-------
TABLE 4.14
National Bureau
of Standards Calibration
Gas Standards
C H in Air	CO in N	NO in N	CO in N
3 8	L	2.	L Z
Nominal Value (ppm) Nominal Value (ppm)	Nmtinal Value (ppm) Nominal Value
3
10
25
1
10
50
50
7
50
100
100
14
100
500
250

500
1000
500
1000

REF. 59

-------
62
12
vehicles. The data listed in Table 4.3 indicate that the cold
transient phase of the CVS-CH test accounted for 87% and 94% of the
total mass of HC and CO, respectively, for a vehicle equipped with an
23
oxidation catalyst. In addition, these data indicate that the varia-
tion in HC and CO during the cold transient phase accounted for 99% of
the total variation in HC and CO during the CVS-CH test.
Data reported by Reference 33 can also be employed to investigate
the effect of ambient temperature on the total mass of HC, CO and NO^
produced during each of the three phases of the CVS-CH test for four
prototype 1975 vehicles equipped with oxidation catalysts. At an
ambient temperature of 75°F, the average value, taken over all four
vehicles, of the total mass of HC and CO collected in the cold transient
bag accounted for approximately 64% and 73% by mass of the pollutants
collected in the three bags, respectively. Under these conditions,
the average composite CVS-CH emissions for these vehicles were 0.49,
5.5 and 2.32 g/mi for HC, CO and NO^, respectively.
In addition to the composite emissions presented in Table 4.5,
modal emission data are also available for the vehicles equipped with
25
2.0 liter CVCC stratified-charge engines. In these vehicles, the
cold transient phase account for approximately 50% to 60% and 307» to
40% of the total mass of HC and CO collected during the CVS-CH test,
respectively.
In the present CVS-CH test procedure,the total mass of pollutant
emissions in the cold transient, hot transient and stabilized portions
of the test are weighted by 0.43, 0.57 and 1.0, respectively. There-
fore, the contribution of the cold transient phase to the computed
vehicular composite emissions in grams per mile is not only dependent on
the total mass of pollutant collected but also on the weighting factors.
4.5 Effect of Ambient Temperature on Exhaust Emissions
It has been previously reported that exhaust emissions on the
CVS-C test for vehicles soaked at either 20°F or 40°F were significantly

-------
63
greater for both vehicles, with and without oxidation catalysts, than
the emissions for the same vehicles when they were soaked in accordance
with the CVS-CH test (e.g., 60°F to 86°F). The Emission Testing Lab-
oratory of Environment Canada studied the effect of ambient temperature
34
on exhaust emissions. The test procedure used during the course of
the tests was very similar to the CVS-C test procedure, and tests were
conducted on five 1972 and 1973 vehicles. Each vehicle was tested 30
to 40 times under cold conditions and 15 to 20 times under baseline
conditions (e.g., 60°F-86°F).
Linear correlations of present emission changes as referenced to
emissions measured at 60°F were found to correlate the data better
than parabolic fits. The average of the emissions from the five
vehicles tested for HC and CO are plotted in Figure 4.4 as a function
of percent deviation from the average emissions at 60°F. The average
HC curve yields a 100% increase in HC emissions at 8°F as compared to
the 60°F emissions. The correlation lines for all five vehicles had
different slopes ranging from approximately 0.62 to 1.5 times the
average slope. The CO average curve yields an increase of 100% in CO
emissions at 20°F as compared to the 60°F emissions. The slopes of
the lines for the five cars were all different with slopes ranging
from approximately 0.3 to 1.6 times the average slope. The average
N0^ emissions from the five vehicles showed a slight decrease in emis-
sions (less than 3%) at 0°F as compared to emissions at 60°F.
The Bartlesville Energy Research Center of the U.S. Department
of Interior has recently completed a series of tests designed to
determine the effect of ambient temperature on exhaust emissions and
fuel economy for twenty production cars from 1967-1973, four prototype
1975 emission cars with oxidation catalysts, one vehicle .equipped with
a diesel engine and one vehicle equipped with a PROCO-type stratified-
33
charge engine.
The experiments were carried out using the CVS-CH test procedure
and the results are summarized in Table 4.15. The effect of ambient

-------
64
FIGURE 4.4 Effect of Ambient Temperature on Exhaust Emissions
REF.

-------
65
TABLE 4.15
Effect of Ambient Temperature on Exhaust Emissions of
Various Vehicles on the CVS-CH Test


Test Temperature, °F

Tests with and without
20
50^
75
110
Air Co
nd;, 110" F 2/
With
- Without
3/
Production vehicles-






Hydrocarbon 	
7.44
5.63
4.66
4.67
4.56
3.98
Carbon monoxide	
81.6
65.4
53.8
63.5
72.3
57.8
Nitrogen oxides	
6.40
5.25
5.06
4.13
5.94
4.44
Aldehydes	
.23
.20
.18
.17
.n
.14
4 /
Hydrocarbon-






Total	
.8.51
5.90
5.01
4.95
4.28
3.61
Non-methane 	
7.76
5.44
4.71
4.67
3.98
3.35
Reactive	
6.55
4.70
4.01
4.33
3.33
2.80
Fuel economy, mpg	
10.8
10.9
11.4
11.5
10.34
11.58
Catalyst equipped—^






Hydrocarbon	
1.31
.81
.49
.50
.54
.50
Carbon monoxide 	
28.0
15.8
5.5
6.1
11.9
6.1
Nitrogen oxides	
3.14
3.05
2.33
2.35
2.66
2.35
Aldehydes	
.035
.035
.040
,021
.010
.020
4 /
Hydrocarbon-





.49
Total	
1.31
.81
.49
.50
' .52
Non-methane	
1.03
.66
.40
.38
. .40
.38
Reactive	
.90
.57
.35
.34
.35
.33
Fuel economy, mpg	
9.6
10.3
10.8
11.4
10.2
11.4
Stratified charge, PROCO-






Hydrocarbon	
.55
.28
.20
.14
-
-
Carbon monoxide	
1.8
.6
.7
.4
-
-
Nitrogen oxides	
1.37
1.25
1.05
1.07


Aldehydes	
.04
.03
.01
.01
-
-
Hydrocarbon






Total	
.55
.28
.20
.14
-
-
Non-methane	
.46
.22
.15
.11
-
-
Reactive	
.40
.20
.13
.10
-
-
Fuel economy, mpg 	
19.3
20.4
21.2
20.5
-
-
Diesel equipped-^


.49



Hydrocarbon 	
.60
.30
.40
-
-
Carbon monoxide	
1.9
1.4
1.4
1.2
- •
-
Nitrogen oxides	
2.06
1.99
1.78
1.88
-
-
Aldehydes	
.05
.04
.05
.05
-
-
Fuel economy, mpg	
17.1
18.6
19.7
20.1
-

y
2/
3/
y
5/
6/
Five 1973 vehicles not tested at 50° F.
Thirteen production cars.
Twenty cars—duplicate tests.
Six cars.
Four cars—duplicate tests.
Single car—duplicate tests.	REF. 33

-------
66
temperature on HC, CO and NOx exhaust emissions for the various types
of vehicles tested are graphically represented in Figures 4.5 to 4.7.
Inspection of these results indicates that ambient temperature has a
significant effect on HC and CO emissions for the twenty 1967-1973
production vehicles. Hydrocarbon and CO emissions increase by 60% and
52%, respectively, when the vehicles were tested at 20°F in comparison
to tests at 75°F. The NO^ emissions from the same group of vehicles
increased by approximately 25% in the same temperature interval.
The HC and CO emissions from the four prototype, catalyst-equipped
vehicles increased by 160% and 4097o, respectively, when the tests were
carried out at 20°F rather than 75°F. The NO emissions for these same
x
vehicles increased by approximately 33% over the same temperature inter-
val .
Hydrocarbon emissions for the diesel and stratified-charge PROCO-
powered vehicle were also found to increase significantly between 75°F
and 20°F. In contrast, the CO emissions from these two vehicles were
relatively constant with ambient temperature. NO^ emissions for these
two vehicles were also shown to be less dependent on ambient tempera-
ture than the emissions from either class of spark-ignition, engine-
powered vehicles.
As indicated earlier, the cold transient phase of the test
accounted for approximately 64% and 73% of the total mass of HC and
CO, respectively, for the four prototype, catalyst-equipped vehicles
when the test was run at 75°F. At 20°F, the cold transient phase
accounted for approximately 89% and 96% of the total HC and CO col-
lected for the same vehicles.
4.6 Durability Test Methods and Procedures
The test methods and procedures associated with the development
of emission-deterioration factors (DF) for the durability-data fleet
have been discussed in Section 4.2. Some concerns have been raised

-------
67
FIGURE 4.5 Effect of Ambient Temperature on HC Exhaust Emissions
During the CVS-CH Test
REF. 33

-------
80
70
60
50
40
30
20
10
0
URE
68
¦ 20 Production Cars, 1967-1973
• 4 Prototype Advanced Emission Cars (all catalyst)
x Diesel
O Stratified Charge (Proco)
— — —
9	r

20	40	60	80
TEST AMBIENT TEMPERATURE (°F)
100
120
4.6 Effect of Ambient Temperature on CO Exhaust Emissions
the CVS-CH Test
REF. 33

-------
69
20 Production Cars 1967-1973
4 Prototype Advanced Emission Cars (all catalyst)
Stratified Charge Car (Proco)
20	40	60	80	100
TEST AMBIENT TEMPERATURE (°F)
120
FIGURE 4.7 Effect of Ambient Temperature on NO Emissions During
the CVS-CH Test	x
REF. 33

-------
70
over the methods that are presently employed to determine the deterio-
ration factors which are subsequently used in certification. It has
been suggested that statistical variability associated with exhaust-
emission tests may cause errors in the determination of deterioration
factors when only a limited number of durability-data cars are tested
56 57
for a particular engine-system combination. * In addition, concern
has been raised over the ratio method that is presently employed to ob-
35
tain the deterioration factors for durability-data vehicles. Both of
these points are discussed below.
Monte Carlo Simulation techniques have been used^ to generate
statistically possible emission test measurements at 5,000 mile, incre-
ments in order to estimate the effect of test-error variability on cal-
culated deterioration factors. In this simulation, fleets of 10, 30
and 50 durability-data vehicles with various mean emission values,
specified deterioration factors and test-to-test variability were hy-
pothesized. Test-point measurements for each vehicle in the fleet were
computed using the test variability for 5,000 through 50,000 miles of
simulated durability mileage accumulation. Subsequently, a least-
square regression line was fitted through each set of simulated data
and a DF was calculated for each vehicle.
Typical results based on the Monte Carlo Simulation computations
for a fleet of 10 identical vehicles indicating the possible variability
in the DF as a function of exhaust-emission, test-to-test variability
are listed in Table 4.15a.^^ Inspection of these results indicates
that the mean value of DF for the 10-vehicle fleet is in reasonable
agreement with the true value of DF as a function of both the exhaust
emissions at 4,000 miles and the test-to-test coefficient of variation.
However, the coefficient of variation of the DF's computed for each 10-
vehicle fleet are generally approximately equal to the corresponding
test-to-test coefficient of variation.
Another Monte Carlo Simulation which included deterioration-
factor variability as well as car-to-car and test-to-test variability
has been reported by Reference 58. The necessary input in this simu-
lation included a 50,000 mile emission point, an absolute deterioration
factor, and absolute values of the test-to-test, car-to-car and

-------
TABLE 4.15a
Monte Carlo Simulation of Possible Variability in DF
as a Function of Test-to-Test Variability
EMISSIONS
AT

TRUE DF = 1.
0



TRUE DF
= 2.C


COV =
10%
COV =
205!,
COV =
30%
COV =
» 10%
COV =
20%
COV =
30%
4,000 MI
(g/mi)
DF
£-2
col t K
DF
-!*
X
DF
7 %
X
DF
s „
7 %
X
DF
7*
X
DF
f %
X
.41
.978
8
1.00
15
1.20
37
2.11
9
2.12
39
2.36
50
1.5
.931
9
1.05
22
.966
32
2.02
6
1.99
26
2.19 .
48
3.1
.949
12
.980
19
1.05
24
2.06
11
2.29
29
1.99
. 22
9.0
1.03
12
.986
17
1.13
24
2.03
9
1.97
21
2.22
34
15
.983
11
.998
19
..901
23
1.99
11
2.24
16
1.70
20
Fleet Size = 10
COV = test-to-test coefficient of variation, %
— it
- = deterioration factor coefficient of variation % from Monte Carlo Simulation
x
DF = mean deterioration factor from Monte Carlo Simulation
REF. 57

-------
72
deterioration-factor variability. These values were used to randomly
adjust 5,000 mile increment emission-data points in order to calculate
statistically possible deterioration factors, 50,000 mile and 4,000
mile intercepts. The results were analzed to obtain the probability
that an emission test would yield adjusted emissions below the
federal standards given a deterioration factor and a 50,000 mile emis-
sion value. Also determined was the adjusted emission value below
which 95% of all adjusted emissions could be expected to lie. These
computations indicated that the various sources of variability con-
tribute to the probability that a given vehicle will pass an emission
test. Further discussion of the technique used and a complete table
of the results obtained may be found in Reference 58.
The Monte Carlo Simulation reported in Reference 57 indicates
that statistical variability associated with the determination of DF's
may reflect the special circumstances of the test rather than the
capability of the underlying technology. This idea is particularly
relevant when it is realized that deterioration factors are often
based on emission tests of a single durability-data car. In order to
reduce these errors, some type of data-averaging procedure over a
larger group of "similar vehicles" might be preferable for deterioration-
factor calculations. However, a too-extensive use of average treats
all vehicles and systems as equal and, hence, attractive technologies
may be counterbalanced by less desirable technologies. At the present
time, the manufacturer has the option of testing more than on "identical
durability-data vehicle" in each engine-system combination, with all the
emission data collected being used to calculate the deterioration
factor. If appropriate allowances could be made for possible malfunc-
tion of a durability-data car in a small fleet of "identical vehicles,"
a data-averaging technique would be statistically desirable but still
might not be cost-effective from the manufacturer's point of view.
It has also been suggested that the ratio method presently em-
ployed to compute DF based on the durability data may unnecessarily

-------
73
penalize durability vehicles with relatively low 4,000 mile extrapolated
emissions.^"*
As described previously, a deterioration factor (DF) for each
pollutant is determined by fitting a least squares best fit straight
line to all durability emission data and defining DF in terms of
exhaust emissions of pollutant i
(DF)^ = at 50,000 miles		(1)
exhaust emissions of pollutant i
at 4,000 miles
where the exhaust emissions at both 4,000 miles and 50,000 miles are
obtained from the least squares fit line. To facilitate discussion
and introduce appropriate nomenclature, a least-squares line repre-
senting a "typical" deterioration-factor test result is shown in
Figure 4.8. The least-squares line obtained from the data is repre-
sented by E(x) = m x + b;- where E(x). is the emissions at x miles and
m and b are the least mean square values of the slope and intercept,
respectively. The equation for the least squares line can be readily
rewritten in terms of the slope m and the interpolated value of the
emissions at 4,000 miles (I).
E(x) = m (x-4,000) + I	(2)
The deterioration factor as computed by EPA for the data in Figure 4.8
would be given by
DF = E(4?000)? = m(5Q'000 - 4,000) + I	(3)
or simplifying
DF = 1 + 46,000 m	(4)
I
Inspection of Equation (4) indicates that the DF, as computed by EPA,
is dependent on both the slope of the deterioration line and the extrap-
olated value of the line at 4,000 miles.
The vehicle under consideration is checked for certification by
testing a corresponding emission-data car at, 4,000 miles and multi-
plying the numerical value (X) of each pollutant measured during this
test by the DF.

-------
74
FIGURE 4.8 Alternate Methods for Evaluating the Deterioration
Emission Control Systems

-------
75
Certification value = DF x X = (1 + 46,000 m) X.	(5)
I
If the certification value, as computed from Equation (5), is less than
the appropriate standard, the vehicle passes; if not, it fails.
If one assumes that the rate of deterioration of the emission-
control system is not a function of either the extrapolated value of
the emissions at 4,000 miles (e.g., m £ m (I)) for the durability-data
vehicle or the calibration of the system, then this technique, as out-
lined in Equation (5), may unnecessarily penalize an emission-data
car whose corresponding durability-data car had a low value of extrapo-
lated emissions at 4,000 miles (I).
35
In order to circumvent this potential problem, Hromi has
suggested another method for determining whether a vehicle passes
certification. This technique is based on a simple difference formu-
lation as given by Equation (6):
Certification Value = X + E(50,000) - E(4,000)	(6)
= X + 46,000 m
The vehicle would pass if the certification value, as computed
from Equation (6), was less than the appropriate emission standard
and fail if it was greater than the standard.
Sufficient data could not be obtained during the course of this
study to determine if m ^ m(I) for various control systems, durability-
car calibrations, etc. If at a later date sufficient data are available
for a range of emission-control systems, it might be appropriate to
consider changing the method of combining durability data with emission
standards as outlined by Equation (6) rather than the method presently
employed (Equation 5).
4.7 Evaporative HC Emissions
In addition to HC exhaust emissions discussed previously, fuel
evaporation losses may contribute significantly to the total HC emis-
sions from Ii)MV. Evaporative emissions differ from exhaust emissions
in that evaporative emissions are formed as a result of physical

-------
76
rather than chemical processes. In general, fuel hydrocarbon can be
emitted as a result of the evaporation process from both the fuel tank
and the carburetion system. A motor vehicle experiences evaporative
hydrocarbon emission as a result of its normal operating procedure.
For example, at the end of every vehicle trip, evaporation may occur
from the carburetor bowl due to heat being transferred from the engine
block. The evaporated vapor may then escape through any available
opening in the carburetor system. This type of hydrocarbon evaporation
is known as "hot soak losses." Other evaporative losses have been
termed "diurnal," "running," and "refueling" losses. The diurnal loss
is the result of the daily temperature rise and the corresponding
evaporation of hydrocarbon fuel from the fuel tank. Running losses
are similar to both the diurnal and hot-soak losses except that the
necessary temperature increase and heat transfer are provided by the
engine when it is operating. Refueling losses occur whenever fuel
vapors are emitted during fueling operations.
A test method has been developed to measure diurnal, running and
hot-soak evaporative emissions. Prior to initiating the test, the
vehicle is soaked for at least 12 hours to insure that the engine has
completely cooled down. Cool fuel is placed in the tank and the di-
urnal losses are collected by heating the fuel from 60°F to 84°F in
one hour. The vehicle is subsequently run on the CVS-CH cycle and
the engine is shut down. The running and hot-soak emissions are col-
lected during the test cycle and the one-hour period following engine
shutdown. The HC emissions collected during this test can be subse-
quently utilized to estimate the grams of HC evaporated per day from
the fuel tank (diurnal losses) and the grams of HC evaporated from the
system during an average urban trip and the one-hour period subsequent
to engine shutdown (running plus hot-soak losses). The present federal
standard for evarporative emissions is 2.0 g/test.
In order to compare the magnitude of exhaust and evaporative HC
emissions, it is necessary to develop a formula that can be utilized

-------
77
to convert evaporative test- results to g/mi. Reference 36 has sug-
gested that an appropriate expression for converting evaporative-
emission results to g/mi is given by:
/evaporative HC emissionsj_ /dirunal losses! /e . ^eg r da \
\ in g/mi	/ \ in g/tes.t )/ * m 63 per ay' +
/ running and hot-soak losses; J X{miles per trip)
\ in g/test	'/	(7)
Suitable average values of miles per day and miles per trip are 34 and
7.5, respectively.
Two different methods have, been developed for collecting and
measuring evaporative HC emissions. Presently, the method used by the
federal government for certification of evaporative emission-control
20
systems is based on measuring the weight of fuel vapor absorbed in
carbon canister traps. The traps are connected to the fuel tank vents
and to the carburetor external vent during the course of the test.
After the test is completed, the traps are disconnected from the
vehicle and weighted to determine if the 2.0 g/test standard is met.
The second test method, the Sealed Housing for Evaporative Determina-
37
tions (SHED) method, is more comprehensive.] Basically, the SHED
method involves placing the vehicle in a sealed enclosure throughout
the test, and the HC concentration in the enclosure at the end of the
test, measured with an FID, is used, to determine the mass of HC
evaporative emissions. Further discussion of the SHED method may be
found in References 38 and 39.
40
A study employing the SHED method to determine HC evaporative
emissions reported that average evaporative, HC emissions for 55
preevaporative-controlled vehicles was about 37 g/test. Based on
these results, it has been estimated that evaporative emissions from
precontrolled vehicles (i.e., pre-1971 model-year vehicles) were
41
approximately 3 g/mi.
The results of evaporative-emission certification tests for
model years 1971-1974, based on the carbon canister trap method, are

-------
78
shown in Table 4.16. It was shown that 60% of the tests reported
as less than 0.1 g were actually reported as 0.0 g, and that in
general, most of the tests reported as 0.0 g corresponded to negative
canister-weight changes. Comparison of the evaporative-emissions data
shown in Table 4.16 with the estimated precontrolled evaporative emis-
sions of 37 g/test seems to indicate that present evaporative-control
systems are very effective, and that evaporative emissions have been
essentially completely controlled.
The evaporative HC emissions results obtained in two recent EPA
42 43
surveillance programs ' for both precontrolled and controlled
evaporative-emissions systems indicate that this conclusion is not
correct. The surveillance program carried out in 1972 measured
evaporative emissions from both controlled and uncontrolled vehicles
43
employing the SHED method. The SHED method was employed during these
tests rather than the carbon-canister-trap method since it takes into
account evaporative emissions from the entire fuel system including
gasket and throttel shaft leakages. In addition, vehicles manufac-
tured prior to the implementation of evaporative controls are not
readily amenable to measurement by the carbon-canister method.
Evaporative-emission measurements were made in Los Angeles for 136
vehicles ranging from 1957 to 1971 model yea^r and twenty-two 1971
model-year vehicles were tested in Denver to assess the effect of
high altitude. The results of these tests are summarized in Table 4.17.
The weighted values of these HC emissions in grams per mile as computed
from Equation (7) are also listed in Table 4.17. Results from the
1972 fiscal year surveillance program included evaporative-emission
tests on twenty-two 1972 model-year vehicles in both Los Angeles and
Denver. These results were also based on the SHED method and the data
as reported in Reference 42 are also listed in Table 4.17.
Inspection of these results indicates that the weighted value of
the uncontrolled vehicles of 2.7 g/mi is in good agreement with the
previously estimated value of 3.0 g/mi. The significant difference in

-------
TABLE 4.16
Summary of Evaporative HC Emission Test Results
for 1971-1974 Certification Test Results
MODEL YEAR NO. OF PERCENT OF TESTS PERCENT OF TESTS MAX. VALUE AVERAGE VALUE FED. STD.

TESTS
LESS THAN 0.1 R
LESS THAN 1.0 R
(g/test)
(g/test)
(g/test)
71
131
32
82
3.65
0.545
6.0
72
370
45
91
1.90
0.307
2.0
73
351
56
94
1.90
0.251
2.0
74
399
45
98
1.90
0.258
2.0
Test based on carbon canister trap method
REF. 42

-------
80
TABLE 4.17
Summary of Evaporative HC Emission Test Results
Obtained During Surveillance Tests^"


L. A.
Data (g)
Denver Data (g)
Weighted
Model
No.



Values
Year
Tests
Diurnal
Hot Soak
Diurnal Hot Soak
g/mi
57-69
102
26.08
14.67
_ -
2.7
70
13
17.75
10.70
-
1.9
71
21
14.87
10.89
-
1.9
71
22
-
-
47.2* 34.8*
6.0*
72
22
12.40
11.80
-
1.9
72
22
"
"
17.4 14.2
2.4
Note:
*Winter grade fuel
(11.7 RVP)
used on all tests. L.
.A. data

up to 1971
used all
types of
fuel (7.8 - 12.0 RVP).

^Experimental data based on SHED method
REF. 42, 43

-------
81
the 1971 and 1972 model-year evaporative emissions obtained in Denver
has been attributed to the differences in fuel volatility that were
employed during the testing. The 1971 Denver vehicles were tested
using commercial winter-grade fuel of high volatility (RVP = 11.-7 psi),
and the 1972 vehicles were tested using the standard test fuel of
summer-grade volatility (RVP = 8.8 psi). Significantly higher
evaporative emissions are to be expected when testing with high-
volatility fuel as compared to lower-volatility fuel. The same situa-
tion occurred during tests of model year 1957-1971 vehicles in Los
Angeles, but the effect on the overall average is less pronounced since
a wide variety of vehicles were tested at random times with both types
42
of fuel. EPA has indicated that the test data obtained during the
FY 1971 surveillance program will be reanalyzed in the near future to
account for the effects of fuel volatility and other procedural factors
on the evaporative emissions.
Standard test fuel was employed in both the Denver and Los Angeles
evaporative-emission tests for the 1972 vehicles, and the differences
in values reported in Table 4.17 have been attributed to the effect
of barometric pressure (i.e., 24.5 in. Hg at Denver as compared to
30.2 in. Hg at Los Angeles).
Comparison of the results listed in Tables 4.16 and 4.17 indi-
cates a significant difference in the assessment of the relative
importance of evaporative HC emissions on the overall strategy designed
to control HC emissions to the atmosphere. The 1975-76 Federal
Interim Standards and the 1975-76 California standards for HC exhaust
emissions are 1.5 and 0.9 g/mi, respectively. While the SHED-method
data are relatively limited, they nonetheless raise a very important
issue. These data indicate that vehicles which are supposed to have
95% effective evaporative-emission controls, have evaporative HC
emissions that are larger than the 1975-76 exhaust-emission standards
when both are compared onfgrams-per-mile basis, it has been estimated
42
that refueling losses amount to approximately an additional 0.4 g/mi.

-------
82
Based on these results, two immediate needs are obvious. First,
an accurate test method and procedure for measuring evaporative HC
emissions must be developed, which corresponds closely to "real life."
It is possible that a careful evaluation of the situation may show
that the SHED method is adequate for this purpose. Second, an exten-
sive study should be undertaken to determine the relative importance
and cost effectiveness of control strategies designed to control
automotive HC emissions based on both exhaust and evaporative HC
emission standards.
4.8 Consideration of Nonreactive HC Exhaust-Emission Standards
Presently there is some question about both the absolute magni-
tude and method of measuring exhaust HC emissions during the federal
certification procedure. This concern is based on the fact that
national primary and secondary HC air-quality standards are specified
in terms of non-methane HC's while the exhaust-emission measurement
techniques and standards include methane. Methane has been excluded
from the air-quality standards since it is not considered to partici-
pate in atmospheric photochemical reactions leading to the formation
of photochemical oxidants and since substantial ambient levels of
methane are known to originate from uncontrollable sources.
The measurement of total HC exhaust emissions rather than non-
methane hydrocarbons was promulgated by the EPA Administrator since:
(1) at the time, adequate techniques for the routine measurement of
methane in LDMV exhaust were not available; (2) the methane fraction
of HC emissions was assumed to be low; and (3) potential emission-
control systems were expected to reduce all exhaust HC components
44
equally.
The introduction of oxidation catalysts as an important HC and
CO emission-control technology and the development of new instrumen-
tation techniques may invalidate some of the above assumptions.

-------
83
Ford has recently petitioned EPA to amend the motor-vehicle and
motor-vehicle-engine-exhaust-emission test procedures to permit ex-
clusion of methane in determining compliance with federal motor-
vehicle-emission standards. The petition includes: (1) information
relating to the non-photochemical reactivity of methane; (2) data
relating the quantities of methane contained in the exhaust of LDMV
equipped with various types of control systems, including oxidation
catalysts; and (3) information concerning the availability of instru-
mentation methods for the measurement of both total and non-methane
HC's in automotive exhaust.
45
Ford's data indicate that methane may represent approximately
10% to 40% of the total HC in the exhaust of catalyst-equipped
vehicles. In addition, a good correlation was shown between total HC
measured with the standard FID technique and a modified FID technique
which has been designed to measure both total HC and methane.
Other laboratries have also reported large concentrations of
methane in automotive exhaust from LDMV equipped with oxidizing
s	26
catalysts. Chrysler has reported data which indicate that mass
fractions of unreactive hydrocarbons, including methane and ethane,
are increased as the exhaust passes through an oxidation catalyst.
33
In addition, data presented in Table 4.15 indicate that the non-
methane fraction of total hydrocarbons in automotive exhaust is ap-
proximately 94% and 81% for six pre-1974, noncatalyst-equipped
vehicles and six catalyst-equipped vehicles, respectively. Data in
p	'
Table 4.15 also indicate that the fraction of reactive hydrocarbons
in the same group of vehicles is approximately 80% and 70% for noncatalyst-
equipped and catalyst-equipped vehicles, respectively. In these data,
nonreactive HC's were assumed to include methane, ethane, propane,
acetylene and benzene. It is important to note that the mass fraction
of aldehydes to total hydrocarbons was not significantly different for
catalyst-equipped and noncatalyst-equipped vehicles.

-------
84
It is generally agreed that oxidation catalysts in emission-
control systems may preferentially reduce the photochemically reactive
portion of the hydrocarbon exhaust. This then leads to the conclusion
that the current "total" hydrocarbon standard may unnecessarily penalize
vehicles equipped with this type of control system. Thus, the
conversion of motor-vehicle exhaust hydrocarbon standards to a
nonreactive basis may be a technically desirable goal. However, it is
believed that such a change would require significant expenditures of
resources by the EPA, motor-vehicle manufacturers, and other motor-
testing laboratories for modification of existing equipment and
procedures. Furthermore, substantial efforts must also be made to
establish revised testing procedures and to determine the magnitudes
of nonmethane hydrocarbon emission standards equivalent to present
total hydrocarbon standards. In order to obtain information associated
45
with these questions and in response to the petition by Ford, the EPA
has recently reported that it is considering changing HC regulations to
include only nonmethane hydrocarbons and has requested that all
interested parties provide information concerning the following points:
(1)	Identification of nonreactive hydrocarbon components of
motor-vehicle exhaust which should be excluded in determining
compliance with motor-vehicle emission standards. EPA has
identified ethane, propane, acetylene, and benzene in addition
to methane.
(2)	Availability of methods for routine measurement of nonreactive
hydrocarbon components in motor vehicle exhaust.
(3)	Quantities of reactive and nonreactive hydrocarbon compounds
in motor vehicle exhaust.
(4)	Impact of total versus reactive hydrocarbon standards.
(5)	Lead time required for implementation of nonreactive hydro-
carbon testing.
(6)	Impact of reactive hydrocarbon standards on other motor
vehicle compliance efforts.

-------
85
A final decision concerning this issue cannot be reached until the
44
additional data requested by EPA have been collected and analyzed
from both a technical feasibility and cost-benefits point of view.
4.9 Summary
The CVS-CH exhaust-emission test procedure plays an important role
in the federal certification process for 1975 and subsequent model,year
LDMV, and, hence, it is important to develop an understanding of the
statistical variability of exhaust-emission tests that can be expected
for vehicles designed to meet these standards. Due to both random and
systematic errors, measurements of exhaust emission from a given LDMV
have poor repeatability. Various.factors including vehicle variability,
emission collection and measurement variability and environmental test
variables contribute to the. poor test reproducibility. In addition to
these variations associated with a given vehicle, variations in
emissions from a group of similar production vehicles are also
encountered.
The results of programs designed to isolate important sources of
variability in CVS-CH exhaust-emission tests, other than vehicle
variability, have indicated that local ambient test conditions, driver
skill, dynamometer design; roll spacing and calibration method, and
systematic errors associated with CVS sampling.and gas analysis
techniques all-have an effect on laboratory.-to-laboratory exhaust-
emission variability. It has been shown that, variations in-barometric
pressure, humidity and test-cell temperature can have a significant
effect on HC, CO and NO^ (K^ corrected) exhaust emissions. It has also
been shown that vehicle soak temperature, within the allowed range of
60 °F to 86 °F, can influence HC and CO exhaust emissions.
Tests on 1975 type vehicles equipped with oxidation catalysts have
shown that the cold transient phase of the CVS-CH test is a major

-------
86
contributor to both the total mass of HC and CO emissions collected
during the test and the overall variability of HC and CO exhaust-
emission measurements. This is not the case with NO^ where approximately
equal masses of NO are collected during the three phases of the test.
Variations in exhaust-emission measurements specified in terms of
the standard deviation as a percent of the mean for a 1975-76 vehicle
in a given cell or from cell to cell at one laboratory can be expected
to range between 1C% to 25%, 15% to 30% and 5%. to 15% for HC, CO and
N0x> respectively. Limited data indicate that the statistical
variability of exhaust-emission tests on CVCC stratified-charge type
vehicles may be lower than that mentioned above. In addition,
systematic errors from laboratory to laboratory of as much as 20% to 30%
of mean emission values have also been reported in certain cases. In
general, exhaust-emission variability tends to increase as exhaust-
emission standards are reduced below 1975-76 levels. Based on these
considerations, it is apparent that significant consequences should not
be attached to a single CVS-CH exhaust-emission test.
It is concluded that exhaust-emission variability from test to test
and laboratory to laboratory can be reduced if the following two
programs are developed and implemented:
(1)	Establishment of mandatory correlation test programs among
governmental, automotive manufacturers' and other test
laboratories that are carrying out exhaust-emission measure-
ments on the CVS-CH test. These correlation programs should
include CVS-CH tests of relatively stable vehicles and tests
of the CVS sampling and gas analysis systems using either
exhaust gas generators or premixed gas cylinders as the gas
source.
(2)	Establishment of close tolerances on ambient test-cell
temperature and humidity for the CVS-CH test method and the
development of correction factors relating barometric pressure
and exhaust emissions.

-------
87
The mandatory correlation test programs should be designed to
determine systematic errors in exhaust emission measurements from cell
to cell and between various laboratories. Once systematic measurement
errors are isolated at a given facility, they can be .corrected or the
measurements can be adjusted to take them into account.
It has been shown that one of the important factors contributing
to emission-test variability is variation in ambient conditions.
Unfortunately, the effect of ambient-condition variations during the
CVS-CH test on the exhaust emissions can vary from vehicle to vehicle
and, hence, correction factors designed to correct measured exhaust
emissions for ambient variations should only be employed when it is
impractical to control ambient test conditions. Therefore, it is
concluded that .the CVS-CH test procedure should be modified to require
close control of both ambient temperature and humidity during the
course of the test. Presently, no controls are placed on humidity,
and ambient temperature is allowed to vary between 68 °F and 86 °F
during CVS-CH tests. Good control of ambient temperature and humidity
will require the installation of air conditioning systems at most
laboratories. Unfortunately, barometic pressure cannot be conveniently
controlled during the course of CVS-CH tests. Since barometric pressure
has been shown to significantly affect exhaust emissions and both
systematic and random variations of barometric.pressure occur between
test laboratories, it is concluded that a program designed to determine
exhaust-emission correction factors for variations in barometric
pressure should be developed and incorporated in the CVS-CH test
procedure as soon as possible. More stringent control of the
temperature range allowed for vehicle soak areas prior to testing would
require significant capital expenditures for air conditioning equipment.
It is recommended that a program be carried out to determine the cost
effectiveness of placing more stringent controls on vehicle soak
temperatures.

-------
88
It has been shown that ambient temperature variations, commonly
encountered in large sections of the nation, can significantly
increase exhaust emissions of HC and CO for both pre-1975 production
vehicles and 1975 prototype models equipped with oxidation catalysts.
As an example of this effect, the HC and CO emission from four
prototype catalyst-equipped vehicles were shown to increase by 160%
and 409%, respectively, over the temperature range 75 °F to 20 °F.
The N0x emissions for these same vehicles increased by approximately
337o over the same temperature interval. Therefore, it can be
anticipated that starting a vehicle at temperatures commonly
encountered during winter in large sections of the nation, e.g., 0 °F -
32 °F, would cause large increases in exhaust emissions.
Until recently, it had been assumed that evaporative HC emissions
had been 95% controlled with respect to uncontrolled evaporative
emissions. This conclusion was based on evaporative-emission tests
employing the carbon-canister trap method. Evaporative HC emissions
measured in two recent surveillance programs, employing the SHED
method for both precontrolled and controlled evaporative emissions
systems, indicate that this conclusion is not correct.
These data indicate that vehicles which are supposed to have 95%-
effective evaporative-emission controls, have evaporative HC emissions
that are larger than the 1975-76 exhaust emission standards when both
are compared on a grams per mile basis.
In light of these results, an accurate test method and procedure
for measuring evaporative HC emissions must be developed, which
corresponds closely to "real life." It is possible that a careful
evaluation of the situation may show that the SHED method is adequate
for this purpose. Also, an extensive study should be undertaken to
determine the relative importance and effectiveness of control
strategies designed to control automotive HC emissions based on both
exhaust and evaporative-emissions standards.

-------
89
Some concerns have been raised over the methods that are presently
employed to determine the deterioration factors of durability-data cars
that are subsequently used in the certification procedure. It has been
shown that the statistical variability associated with the determination
of deterioration factors may reflect the special circumstances of the
test rather than the capability of the underlying technology. This idea
is particularly relevant when it is realized that deterioration factors
are often obtained from emission tests on a single durability data car.
An averaging procedure over a larger group of "similar vehicles" might
be preferable for deterioration factor calculations. However, if
averaging is used too extensively, the averaging will treat all vehicles
and systems on equal texms, and attractive technologies may be counter-
balanced by less desirable technologies.
Since the deterioration of emission-control systems has been
assumed to take place in a linear fashion, it has been suggested that a
simple difference technique rather than the ratio technique, which is
presently employed for determining deterioration factors, might be more
appropriate. Sufficient data were not available to determine which of
these two techniques .is more appropriate.
Finally, the possibility of modifying the present HC exhaust-
emissions standards and measuring techniques to include only reactive
HC's has been discussed. This proposal is presently under consideration
by EPA and it is suggested that the resolution of this issue should be
deferred until the additional data collected by EPA have been analyzed
from both a technical feasibility and cost-benefits point" of view.

-------
5.0 FUEL ECONOMY TEST METHODS AND PROCEDURES
5.1 Introduction
Due to the present and projected energy shortages, considerable
national attention has been focused recently on the fuel economy of
LDMV. The development of a rational energy conservation strategy
for the transportation sector requires, among other things, a detailed
understanding of the factors that influence automotive fuel economy
and the development of standardized fuel economy test methods and pro-
cedures. The short-term need is to assess, approximately, the validity
of present fuel economy tests. The medium-term need is to develop a
standardized fuel economy test method and procedure which are repre-
sentative of average nationwide driving patterns and mileage accumula-
tion. The long-term need is for a more accurate data base which can be
employed to make projections of the effects of changes in vehicle tech-
nology and use patterns on total U.S. fuel consumption.
Until recently, automotive fuel economy has been primarily the
concern of the automotive manufacturers. However, in the past few
years, both the public and the government have taken a much more
active interest in this area due to the general unavailability and
increasing costs of automotive fuel. Unfortunately, at the present
time, generally accepted standardized fuel economy test methods and
procedures are not available. Due to the significance of this issue,
it is important that standardized fuel economy test methods and pro-
cedures, based on sound engineering and scientific considerations, be
developed as soon as possible.
The present section has been written in an effort to summarize
and evaluate information presently available concerning the important
factors affecting fuel economy, advantages and disadvantages of alter-
nate fuel economy test methods and procedures, statistical variability
of fuel economy measurements and to discuss the effect of cold-start
and ambient temperature of LDMV fuel economy.
90

-------
91
5.2 Factors Affecting Fuel Economy
The important factors influencing fuel economy of a LDMV include
the driving cycle, the vehicle characteristics, the installation of
emission-control systems, the driving habits of the consumer and the
ambient conditions, including the effect of cold start. The fuel
economy of a given vehicle is strongly dependent on the driving cycle
employed during the test. The fuel economy tends to be maximized
under steady-state cruise conditions. In general, steady-state cruise
fuel economies tend to increase as the speed is increased from low
values, reach a peak in the range of 30 to 50 mph and then decrease
significantly at higher speeds. The effect of steady cruise speeds on
the fuel economy of a 2,100 lb subcompact, a 3,500 lb intermediate and
a 5,200 lb luxury sedan is shown in Figure 5.1.^ Figure 5.2 compares
the 40 to 70 mph fuel economy of these three vehicles with the fuel
46
economy obtained over an urban driving cycle used by Chrysler.
Inspection of these results indicates that the best fuel economy for
each vehicle (at 40 mph cruise) is higher by more than a factor of
two than its poorest (cold urban cycle) fuel economy. In addition,
the fuel economy of a given vehicle is approximately 507o greater for
a typical highway cycle than for a typical urban cycle.
The design characteristics of a vehicle strongly influence its
fuel economy. In general, the fuel economy of a vehicle on any given
driving cycle is affected by: (1) inertia forces that are dependent
on the vehicle's mass and acceleration; (2) road load that includes
rolling resistance and aerodynamic drag; (3) engine efficiency;
(4) drive-train efficiency, including the effects of vehicle trans-
mission and rear axle ratio; (5) accessories, including air condition-
ing, power steering, etc.; and (6) emission-control devices. Many of
the above factors are interrelated, e.g., heavy vehicles tend to have
larger frontal areas than lighter vehicles, and, hence, both the
inertia forces and aerodynamic drag are often increased on heavy

-------
92
Subcompact,
40i—
| 30-
>-
2
o
2 20-
o
o
LLI
-I
LLI
2 io-
°	30	40	5tT	60	70
SPEED (mph)
FIGURE 5.1 Cruise Road Load Fuel Economy Versus Speed for Three
Vehicles
REF. 46

-------
93
40
I 30
>
S
O
B 20
o
111
	 40 mph
Road
Load
10
70 mph
Warm
Cold
— Urban
40.7
26.1
21.2
17.1
Road
Load
40 mph
70 mph
Warm
Cold
Urban
22.3
16.7
11.5
10.0
Road
Load
40 mph
70 mph
Warm
Cold
Urban
17.2
12.7
8.4
6.8
SUBCOMPACT
4-CYL ENGINE
MANUAL TRANSMISSION
Approx. 2,100 lb
INTERMEDIATE
V-8 ENGINE
AUTOMATIC TRANSMISSION
Approx. 3,500 lb
LUXURY SEDAN
V-8 ENGINE
AUTOMATIC TRANSMISSION
Approx. 5,200 lb
FIGURE 5.2 Fuel Economy Ranges (Urban Cycle to Road Load) for Three
Vehicles on Different Driving Cycles
REF. 46

-------
94
vehicles when compared to light vehicles. Also, in a similar manner,
the addition of accessories to ,a vehicle, such as an air conditioner,
not only requires energy for its operation but also increases the
inertia forces that the vehicle must overcome. Therefore, it is not a
simple matter to optimize simultaneously all of the important parame-
ters influencing fuel economy to achieve the "best" fuel economy.
Indeed, it is often even difficult to obtain a base line in fuel econ-
omy for comparison purposes since car weights have fluctuated, base
engines have changed, some models have been introduced and others
dropped etc.
Based on a multiple regression analysis on the measured fuel
economies of over 1,400 vehicles as determined for a cold-start urban
driving cycle, it has been concluded that vehicle inertia weight is
2
the single most important vehicle parameter affecting fuel economy.
These results indicated that a typical vehicle in the 5,000 lb inertia-
weight class has approximately a 50% lower fuel economy than a typical
vehicle in the 2,500 lb inertia-weight class when the CVS-C driving,
cycle is used for the comparison.
Since vehicle weight is seen to be an important factor in fuel
economy, comments on trends in vehicle weight are relevant. The
47
results of a recent study indicate that the curb weight of passenger,
cars in all market classes has steadily increased with time in the
period between 1956 and 1974. (Note: Inertia weight is generally
defined as curb weight plus 300 lb.) The conclusions reached in
Reference 47 are summarized below:
1. Passenger cars in all market classes have shown a marked and
steady increase in curb weight with time. This curb-weight
increase trend is independent of manufacturer! For example,
Chevrolet and Ford standard-size cars increased approximately
1,100 lb (33%) and 980 lb (29%), respectively, between 1956 and
1974. In the intermediate class, the Fairlane/Torino series
increased curb weight by approximately 1,100 lb (36%) from 1962-
to 1974;' the Chevelle increased curb weight 900 lb (28%) from 1966
to 1974. In the compact class, from 1962 to 1974, the, Chevy II/
Nova series increased curb weight by 940 lb (36%), while the
Valiant increased by 620 lb (24%).

-------
95
2.	The Intermediate-class car of 1974 weighs about the same as the
standard size car of 1970 (approximately 4,200 lb curb weight).
Similarly, the compact car of 1974 has about the same curb weight
as the Intermediate car of 1966 to 1970 (approximately 3,300 to
3,600 lb).
3.	The overall sales-weighted curb weights of U.S. passenger cars
dropped sharply from the 1958 level (approximately 3,700 lb) to
approximately 3,450 lb in the 1960 to 1964 period. This was due
to the introduction-of compacts in 1960 and high sales of both
compacts and Intermediates in that period. Since 1964, sales-
weighted curb weights have risen steadily, reaching approximately
3,650 lb in 1973.
4.	The overall U.S. sales-weighted inertia test-weight average
(including domestic and foreign cars) has the same general pattern
as curb-weight variation with time. It dropped sharply from the
1958 level (3,967; lb) to its lowest value of 3,712 lb in 1961.
Since 1961, there has been a steadily rising sales-weighted iner-
tia test-weight trend, reaching a new high value of 3,968 lb in
1973.
5.	Curb and inertia test-weight values for domestic passenger cars
surpassed their 1958 levels in 1970 and appear to be on a still-
rising trend. For example, the sales-weighted inertia test-
weight average, for domestic cars only, was 4,223 lb in 1973, com-
pared with 4,096 lb in 1958.
Figure 5.3, which was taken from Reference 47, clearly indicates
the general pattern of these weight trends.
The influence of inertia weight on vehicle fuel economy measured
on a cold urban driving cycle for 1973 model-year vehicles and an
average of pre-emission-controlled (1957-1967) vehicles was also
reported in Reference 47. These results are listed in Figure 5.4.
Inspection of Figure 5.4 indicates that the fuel economy penalty of a
heavy car, from all causes, is very large when compared to light vehi-
cles. It should be noted that although this graph shows a general
trend with higher fuel economy at lower inertia weights, the trend is
increased by other external factors, such as the consideration that
lighter cars tend to have smaller engines, manual transmissions and
fewer accessories. These external factors also tend to increase the
fuel economy of the lighter cars with respect to heavier cars. Also,

-------
96
4500
= 4000
H
X
o
111
§
CO
tr
D
° 3500
3000
Standard
Size
Chevrolet
Standard
Size
Ford
1958
1962	1966
MODEL YEAR
1970
1973
FIGURE 5.3 Vehicle Weight Versus Model Year for Two Standard Size
Vehicles
REF. 47

-------
2000 2500 3000 3500 4000 4500 5000 5500
INERTIA WEIGHT (lb)
FIGURE 5.4 Fuel Economy of Vehicles on the CVS-C Cycle as a Function
of Inertia Weight
REFo 47

-------
98
Figure 5.4 indicates 'that a given weight increase has a more signifi-
cant effect on fuel economy for lighter cars than for heavier cars.
The significance of inertia weight as related to vehicle fuel
economy for any given driving cycle on a level road can be readily
explained by considering the power that must be supplied to- the drive
wheels of the vehicle in order to overcome inertia and road load.
Power is required to accelerate, the vehicle from one speed to another,
and once the vehicle is in motion, the drive wheels must provide power
to overcome the road load. Road load is usually considered to be due
to the sum of vehicle rolling resistance and'aerodynamic drag.
The acceleration horsepower is readily computed in terms of the
acceleration of.the vehicle multiplied by the vehicle speed
(HP)	= m dv v	/1N
acc	(1)
where
(HP)acc = horsepower required to overcome acceleration load
m = mass of vehicle
v = vehicle speed
= vehicle acceleration
dt
Various correlations have been suggested to determine the power
required to overcome vehicle rolling resistance. Many organizations
have recommended that the power required to overcome rolling resis-
tance, (HP)roj^, can be accurately represented by the following rela-
tionship
(HP) roll = mfv	(2)
where m and v are as defined above and f is the tire rolling resistance
coefficient. The parameter f is difficult to determine accurately and
tends to depend on a number of parameters, including tire type, vehi-
cle speed, ambient conditionsy tire-inflation pressure, road-surface
conditions, etc. There are three common types of tires available--

-------
99
bias, bias-belted and radial. Typical rolling resistance coefficients
48
for these three types of tires vary as shown in Figure 5.5.
The power to overcqme aerodynamic drag, (HP) , is given by
n * 3	aero
(HP)aero " h/° ^
where
= density of air surrounding the vehicle
C^j = vehicle drag coefficient
A = projected vehicle frontal area
The drag coefficient, C^, is in the range of 0.4 - 0.6, and the pro-
duct CjjA ranges from approximately 8 to 15 ft for most LDMV.
The total power required to overcome road load can be computed by
adding Equations (2) and (3)
road = W)roll + <"*>„«„ " »«v + kfiW
load
Inspection of Equations (1) and (4) indicates that the vehicle
inertia weight has a direct contribution to two out of the three power
terms. The third term,	, is also indirectly influenced by the
vehicle's weight in that heavier vehicles may have larger frontal sur-'
face areas, A, than lighter vehicles.
The relative magnitudes of the horsepower required to overcome
rolling resistance, aerodynamic drag and various acceleration rates
for a "typical" 4,500 lb inertia-weight vehicle are shown in Figure
5.6. In these computations, the rolling resistance coefficient f,
the CjA term and the air density at standard conditions were taken to
be .015, 13 ft^ and .074 lb/ft^, respectively. In order to compare
(HP) with the other power requirements, the acceleration horsepower
d oo	2
for two acceleration rates of 4.84 ft/sec (maximum acceleration rate
2
in CVS-CH cycle) and 2.0 ft/sec is also plotted in Figure 5.6.
In general, due to the magnitude of the coefficients and the
cubic dependence on velocity, the aerodynamic horsepower, (^)aero>
is insignificant in urban low-speed driving when compared to the

-------
100
FIGURE 5.5 Tire Rolling Resistance as a Function of Speed
REF. 48

-------
101
SPEED (mph)
FIGURE 5.6 Horsepower Required to Overcome Inertia, Aerodynamic and
Rolling Loads as a Function of Speed for a Specific Vehicle

-------
102
rolling resistance horsepower, (HP)However, the importance of
aerodynamic drag ori total road-load horsepower becomes very signifi-
cant at high speeds., Thus, at high speeds the aerodynamic styling of
the vehicle will have a significant effect on both the vehicle's fuel
economy and its maximum speed.
Obviously, the aerodynamic drag must also be a function of the
surrounding air velocity in the form of head winds and crosswinds
which can significantly affect fuel economy. Inspection of Figure 5.6
also indicates.that the instantaneous power required to accelerate the
2
vehicle at 4.84 ft/sec is very significant when compared to the road-
load power.at the same speed. Even relatively moderate accelerations,
2
2.ft/sec , require significant instantaneous power outputs when com-
pared to road load in order to overcome the inertia effects. Since
the product C^A is not a linear function of a vehicle's mass, the
relative importance of aerodynamic drag tends to increase for lower
inertia-weight vehicles.
The spark-ignition, internal-combustion engine does not effi-
ciently convert the energy released from the combustion process to
mechanical work available at the drive shaft. In the theoretical
limit, an ideal spark-ignition engine with a compression ratio of
approximately 8.2 could attain a maximum thermal efficiency in the
range of 57%. However, this ideal efficiency represents an engine
operating without heat losses to the cooling water or lubricating oil,
without energy losses due to friction drive" train or pumping losses
and without combustion inefficiencies etc. Due to these and other
factors, actual spark-ignition engines only have thermal efficiencies
in the range of 20%. An estimate of the percent of energy liberated
by the combustion process available to propel the vehicle and drive
the vehicles accessories along with an estimate of the relative mag-
nitude of the losses as a function of vehicle cruise sp.eed are shown
49
in Figure 5.7. In general, increasing engine compression ratio
within operational limits results in both fuel economy and acceleration

-------
103
y Heat Rejected
Cooling Water
Heat Available to
y Propel Vehicle
and Accessories
50	60
SPEED (mph)
FIGURE 5,7 Energy Balance for a Typical Spark-Ignition Engine as a
Function of Cruise Speed *(Pumping losses occur during the intake
stroke and partial throttle setting and result from the piston pushing
against the atmospheric pressure in the crankcase offset by partial
vacuum in the combustion chamber.)
REF. 49

-------
104
gains, whereas, an increase in engine displacement tends to result in
fuel economy losses and acceleration gains. An increase in engine dis-
placement does not necessarily result in fuel economy losses since an
increase in displacement could be accompanied by a reduction of the
axle ratio. A reduction of the axle ratio would result in lower engine
speed at the same vehicle speed which could tend to increase the fuel
economy of a given vehicle. In general, a numerically lower axle ratio
will allow for a better fuel economy because the engine will turn
slower for a given vehicle speed at the same power and will have less
internal friction.
The axle is only one of the driveline characteristics that can
affect fuel economy. Another source of fuel economy loss is within
the automatic transmission with its associated torque converter slip-
page. A change from a high-slip loose to a low-slip, tight-torque
converter can increase fuel economy. The basic fuel economy differ-
ence between a manual and automatic transmission should also be con-
sidered. In general, a manual-transmission-equipped vehicle will
obtain better fuel economy than the equivalent automatic-transmission-
equipped vehicle. However, depending on the test cycle, the vehicle
speed, and the rear-axle ratio, it is possible for an automatic-
transmission-equipped vehicle to obtain better fuel economy than a
manual-transmission vehicle.
The state of a vehicle's maintenance can also affect its fuel
economy. Proper tire inflation and wheel alignment can reduce rolling
resistance and improve fuel economy. In addition, the state of a
vehicle's tune and other calibration factors can significantly affect
fuel economy.
The addition of convenience accessories such as air conditioning,
power steering, power brakes, heater and defroster blowers, power win-
dows, power seat adjusters, lighting systems, etc. can also signifi-
cantly lower a given vehicle's fuel economy. For example, the esti-
mated penalty in fuel economy over an unequipped car for air

-------
105
conditioning is 9% to 20%. In general, all power-assist devices decrease
the vehicle's fuel economy because they require additional power to
operate and add weight to the vehicle.
It is generally acknowledged that the fuel economy of many vehicles
has decreased due to the introduction of emission controls. However,
the magnitude of this decrease is a relatively controversial subject.
2
EPA has compared the fuel economies, measured on the CVS-C test cycle,
of pre-controlled vehicles (1957-1967 average) to the -fuel economy of
1973 vehicles as a function of inertia-weight class. Their results
indicated the fuel economy of vehicles in all inertia-weight classes up
to 3,500 lb obtained slightly improved fuel economy even in spite of
the installation of emission controls while vehicles in the inertia-
weight class 4,000 lb and above have poorer fuel economy (ranging from
approximately 14% to 18%) than comparable uncontrolled vehicles.. Other
authors have contributed a greater fuel economy penalty to emission
controls.
Since vehicles are not generally driven by the consumer in the
steady-state cruise mode' and an individual driver strongly influences
the rate of accelerations and decelerations, the fuel.economy obtained
by different drivers over the same route under identical traffic con-
ditions will vary significantly.
Significant differences have been reported for fuel economy of
vehicles for a trip started from a "cold-start condition" as compared
to the fuel economy for the same trip and vehicle when it is started
from a "hot-start condition." Data, which will be discussed in more
detail in Section 5.6, have indicated that short trips in the range of
3 to 5 mi initiated from a cold-start condition can result in fuel
economy penalties of as much as 20% to 35% when compared to fuel econ-
omy of a fully warmed-up vehicle. In addition, variations in ambient
conditions affect vehicle fuel economy.
The preceding discussion has been designed to provide an overview
rather than a dissertation on large numbers of important factors that

-------
106
influence the fuel economy of a LDMV. Based on these discussions, it
can be concluded that the relative fuel economy of various vehicles
can only be reported with any degree of certainty if standardized fuel
economy test methods and procedures are developed.
5.3 Test Methods and Procedures
The critical elements in any standardized fuel economy test method
and procedure must include: a representative standard driving cycle or
cycles; a method to determine the mass or volume of fuel consumed and
the distance traveled during the test; and a procedure for simulating
vehicular loads encountered during the course of the driving cycle.
Questions concerning the development and choice of appropriate
fuel economy driving cycles have been discussed in Section 3.4. These
considerations led to the conclusion that since driving patterns change
with time and two drivers would tend to drive the same route in a dif-
ferent manner, it is impossible to develop an absolutely typical fuel
economy driving cycle. However, it was recommended that two standard-
ized fuel economy driving cycles, one designed to represent urban fuel
economy and one designed to represent highway driving, be adopted to
evaluate the relative fuel economies of various automotive designs and
to compare the relative fuel economies of one vehicle versus another
in order to allow rational consumer choice.
Two techniques have been commonly applied to measure fuel economy.
In the first technique, a direct measurement of the mass or volume of
fuel used during the course of the test is obtained from appropriate
instrumentation and this data is used in conjunction with the distance
traveled and conversion factors to compute the vehicle fuel economy in
miles per gallon (mpg). The second technique may be termed the Carbon
Mass Balance Method. When employing this technique, the H.C, CO and
CO2 emissions are collected and measured during the course of the driv-
ing cycle as in the CVS-CH test procedure. Since the mass of carbon

-------
107
per gallon of gasoline is known, the measurement of the vehicle emis-
sions during the course of the test can be employed"in conjunction with
the distance traveled to compute a fuel economy based on the concept
that carbon mass is conserved. The major assumptions required when
using this technique include:
1.	The carbon contained in the HC, CO and CO2 in the exhaust is
the only carbon in the exhaust. This means that other carbon-
containing compounds, such as oxygenated hydrocarbons that are
not detected by a Flame Ionization Detector (FID) and carbo-
naceous particulates, are ignored.
2.	All of the carbon that is measured in the exhaust in the form
of HC, CO and CO2 came from the fuel; there are no other
sources of carbon.
3.	All the fuel consumed during the test can be accounted for by
the carbon in the exhaust. This means that all of the fuel
that leaves the tank during the test is assumed to pass
through the engine and that no carbon leaks out of the exhaust
system before being analyzed or evaporates from the vehicle.
Both of these test methods for fuel economy measurements have
advantages and disadvantages and are subject to significant errors if
careful experimental techniques and procedures are not followed. How-
ever, under conditions where both techniques can be simultaneously
employed, either method can be used to obtain accurate measurements of
fuel economy. An .early study designed to examine the correlation
between measured fuel economies based on, the Carbon Mass Balance and
weighing techniques was reported in Reference 2. In this study, eight
CVS-C tests were conducted at the EPA Motor Vehicle Emission Laboratory
in Ann Arbor, Michigan, on three different vehicles. The average dif-
ference in fuel economy (calculated fuel economy minus weighted fuel
economy, divided by weighted fuel economy) was found to be 4.57® with
the calculated fuel economy being higher than the weighed fuel economy.
The individual differences ranged from 2.6% to 8.17o. Another investi-
gation was performed using data from the work reported in Reference 2.
The same calculation was performed on 245 sets of data for which there

-------
108
were both a fuel weight and HC, CO and CO^ data. The same type of
calculation yielded a 3.37o difference with the standard deviation of
the difference being 8.1%. Conversations with a number of individuals
from various laboratories during the course of the present study sug-
gested that excellent correlation can presently be obtained between the
two fuel economy test methods discussed above.
Recent data obtained from an automotive company"*^ in which three
vehicles were tested nine times each on both the CVS-CH cycle and the
EPA highway cycle have indicated that the standard deviation as a per-
cent of mean for fuel economy measurement based on fuel-metering tech-
niques and the Carbon Mass Balance method are approximately equal.
These results are given by:
s	s
X	X
Carbon Balance	Fuel Meter
Cycle Method	Method
CVS-CH 2.77,	1.97.
EPA's Highway 1.6%	1.9%
All of these results indicate that either the fuel meter method or
the Carbon Mass Balance Method can be successfully employed to accu-
rately determine fuel consumption if careful experimental techniques
and procedures are developed and followed during testing. Alternate
satisfactory techniques are readily available to determine distance
traveled during the test.
Once the fuel economy driving cycles and procedures for determin-
ing fuel consumption and distance traveled have been defined, one must
specify how the appropriate inertial and road loads are to be applied
to the vehicle as it traverses the specified driving cycle. Basically,
two alternate procedures have been suggested. First, there is the
obvious dynamic technique of simply driving the vehicle on the road or
on a test track with appropriate instrumentation to allow the driver to

-------
109
follow the given velocity versus time cycle. The second technique that
has been proposed is to simulate the actual loading of the vehicle dur-
ing the course of the driving cycle with the aid of a chassis dynamome-
ter. It should be noted that there is no inherent technical reason for
eliminating the use of chassis dynamometers from fuel economy testing
if it can' be shown that they accurately represent the actual inertial,
aerodynamic rolling resistance and accessory loads experienced by the
vehicle as it traverses the driving cycle. Both the road or track and
dynamometer procedures have inherent advantages and disadvantages as
outlined below.
The advantages of a road or track test procedure include:
1.	The total load, including inertia loads, road loads and the
effects of accessory loading are accurately simulated.
2.	Fuel economy tests may be carried out anywhere a suitable
level stretch of pavement can be found.
3.	Assuming that an appropriate road or track is available, rela-
tively low expenditure for capital is required.
Disadvantages of road and track testing include:
1.	Testing cannot be carried out under adverse weather conditions.
2.	Correlation parameters, designed to take into account the
effects of ambient variables on fuel economy measurements,
must be developed and applied to the base data in order to
obtain fuel economy measurements at standard conditions.
3.	Some difficulties are encountered with reproducibility of
vehicle speed versus time .especially when complex cycles, such
as the UDDS or proposed EPA highway cycle, are employed.
4.	There is a lack of a satisfactory evaluation of the effect of
cold start on fuel economy.
5.	Fuel economy and exhaust emissions cannot be simultaneously
measured.

-------
110
The advantages of test methods based on the use of a chassis dyna-
mometer include:
1.	The test can be conducted independent of local weather condi-
tions.
2.	Since test cell conditions can be controlled, correction fac-
tors for the effect of ambient conditions on experimental
results are minimized.
3.	The same driving cycle (speed versus time trace) can be fol-
lowed within specified close tolerances.
4.	Fuel economy and exhaust emissions can be simultaneously mea-
sured .
There are also a number of disadvantages to employing a chassis
dynamometer test procedure. These include:
1.	Total load simulated on the chassis dynamometer may not accu-
rately duplicate the rolling resistance and aerodynamic drag
experienced by the vehicle on the road.
2.	In general, the cooling fan airflow characteristics in a dyna-
mometer facility do not exactly reproduce the airflow charac-
teristics of a moving vehicle. Therefore, the effect of
vehicle warm-up may be slightly different than that encoun-
tered on the road.
3.	The present method of accounting for the existence of air con-
ditioning, by setting in the dynamometer a 10% increase in
road load horsepower at 50 mph, may not exactly duplicate the
overall effects of air conditioning on fuel economy.
4.	The large discrete inertia-weight differences, 250 lb and
500 lb on the chassis dynamometer presently employed for fuel
economy testing that are used to simulate load may bias the
simulated load too high or too low for a given vehicle due to
the vehicle's true inertia weight being too close to the cut-
off point between inertia-weight classes. Due to the manner
in which inertia-weight loads are presently set in the CVS-CH
test method, a vehicle at the upper limit of a given weight
class receives a fuel economy advantage over its expected
road fuel economy in the range 57o to 8%. Conversely, a vehi-
cle at the low end of the inertia-weight class receives a
penalty in the range 5% to 8%. These computations were
obtained using EPA's^ simple correlation of fuel economy
versus inertia weight.

-------
Ill
EPA has simultaneously measured the fuel,economy and emissions of
LDMV operating on both the CVS-C and CVS-CH test cycles. The test
cycles have been driven on a chassis dynamometer and the test methods
and procedures have been similar to the one utilized during the course
of emission testing. The fuel consumed during the test has been gener-
ally measured by the Carbon Mass Balance Method, and the fuel economy
has been computed by dividing the cycle distance by the volume of fuel
consumed, during the course of the test. Data obtained in this manner
have taken into account effects due to both cold-start and hot-start
conditions, and the results have been reported as representative of
urban fuel economy.
In addition to measuring and reporting urban fuel economy as
determined during the CVS-CH test, EPA plans to measure and report
highway fuel economies for 1975 model-year vehicles. The EPA highway
cycle (see Section 3.4) will be employed for these tests. As in the
case of the urban fuel economy data, the cycle will be driven on a
chassis dynamometer, the fuel consumed will be measured by the Carbon
Mass Balance Method, and the vehicle's highway fuel economy will be
computed by dividing the length of the highway cycle by the fuel con-
sumed. These results will correspond to a hot-start highway driving
cycle since the highway cycle will be driven after the CVS-CH cycle is
completed.
18
The SAE has recently adopted a set of recommended practices for
fuel economy measurements based on a road and track test procedure..
Reference 18 recommends four driving cycles (see Section 3.4), all of
which are to be driven on a road or track. Complete specifications of
the recommended practice may be found in Reference 18. All tests are
to be carried out on a fully warmed-up vehicle. Fuel economy will be
based on the average fuel consumed and distance traveled on two suc-
cessive runs over the course. The test is to be repeated until two
successive tests repeat within 27» for fuel consumption and 17„ for
elapsed time. The measured fuel economy is subsequently corrected to

-------
112
standard conditions by multiplying the measured value by empirical cor-
rection factors designed to account for fuel properties and ambient
variations.
5.4 Statistical Variation of Fuel Economy Measurements
As in the case of exhaust emissions, information concerning the
statistical variability of fuel economy measurements is of importance.
Information concerning the variability of measured fuel economy in a
given test cell or test track, the variability of measurements from
cell to cell at a given laboratory and the variability of measurements
from laboratory to laboratory for a given vehicle are of interest.
Another major concern in fuel economy measurements is the problem asso-
ciated with the estimation of the fuel economy of an entire class of
vehicles based on tests carried out with a single or small number of
vehicles from that class. This type of error in reported fuel economy
is due mainly to the vehicle-to-vehicle variability. Data obtained
during this study concerning statistical variations in fuel economy
measurements are summarized below.
Due to the recently developed concern with fuel economy, only a
small amount of the data required to examine the variabilities men-
tioned above is readily available. However, since a large amount of
the exhaust emission data has included CC^ and since CC^ is by far the
predominate factor in determining fuel economy based on the Carbon
Mass Balance Method, the statistical variability of CC^ emission data
may be readily used to estimate the statistical variability of fuel
economy. Whenever possible, actual reported fuel economy data will be
quoted.
22
As previously noted in Section 4.3, EPA has reported the results
of 30 CVS-CH tests designed to determine the reproducibility of both
emissions and fuel economy. The Carbon Mass Balance Method was used to
determine fuel economy. The results of these tests give an average

-------
113
fuel economy and standard deviation as percent of the mean of 10.31 mpg
and 2.6%, respectively. Since the CVS-CH procedure has a cold start
and a hot start, the emission data for this vehicle in a given cell can
be used to calculate both hot-start and cold-start fuel economy. Emis-
sions from bag 1 were used to calculate a cold-start fuel economy while
emissions from bag 3 were used to compute the hot-start fuel economy.
The hot-start fuel economy and standard deviation as percent of the
mean were computed to be 11.85 mpg and 2^5%, respectively, as compared
to the cold-start values of 10.03 mpg and 2.67«.
Another study, previously discussed in Section 4.3, reported
emissions for 16 CVS-CH tests of a Chevrolet Impala equipped with an
23
oxidizing catalyst in a single test cell. As shown in Table 4.3, the
COg standard deviation as percent of the mean was reported to be 1.2%.
As previously stated, this indicates that the fuel economy variation as
percent of the mean was also approximately 1.2%.
The fuel economy variation of a 1973 model-year production vehicle
has been reported for a number of different cells at the same labora-
26
tory. A CVS-H test procedure was followed and the fuel economy was
computed based on-the Carbon Mass Balance Method. The range of data
obtained -in five different test cells is given by
Mean Fuel Economy
Number of Tests For Each Cell s
at Each Cell	x, mpg		x
13-18	10.83-11.51	2.8%-5.5%
A vehicle equipped with an oxidizing catalyst has been tested sev-
eral times over the CVS-CH cycle in various cells at Chrysler Corpora-
26
tion and three times in one test cell at EPA in Ann Arbor. Fuel
economy was not reported, but the C02 standard deviation as percent of
mean (see Table 4.6) was 10.6% arid 2.1% at Chrysler -and EPA, respec-
tively.

-------
114
An automobile manufacturer50 has reported that the test-to-test
fuel economy standard deviation of a single vehicle tested on a track
procedure as 1.1% of the mean. This variability may be due to the fact
that road fuel economy data are often rejected as part of the standard-
ized test procedures if the average fuel economy for two runs up and
down the track is not within specified close tolerances.
Once the cell variation has been established, the cell-to-cell
variation in a given laboratory may be examined. Some REPCA I data for
cell-to-cell variability of fuel economy as measured on the CVS-H cycle
26
have been obtained. In addition, similar data from another labora-
tory have been obtained for a 1973 model production vehicle on a CVS-H
26
cycle. These data are summarized below:
Total Number Total Number Total Mean Fuel s
Laboratory	of Tests	of Cells Economy x, mpg x
General Motors	70	4	13.61	1.8%
Chrysler	90	6	11.08	4.4%
The next level of complication is to determine a representative
laboratory-to-laboratory variability of fuel economy measurements. The
SAE Fuel Economy Task Force has carried out a round robin fuel economy
program for three vehicles at various laboratories. The tests were
carried out using the CVS-C test procedure and fuel economy was deter-
26
mined by the Carbon Mass Balance Method. The results of this study
are summarized in Table 5.1.
The results of another correlation study50 are listed in Table 5.2.
The tests were carried out on three vehicles operating on the CVS-H
test. The overall results listed in Table 5.2 are based on a simple
pooling of all data.
Reference 50 has reported data relating to the vehicle-to-vehicle
variation in fuel economy among samples of identical vehicles. A test-
track procedure was used to determine the desired fuel economy and the

-------
115
Table 5.1 Summary of Round Robin Fuel Economy Test Program
Compact	Compact	Full Size
6 Cyl. V-8 V-8
No. of Laboratories	5	5	6
No. of Tests	8	7	8
Mean CVS Fuel Economy	15.6 MPG	11.8 MPG	11.0 MPG
Standard Deviation of Tests	.95 MPG	.45 MPG	.97 MPG
Percent Std. Dev. of Tests	6.1%	3.8%	8.8%
Range of Fuel Economy	14.3-16.7	11.3-12.4	9.6-11.0
Dynamometer Horsepowers Set Per Federal Register
REF. 26

-------
116
Table 5.2 Summary of CVS-H Fuel Economy within, between and over
Various Laboratories^"
Vehicles


2.
1973
5L Capri
1974
351W Monte go
460
1974
T'Bird

Facilities
X
s
s/x
X
s
s/x
X
s
s/x
1.
Environmental
Protection Agency
20.0
.54
.027
12.6
.24
.019
10.5
.61
.058
2.
Ford Emission Test
Laboratory(ETL-75)
19.7
.56
.028
12.5
.12
.010
10.5
.16
.015
3.
Ford Emission Test
Laboratory(ETL-74)
19.7
.59
.030
12.4
.15
.012
10.6
.23
.022
4.
Ford Allen Park
Testing Laboratory
19.6
.30
.015
12.6
.19
.015
10.7
.10
.009
5.
Ford Los Angeles
Laboratory
18.7
.21
.011
-
-
-
9.7
.15
.016
6.
California Air
Resources Board
18.9
.22
.012
-
-
-
10.5
.17
.016
Overall	19.5 .64 .033 12.5 .20 .016 10.4 .51 .049
Tests at EPA and ETL-75 were conducted with 1975 CO instrumentation.
Testing at ETL-74 and remainder of facilities were with 1974 CO in-
strumentation.
x - fuel economy in mpg
s - standard deviation in mpg
s - standard deviation as percent of mean
x
REF. 50

-------
117
data for a large number of measurements indicated that the vehicle-to-
s	•1
vehicle variation in fuel economy, x, is 3.5%. Reference 50 also
reported the results of a study conducted to determine the variability
of results from both test-track and laboratory fuel economy tests. The
study concluded that with a single vehicle the true population mean can
be established within + 6.5% for the Ford City-Suburban cycle (test
track) and + 11.02% for the CVS-C cycle (dynamometer) with 90% confi-
dence. The sample sizes required to predict mean fuel economy for a
given car line within + 3% of the true population mean at 90% confi-
dence was estimated to require testing of five vehicles for the test-
track cycle compared to 14 vehicles for the laboratory dynamometer
25
cycle. Another source has reported that the fuel economy and stan-
dard deviation as percent of mean for CVS-CH tests on 38 different 1.5-
liter, stratified-charge CVCC production vehicles was 25.2 mpg and 3.6%,
respectively.
5.5 Reliability of Fuel Economy Measurements Based on Chassis Dyna-
mometer Tests
The ability of any fuel economy test procedure, incorporating a
chassis dynamometer as one of its elements, to accurately determine the
fuel economy of a vehicle over a specified.driving cycle has been ques-
tioned. It has been previously stated in Section 5.3 that there is no
inherent technical reason for eliminating the use of chassis dynamome-
ters from fuel economy testing if it can be shown that they adequately
simulate the actual inertial, aerodynamic, rolling resistance and
accessory loads experienced by a vehicle as.it traverses the driving
cycle. Unfortunately, the results of definitive studies designed to
address this question are not presently available. However, due to the
importance of this issue, the limited data which are presently avail-
able concerning the repeatability and accuracy of fuel,economy test
procedures incorporating a chassis dynamometer as the elements for both
urban and highway driving cycles are discussed below.

-------
118
The test results presented in Section 5.4 indicate that fuel econ-
omy measurements employing chassis dynamometers for an urban driving
cycle (e.g., CVS-H, CVS-C and CVS-CH tests) are reasonably repeatable.
In fact, the data indicate that the variability of measured fuel economy
in terms of the standard deviation as a percent of the mean in a given
cell, from cell to cell at a given laboratory and from laboratory to
laboratory is in the range of 2% to 8%. The limited data obtained for
reproducibility studies on the EPA highway cycle have indicated a simi-
lar variability.
Questions relating to the ability of a chassis dynamometer to ade-
quately represent the inertia and road loads as a function of speed for
a given vehicle are also of importance. In addition, it would be desir-
able to have data relating to the importance of variations in loading on
fuel economy measurements. It is generally agreed that dynamometers
with twin small-diameter rolls, such as the Clayton-type units presently
utilized by EPA, do not provide the same tire-roll loading effects as
those found on the road. It is possible to minimize inaccuracies in
dynamometer loading if the load on the dynamometer is set so that vehi-
cle coast-down time on the rolls matches vehicle coast-down times
obtained on the road. In addition, the Clayton-type units are generally
limited in load adjustment and inertia-weight increments. The Federal
Register presently specifies 250 lb inertia-weight increments for low
inertia-weight vehicles in the range 1,500 to 2,750 lb and 500 lb
inertia-weight increments in the range 3,000 to 5,000 lb. The effect
that this test procedure has on dyno versus expected road fuel economy
has been discussed in Section 5.3.
Large-roll electric dynamometers provide independent adjustment of
simulation for aerodynamic drag and rolling resistance plus excellent
vehicle mass simulation. However, these dynamometers are significantly
more expensive than small-roll units.

-------
119
Data reported in Reference 29 have indicated that the dynamometer
load applied by a Clayton-type dyno while driving the CVS-H cycle with
three different inertia-weight vehicles was a function of both dyno-roll
spacing and dyno-calibration method. The total work (measured by drive
shaft torque) during the CVS-H test using 20 in. roll spacing versus
17% in. roll spacing on all inertia-weight vehicles was higher by an
average of 1.3%. However, analysis of the data indicated that roll
spacing had no significant effect on CO^ emissions and, hence, it can be
concluded that roll spacing did not have a significant effect on fuel
economy.
The dynamometer calibration technique for setting road load at
50 mph was also shown to affect total work required to drive the CVS-H
cycle. Differences in work ranging from 1.2% less total work for the
3,000 lb inertia-weight vehicle to 2.4% more work for the 5,500 lb vehi-
cle were reported as a function of dynamometer-calibration method.
Another manufacturer"*^ has' reported that' the calibration of the
indicated load on a given Clayton-type dynamometer is not as reproduci-
ble as might be desired and that this factor may cause variations of as
much as 10% in fuel economy measured for low inertia-weight vehicles on
the CVS-CH test.
The SAE-recommended urban, suburban and 70-mph interstate driving
cycles fuel economies were determined at three different test tracks,
one rcau site and on a large roll chassis dynamometer. The results of
52
these tests are summarized in Table 5.3. The road loads on the large-
roll, mechanical-chassis dynamometer were set by matching as closely as
possible the coast-down times of the vehicles on the River Road Highway
to the dyno coast-down times. Coast-down times could be accurately
matched for the two heavy vehicles, but due to internal friction in the
dynamometer, the road coast-down times for the light vehicle could not
be simulated. Therefore, numerous tests at different dyno loads were
carried out for the low inertia-weight vehicle. Due to this problem,

-------
Table 5.3 Fuel Economy Measurements as Measured on Both Road Track Tests and Dynamometer Tests
Vehicle
Company Load Facility
Corrected Fuel Economy
SAE Urban
Cycle
SAE Suburban
Cycle
SAE 70 mph Interstate
Cycle
(mpg)
(mPg)
(mpg)
1973 Dodge Charger
IW = 4,400
CID = 318-2V
1973 Buick Century
IW = 4,620
CID = 350-4V
1973 Ford Pinto
IW = 2,740
CID => 122-2V
Chrysler
GM
Ford
Chrysler
Shell
Shell
Chrysler
GM
Ford
Chrysler
Shell
Shell
Shell
Chrysler
GM
Ford
Chrysler
Shell
Shell
Shell
Shell
Shell
Track
Track
Track
Track
River-Road
Highway
Dyno
Track
Track
Track
Track
River Road
Dyno-normal
load
Dyno-light
Track
Track
Track
Track
River Road
Dyno-Load
Dyno-Load2
Dyno-Load3 *
Dyno-Load4
*
10.2
10.3
10.2
10.0
10.1
9.9
9.9
10.1
9.7
9.9
10.3
10.0
9.8
15.4
15.8
15.5
15.6
16.2
14.3
14.9
14.7
15.2
17.3
17.4
16.7
17.2
16.5
16.3
15.5
15.6
15.2
15.5
15.9
15.6
15.6
22.4
22.5
22.4
22.8
22.7
20.0
20.8
22.6
25.4
15.8
16.1
15.4
15.8
15.1
15.0
14.0
13.8
13.7
14.0
14.0
14.3
14.6
18.3
18.8
18.5
18.9
18.7
16.1
18.3
19.2
23.0
*Dyno-Loadj to Dyno-Load^ represent decreasing dynamometer loading.
REF. 52

-------
121
one should not expect to obtain good agreement between the fuel econo-
mies measured on the track and the dynamometer for the low-weight vehicle.
It should be noted that the dyno was designed and constructed for test-
ing of vehicles near 4,000 to 5,000 lb and that the dyno could have been
designed to minimize internal friction which would allow coast-down
times for low inertia-weight vehicles to be accurately reproduced.
Inspection of the data, reported on all three driving cycles for
the two heavy vehicles, indicates that excellent agreement within 1% to
1
47o of the test-track mean can be obtained on a large-roll chassis dyno-
mometer if care is taken to set the dyno load to reproduce actual road
coast-down times. The results for the low inertia-weight vehicle indi-
cate that significant errors in fuel economy can occur for dynamometer
tests if road load is not adequately simulated by the dynamometer.
Another laboratory has reported fuel economies of eight vehicles
for the EPA highway cycle as measured on both a dynamometer and a test
26
track. The fuel economy of the vehicles driven on the SAE 55-mph
interstate cycle was also measured on the test track. These data are
presented in Table 5.4, and the measured dyno versus track results for
the EPA highway cycle are graphically represented in Figure 5.8. The
dynamometer data were obtained on a Clayton-type dyno, and the road load
at 50 mph was set according to the- "cook book" method established in the
Federal Register.
Inspection of these data indicates that, in all cases, the fuel
economy measured on the dynamometer is larger than the value obtained on
the track. Percentage differences in the values (fuel economy dyno
minus fuel economy track divided by fuel economy dyno) range from
approximately 4% to 167«. It is also important to note that fuel econo-
mies measured on the track for these eight vehicles as measured on the
SAE 55-mph interstate cycle are in good agreement with values obtained
on the EPA highway cycle.
Various sources have analyzed the energy required to drive a vehi-
cle over the CVS-CH cycle, and it has been shown that inertia forces and

-------
Table
5.4 Comparison of Fuel Economies for Highway Cycles
Measured on a


Track and
on a Dynamometer
Fuel
Economy
(mpg)



Dyno Test

Track Tests
Body
Engine
D'ispl.
(CID)
Inertia
Weight
EPA
Highway
Cycle
EPA
Highway
Cycle
SAE
Interstate 55
Cycle
A1 Body
225
3,500
25.7.
22.6.
21.5
B1 Body
318
4,000
20.9
17.5
17.;4
B' Body
360
5,000
18;.l
16.6
16:5
B1 Body
360
5,000
18.6
15.9
16.1
C' Body
400
5,000
17.4
16.2
16.2
C1 Body
400
5,000
14.6
14.0
14.4
C1 Body
440
5,000
16.7
-
15; 8
C' Body
440
5,000
17.6
16.4
16.4
^Dyno-load set as per Federal Register.
REF. 26

-------
123
6 8 10 12 14 16 18 20 22 24 26 28
ROAD TEST-EPA HIGHWAY (mpg)
FIGURE 5.8 Fuel Economy for EPA Highway Cycle as Measured on Both a
Track and a Dynamometer
REF. 26

-------
124
49 53
idle conditions dominate fuel consumption. ' Therefore, potential
errors in dynamometer road-load encountered during CVS-CH tests should
not significantly affect the measured fuel economy. Similar computa-
tions concerning the, energy required to traverse the EPA highway cycle
indicate that rolling 'resistance and aerodynamic drag may constitute as
much as 65% to 75% of the energy required. Therefore, errors associat-
ed with dynamometer road loads may significantly affect the measured
fuel economies on the EPA highway cycle.
5.6 The Affect of Cold-Start and Ambient Temperature on Fuel Economy
As discussed in Section 3.2, many automobile trips are very short,
and in the U.S., more vehicle miles are driven for trips within 0.5
miles of 4.5 miles than trips over any other 1-mile interval. In addi-
tion, a significant number of vehicle trips are initiated from a cold-
start condition which is defined as the initiation of a trip after the
vehicle has remained in a stationary position for a period sufficiently
' • >
long so that it is essentially in' thermal equilibrium with local ambi-
I
ent conditions. The severe fuel economy penalties associated with
short trips initiated from cold-start are discussed below. In addition,
data relating to fuel economy for a specified driving cycle as a func-
tion of ambient temperature are presented;
When considering trips of various lengths, it has been proposed
that two phases of an urban-cycle fuel economy, cold and warm, must be
'46
considered. The cold phase of urban operation includes approximately
the first 10 miles of travel and the warm phase is represented,by trips
longer than 10'miles^ in duration. During the cold-phase operation, the
choke and lubricant warm-up are considered to be i^jor factors in
determining fuel economy. The warm phase may be expected to have sig-
nificant improvements over the cold phase in fuel economy. One method
of graphically illustrating the expected fuel economy due to vehicle
warm-up versus trip length is shown in Figure 5.9. This figure

-------
90
80
70
60
50
40
30
20
10
o Pre-Emissions
Controlled Vehicles (Ref. 54)
*~ a 1973 Vehicles
SAE Task Force Data (Ref. 50)
to
m
I	I
J	I	I	I	L
I
J	I	L
8 9 10 11 12
TRIP LENGTH (mile)
13 14 15 16 17 18
5.9 Effect of Trip Length on Cold-Start Fuel Economy Penalty
Urban Driving Cycle

-------
126
indicates that trip length and necessary warm-up have significant
effects on fuel economy.
The original data in Figure 5.9, in terms of percent of fully
warmed-up fuel economy versus trip lengths, were obtained from Refer-
ence 54. These data were obtained for a fleet of pre-emission-control
vehicles which were equipped with automatic transmissions and automatic
chokes. The vehicles were operated on a simulated city-traffic sched-
ule with a cold start under varying weather conditions. From this
graph, it is seen that the average fuel economy for a trip of approxi-
mately four miles is only about 75% of;the expected fully warmed-up
fuel economy. As is seen, the vehicle's average fuel economy approaches
the fully warmed-up value only if the trip length is relatively long.
Recently, questions have been raised concerning the validity of
these results for vehicles equipped with modern emission controls. In
conjunction with the work of the SAE Fuel Economy Task Force, typical
cold-start fuel economy data were obtained for a 1974 sports compact
• 50
and a 1973 intermediate vehicle. Two conditions were studied: one
where the vehicle soak temperature was in the range from 25 ° F to 45 ° F
(outdoor soak) and one where the soak range was from 65 °'F to 70°F
(indoor soak). The data averaged for a number of urban cycle tests
conducted on two vehicles of each type are also plotted in Figure 5.9.
Inspection of these results indicates that vehicles with advanced
emission-control systems show the same cold-start fuel economy penal-
ties for various trip lengths as vehicles •without emission controls.
This result may be due to the .fact that the warm-up of many non-emission
control related components (e.g., water jacket, transmission, rear end,
tires, etc.) may significantly affect cold-start fuel economy.
Since the warm-up effect as discussed is thermal, ambient tempera-
ture could also be expected to influence fuel economy. Data reported
in Reference 54 indicate that a vehicle will warm up more quickly as
.the ambient temperature increases. These results indicate that a driver
in a cold climate who drives his vehicle for many short trips can expect

-------
127
to receive an average fuel economy for his vehicle which is significant-
ly below that which could be obtained in longer trips and warmer cli-
mates. Reference 54 also indicates that a percentage of fully warmed-
up fuel economy can be calculated for a highway-type driving cycle.
The highway-type warm-up schedule curve reported in Reference 54 falls
only slightly above that of the urban or city warmed-up schedule. Note
that Figure 5.9 has only been discussed in terms of cold-start condi-
tions. Reference 55 suggests that the trend indicated in Figure 5.9
would also apply to vehicles which are in a warmed-up condition before
the trip is started. It would be expected that for a short trip, the
average fuel economy of a hot start would be greater than that of the
equivalent fuel economy of a cold start for the same driving cycle and
trip length.
The influence of ambient temperature on automotive fuel economy
is also of interest. The fuel economy of various classes of vehicles
on the CVS-CH test cycle as a function of ambient soak and operating
temperature as reported by Reference 33 is shown in Figure 5.10. In-
spection of these results indicates that a reduction in ambient temper-
ature from 75°F to 20°F results in an average fuel economy penalty of
approximately 5% for the twenty 1969 to 1971 model production vehicles
and 11% for the prototype catalyst equipped vehicles. Fuel economy
penalties of 13% and 9% between 20°F and 75°F were reported for the
diesel-equipped and stratified-charge PROCO vehicles, respectively.
Since warm-up is seen to have a significant effect on the average
fuel economy of a vehicle and since short trips and cold starts are
known to occur with high frequencies as previously discussed, a fuel
economy test which includes both a cold and a hot start in an urban
cycle would more closely parallel real urban fuel economy. This idea
has been incorporated into the CVS-CH test procedure. Note that the
hot start of the CVS-CH procedure occurs after a cold start and the
first 7.45 miles of the cycle have been driven. Thus, it can be seen

-------
128
Stratified Charge (Proco)		___ _ _______
20 Production Cars 1969-1971
4 Prototype Advanced Emission Cars (all catalyst)
I
JL
_L
20	40	60	80	100
TEST AMBIENT TEMPERATURE (?F)
120
FIGURE 5.10 Effect of Ambient Temperature on Fuel Economy for Various
Types of.Vehicles on the CVS-CH Test Cycle
REF. 33

-------
129
by referring to Figure 5.9 that the hot start of the CVS-CH procedure
is initiated from a slightly less than fully warmed-up condition.
5.7 Summary
Some of the most important factors that influence the fuel economy
of LDMV include the driving cycle and the driving habits of individ-
uals, vehicle characteristics including inertia weight, the effect of
cold-start operation, particularly on short trips, and the state of
maintenance of the vehicle. It has been suggested that since'driving
patterns change with time and that two drivers would tend to drive the
same route in a different manner, it is impossible to develop an abso-
lutely typical fuel economy driving cycle. However, since fuel economy
is strongly dependent on the driving cycle, it is important to develop
standardized fuel economy driving cycles and test methods and proce-
dures in order to evaluate the relative fuel economies of various auto-
motive designs and to compare the relative fuel economies of one vehi-
cle versus another. In order to obtain a reasonably accurate determi-
nation of the fuel economy of a vehicle, it is concluded that two
standard LDMV fuel economy driving cycles, one designed to represent
urban driving and one designed to represent highway driving, are needed.
It has been shown that severe fuel economy penalties are incurred
when an LDMV is driven on a short trip that is initiated from'a cold-
start condition. In addition, it has been shown that many automotive
trips are very short and that more vehicle miles are driven for trips
within 0.5 miles of 4.5 miles than trips over any other 1-mile inter-
val. Therefore, it is concluded that in order to obtain economies that
are in good agreement with expected real-life, urban fuel economies, an
urban fuel economy test method and procedure should include both a cold-
start and hot-start phase. The driving cycle associated with the
CVS-CH test method meets this requirement.

-------
130
The critical elements in any standardized fuel economy test method
and procedure include: a representative standard driving cycle or
cyclesj a method to determine the mass or volume of fuel consumed and
the distance traveled during the test; and a procedure for simulating
vehicular loads encountered during the course of the driving cycle.
Since the driving cycle associated with the CVS-CH test1method
meets the criteria for an appropriate urban fuel economy driving cycle
and a significant data base of urban fuel economies has been collected
with this cycle, it is-concluded that the CVS-CH-driving cycle ,can be
used as the standard urban fuel, economy driving cycle. In addition, it
is suggested that, either the EPA highway cycle or a combination of the
SAE,suburban and interstate cycles can be employed as the standard-
highway fuel economy, cycle.
Two methods haye been utilized to measure fuel consumption. The
first method is based on the direct measurement of mass or volume of
fuel consumed during the course of the test. The second method,..the
Carbon Mass Balance Method, is based on the collection ..and measurement
of carbon containing compounds in the vehicle's .exhaust.. Since the
mass of carbon per gallon of fuel .is known, the mass, of carbon in the
exhaust can be readily employed to compute the mass or volume of. fuel
consumed during the test. rWhen applying either method, the distance
traveled during the test- is measured by.alternate satisfactory tech-
niques and fuel economy is readily computed. Both of these test meth-
ods for fuel consumption measurements have advantages and disadvantages
and are subject to significant errors if careful experimental tech-
niques and procedures are not followed. However, under- conditions
where both techniques can be simultaneously employed, either method
can be used to obtain, accurate measurements of fuel consumption.,
Two procedures!have -been proposed.for simulating vehicular loads
encountered during the course of the driving cycle. One method is
based on driving the vehicle over a suitable road or test^track, and

-------
131
the second method is based on driving the vehicle on a chassis dynamome-
ter. Both of these procedures have distinct advantages and disadvan-
tages. It has been concluded that there is no inherent technical reason
for eliminating the use of chassis dynamometers from fuel economy test-
ing if it can be shown that they adequately simulate the actual iner-
tial, aerodynamic, rolling resistance and accessory loads experienced
by a vehicle as it traverses the driving cycle. Unfortunately, the
results of definitive studies designed to evaluate the ability of chas-
sis dynamometers to accurately simulate vehicle loads under a wide
variety of conditions and to evaluate errors in fuel economy measure-
ments associated with specified errors in dynamometer load simulation
have not been reported.
Based on the limited data presently available, one can speculate
that fuel economy tests incorporating chassis dynamometers, with suit-
ably fine inertia-weight controls, can be successfully employed to
determine fuel economies for urban driving cycles where inertia forces
and idle conditions rather than road loads dominate LDMV fuel consump-
tion. In addition, limited data have indicated that fuel economy data
obtained on chassis dynamometers and road or track tests can be made to
correlate well even for ftonurban driving cycles, where road loads domi-
nate vehicle fuel consumption, if care is taken to set the dynamometer
loading so that vehicle coast-down times obtained on the dynamometer
closely match vehicle coast-down times obtained' on the road. The
results of another study Indicate that fuel economies for the EPA high-
way driving cycle as measured on a Clayton-type dyno, where the road
load at 50 mph was set according to the "cook book" method established
in the Federal Register, did not correlate well with results obtained
on a test track. In all cases, the fuel economies measured on the dyno
were larger than the corresponding values measured on the track and dif-
ferences between the two values ranged from 47o to 16%. These results
indicate that certain vehicles may obtain significantly higher fuel

-------
132
economies than are actually warranted on the EPA highway cycle using
present EPA fuel economy test'methods and procedures.
Fuel.economy measurements are.considerably more reproducible .than
emissions measurements^ Data obtained during the course.of this study
indicate that fuel economy measurements employing chassis dynamometers
for an urban driving cycle (e.g., CVS-H, CVS-C and CVS-CH tests) are*
reasonably repeatable., In fact, the data indicate that the variability
of measured fuel economy in terms of the standard deviation as' a per-
cent of the mean in a given cell, from cell to cell at a given labora-
tory and from laboratory,to laboratory, is in the range of 2% to 8%..
The limited.data obtained for reproducibility studies on the EPA high-
way cycle have indicated,a similar variability.
It has been reported that a large number of fuel econony measure-
ments. carried .out on test tracks, have indicated that vehicle-to-vehicle
variation in fuel economy for identical vehicles is. of the order of
3.5% of the,mean fuel economy. The results,of afstudy designed to
determine the variability'of results from both test-track and chassis-
dynamometer fuel" economy tests have also-been reported. The?study con-
cluded that with a single vehicle the true population'mean. can.be -
established within approximately 7%'for a city/suburban driving cycle
on a test track and approximately 117» for the urban CVSrC test cycle on
a chassis dynamometer with 90% confidence. The sample;sizes required
to predict mean fuel economy for a given car. line within 3% of the true
population mean at 907o confidence were estimated to,require testing of
5 vehicles for the test-track cycle compared to 14.vehicles for the
laboratory chassis-dynamometer cycle.
Due to the potential significance that may be attached to reported
fuel economies,of individual vehicles, it is suggested that reported
fuel economies be based on .the results of several tests.

-------
6.0 EVALUATION OF THE DATA
6.1	Summary
The important aspects of present and projected emission and fuel
economy test methods and procedures for LDMV have been considered.
The CVS-CH exhaust emission test procedure plays an important role in
the certification process for 1975 and subsequent model-year LDMV.
Hence, the statistical variability of CVS-CH exhaust emission tests for
these vehicles and the relative magnitudes of the various factors
affecting both systematic and random errors encountered during CVS-CH
tests were discussed. Additional questions concerning the effect of
ambient temperature on exhaust emissions, the suitability of present
exhaust emission-control durability test methods and procedures and the
possible modifications of the present HC exhaust-emission standards and
measurlngi techniques designed to account for only reactive HC were also
addressed. Finally, recent data relating to the effectiveness of pre-
sent evaporative HC emission-test methods and procedures and LDMV
evaporative-control systems were presented and discussed.
At the present time, generally accepted standardized fuel economy
test methods and procedures are not available. Due to the significance
of this issue, it is important that standardized test methods and proce-
dures, based on sound engineering and scientific considerations, be
developed as soon as possible. In the general area of fuel economy,
the important factors affecting LDMV fuel economy, the important ele-
ments in any standardized fuel economy test method and procedure, the
currently proposed test methods and procedures and the statistical
variability of fuel economy measurements were considered in some
detail.
6.2	Conclusions
The major conclusions reached during this study have been summa-
rized in Section 1.2. A more detailed summary of the major conclusions
and the supporting data are given below:
133

-------
134
Recent data, obtained utilizing the SHED method, indicate that
evaporative HC emissions from vehicles equipped with present-
technology, evaporative-control systems are of the order of
1.9 g/mi which is larger than the 1975-76 federal HC exhaust
emission standards of 1.5 g/mi. These results were obtained
on systems in which evaporative HC emissions were reportedly
95% controlled with respect to uncontrolled HC. evaporative
emission levels of approximately 3.0 g/mi. Therefore, an accu-
rate test method and procedure for measuring evaporative HC
emissions must be developed and implemented as soon as pos-
sible. Careful evaluation of the situation may show that the
SHED method is adequate for this purpose.
The driving cycle associated with the CVS-CH test was found to
represent an average urban trip "as well as it needs to" for
purposes of determining LDMV exhaust emissions.
Significant consequences should not be attached to a single
CVS-CH exhaust-emissions test since both random and systematic
errors contribute, to poor repeatability of exhaust-emission
test results for a given LDMV. Various factors, including
vehicle variability, emission collection and measurement vari-
ability and environmental test variables, contribute to the
poor test reproducibility. Variations in exhaust-emission
measurements specified in terms of the standard deviation as a
percent of the mean for a 1975-76 vehicle in a given cell or
from cell to cell at one laboratory for most engine-control
system configurations can be expected to range between 10% to
257o, 15% to 30% and 5% to 15% for HC, CO and NO , respectively.
X
Significant systematic errors in, mean emission values of a
given test vehicle of as much as 20% to 30% have been reported
between various emission-testing laboratories. Establishment
of mandatory correlation'test programs among governmental,
automotive manufacturers' and other test laboratories that are

-------
135
carrying out exhaust emission measurements on the CVS-CH test
could significantly reduce these errors. These correlation
programs should include CVS-CH tests of relatively stable vehi-
cles and tests of the CVS sampling and gas analysis systems
using either exhaust-gas generators or premixed-gas cylinders
as the gas source. Once systematic measurement errors are iso-
lated at a given facility, they can be corrected or the mea-
surements can be adjusted to take them into account.
5.	Fluctuations in test-cell temperature and humidity, within the
ranges specified in the CVS-CH test method, and variations in
barometric pressure have been shown to significantly affect
measured exhaust emissions. Presently, no controls are placed
on humidity and ambient temperature 'is allowed to vary between
680F and 86°F during CVS-CH tests. Modification of the
CVS-CH-test procedure incorporating close control of both ambi-
ent temperature and humidity would decrease the statistical
variation in exhaust-emission measurements. Unfortunately,
barometric pressure cannot be conveniently controlled during
the course of CVS-CH tests. Since both systematic and random
variations of barometric pressure occur between test labora-
tories, appropriate correction factors relating variations, in
barometric pressure to exhaust emissions should be applied to
CVS-CH tests.
6.	Ambient temperature variations, commonly encountered in large
sections of the nation during winter (e.g., 0°F-32°F), can
significantly increase exhaust emissions of HC and CO, for
both pre-1975 production vehicles and 1975 prototype models
equipped with oxidation catalysts, above the emissions mea-
sured during the course of the CVS-CH test. As an example of
this effect, the HC and .CO emissions from four prototype
catalyst-equipped vehicles were shown to increase by, 160% and
and 4097o, respectively, when the vehicles were tested at 20 ° F

-------
136
in comparison to 75 °F. The NO^ emissions for these same vehi-
cles increased by approximately 33% over the same temperature
interval.
7.	Statistical variability associated with the evaluation of dete-
rioration factors, often obtained from tests on a single
durability-data car, may reflect the special circumstances of
the test rather than the capability of the underlying technol-
ogy. An averaging procedure over a larger group of similar
vehicles might be preferable. However, averaging, if used too
extensively, treats all vehicles and systems as equal, and
attractive technologies are counterbalanced by less desirable
technologies.
8.	The critical elements in any standardized fuel economy test
method and procedure include: a representative standard driv-
ing cycle or cycles; a method to determine the mass or volume
of fuel consumed and the distance traveled during the test;, and
a procedure for simulating vehicular loads encountered during
the course of the driving cycle.
9.	Significant vehicular miles are driven in both urban and non-
urban environments and fuel economy in these two driving modes
may differ by approximately 50%. Two fuel economies, one
designed to represent urban driving and one designed to repre-
sent highway driving, will provide the required basis for eval-
uating the relative fuel economies of various automotive
designs and comparing the relative fuel economies of one vehi-
cle versus another.
10. It has been shown that severe fuel economy penalties are
incurred when an LDMV is driven on a short trip that is initi-
ated from a cold-start condition. Since a significant number
of urban trips are very short and initiated from a cold start,
an urban fuel economy driving cycle should include a cold-
start phase.

-------
137
11.	The CVS-CH driving cycle is a sufficiently accurate represen-
tation of urban driving patterns to be used as the urban fuel
economy driving cycle.
12.	A fuel economy driving cycle similar to the EPA highway cycle
or a combination of the SAE suburban and interstate cycles may
be appropriate"for measurements of highway driving fuel econo-
mies.
13.	Either the Carbon Mass Balance Method or the direct measure-
ment of the mass or volume of fuel consumed can be employed to
obtain accurate measurements of fuel consumption if careful
experimental techniques and procedures are developed and fol-
lowed. Both of these test methods for fuel-consumption mea-
surements have disadvantages and advantages and are subject to
significant errors if careful experimental techniques and pro-
cedures are not followed.
14.	Two procedures have been proposed for simulating vehicular
loads encountered during the course of the driving cycle. One
method is based on driving the vehicle over a suitable road or
test track, and the second method is based on driving the cycle
on a chassis dynamometer. Both of these procedures have dis-
tinct advantages and disadvantages.
15.	There is no inherent technical reason for eliminating the use
of chassis dynamometers from fuel economy testing if it can be
shown that they adequately simulate the actual inertial, aero-
dynamic, rolling resistance and accessory loads experienced by
a vehicle as it traverses the driving cycle. Unfortunately,
the results of definitive studies designed to evaluate the
ability of chassis dynamometers to accurately simulate vehicle
loads under a>wide variety of conditions and to evaluate any
errors in fuel economy.measurements associated with specified
errors in dynamometer load simulation have not been reported.

-------
138
16.	Limited data presently available indicate that fuel economy
tests incorporating chassis dynamometers, with suitably fine
inertia-weight-controls, can be successfully employed to deter-
mine fuel- economies for urban driving cycles where inertia
forces and-idle conditions, rather than road loads,, dominate
LDMV fuel consumption.
17.	Limited data have indicated>that fuel economy data obtained on
chassis dynamometers and road or'track tests can be made;to
correlate wellveven for nonurban driving cycles, where road
loads dominate vehicle fuel consumption, if care is taken to
set the dynamometer loading so that vehicle coast-down times
obtained on the dynamometer closely match vehicle coast-down
times obtained on the road. However, the results of another
study indicate that fuel economies for the EPA highway driving
cycle as measured on a Clayton-type ,dyno, where the road load .
at 50 mph was set according to the "cook book" method estab-
lished in the Federal Register, did not correlate well with
results obtained on a test track. In all cases, „the fuel
economies measured on the dyno were larger than the corre-
sponding values measured ,on the track and differences between
the two values ranged from 4% to 16%. These results indicate
that certain vehicles may obtain significantly higher fuel
economies than aire actually , warranted on the EPA highway cycle
using present EPA fuel economy,test methods and procedures.
18.	Fuel economy measurements are considerably more reproducible
than' emission measurements.- Data indicate that fuel economy
measurements employing chassis dynamometers for an urban driv-
ing' cycle (e.g., CVS-H, CVS-C, and CVS-CH tests) are reason-
ably- irepeatable. In fact, the data indicate that the vari-
ability, of measured fuel economy in terms of the standard
deviation as a percent of the mean in a given cell, from cell
to cell at a given laboratory and from laboratory to laboratory

-------
139
is in the range of 2% to 8%. The limited data obtained for
reproducibility studies on the EPA highway cycle have indi-
cated a similar variability. However, due to the potential
significance that may be attached to reported fuel economies
of individual vehicles, reported fuel economies should be
based on the results of several tests.

-------
REFERENCES
1.	Federal Highway Admin., "Nationwide Personal Transportation Study,"
Report Nos. 1-9, U.S. Department of Transportation, April 1972 to
November 1973.
2.	Austin, T.C., and K.H. Hellman, "Passenger'Car Fuel Economy Trends
and Influencing Factors," SAE Paper No. 730790, September 1973.
1
3.	Teague, D.M; "Los Angeles Traffic Pattern Survey." Vehicle Emis-
sions . SAE Progress in Technology Series,,VI (New York, 1964)
pp. 17t38, '44.
4.	Haas,.G.C., M.,P. Sweeney, and J.N. Pattison "Laboratory Simulation
of Driving Conditions in the Los Angeles Area." Vehicle Emissions,
SAE -Progress in Technology Series, XII (New York, 1968) pp. 317-324.
5.	Kearm, D.H., and R.L. Lamoureux, "A survey of average driving pat-
terns in the Los Angeles urban area." TM-(L)-4119/000/01, Feb-
ruary 28, 1969.
6.	Kruse, R.E. and T.A. Huls, "Development of the federal urban driv-
ing cycle," SAE Paper No. 730553, May 1973.
7.	The Federal Register, XXXV, 136 (July 15, 1970).
8.	APRAC Project CAPE-10 Final Report, Coordinating Research Council,
New York.
9.	Lamureux, R.L., "Driving and vehicle use patterns in major metro-
politan cities - Phase I," APCA Paper No. 72-173, June 1972.
10.	Smith, M. and M.J. Manos, "Determination and evaluation of urban
vehicle operating patterns," APCA Paper No. 72-177, June 1972,.
11.	Smith, M., and D.M. Weston, "A technique for generating representa-
tive chassis dynamometer test cycles," APCA Paper No. 72-165,
June 1972.
12.	"Feasibility of Meeting the'1975-76 Exhaust Emission Standards in
Actual Use," Panel on Testing, Inspection and Maintenance for the
Committee on Motor Vehicle Emissions of the National Academy of
Sciences, June 1973.
13.	The Federal Register, XXXVIII, 84 (May 2, 1973).
14.	"New Motor Vehicles and Engines; Air Pollution Control - 1974 Model
Year Test Results," Federal Register. XXXIX, 40 (February 27, 1974),
7664.
140

-------
141
15.	Kruse, R.E. and C.D. Paulsell, "Development of a highway driving
cycle for fuel economy measurements," Environmental Protection
Agency, Emission Control Technology Division, Ann Arbor, Michigan,
March 1974.
16.	Paulsell, C.D., "Amendments to the report on development of a high-
way driving cycle for fuel economy measurements," Environmental
Protection Agency, Emission Control Technology Division, Ann Arbor,
Michigan, April, 1974.
17.	LaPointe, C., Ford Motor Company, private communication, June 1974.
18.	Society of Automotive Engineers Recommended Practive, "Fuel Economy
Measurement - Road Test Procedure," SAE J1082, (New York, April 1974).
19.	Cole, E.N., "Statement to Senate Public Works Committee by General
Motors Corporation, Attachment 7," November 5, 1973.
20.	Code of Federal Regulations, Title 40 Protection of Environment,
Office of the Federal Register, July 1, 1973, Revised.
21.	The Federal Register. XXXIX, 92, (May 10, 1974) 16904.
22.	Hellman, K., EPA Ann Arbor, private communication to R.A. Matula,
1974.
23.	Elder, C., General Motors Corporation, private communication to
R.A. Matula, 1974.
24.	General Motors Request for Suspension of the 1976 Emission Stan-
dards, Appendix 27, June 20, 1973.
25.	Report on Honda Emission-control Systems for 1975 and Subsequent
Model Years, Prepared for the Committee on Motor Vehicle Emissions
of the National Academy of Sciences, May 24, 1974.
26.	Heinen, C., Chrysler Corporation, private communication to R.A.
Matula, June 1974.
27.	Klingenberg, H., M. Fock, K. Lies, and L. Pazsltka, "A critical
study of the United States exhaust emission certification - test -
error analysis for the test procedure," APCA Paper No. 74-242,
Presented at the 67th Annual Meeting of the Air Pollution Control
Association, Denver, Colorado, June 1974.
28.	"The method of emission averaging as it relates to the 1975 federal
emission standards," Preliminary Report of AQC-Ad Hoc Statistical
Panel of MVMA, Feburary 26, 1973.

-------
142
29.	Ford Emission Cell Correlation Program. Vols. 1-3, (Ford Motor
Company 1974.)
30.	Hocker, A.J., "Effect of' cold soak temperature on emissions of AMA
cars," California Air Resources Board, Project No. M-257, August 22,
1973.
31.	"CVS cross 'reference service - 1973," Report No. 3, Scott Research
Laboratories, Inc., Plumsteadville, Pa., April 4, 1974.
32.	"Nitric oxide cross reference service - 1973," Report No. 3, Scott
Research Laboratories, Inc., Plumsteadville, Pa., April 18, 1974.
33.	Hum, R.W., Barlesville Energy Research Center, Bureau of Mines,
U.S. Department of the Interior, private communication June 27,
1974.
34.	Polak, J.C., "Cold weather emissions," Workshop on the Influence
of Cold Weather on Automobile Emissions, Ottawa, Canada, October
1973.
35.	Hromi, J.D., "Some aspects of determining new motor vetiicie emis-
sion levels," Presented at the Conference on Statistics and the
Environment, Washington, D.C., March 6-8, 1974.
36.	Heywood, J.B., "A preliminary assessment of automotive evaporative
hydrocarbon emissions," Department of Mechanical Engineering,
Massachusetts Institute of Technology, October 29, 1973.
37.	"Sealed Housing' for Evaporative Determinations (SHED) Technique,"
SAE Procedure J171, September 1970.
38.	Martens, S.W., and K.W. Thurston, "Measurement of total vehicle
evaporative emissions," SAE Paper No. 680125, January 1968.
39.	Martens, SJW., "Evaporative emission measurements with the SHED -
a second progress report," SAE Paper No. 690502, May 1969.
40.	Nelson, E.E., "Hydrocarbon-control for Los Angeles by Reducing
gasoline volatility," SAE Paper No. 690087, Vehicle Emissions.
Part III. (SAE," 1971), pp. 775-801.
41.	Kramer, R.L,, and N.P^ Cernansky, '"Motor vehicle emission rates,"
U.S. Dept." HEW, National Air Pollution Control Administration,
Office of Criteria and Standards Report, August 15, 1970.

-------
143
42.	Paulsell, C.D., "Mobile source evaporative emissions - a draft,"
EPA, Ann Arbor, Michigan, March, 1974.
43.	"Automobile exhaust emission surveillance, a summary," EPA Report
APTD-1544, prepared by Calspan Corp. under Contract No. 68-01-0435,
May, 1973.
44.	The Federal Register. XXXIX, 92, May 10, 1974, 16904.
45.	Ford Motor Company - Petition for Amendment of Part 85 of Title 40
of the Code of Federal Regulation Control of Air Pollution from
New Motor Vehicles and New Motor Vehicle Engines.
46.	Heubner, Jr., G.J., and D.J. Gasser, "Energy and the automobile -
general factors affecting vehicle fuel consumption," SAE Paper
No. 730518, 1973.
47.	Hutchins, F.P. (Project Officer) "Passenger car weight trend
analysis - volume II - technical discussion," EPA Report No.
460/3-73-006b, January 1974.
48.	Elliott, D.R., W.K. Klamp, and W.E. Kraemer, "Passenger tire
power consumption," SAE Paper No. 710575, June 1971.
49.	U.S. Department of Transportation, private communication to R.A.
Matula, June, 1974.
50.	Nolan, J., Ford Motor Company, private communication to R.A. Matula,
May 1974.
51.	SAAB-SCANIA, private communication to R.A. Matula, June, 1974.
52.	Baker, J.B., Memo to SAE Fuel Economy Measurement Procedures Task
Force, April 1, 1974.
53.	LaPointe, C., "Factors affecting vehicle fuel economy," SAE Paper
No. 730518, 1973.
54.	Scheffler, C.E., and G.W. Niepoth, "Customer fuel economy esti-
mated from engineering test," SAE Paper No. 650861, 1965.
55.	"A report on automobile fuel economy," U.S. EPA, Office of Air
and Water Programs - Office of Mobil Source Air Pollution Control,
October 1973.
56.	General Motors Request for Suspension of 1975 Federal Emissions
Standards, Appendix 27, March 5, 1973.

-------
144
57.	Bowditch, F.W., General Motors Corporation, private communication
to R.A. Matula, September 1974.
58.	EPA Methodology Report, Appendix B, Decision of the Administrator
on Remand from the United States Court of Appeals for the District
of Columbia, April 11, 1973.
59. Kirchhoff, W., National Bureau of Standards, private communica-
tion to R. A. Matula, June i974.

-------
APPENDIX A
Organizations Site-Visited or Interviewed
1.	Environmental Protection Agency	January 23, 1974
Emission Control Technology Division
Ann Arbor, Michigan
2.	Fuel Economy Measurement Procedures	February 6, 1974
Task Force
Society of Automotive Engineers
Romulus, Michigan
3.	Environmental and Safety Relations	February 12, 1974
Chrysler Corporation
Detroit, Michigan
4.	World Headquarters
Ford Motor Company
Dearborn, Michigan
5.	Environmental Activities Staff
General Motors Corporation
Warren, Michigan
6.	Transportation Systems Center
Department of Transportation
Cambridge, Massachusetts
7.	Volkswagen of America, Inc.
(meeting in Washington, D.C.)
8.	Environmental Protection Agency
Emission Control Technology Division
and
Surveillance Branch
Ann Arbor, Michigan
9.	Mobil Research and Development Corporation	May 8, 1974
Paulsboro, New Jersey
10.	California Air Resources Board	May 13, 1974
El Monte, California
11.	Clayton Manufacturing Company	May 13, 1974
El Monte, California
12.	Automotive Power Systems Evaluation Study	May 13, 1974
Jet Propulsion Laboratory
Pasadena, California
February	13, 1974
February	14, 1974
March 1,	1974
March 5,	1974
April 3,	1974
145

-------
146
13.	Olsen Laboratories, Inc.
Anaheim, California
14.	Emission Development
American Motors
Detroit, Michigan
15.	Vehicle Emissions Planning	June 5, 1974
Chrysler Corporation
Detroit , .Michigan
16.	World Headquarters	June 6, 1974
Ford Motor Company
Dearborn, 'Michigan
17.	Proving Ground	June 7, 1974
General Motors Corporation
Melford, Michigan
18.	Bartlesville Energy Research Center	June 11, 1974
Bureau of Mines
Bartlesville, Oklahoma
(Interview in Denver, Colorado)
19.	Department of the Environment	June 17, 1974
London, Eng1and
20.	Ministere-de l'Amenagemerit du Territoire	June 18, 1974
Paris, France
21.	Vereiriigung Deutscher'Automobilhersteller	June 19, 1974
Frankfurt, West Germany
(Interview in Geneva, Switzerland)
22.	Advance Engineering Department	June 24,' 1974
, Saab-Scania Aktiebolag
Trollhattan, Sweden
23.	Ministero dei Transport! e dell	June 25, 1974
Aviazione Civile
Roma,. Italia
24.	Environmental Activities Staff'	September 5, 1974
General Motors Corporation
Warren, Michigan
(Interview in Philadelphia, Pennsylvania)
May 14, 1974
June 5, 1974

-------
147
Additional information was obtained from the following Foreign
Manufacturers' presentations to CMVE meeting in Washington, DC,
May 21-24, 1974:
a.	Daimler-Benz AG
b.	Fiat, S.p.A./Ferrari
c.	Honda Motor Company
d.	Nissan Motor Company, Ltd.
e.	Adam Opel AG
f.	Peugeot, Inc.
g.	Regie Nationale des Usines Renault
h.	Saab-Scania Aktiebolag
i.	Toyo Kogyo Company, Ltd.
j.	Toyoto Motor Company, Ltd.
k.	Volkswagenwerk AG
1.	AB Volvo

-------
148
APPENDIX B
General Questions to Organizations Prior to Site Visit
from the Consultant on Testing
of the
Committee on Motor Vehicle Emissions
A.	Test Cycles
1.	Discussion of the advantages and disadvantages of the application
of the 1975 FTP as a basis for emissions standards testing and
for the determination of urban fuel economy.
2.	Summary of any studies that have been carried out to evaluate
typical driving patterns in urban, nonmetropolitan and rural
areas. Items of particular interest include distribution of
number of trips versus trip length, percentage of trips
initiated from "cold start" versus "hot start," average trip
speed, number of stops per mile, etc.
3.	Constructive evaluation of EPA's recently proposed nonmetropol-
itan fuel economy driving cycle.
B.	Test Methods and Procedures
1. Emissions
a.	Discussion of any special shortcomings or problems associated
with present emissions test procedures, special emphasis
should be given to CVS sampling system, instrumental
limitations for pollutant measurements, etc.
b.	Discussion of relative advantages and disadvantages of
modifying HC standards to incorporate only a non-methane
hydrocarbon standard.
c.	Summary of work to date on development of test methods and
procedures developed to measure l^S, SO and sulfates in
automotive exhaust.	x
d. Discussion of the accuracy of present gas calibration
standards used in emission testing and its effect on
accuracy of emission measurements.
e.
Discussion of the contribution of various factors to the
statistical variability of emission tests.

-------
149
2.	Fuel Economy
a.	Discussion of Carbon Mass Balance Methods versus Weighing
Methods to determine fuel economy.
b.	Discussion of chassis dynamometer versus road/track test
procedures for the determination of fuel economies.
c.	Critical evaluation of the use of chassis dynamometer
facilities to simulate road loads for fuel economy tests.
Items of particular interest include the dynamometer'.s
ability to accurately.simulate rolling resistance-and
aerodynamic drag. ' In addition, problems associated with the
effect of cooling fan characteristics on vehicle warm-up,
etc. should be discussed.
d.	Discussion of the contribution of various factors to the
statistical variability of.fuel economy tests.
e.	Discussion of the relative importance of rolling resistance
and aerodynamic drag as they apply to vehicle road load as a
function of vehicle speed.
3.	Vehicle Driveability
a.	Discussion of qualitative test methods to determine vehicle
driveability.
b.	Discussion of quantitative test methods to determine vehicle
driveability.
C. Experimental Results
1. Emissions
a.	Data on any inner-lab and intra-lab reproducibility and/or
correlation testing programs on vehicular emissions.
b.	Data on statistical variability of emissions from a single
vehicle undergoing many tests and/or from a group of similar
production model vehicles.
c.	Data detailing the effect of ambient conditions (special
consideration to temperature) on emissions.
d.	Data summarizing H S, SO and sulfate emissions from catalyst
equipped vehicles.	x

-------
150
e. Data on HC evaporative emissions as obtained in SHED tests.
Comparison of total HC evaporative emissions as obtained in
SHED tests versus "cannister" tests.
2.	Fuel Economy
a.	Data on any inner-lab and intra-lab reproducibility and/or
correlation testing programs on vehicular fuel economy.
b.	Data on statistical variability of fuel economy from a
single vehicle undergoing many tests and/or from a group of
similar production model vehicles.
c.	Data detailing the effect of ambient conditions (special
consideration to temperature) on fuel economy.
d.	Data on fuel economy versus trip length and any data
relating to the relative fuel economies of a specified trip
when the trip is initiated from a "cold start" versus a
"hot start."
e.	Data on fuel economy penalties associated with vehicle
accessories, transmission type, etc.
3.	General Considerations
a.	Data on road load on vehicles versus vehicle speed.
i. Rolling resistance versus speed
ii. Aerodynamic drag versus speed
b.	Data on relative importance of inertia forces and road
load as a function of vehicle speed and acceleration.

-------
GLOSSARY
A/F		air/fuel
AMA	Automobile Manufacturers Association
AMC	American Motors Corporation
ASIA	Automotive Service Industry Association'
AT	 automatic transmission
CCS		direct-fuel-i'njected, stratified-charge engine such as
the Ford PROCO or Texaco TCCS
CID	 cubic inch displacement
CMVE	Committee on Motor Vehicle Emissions
CNG	 compressed natural gas
CR	 compression (or expansion) ratio
CVCC	 prechamber, dual-carburetor stratified-charge engine such
as the Honda Motor Company Compound Vortex-Controlled
Combustion Engine
CVS	 constant volume sample emissions test procedure
CVS-CH	 constant volume sample--cold and hot start emissions test
procedure
Cyl		cylinder
DF		deterioration factor
Dual Cat....	a reducing catalyst followed by an oxidizing catalyst
EFI		electronic fuel injection
EGR		exhaust gas recirculation
EPA		U.S. Environmental Protection Agency
FI		fuel injection
FTP		Federal Test Procedure
GM		General Motors Corporation
g/mi		grams/mile of emissions
GT		gas turbine
HC		various hydrocarbon compounds
HCAT		oxidizing catalyst system for removal of HC and CO
ICE		internal combustion engine.
IFC		integrated fuel control (carburetor)
JPL	 Jet Propulsion Laboratory, California Institute of
Technology
151

-------
152
LBS		lean-burn engine system
LDMV		light-duty motor vehicle
LPG		liquefied petroleum gas
M/C		Manufacturing and Costs Panel of CMVE
Modif		standard internal-combustion engine modified with emissions
controls such as spark retard and air pump
MON		motor octane number
mpg		miles per gallon fuel economy
MT		manual transmission
MVMA	Motor Vehicles Manufacturers Association
NIASE		National Institute for Automotive Service Excellence
NO^		various nitrogen oxide compounds
P		production
PCV		positive crankcase ventilation
pp...,		pre-production
PROCO		Ford--Programmed Combustion Process (direct-fuel-injected,
spark-ignited, open-chamber, stratified-charge engine)
R		research
RE		rotary engine
RON	research octane number
SAE		Society of Automotive Engineers
SHED		sealed housing for evaporative (emissions) determinations
std. dev....	standard deviation
TCCS	Texaco controlled combustion system (direct-fuel-injected,
spark-ignited, open-chamber, stratified-charge engine)
3-way cat...	combined oxidizing and reducing catalyst
HDDS			urban dynamometer driving cycle
VMT	vehicle miles traveled

-------