ECTD, SDSB, J. Kearis, October 1982.
-------
NOTE: THE APPENDICES TO THIS REPORT ARE AVAILABLE UPON
REQUEST.
------- |