EPA-AA-TEB-EF-90-4
Technical Report
Fuel Volatility Effects
on Exhaust Emissions
By
Celia Shih
August 1990
NOTICE
Technical Reports do not necessarily represent final EPA
decisions or position. They are intended to present technical
analysis of issues using data which are currently available.
The purpose in the release of such reports is to facilitate the
exchange of technical information and to inform the. public of
technical developments which may form the basis for a final EPA
decision, position or regulatory action.
Test and Evaluation Branch
Emission Control Technology Division
Office of Mobile Sources
Office of Air and Radiation
U.S. Environmental Protection Agency
-------
1.0 BACKGROUND
In MOBILE4, the effects of fuel volatility, expressed as
Reid Vapor Pressure (RVP) in pounds per square inch (psi), on
vehicles' exhaust emissions were modeled using data from EPA's
Emission Factor Programs (EFP) conducted in Ann Arbor and test
programs performed by Automotive Testing Laboratories, Inc.
(ATL) at East Liberty, Ohio. The results of these test
programs were summarized in two reports [1,2]. The American
Petroleum Institute (API) submitted some of their test data
(see discussions in Data Sources section below) before the
release of MOBILE4. These data were reviewed and found not to
be contradictory to the MOBILE4's assumptions. The algorithm
used in MOBILE4 has been documented [3], and is summarized in
the following paragraphs:
(1) When ambient temperatures are less than 41°F, it is
assumed that there is no RVP effect on vehicles' exhaust
emissions. This assumption is applied to all model years and
all gasoline-powered vehicle types.
(2) There were no data to characterize the RVP effects
on exhaust emissions for fuel volatilities less than the
certification level of 9.0 psi, especially with ambient
temperatures below 75°F. The MOBILE4 model has assumed that
the effects of low RVP fuels on exhaust emissions were the same
as those of 9.0 psi fuel (i.e., no effect).
(3) Vehicles tested at ambient temperatures of 75°F and
up have shown exhaust emissions increases when the fuel
volatility is higher than the certification level of 9.0 psi.
Therefore, various RVP adjustment factors are used for
different vehicle types and model year groups. For example,
adjustments are made to exhaust HC and CO composite emissions
for 1970-79 light-duty gasoline-powered vehicles (LDGVs),
1970-80 LDGTls, and 1979-80 LDGT2s when the ambient temperature
is higher than 40°F. A different set of correction factors is
used to adjust exhaust HC and CO composite emissions for 1980+
LDGVs, 1981+ LDGT1S/LDGT2S, and 1985+ HDGVs at ambient
temperatures between 41°F and 75°F. For ambient temperatures
higher than 75°F, and model years 1980+ LDGVs, 1981+
LDGTls/LDGT2s, and 1985+ HDGVs, a combined temperature and fuel
volatility correction factor is used for each bag of the FTP
exhaust emissions.
The purpose of this report is to assemble all data
available as of May 1, 1990 to re-evaluate these MOBILE4
assumptions, especially in the areas of low RVP fuels (less
than 9.0 psi) at temperatures less than 75°F.
-1-
-------
2.0 DATA SOURCES
2.1 Pre-1980 Vehicles
(1) In 1975, Chevron Research [4] tested a total of
seven vehicles (model years 1973 to 1976) at two fuel
volatility levels and at two ambient temperatures using a
randomized test sequence. The two fuels were the equivalent of
8.5 and 6.5 psi RVP*. The two ambient temperatures were 75 and
55°F, with relative humidities of 42 and 73 percent,
respectively. One of the test vehicles was tested twice, once
with the original carburetor and once with a replacement
carburetor. All other vehicles were tested in "as-received"
condition after their ignition and emission control systems
were checked and failed components repaired. Out of 32 tests
("8" vehicles x 2 fuel RVPs x 2 temperatures), 20 were run in
replicates with the average emissions listed in their report.
Descriptions of the test vehicles are included in Table Al of
the Appendix to this report. Note that they were all
California vehicles.
(2) In 1979, Exxon Research and Engineering Company [5]
tested a total of eight 1974 through 1977 model year vehicles
at six levels of fuel RVP and at the FTP ambient temperature of
75°F. The test fuels used were 11.8, 9.4, 9.1, 8.8, 6.8, and
6.6 psi RVP. There were two 9.1 psi fuels used in the
program: one being a "high octane Indolene" used for
certification, and one being a 50-50 blend of 6.5 and 11.8 psi
fuels, with many more tests on the latter fuel than the
former. At each fuel volatility level, most of the vehicles
were tested two or three times. Among the test fleet, three of
them were California vehicles, and two of them had their
engines overhauled previously. Descriptions of the test
vehicles are given in Table A2 of the Appendix.
(3) In 1988, the California Air Resources Board (GARB)
included one model year 1978 vehicle in their five-car
evaluation project [6] with two fuel volatilities at FTP
ambient temperature of 75°F. The two test fuels were 8.6 and
7.2 psi RVP. Each vehicle was tested twice at each fuel
volatility level. Description of this test vehicle is included
in Table A3 of the Appendix.
* With 11.8 and 8.3 Front End Volatility Index (FEVI) rating
volatilities, respectively. This FEVI rating volatility
is defined as [RVP + 0.13 * (% evaporated at 158°F)].
—2 —
-------
2.2. Post-1980 Vehicles
(1) In 1988, Chevron Research conducted two test
programs in an effort to evaluate the effects of reduced RVP on
exhaust emissions [7,8]. The first portion of the program
involved a total of fourteen vehicles tested during two phases,
each comparing exhaust emissions with two different RVP fuels
[4]. In Phase I, 11.4 and 8.1 psi fuels were used with an
ambient temperature of 43°F. In Phase II, 8.4 and 6.1 psi
fuels were used with an ambient temperature of 55°F.
Descriptions of the test vehicles are included in Table Bl of
the Appendix. Note that all were California vehicles.
(2) Four of the test vehicles in the 1988 GARB five-car
evaluation project (discussed in A.3 above) were 1981+ model
year vehicles, as described in Table A3 of the Appendix.
(3) In 1987, ATL, under the sponsorship of API, tested
six vehicles at four fuel volatility levels and at three
ambient temperatures [9] using a randomized test sequence. The
four fuels were 10.5, 9.0, 8.0, and 6.5 psi in fuel RVP. The
ambient temperatures were 42, 55, and 80°F. Out of 72 tests (6
vehicles x 4 fuels x 3 temperatures), all were run twice, with
4 tests run three times. Descriptions of these six vehicles
are included in Table B2 of the Appendix.
(4) In addition to those EFP data included in MOBILE4
analyses [1,2,3], EPA has since tested ten model year 1987
fuel-injected vehicles at ambient temperature of 50°F and at
three fuel volatility levels: 14.6, 11.7, and 9.0 psi RVP.
Descriptions of these ten vehicles are included in Table B3 of
the Appendix.
-3-
-------
3.0 DATA ANALYSIS
The following criteria were used in analyzing the data:
(1) Each test program was analyzed separately.
(2) Within each test program, data from each ambient
temperature level were examined.
(3) Since the sample sizes were relatively small (at
each temperature level and for each test program), a
statistical significance level of 0.10 was used.
(4) One of the test vehicles from the 1975 Chevron
program (1973 Pontiac Catalina) was treated as if it were two
vehicles, since it was tested with the original carburetor and
tested again with a replacement carburetor (see A.I, section
2.0) .
(5) Between the two 9.1 psi fuels used in the Exxon
program (as discussed in A.2, section 2.0), emissions from the
"high octane Indolene" were used in this analysis, since this
is the type of fuel used for certification.
(6) The model year 1978 vehicle tested by GARB was
separated from the other four 1983+ vehicles in the analysis,
since it was built to comply with a different evaporative
emission standard.
(7) One of the test vehicles from the 1988 Chevron
program (1983 Toyota SR-5 pickup) was excluded from the current
analysis, as the vehicle was of a different type (LDGT1).
Three types of statistical procedure were used to examine
the impact of fuel volatility on exhaust emissions: paired-t
test, student-t test, and regression analysis. Descriptions of
the statistics and results are summarized in the following
sections.
-4-
-------
3.1 Paired-t Test
The paired-t test is a statistical procedure used to
examine the effects (exhaust emissions) of a certain treatment
(fuel volatility) which has been applied to the same sample
population (test vehicle) under similar conditions (i.e., at
the same ambient temperature, FTP speed cycle, etc.). This is
a more powerful statistical tool than the commonly used
student-t test to be discussed later. In general, the
statistic used in paired-t test has the form [10]:
T = D / s(D)
where: D = mean of the differences Dt/ for i = l,2,...,n,
s(D) = standard deviation of the differences.
This T statistic is distributed as a t-distribution with (n-1)
degrees of freedom. The hypothesis is that the mean of the
differences (D) equals zero. This hypothesis is to be rejected
if the absolute value of T is greater than a critical value
estimated from the t-distribution, based on a significance
level of 0.10 and sample size n.
For each pollutant, Dt was calculated as the difference
between emissions from a high RVP fuel and those from a low RVP
fuel on each vehicle at a given temperature:
DI = Edigh RVP ~~ ELOW RVP
If replicate tests on a given vehicle were available, the
average of the emissions measured on each test were used.
Means and standard deviations of the differences (Dt) at each
temperature level from each test program were calculated and
summarized. Results from pre-1980 vehicles are presented in
Table 1, and results from 1980+ vehicles are shown in Table 2.
Also listed in Tables 1 and 2 are the probabilities
calculated from the t-distribution. Probabilities designate
the numerical value associated with the chances that D being
different from zero are caused by random error. Therefore, if
the probability value is less than or equal to 0.10, this
implies that D is significantly different from zero, or that
the difference between emissions from a high and from a low RVP
fuel is statistically significant.
The average percent emission change is calculated from the
percent emission change for each vehicle:
Di / EHigh RVP * 100.0
-5-
-------
Therefore, a negative average percent emission change would
indicate that on the average the emissions are increased by
using a lower RVP fuel.
In evaluating paired-t test results, it should be noted
that the effect of fuel volatility on exhaust emissions is
significant if and only if there is consistency among all
statistics being evaluated.
For example, a determination that a lower RVP fuel (say
6.5 psi compared to 8.0 psi) would increase exhaust HC
emissions at 55°F, results from pairs of test results at these
two fuel volatility levels from the API/ATL program, where:
(1) the means (D's) of the exhaust HC emission
differences are negative,
(2) the probabilities that D's being equal to zero are
less than or equal to the chosen significance level,
(3) the average percent emission change values are
negative (consistent with the mean emission differences), and,
(4) there is a consistent trend in the occurrences of
the above three criteria at similar temperature levels and
across the test programs. For example, if the exhaust HC
emission differences for 8.4 vs. 6.1 psi RVP fuels were
significant at 55°F from the Chevron test program, it is
logical to expect that the emission differences for similar
volatility levels (8.0 vs. 6.5 psi RVP fuels) should also be
significant at similar temperatures (55°F) from the API/ATL
test program.
-6-
-------
3.1.1. Pre-1980 Vehicles
On examining test results from pre-1980 vehicles, the 1978
vehicle tested by GARB at 75°F showed very large percent
emission increases from using a lower RVP fuel. Since this was
the only pre-1980 vehicle tested by GARB, and the only 8.6 vs.
7.2 psi RVP emission levels, no statistical comparison can be
made, and thus no conclusion can be drawn.
For other pre-1980 vehicles, as can be seen from Table 1,
the majority of the paired-t significance levels are greater
than 0.10, with the following exceptions:
Test
Program
Chevron
Exxon
Exxon
Exxon
op
55
Fuel RVPs
(psi)
11.8/8.3
75 11.8/6.6
75 8.8/6.6
75 6.8/6.6
Pollutant(s) that showed
Significant Differences
HC,CO
NOx
HC,NOx
NOx
This implies that, except for the above cases, there is no
significant difference in emissions when different volatility
fuels are used.
At each temperature level, the average percent emission
changes, either increasing or decreasing by using a lower RVP
fuels, are also examined. The majority of the emission changes
are small (say, within 10 percent), with the following
exceptions:
Test
Program
Chevron
Exxon
Exxon
Exxon
Exxon
Exxon
Exxon
Exxon
55
75
75
75
75
75
75
75
Fuel RVPs
(psi)
11.8/8.3
11.8/6.8
11.8/6.6
9.4/9.1
9.4/6.8
9.4/6.6
9.1/8.8
8.8/6.6
Pollutant(s) that showed
10% or more Emission Change
Increase Decrease
HC,CO
NOx
HC
CO
CO
CO
CO
CO
NOx
Note that the percent emission changes from the Chevron program
at 55°F for both HC and CO (11.8 vs. 8.3 psi RVP) and the Exxon
program at 75°F (8.8 vs. 6.6 psi RVP) for HC and NOx were also
statistically significant.
-7-
-------
However, there is no consistency in the signs of the mean
emission differences. For example, at 75°F, mean emission
differences from the Chevron program at 11.8 vs. 8.3 psi RVP
for both HC and CO are negative (i.e., the HC and CO emissions
from the lower RVP fuel were higher than those from the higher
RVP fuel). But from the Exxon program at the same temperature
level and similar fuel volatility levels (11.8 vs. 8.8 psi),
the mean HC and CO differences are positive (i.e., the HC and
CO emissions from the lower RVP fuel were lower than emissions
from the higher RVP fuel).
Therefore, it is concluded that, based on the current data
from pre-1980 vehicles, the fuel volatility effect on vehicle's
exhaust emissions at 75°F may not be significant and
consistent. The Chevron test program data at 55°F (11.8 vs.
8.3 psi) showed both significant and consistent fuel volatility
effect on exhaust HC and CO emissions.
3.1.2 1980+ Vehicles
For 1980 and later model year vehicles, as can be seen
from Table 2, the majority of the paired-t probabilities are
also greater than 0.10, with the following exceptions:
Test Fuel RVPs Pollutant(s) that showed
Program °F (psi) Significant Differences
API/ATL 42 9.0/8.0 NOx
EPA 50 14.6/11.7 NOx
EPA 50 14.6/9.0 NOx
Chevron 55 8.4/6.1 HC,CO,NOx
API/ATL 55 8.0/6.5 HC
API/ATL 80 10.5/9.0 HC
API/ATL 80 10.5/8.0 HC
At each temperature level, the average percent emission
changes, either increasing or decreasing by using a lower RVP
fuels, are also examined. The majority of the emission changes
are within 10 percent, with the following exceptions:
-8-
-------
Pollutant(s) that showed
Test Fuel RVPs 10% or more Emission Change
Program ^F (psi) Increase Decrease
EPA 50 14.6/11.7 CO
EPA 50 14.6/9.0 - NOx
Chevron 55 8.4/6.1 HC
API/ATL 80 10.5/9.0 - CO
API/ATL 80 10.5/8.0 - HC,CO
API/ATL 80 10.5/6.5 - CO
At the 42/43°F temperature level, only NOx emissions from
the API/ATL test program for 9.0 vs. 8.0 psi RVP fuels showed a
statistically significant mean emission decrease. The average
percent emission change values are all small (within ten
percent increase or decrease). Therefore, it is concluded that
at 42/43°F temperature level, there is no conclusive evidence
of RVP effect on exhaust emissions.
At 50/55°F, only the 1988 Chevron program data (8.4 vs.
6.1 psi RVP fuels) showed significant mean emission differences
(increases) for all three pollutants, and close to an eleven
percent HC emission increase. For similar fuel volatility
levels (at 8.0 vs. 6.5 psi fuels), the HC emission increases
from API/ATL program was also significant. Therefore, by the
consistency criterion, it is concluded that the fuel RVP effect
on HC emissions at 50/55°F may be significant. There is no
significant (and no consistent) RVP effect on CO or NOx
emissions.
At the 75-80°F temperature range, the probability
statistics showed significance for HC emissions reduction by
reducing fuel volatilities from 10.5 to 9.0, and 10.5 to 8.0
psi RVP. However, the significance is lessened when lower than
9.0 psi fuel is used. Also, the emission increase on CO
emissions when 10.5 psi RVP fuel is used in the API/ATL program
is ten or more percent, although their associated probabilities
are not significant. The API/ATL test results are consistent
with the MOBILE4 assumption — that there is a small fuel
volatility effect on exhaust HC/CO emissions when fuel
volatilities are higher than the certification level of 9.0 psi
and ambient temperatures are 75°F and higher.
-9-
-------
3.2 Student-t Test
Since some of the test programs (such as Chevron and
API/ATL) used randomized test sequences, the paired-t test
procedure may not be appropriate. For this reason, the
student-t test procedure was also used for evaluation. The
student-t test is a statistical procedure used to compare two
sample means and their variances. Therefore, at each
temperature level from each test program, the average exhaust
emissions from a high volatility fuel can be compared to the
average emissions from a low volatility fuel, if the variances
of the two samples are statistically equal [10]:
T = (E1 - E2) / SD
where: Ei = average emissions from a high volatility fuel,
E2 = average emissions from a low volatility fuel,
SD = standard deviation.
The T statistic is distributed as a t-distribution with
(nl+n2-2) degrees of freedom. The hypothesis is that the two
sample means are equal (or, Ei - E2 = 0). This hypothesis
is to be rejected if the absolute value of T is greater than a
critical value estimated from the t-distribution, based on a
significance level of 0.10 and the two sample sizes (nl and n2).
The two sample emission variances can be compared based on:
F = SI2 / S22
where:
SI2 = emission variance from a high volatility fuel, and
S22 = emission variance from a low volatility fuel.
This F statistic is distributed as a f-distribution with (nl-1)
and (n2-l) degrees of freedom. The hypothesis is that the
ratio of the two variances equals unity. This hypothesis is to
be rejected if the numerical value of F is greater than a
critical value estimated from the f-distribution, based on a
significance level of 0.10 and the two sample sizes (nl and n2).
For each pollutant, means and standard deviations were
calculated and summarized at each temperature level from each
test program. Results from pre-1980 vehicles are presented in
Table 3, and results from 1980+ vehicles are shown in Table 4.
-10-
-------
Also listed in Tables 3 and 4 are three probabilities:
(1) The probability that the two average emissions being
different from zero is caused by random error — denoted as
"t-probability."
(2) The probability that the ratio of the two emission
variances being different from unity is caused by random error
— denoted as "f-probability."
(3) The probability that the average emissions from a
high RVP fuel are lower than the average emissions from a low
RVP fuel. Note that this probability has the minimum value of
0.0 and the maximum value of 1.0, and is not calculated when
the sample size is less than 5.
In evaluating student-t test results, criteria similar to
those used for evaluating the paired-t test are also used.
These criteria include: average emissions being significantly
different (i.e., the t-probability value needs to be less than
or equal to 0.10), and a consistent trend in the occurrences of
significance at each temperature level and across the test
programs.
An additional criterion is used in evaluating student-t
test results. The numerical values associated with the
probability that the mean emissions from a high RVP fuel being
lower than the mean emissions from a low RVP fuel should be
closer to the two extremes, either larger than 0.75 or smaller
than 0.25. If'this probability value is 0.50, this would imply
that there is a 50-50 chance that the average emissions from
one RVP level fuel will be lower (or higher) than the average
emissions from another RVP level fuel, hence the fuel
volatility effect on emissions is not significant. A note of
caution is that even if the probability values are close to the
two extremes, most of the test programs examined here had
relatively small sample sizes (ranging from 4 to 13).
3.2.1 Pre-1980 Vehicles
For pre-1980 vehicles, as can be seen from Table 3, all t-
and f-probabilities are greater than 0.10, with the majority of
the t-probabilities greater than 0.90. The probabilities that
the high RVP fuel emissions being lower than the low RVP fuel
emissions are mostly between 0.30 and 0.70 (with only one
exception: from Exxon program at 75°F, the probability that
the average NOx emissions from 8.8 psi fuel being lower than
the emissions from 6.6 psi fuel is 0.27). Therefore, it is
concluded that there is no statistically significant fuel RVP
effect on exhaust emissions at any temperatures.
-11-
-------
3.2.2 1980+ Vehicles
For 1980+ vehicles, as can be seen from Table 4, all
t-probabilities at each temperature level are greater than
0.10. The majority of the f-probabilities are also greater
than 0.10, with the following exceptions:
Pollutant(s) that showed
Test Fuel RVPs Significant Differences
Program ^F (psi) in Emission Variances
EPA 50 14.6/11.7 HC,CO
EPA 50 11.7/9.0 HC,CO
API/ATL 80 10.5/8.0 CO
API/ATL 80 10.5/6.5 CO
API/ATL 80 9.0/8.0 CO
Note that the significant differences in emission variances are
simply an indication that the degrees of dispersion within one
sample are different from those in another sample.
The probabilities that the high RVP fuel emissions are
lower than the low RVP fuel emissions are also mostly between
0.30 and 0.70, with the following exceptions:
Probability
Test Fuel RVPs that 1st Mean
Program ^F (psi) Pollutant is < 2nd Mean
EPA 50 14.6/11.7 NOx 0.21
EPA 50 14.6/9.0 NOx 0.10
EPA 50 11.7/9.0 HC 0.29
EPA 50 11.7/9.0 CO 0.26
EPA 50 11.7/9.0 NOX 0.29
API/ATL 80 10.5/8.0 HC 0.23
API/ATL 80 10.5/8.0 CO 0.16
API/ATL 80 10.5/6.5 CO 0.19
API/ATL 80 9.0/8.0 CO 0.29
Note that all the above exceptions lead to the conclusion
that the emissions from a high RVP fuel are more likely to be
higher than the emissions from a low RVP fuel.
All these results suggest that the average exhaust
emissions (HC, CO, and NOx) from a high RVP fuel are not
statistically different than the average emissions from a lower
RVP fuel.
-12-
-------
4.0 REGRESSION ANALYSIS
Regression analysis is a statistical tool used to
determine the relation between a set of independent variables
(fuel volatility and ambient temperature) and a dependent
variable (exhaust emissions) so that the latter can be
estimated from the former. Since the main purpose of this
report is to examine the fuel volatility effect on exhaust
emissions, the regression analysis here serves as a final step
to see if there exists any effect on exhaust emissions under a
combination of factors such as fuel volatility, ambient
temperature, and the interaction of the two.
The results presented here are in a simplified form.
Basically, the regression model has the following form:
E = exp [ a + b*RVP + c*Temp + d*RVP*Temp + e*RVP2
+ f*Temp2 + g*(RVP*Temp)z ]
where: E = exhaust emissions in g/mi,
RVP = fuel volatility in psi, and,
Temp = ambient temperature in °F.
A backward selection process is used to determine the final
equation form. That is, a full set of all independent
variables, whenever appropriate, is used in the first step of
regression analysis. The resulting coefficients are checked
for significance. The most insignificant variable (with the
coefficient having the largest numerical value in significance
and greater than 0.10) is to be removed before the execution of
the next step of regression analysis. This selection process
will continue until all the variables remaining have
coefficients with significance levels less than or equal to
0.10. Therefore, at the end of the final step, the remaining
independent variables are those that would contribute
significantly in the prediction of the dependent variable.
-13-
-------
4.1 Pre-1980 Vehicles
The 1975 Chevron and the 1979 Exxon programs data were
used in the regression analysis. Note that there were two
temperature levels (55 and 75°F) and two fuel volatility levels
(8.3 and 11.8 psi RVP) in the 1975 Chevron test program.
Therefore, only the constant term (a in the above equation) and
three first degree independent variables (RVP, Temp, and
RVP*Temp) were used in the initial step of the backward
selection process. The 1979 Exxon program data were at 75°F
only (but with fuel volatilities ranging from 6.6 to 11.8 psi
RVP), thus only the constant term and two RVP related variables
(RVP, RVP2) were used in the regression analysis. Results
are summarized in the following:
Test Program
1975 Chevron
1979 Exxon
Pollutant Significant Variables
HC
CO
NOx
HC
CO
NOx
Constant
Temp
Constant
Constant
Constant
Constant
As can be seen from the above, for pre-1980 vehicles fuel
volatility is a non-significant independent variable for all
pollutants across both test programs.
4.2 1980+ Vehicles
Data from two test programs were used in the regression
analysis: the 1988 Chevron and the 1987 API/ATL programs.
Note that there were two temperature levels (43 and 55°F) and
two fuel volatility levels (8.1 and 11.4 psi RVP) in the 1988
Chevron test program. Therefore, only the constant term
(denoted as "a" in the above equation) and three first-degree
independent variables (RVP, Temp, and RVP*Temp) were used in
the initial step of the backward selection process. The 1987
API/ATL program had three levels of temperature (42, 55, and
80°F) and four levels of fuel volatility (6.5, 8.0, 9.0, and
10.5 psi RVP), thus the entire seven independent variables
(defined previously) were used in the regression analysis.
Results are summarized in the following:
-14-
-------
Test Program Pollutant Significant Variables
1988 Chevron HC RVP, RVP*Temp
CO Temp
NOx Constant
1987 API/ATL HC RVP, RVP*Temp
CO Temp2
NOx Temp2
As can be seen, for 1980+ vehicles fuel volatility is_ a
significant independent variable for HC emissions across both
test programs.
-15-
-------
5.0 CONCLUSIONS
The effect of fuel volatility on exhaust emissions based
on data not included in the MOBILE4 model was examined. Three
types of statistical procedure were used to examine the impact
of fuel volatility on exhaust emissions: paired-t test,
student-t test, and regression analysis. Results are
summarized in the following:
Statistical
Procedure
Paired-t
Student-t
Regression
Paired-t
Student-t
Regression
Results
Pre-1980 Vehicles
a) The 1975 Chevron program data at 55°F showed
significant fuel volatility effect (11.8 vs.
8.3 psi fuel) on HC/CO emissions, resulting in
higher emissions when low RVP fuel was used.
b) For all other test programs at 75°F the RVP
effects were non-significant.
No statistically significant fuel RVP effect on
exhaust emissions at any temperatures.
Fuel volatility is not a significant independent
variable for all pollutants.
1980-t- Vehicles
a) At 42/43°F,
non-significant.
the fuel RVP effect
was
b) At 50/55°F, both Chevron and API/ATL data
showed significant fuel RVP effect (8 vs. 6 psi
RVP) resulting higher HC emissions under lower
RVP fuel.
c) At 75/80°F, API/ATL data showed significant
HC/CO reductions with lower RVP fuel.
Some showed significant differences in the
variances. No significant differences on the
average emissions. Overall, the emissions from
one level RVP fuel are likely to be the same as
the emissions from another level of RVP fuel.
For Chevron and API/ATL programs, fuel RVP is a
significant independent variable for HC
emissions.
-16-
-------
The algorithm used in MOBILE4 was based on the use of the
more powerful paired-t test procedure, since in EPA's EF
testing program vehicles were tested with the higher in-use RVP
fuel first. Both Chevron and API/ATL program data were based
upon randomized test sequence results. The conclusions from
the student-t test would be more applicable for those studies.
Even if the paired-t test results were examined for Chevron and
API/ATL program data, the majority of the data had shown no
fuel volatility impact on exhaust emissions, with significant
fuel RVP impact only at 50/55°F based on a relatively small
sample size.
The pre-1980 vehicles were equipped with oxidation
catalysts and the cold start was controlled mostly by a choke.
It is reasonable to believe that the cold start portion of the
FTP test might result in slightly higher exhaust emissions by
using a lower RVP fuel, especially when the ambient temperature
is relatively low and when vehicles may not be in a "top-notch"
condition.
For 1980+ vehicles, however, higher exhaust HC emissions
occurred only when 8.0 and 6.0 psi fuels were compared and the
ambient temperature was at 50/55°F. The sample sizes of these
two test programs (Chevron and API/ATL) that showed significant
fuel RVP impacts were relatively small (with N=13 and 6,
respectively). Further, data from recent fuel surveys [11]
showed that only very few brands of gasoline with fuel
volatility lower than 8.0 psi were commercially available (in
cities like Albuquerque, Las Vegas, Los Angeles, and Phoenix,
during the summer months). The lowest fuel volatility surveyed
was 7.5 psi RVP from Phoenix in the summer of 1988, for an
unleaded premium gasoline.
As suggested by the data, using fuels with volatilities at
6.0 or 6.5 psi RVP could result in a small exhaust HC emission
increase at 50/55°F ambient. But the benefit of using a higher
psi fuel to offset this small exhaust HC emissions increase for
the early morning trip even in the high temperature ozone
season is probably cancelled out by the higher evaporative hot
soak HC emissions generated in the midday when the majority of
daily trips occur and the ambient temperatures can be over 90
degrees Fahrenheit.
To summarize, in analyzing the currently available data,
there were inconsistencies in results. Under the unlikely
combination of relatively low temperature and less than 8.0 psi
RVP fuel volatilities, a small increase in exhaust HC emissions
is negligible when compared with the benefit of a relatively
larger decrease in evaporative hot soak emissions. For these
reasons, the MOBILE4 model assumed no fuel volatility effect on
exhaust emissions when the fuel RVP is below the certification
level of 9.0 psi.
-17-
-------
This assumption is adequate for most of the areas with
moderate ambient temperature profiles in high ozone season
(with daily minimum of 60 to the maximum of 84°F) . The
combination of lower (say, 50/55°F) ambient temperature and
lower (6.0) psi RVP fuel is expected to be rare. Even in areas
with more extreme ambient temperature profiles (say, 70 to
96°F), the suggested fuel volatility required by 1992 is
7.8 psi RVP in EPA's Final Rule for Phase II Fuel Volatility
Control.
It is concluded that the assumptions used in MOBILE4 are
adequate for all situations likely to be encountered in real
world modeling. There may be uncertainties when estimating
exhaust emissions in areas with low RVP fuels (less than 9.0
psi) at temperatures greater than 40°F. More data should be
collected and analyzed before these uncertainties can be
addressed.
-18-
-------
Table 1
Paired-t Test Results
Pre-1980 Vehicles
Amb
Temp.
(°F)
55
75
75
75
75
75
75
Fuel
RVP
(psi)
11.8
vs.
8.3
8.6
vs.
7.2
11.8
vs.
8.3
11.8
vs.
9.4
11.8
vs.
9.1
11.8
vs.
8.8
11.8
vs.
6.8
N
Mean
Emission
Difference
r Poll (g/mi)
Probabilities
that the
Differences
Standard are from
Deviation Random Error
Chevron Data (1973-76 Vehicles)
8
8
7
HC
CO
NOX
-0.1561
-1.6387
-0.0329
GARB Data (
1
1
1
HC
CO
NOx
-0.2220
-2.9700
0.2100
0.16214
2.39650
0.08321
1978 Vehicle)
0.0296*
0.0943*
0.3364
-
Chevron Data (1973-76 Vehicles)
8
8
7
8
8
8
8
8
8
8
8
8
8
8
8
HC
CO
NOx
Exxon
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
-0.0359
-0.7213
-0.0114
0.15777
2.81200
0.09651
0.5406
0.4917
0.7647
Data (1974-77 Vehicles)
-0.0186
-0.3258
0.0164
0.0185
0.3456
0.0729
0.0189
0.5236
-0.1018
-0.0205
0.8361
0.0195
0.16978
1.04230
0.27891
0.07007
1.23500
0.28561
0.18292
2.85160
0.26899
0.23953
1.59310
0.14529
0.7654
0.4060
0.8728
0.4795
0.4546
0.4939
0.7789
0.6195
0.3202
0.8157
0.1813
0.7155
Average
Percent
Emission
Change
-18.87
-12.70
-2.00
-25.00
-29.38
17.65
-5.60
1. 15
-0.38
4.95
-7.69
-0.91
1.25
4.64
1.89
8.46
4.09
-7.56
5.98
10.73
-1.62
* Probability value is less than 0.10, an indication that the
emission differences are significant.
-19-
-------
Table 1 (Continued)
Paired-t Test Results
Pre-1980 Vehicles
Amb
Temp.
(°F)
75
75
75
75
75
75
75
75
75
Fuel
RVP
(psi)
11.8
vs.
6.6
9.4
VS.
9.1
9.4
vs.
8.8
9.4
vs.
6.8
9.4
vs.
6.6
9.1
vs.
8.8
9.1
vs.
6.8
9.1
vs.
6.6
8.8
vs.
6.8
N
Mean
Emission
Difference
r
Poll
(g/mi)
Exxon Data
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
-0.0488
0.4936
0.1274
0.0371
0.6714
0.0565
0.0375
0.8494
-0.1181
-0.0019
1.1619
0.0031
-0.0301
0.8194
0.1110
0.0004
0.1780
-0.1746
-0.0390
0.4905
-0.0534
-0.0673
0.1480
0.0545
-0.0394
0.3125
0.1213
Standard
Deviation
(Continued)
0.21001
2.40380
0.07821
0.19777
1.46260
0.29223
0.13328
3.63790
0.35424
0.14134
2.25150
0.30659
0.15198
3.19480
0.30274
0.20357
2.85570
0.37707
0.26185
1.21800
0.24638
0.23439
2.26850
0.27382
0.07394
1.98830
0.24735
Probabilities
that the
Differences
are from
Random Error
0.5324
0.5796
0.0025*
0.6119
0.2353
0.6015
0.4523
0.5301
0.3770
0.9711
0.1878
0.9778
0.5925
0.4917
0.3342
0.9960
0.8650
0.2316
0.6862
0.2922
0.5594
0.4438
0.8588
0.5910
0.1757
0.6701
0.2082
Average
Percent
Emission
Change
-1.02
10.73
6.95
-5.28
10.14
1.57
2.90
8.41
-7.86
0.64
14.53
-2.42
-7.69
14.25
5.90
6.14
-2.57
-11.15
3.35
5.24
80
-4
-3.66
6.23
3.59
-2.51
6.66
4.42
* Probability value is less than 0.10, an indication that the
emission differences are significant.
-20-
-------
Table 1 (Continued)
Amb
Temp.
75
75
Fuel
RVP
(psi) N
8.8
vs.
6.6
6.8
vs.
6.6
8
8
8
8
8
8
Paired-t Test Results
Pre-1980 Vehicles
Mean
Emission
Difference
Poll (g/mi)
Standard
Deviation
Probabilities
that the
Differences
are from
Random Error
Exxon Data (Continued)
HC
CO
NOx
HC
CO
NOx
-0.0676
-0.0300
0.2291
-0.0283
-0.3425
0.1079
0.06025
1.19070
0.27782
0.09717
1.76110
0.10130
0.0156*
0.9452
0.0524*
0.4380
0.5994
0.0196*
Average
Percent
Emission
Change
-10.62
6.87
11 . 53
-8.83
-1.25
7. 58
* Probability value is less than 0.10,
emission differences are significant.
an indication that the
-21-
-------
Table 2
Paired-t Test Results
1980+ Vehicles
Amb
Temp.
Fuel
RVP
(psi)
N
Probabilities
Mean that the
Emission Differences
Difference Standard are from
Poll (g/mi) Deviation Random Error
Chevron Data (
43
11.
vs
8.
4
t
I
11
11
11
HC
CO
NOx
-0
-0
-0
.0245
.2940
.0154
API/ATL Data (
42
42
42
42
42
42
10.
vs
9.
10.
vs
8.
10.
vs
6.
9.
vs
8.
9.
vs
6.
8.
vs
6.
5
,
0
5
,
0
5
,
5
0
,
0
0
.
5
0
.
5
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
-0
-0
-0
0
-0
-0
0
-0
-0
0
-0
0
0
0
0
0
0
0
.0158
.1868
.0242
.0058
.6477
.0042
.0212
.0910
.0003
.0217
.4608
.0200
.0370
.0958
.0238
.0153
.5567
.0038
1981-83 Vehicles)
0
0
0
.05334
.70524
.05239
0
0
0
.1593
.1969
.3537
1983-86 Vehicles)
0
1
0
0
0
0
0
1
0
0
1
0
0
1
0
0
1
0
.05272
.41350
.08108
.06246
.85635
.09735
.13222
.40310
.09061
.05820
.25630
.02258
.13564
.32390
.06949
.13396
.30050
.06911
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.4950
.7592
.4981
.8281
.1231
.9206
.7111
.8800
.9932
.4036
.4101
.0822*
.5336
.8662
.4391
.7904
.3424
.8972
Average
Percent
Emission
Change
-4.29
-3.27
-1 . 10
-2.32
-0.49
-4.20
-0.34
-7.91
-1.42
-3.51
-2.40
-0.49
1.85
-8.53
2.79
-1.21
-3.02
3.23
-3.15
4.48
0.42
* Probability value is less than 0.10, an indication that the
emission differences are significant.
-22-
-------
Table 2 (Continued)
Paired-t Test Results
1980+ Vehicles
Amb
Temp.
Fuel
RVP
(psi)
N
Probabilities
Mean that the
Emission Differences
Difference Standard are from
Poll (q/mi) Deviation Random Error
EPA Data (1987 Vehicles)
50
50
50
14.
vs
11.
14.
vs
9.
11.
vs
9.
6
,
7
6
t
0
7
t
0
10
10
10
10
10
10
10
10
10
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
-0
-3
0
0
1
0
0
4
0
.1197
.2380
.0760
.0901
.0070
.1260
.2098
.2450
.0500
Chevron Data (
55
8.
vs
6.
4
(
1
11
11
11
HC
CO
NOx
-0
-0
-0
.0453
.6707
.0192
API/ATL Data (
55
55
55
10.
vs
9.
10.
vs
8.
10.
vs
6.
5
,
0
5
,
0
5
.
5
6
6
6
6
6
6
6
6
6
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
0
-0
0
0
0
-0
-0
0
-0
.0008
.0033
.0100
.0200
.1317
.0250
.0200
.0217
.0100
0.
10.
0.
0.
3.
0.
0.
14.
0.
55136
63600
10501
19988
79760
13377
71109
12000
11106
0
0
0
0
0
0
0
0
0
.5097
.3609
.0479*
.1878
.4234
.0155*
.3752
.3666
.1883
1981-83 Vehicles)
0.
0.
0.
03536
90075
03536
0
0
0
.0017*
.0331*
.1022*
1983-86 Vehicles)
0.
0.
0.
0.
1.
0.
0.
1.
0.
04873
95143
09050
05916
04910
07113
08155
32410
09077
0
0
0
0
0
0
0
0
0
.9682
.9935
.7975
.4453
.7709
.4286
.5742
.9696
.7981
Average
Percent
Emission
Change
-3.02
-10.08
7.48
5.77
5.39
13.42
1.66
5.89
5.97
-10.56
-7.91
-3.61
0.24
-2.13
1.81
2.51
0.13
-2.59
-4.33
-3.39
-0.95
* Probability value is less than 0.10, an indication that the
emission differences are significant.
-23-
-------
Table 2 (Continued)
Paired-t Test Results
1980+ Vehicles
Amb
Temp.
iin
55
55
55
75
80
80
80
80
Fuel
RVP
(psi)
9.0
vs.
8.0
9.0
vs.
6.5
8.0
vs.
6.5
8.6
vs.
7.2
10.5
vs.
9.0
10.5
vs.
8.0
10.5
vs.
6.5
9.0
vs.
8.0
N
Probabilities
Mean that the
Emission Differences
Difference Standard are from
r Poll (g/mi) Deviation Random Error
API/ATL Data (Continued)
6
6
6
6
6
6
6
6
6
4
4
4
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
GARB
HC
CO
NOx
0
0
-0
-0
0
-0
-0
-0
0
.0192
.1350
.0350
.0208
.0250
.0200
.0400
.1100
.0150
0.
0.
0.
0.
0.
0.
0.
0.
0.
Data (1983-88
-0
-0
0
.0095
.2375
.0025
API/ATL Data (
6
6
6
6
6
6
6
6
6
6
6
6
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
0
0
-0
0
1
-0
0
1
-0
0
0
0
.0300
.6050
.0167
.0558
.1400
.0108
.0208
.0013
.0258
.0258
.5350
.0058
0.
0.
0.
05181
96958
09165
04488
78920
06388
04940
64012
11962
Vehicles)
01555
43485
03304
0
0
0
0
0
0
0
0
0
0
0
0
.4064
.7469
.3925
.3070
.9412
.4777
.1041*
.6913
.7711
.3089
.3546
.8893
1983-86 Vehicles)
0.
1.
0.
0.
1.
0.
0.
1.
0.
0.
1.
0.
03271
10700
08920
05748
59360
10500
03917
59720
07883
04364
26230
09157
0
0
0
0
0
0
0
0
0
0
0
0
.0746*
.2383
.6664
.0632*
.1401
.8105
.2494
.1852
.4586
.2067
.3468
.8821
Average
Percent
Emission
Change
1.97
1.85
65
-5
-4.30
-0.76
-3.54
-6.75
-3.77
1.15
-4.14
-2.46
-3.03
7.82
13.88
-2.92
13.19
20.55
-2.93
7.51
16.86
-3.78
5.61
5.55
-0.35
* Probability value is less than 0.10, an indication that the
emission differences are significant.
-24-
-------
Table 2 (Continued)
Amb
Temp.
80
80
N
Paired-t Test Results
1980+ Vehicles
Mean
Emission
Difference
Poll (g/mi)
Standard
Deviation
Probabilities
that the
Differences
are from
Random Error
API/ATL Data (Continued)
9.0
vs.
6.5
8.0
VS.
6.5
6 HC -0.0092
6 CO 0.3963
6 NOx -0.0092
6 HC
6 CO
6 NOX
-0.0350
-0.1387
-0.0150
0.05572
1.09570
0.05435
0.08283
0.86434
0.06964
0.7036
0.4162
0.6966
0.3481
0.7105
0.6203
Average
Percent
Emission
Change
-0.90
2. 69
-1.02
-8.50
-7.51
-1.21
-25-
-------
Table 3
Student-t Test Results
Pre-1980 Vehicles
Amb
Temp.
(°F) Poll
55
75
75
75
75
75
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
11.8
8.3
11.8
8.3
11.8
8.3
11.8
8.3
11.8
8.3
11.8
8.3
11.8
9.4
11.8
9.4
11.8
9.4
11.8
9.1
11.8
9.1
11.8
9.1
11.8
8.8
11.8
8.8
11.8
8.8
11.8
6.8
11.8
6.8
11.8
6.8
Average
Emissions
N (g/mi)
t-prob. Variance f-prob.
0.4108
0.4496
0.4659
0.4708
0.3298
0.4630
Chevron Data (1973-76
8
8
8
8
8
8
8
8
8
8
8
8
Exxon
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
1.07
1.23
16.51
18.15
1.41
1.44
0.97
1.01
11.29
12.01
1.42
1.43
Data
1.13
1.15
11.42
11.74
1.79
1.77
1.13
1.11
11.42
11.07
1.79
1.72
1.13
1.11
11.42
10.89
1.79
1.89
1.13
1.15
11.42
10.58
1.79
1.77
0.7036
0.7292
0.9290
0.9322
0.8763
0.9766
(1974-77
0.9734
0.9517
0.9597
0.9713
0.9480
0.8183
0.9731
0.9181
0.7723
0.9716
0.8715
0.9511
Vehicles)
0.5894
0.7034
81.8400
90.3830
0.3878
0.4147
0.6657
0.7050
68.5760
96.9190
0.4328
0.4654
Vehicles)
1.0317
1.3688
110.7300
112.9500
0.4538
0.3554
1.0317
1.0045
110.7300
105.6700
0.4538
0.3217
1.0317
1.3846
110.7300
89.5090
0.4538
0.4981
1.0317
1.5271
110.7300
95.4060
0.4538
0.3275
0.3593
0.4899
0.3777
0.4864
0.4762
0.3306
0.3538
0.3930
0.4527
0.3088
0.4246
0.3389
Prob.
1st Mean
is < 2nd
0.64
0.63
0.53
0.53
0.56
0.51
0.51
0.52
0.48
0.49
0.48
0.41
0.49
0.46
0.61
0.51
0.44
0.48
-26-
-------
Table 3 (Continued)
Student-t Test Results
Pre-1980 Vehicles
Arab Fuel Average
Temp. RVP Emissions
(°F) Poll (psi) N (g/mi)
75 HC 11.8 8
6.6 8
CO 11.8 8
6.6 8
NOx 11.8 8
6.6 8
75 HC 9.4 8
9.1 8
CO 9.4 8
9.1 8
NOx 9.4 8
9.1 8
75 HC 9.4 8
8.8 8
CO 9.4 8
8.8 8
NOx 9.4 8
8.8 8
75 HC 9.4 8
6.8 8
CO 9.4 8
6.8 8
NOx 9.4 8
6.8 8
75 HC 9.4 8
6.6 8
CO 9.4 8
6.6 8
NOx 9.4 8
6.6 8
75 HC 9.1 8
8.8 8
CO 9.1 8
8.8 8
NOx 9.1 8
8.8 8
Prob.
1st Mean
t-prob. Variance f-prob. is < 2nd
Exxon Data (Continued)
1.13
1.18
11.42
10.92
1.79
1.66
1.15
1.11
11.74
11.07
1.77
1.72
1.15
1.11
11.74
10.89
1.77
1.89
1.15
1.15
11.74
10.58
1.77
1.77
1.15
1.18
11.74
10.92
1.77
1.66
1.11
1.11
11.07
10.89
1.72
1.89
0.9311
0.9260
0.7011
0.9466
0.8996
0.8488
0.9499
0.8683
0.7230
0.9976
0.8232
0.9916
0.9600
0.8782
0.7219
0.9995
0.9718
0.5940
1.0317
1.4230
110.7300
107.5400
0.4538
0.3919
1.3688
1.0045
112.9500
105.6700
0.3554
0.3217
1.3688
1.3846
112.9500
89.5090
0.3554
0.4981
1.3688
1.5271
112.9500
95.4060
0.3554
0.3275
1.3688
1.4230
112.9500
107.5400
0.3554
0.3919
1.0045
1.3846
105.6700
89.5090
0.3217
0.4981
0.3410 0.53
0.4851 0.47
0.4258 0.36
0.3467 0.47
0.4661 0.45
0.4494 0.43
0.4941 0.48
0.3834 0.44
0.3337 0.63
0.4445 0.50
0.4148 0.42
0.4584 0.50
0.4802 0.52
0.4750 0.44
0.4504 0.37
0.3413 0.50
0.4161 0.49
0.2891 0.69
-27-
-------
Table 3 (Continued)
Student-t Test Results
Pre-1980 Vehicles
Amb
Temp.
(°F) Poll
75
75
75
75
75
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
9.1
6.8
9.1
6.8
9.1
6.8
9.1
6.6
9.1
6.6
9.1
6.6
8.8
6.8
8.8
6.8
8.8
6.8
8.8
6.6
8.8
6.6
8.8
6.6
6.8
6.6
6.8
6.6
6.8
6.6
Average
Emissions
N (g/mi)
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
Prob.
1st Mean
t-prob. Variance f-prob. is < 2nd
Exxon Data (Continued)
1.11
.1.15
11.07
10.58
1.72
1.77
1.11
1.18
11.07
10.92
1.72
1.66
1.11
1.15
10.89
10.58
1.89
1.77
1.11
1.18
10.89
10.92
1.89
1.66
1.15
1.18
10.58
10.92
1.77
1.66
0.9457
0.9234
0.8541
0.9046
0.9775
0.8578
0.9489
0.9491
0.7115
0.9170
0.9953
0.5033
0.9636
0.9467
0.7244
1.0045
1.5271
105.6700
95.4060
0.3217
0.3275
1.0045
1.4230
105.6700
107.5400
0.3217
0.3919
1.3846
1.5271
89.5090
95.4060
0.4981
0.3275
1.3846
1.4230
89.5090
107.5400
0.4981
0.3919
1.5271
1.4230
95.4060
107.5400
0.3272
0.3919
0.2971 0.53
0.4481 0.46
0.4909 0.57
0.3287 0.54
0.4911 0.49
0.4006 0.43
0.4503 0.52
0.4675 0.48
0.2969 0.36
0.4861 0.54
0.4075 0.50
0.3799 0.27
0.4641 0.52
0.4393 0.53
0.4095 0.37
-28-
-------
Table 4
Student-t Test Results
1980+ Vehicles
Amb
Temp.
(°F) Poll
43
42
42
42
42
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
11.4
8.1
11.4
8.1
11.4
8.1
10.5
9.0
10.5
9.0
10.5
9.0
10.5
8.0
10.5
8.0
10.5
8.0
10.5
6.5
10.5
6.5
10.5
6.5
9.0
8.0
9.0
8.0
9.0
Average
Emissions
N (q/mi)
t-prob. Variance f-prob.
0.4644
0.4688
0.4379
8.0
Chevron Data
13
13
13
13
13
13
0.54
0.56
10.22
10.47
0.55
0.56
API/ATL Data
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
0.70
0.71
10.47
10.65
0.76
0.79
0.70
0.69
10.47
11.11
0.76
0.77
0.70
0.68
10.47
10.56
0.76
0.76
0.71
0.69
10.65
11.11
0.79
0.77
(1981-83
0.8817
0.9315
0.9323
(1983-86
0.9424
0.9532
0.8378
0.9784
0.8277
0.9721
0.9132
0.9753
0.9978
0.9199
0.8815
0.8650
Vehicles)
0.1198
0.1263
52.1010
54.5590
0.1421
0. 1558
Vehicles)
0.1375
0.1366
26.8440
30.9760
0.0408
0.0386
0.1375
0.1281
26.8440
23.5510
0.0408
0.0403
0.1375
0.0775
26.8440
22.3920
0.0408
0.0425
0.1366
0.1281
30.9760
23.5510
0.0386
0.0403
0.4971
0.4395
0.4770
0.4699
0.4446
0.4955
0.2720
0.4236
0.4821
0.4728
0.3855
0.4815
Prob.
1st Mean
is < 2nd
0.56
0.53
0.53
0.53
0.52
0.58
0.49
0.58
0.51
0.46
0.51
0.50
0.46
0.56
0.44
-29-
-------
Table 4 (Continued)
Student-t Test Results
1980+ Vehicles
Arab Fuel
Temp. RVP
(°F) Poll (psi)
42
42
50
50
50
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
, HC
CO
NOx
HC
CO
NOx
9.0
6.5
9.0
6.5
9.0
6.5
8.0
6.5
8.0
6.5
8.0
6.5
14.6
11.7
14.6
11.7
14.6
11.7
14.6
9.0
14.6
9.0
14.6
9.0
11.7
9.0
11.7
9.0
11.7
9.0
Average
Emissions
N (q/mi)
t-prob.
Variance
API/ATL Data (Continued)
6 0.71
6 0.68
6 10.65
6 10.56
6 0.79
6 0.76
6 0.69
6 0.68
6 11.11
6 10.56
6 0.77
6 0.76
EPA Data
10 0.81
10 0.93
10 12.98
10 16.22
10 0.90
10 0.83
10 0.81
10 0.72
10 12.98
10 11.97
10 0.90
10 0.78
10 0.93
10 0.72
10 16.22
10 11.97
10 0.83
10 0.78
0.8486
0.9750
0.8417
0.9356
0.8446
0.9746
0.1366
0.0775
30.9760
22.3920
0.0386
0.0425
0.1281
0.0775
23.5510
22.3920
0.0403
0.0425
(1987 Vehicles)
0.7459 0.3468
0.6204
0.4124
0.7187
0.7876
0.1837
0.5582
0.5060
0.5610
0.9768
78.5890
334.3400
0.0469
0.0351
0.3468
0.2596
78.5890
57.0110
0.0469
0.0361
0.9768
0.2596
334.3400
57.0110
0.0351
0.0361
Prob.
1st Mean
f-prob. is < 2nd
0.2744
0.3652
0.4592
0.2972
0.4786
0.4776
0.3365
0.3201
0.3519
0.0307*
0.0073*
0.4833
0.43
0.49
0 .43
0.47
0.43
0.49
0.0695* 0.62
0.0211* 0.68
0.3365 0.21
0.37
0.40
0. 10
0.29
0.26
0.29
* Probability value is less than 0.10, an indication that the
variance differences are significant.
-30-
-------
Table 4 (Continued)
Student-t Test Results
1980+ Vehicles
Amb
Temp.
°F) Poll
55
55
55
55
55
HC
CO
NOx
HC
CO
NOX
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
8.4
6.1
8.4
6.1
8.4
6.1
10.5
9.0
10.5
9.0
10.5
9.0
10.5
8.0
10.5
8.0
10.5
8.0
10.5
6.5
10.5
6.5
10.5
6.5
9.0
8.0
9.0
8.0
9.0
8.0
Average
Emissions
N ( q/mi )
Chevron
13
13
13
13
13
13
API/ATL
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
Data
0.36
0.40
6.18
6.75
0.63
0.65
Data
0.54
0.54
7.95
7.95
0.79
0.78
0.54
0.52
7.95
7.82
0.79
0.80
0.54
0.56
7.95
7.93
0.79
0.80
0.54
0.52
7.95
7.82
0.78
0.82
t-prob
(1981-83
0.6719
0.7344
0.9278
(1983-86
0.9965
0.9990
0.9386
0.9125
0.9602
0.8453
0.9177
0.9933
0.9378
0.9180
0.9585
0.8039
Variance
Vehicles)
0.0465
0.0573
15.2730
20.2720
0.2009
0.2071
Vehicles)
0.0996
0.1085
20.6770
19.3310
0.0383
0.0578
0.0996
0.0892
20.6770
19.0470
0.0383
0.0552
0.0996
0.1142
20.6770
17.1600
0.0383
0.0554
0.1085
0.0892
19.3310
19.0470
0.0578
0.0552
f-prob.
0.3617
0.3158
0.4795
0.4636
0.4715
0.3315
0.4537
0.4652
0.3494
0.4420
0.4214
0.3480
0.4177
0.4937
0.4803
Prob.
1st Mean
is < 2nd
0.66
0.63
0.53
0.50
0.50
0.47
0.46
0.48
0.57
0.54
0.50
0.53
0.46
0.48
0.59
-31-
-------
Table 4 (Continued)
Student-t Test Results
1980-1- Vehicles
Amb
Temp.
(°F)
55
55
75
80
80
Poll
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
Fuel
RVP
(psi)
9.0
6.5
9.0
6.5
9.0
6.5
8.0
6.5
8.0
6.5
8.0
6.5
8.6
7.2
8.6
7.2
8.6
7.2
10.5
9.0
10.5
9.0
10.5
9.0
10.5
8.0
10.5
8.0
10.5
8.0
Average
Emissions
N (q/mi) t-prob.
Variance
API/ATL Data (Continued)
6
6
6
6
6
6
6
6
6
6
6
6
CARB
4
4
4
4
4
4
0.54
0.56
7.95
7.93
0.78
0.80
0.52
0.56
7.82
7.93
0.82
0.80
Data (
0.33
0.34
4.04
4.27
0.45
0.45
API/ATL Data
6
6
6
6
6
6
6
6
6
6
6
6
0.35
0.32
4.35
3.75
0.89
0.90
0.35
0.30
4.35
3.21
0.89
0.90
0.9160
0.9921
0.8871
0.8324
0.9652
0.9142
0.1085
0.1142
19.3310
17.1600
0.0578
0.0554
0.0892
0.1142
19.0470
17.1600
0.0552
0.0554
1983-88 Vehicles)
0.9595 0.0640
0.9282
0.9897
(1983-86
0.6989
0.6275
0.9125
0.4307
0.2858
0.9401
0.0647
11.2500
14.3310
0.0733
0.0647
Vehicles)
0.0192
0.0149
4.9908
3.7726
0.0719
0.0593
0.0192
0.0086
4.9908
1.1420
0.0719
0.0467
Prob.
1st Mean
£-prob. is < 2nd
0.4783
0.4496
0.4819
0.3967
0.4558
0.4984
0.4967
0.4235
0.4605
0.3944
0.3831
0.4193
0.1989
0.0657*
0.3237
0.54
0.50
0.55
0.58
0.52
0.46
0.36
0.33
0.54
0.23
0. 16
0.53
* Probability value is less than 0.10,
variance differences are significant.
an indication that the
-32-
-------
Amb
Temp.
(°F) Poll
80
80
80
80
Table 4 (Continued)
Student-t Test Results
1980+ Vehicles
N
Average
Emissions
(q/mi)
t-prob. Variance
Prob.
1st Mean
f-prob. is < 2nd
API/ATL Data (Continued)
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
10.5
6.5
10.5
6.5
10.5
6.5
9.0
8.0
9.0
8.0
9.0
8.0
9.0
6.5
9.0
6.5
9.0
6.5
8.0
6.5
8.0
6.5
8.0
6.5
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
0
0
4
3
0
0
0
0
3
3
0
0
0
0
3
3
0
0
0
0
3
3
0
0
.35
.33
.35
.35
.89
.91
.32
.30
.75
.21
.90
.90
.32
.33
.75
.35
.90
.91
.30
.33
.21
.35
.90
.91
0
0
0
0
0
0
0
0
0
0
0
0
.8137
.3521
.8680
.6883
.5676
.9659
.9131
.6763
.9507
.6514
.8331
.9150
0
0
4
1
0
0
0
0
3
1
0
0
0
0
3
1
0
0
0
0
1
1
0
0
.0192
.0253
.9908
.3246
.0719
.0658
.0149
.0086
.7726
.1420
.0593
.0467
.0149
.0253
.7726
.3246
.0593
.0658
.0086
.0253
.1420
.3246
.0467
.0658
0
0
0
0
0
0
0
0
0
0
0
0
.3838
.0859*
.4628
.2794
. 1078*
.3994
.2874
.1377
.4560
.1299
.4373
.3576
0.41
0.19
0.56
0.35
0.29
0.48
0.54
0.35
0.52
0.67
0.58
0.54
* Probability value is less than 0.10, an indication that the
variance differences are significant.
-33-
-------
References
1. Shih, Celia, "Analysis of Impact of Fuel RVP on Exhaust
Emissions at 75°F Ambient Temperature,"
EPA-AA-TEB-EF-88-01, May 12, 1988.
2. "Reference Material for Phase II Volatility Control FRM
Docket," EPA Memorandum from Celia Shih, TEB, to Tad
Wysor, SDSB, March 26, 1990, Attachment C.
3. "Reference Material for Phase II Volatility Control FRM
Docket," EPA Memorandum from Celia Shih, TEB, to Tad
Wysor, SDSB, March 26, 1990, Attachment D.
4. Welstand, J. S., "Effects of Reduced Volatility Gasoline
on Exhaust Emissions, Driveability, and Fuel Economy,"
June 10, 1983, Chevron Research Company, Richmond,
California.
5. "The Effect of Fuel Volatility Variations on Evaporative
and Exhaust Emissions," API Publication 4310, Exxon
Research and Engineering Company, Linden, New Jersey, May
1979, prepared for the American Petroleum Institute.
6. "Evaluation of Test Results from the 8 RVP Fuel Evaluation
Project," State of California Air Resources Board
Memorandum from Dave Coel to Steve Albu, August 18, 1988.
7. Gething, J. A., and J.S. Welstand, "Summary Report on the
Effects of Reduced RVP Gasoline on Early 1980's Model
Vehicle Exhaust Emissions and Their Interaction with
Temperature," February l, 1988, Chevron Research Company,
Richmond, California.
8. Welstand, J. S., Chevron Research Company, Personal
Communication with Celia Shih, EPA, April 5, 1990.
9. "Emissions and Vehicle Performance with Lower RVP Fuels,
Final Report," Automotive Testing Laboratories, Inc., East
Liberty, Ohio, January 27, 1988, sponsored by the American
Petroleum Institute.
10. Neter, John and William Wasserman "Applied Linear
Statistical Models," Richard D. Irwin, Inc., 1974.
11. "National Fuel Surveys," Motor Vehicle Manufacturers
Association of the United States, Inc., both summer and
winter of 1986, 1987, and 1988.
-34-
-------
APPENDIX
-35-
-------
Table Al
List of Test Vehicles
1975 Chevron Research Test program
(All California Vehicles)
Veh.tt
l
2A*
2B*
3
4
5
6
7
Model
Year Make/Model
Engine Garb
CID Bbl
1973 Plymouth Fury 360 2
1973 Pontiac Catalina 400 2
1973 Pontiac Catalina 400 2
1974 Volkswagon Super Beetle 98 l
1975 Chevrolet Impala 350 4
1975 Ford LTD 400 2
1975 Ford Pinto 140 2
1976 Oldsmobile Cutlass 350 4
Transmission
Auto/Manna1
A
A
A
M-4
A
A
A
A
Vehicles #2A and tt2B were the same vehicle, tt2A was tested
with the original carburetor and tt2B was tested with a
replacement carburetor.
-36-
-------
Table A2
List of Test Vehicles
1979 Exxon Research Test Program
(All Automatic Transmission Vehicles)
Veh.tt
A
B
C
D
E
F
G
H
Model
Year
1976
1975
1974
1974
1977
1976
1975
1974
Make/Model
Ford LTD
Olds
Valiant
Vega
Granada
Buick
Dodge
Grand Prix
Engine Carb
CID Bbl
400
455
225
140
302
350
318
400
2
4
1
2
2
2
2
4
Catalyst
Equipped
yes
yes
no
no
yes
yes
yes
no
— * Vehicles B, E, and G were California vehicles. The
engines of vehicles B and G had been overhauled
previously, at 3,000 and 6,000 miles, respectively,
prior to the testing.
-37-
-------
Table A3
List of Test Vehicles
1988 CARB Five-Car Evaluation Project
(All California Vehicles)
Model ft of Fuel
Veh.tt Year Make/Model Engine Cyl System
1 1987 Olds Delta 88 3.8L 6 PFI
2 1987 Toyota Camry 2.0L 4 PFI
3 1988 Ford Tempo 2.5L 4 TBI
4 1983 Toyota Tercel 1.5L 4 Carb
5* 1978 Chevrolet Caprice 5.OL 8 Carb
* Vehicle was analyzed separately, since this vehicle was
built for compliance with a different evaporative emission
standard.
-38-
-------
Table Bi
List of Test Vehicles
1988 Chevron Research Test Program
(All California Vehicles)
Model
Veh.tt Year
1
2*
3
4
5
6
7
8
9
10
11
12
13
14**
1981
1982
1982
1983
1982
1983
1981
1983
1983
1983
1983
1982
1983
1983
Make/Model
Phase Engine
Dodge Aries
Chev Cavalier
Olds Ciera
Honda Accord
Ford Escort
Chev Cavalier
Datsun 210
Datsun Maxima
Olds Cutlass
Mazda GLC
Toyota Corolla
Toyota Corolla
Lincoln Town Car
Toyota SR-5 Pickup
I, II
I, II
I, II
I, II
I, II
I
I, II
I, II
I, II
I, II
I
II
II
I, II
L-4
L-4
L-4
L-4
L-4
L-4
L-4
L-6
V-6
L-4
L-4
L-4
V-8
L-4
Displ
(Liter)
2.6
1.8
2.5
1.8
1.6
2.0
1.2
2.4
3.8
1.5
1.6
1.8
5.0
2.4
Trans
Fuel Auto/
System Manual
Garb
Garb
TBI
Garb
Garb
TBI
Garb
PFI
Garb
Garb
Garb
Garb
TBI
Garb
A
A
A
M-5
A
A
M-5
A
A
M-4
M-5
M-5
A
M-5
* Vehicle was tested with the original carburetor on Phase I, and
with the replaced carburetor on Phase II.
** Vehicle was excluded from the analysis, since this vehicle was
categorized as a truck.
-39-
-------
Table B2
List of Test Vehicles
1987 API/ATL Test Program
(All Federal Vehicles with Automatic Transmission)
Model Fuel
Veh.tt Year Make/Model Engine System
CD-I 1985 Olds Cutlass 3.8L Garb 2V
CD-2 1986 Pontiac Grand AM 3.OL PFI
CD-3 1983 Mercury Cougar 3.8L Garb 2V
CD-4 1985 Ford T-Bird 5.OL TBI
CD-5 1985 Plymouth Horizon 2.2L Garb 2V
CD-6 1985 Chev Cavalier 2.OL TBI
-40-
-------
Table B3
List of Test Vehicles
EPA/E&D Test Program
(All Federal Vehicles)
Model Fuel
Veh tt Year Make/Model CID System
1 1987 Pontiac 6000 151 TBI
3 1987 Ford Taurus Wagon 182 PFI
6 1987 Dodge Shadow 135 TBI
7 1987 Ford Taurus Wagon 182 PFI
10 1987 Olds Cutlass 173 PFI
19 1987 Chrysler Lebaron 135 PFI
20 1987 Mercury Cougar 231 TBI
35 1987 Chev Cavalier 121 TBI
43 1987 Pontiac 6000 151 TBI
56 1987 Chev Cavalier 121 TBI
-41-
------- |