United States Air and Radiation EPA420-P-99-023
Environmental Protection June 1999
Agency M6.EVP.006
vvEPA Estimating Weighting
Factors for Evaporative
Emissions in MOBILE6
> Printed on Recycled Paper
-------
EPA420-P-99-023
- Draft -
Estimating Weighting Factors
for Evaporative Emissions in MOBILE6
Larry C. Landman
Document Number M6.EVP.006
June 10, 1999
U.S. EPA
Assessment and Modeling Division
National Vehicle Fuel and Emissions Laboratory
2000 Traverwood Drive
Ann Arbor, Michigan 48105-2425
734-214-4939 (fax)
mobile@epa.gov
NOTICE
These reports do not necessarily represent final EPA
decisions or positions. They are intended to present
technical analysis of issues using data which are currently
available. The purpose in release of these 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.
-------
ABSTRACT
In previous documents (M6.EVP.001, M6.EVP.002, and
M6.EVP.005), EPA proposed methods of estimating vehicles' resting
loss and diurnal emissions based (in part) on the vehicles'
performance (pass or fail) on the purge and pressure tests. EPA
plans to compute model-year and age specific average resting loss
and diurnal emissions by weighting together the emissions of
passing and failing vehicles according to their frequency in the
in-use fleet. This document describes this approach and EPA's
proposed estimates of pass and fail rates as functions of vehicle
age.
Please note that EPA is seeking any input from stakeholders
and reviewers that might aid us in modeling any aspect of resting
loss or diurnal evaporative emissions.
Comments on this report and its proposed use in MOBILE6
should be sent to the attention of Larry Landman. Comments may be
submitted electronically to mobile@epa.gov, or by fax to (734)
214-4939, or by mail to "MOBILE6 Review Comments", US EPA
Assessment and Modeling Division, 2000 Traverwood Drive, Ann
Arbor, MI 48105. Electronic submission of comments is preferred.
In your comments, please note clearly the document that you are
commenting on, including the report title and the code number
listed. Please be sure to include your name, address,
affiliation, and any other pertinent information.
This document is being released and posted. Comments will be
accepted for sixty (60) days. EPA will then review and consider
all comments received and will provide a summary of those
comments, and how we are responding to them.
-------
TABLE OF CONTENTS
Page Number
1.0 Introduction 1
2.0 Data Sources 2
3.0 Analysis 6
3.1 Strata Based on Purge and Pressure Tests ... 6
3.1.1 Vehicles Failing the Pressure Test . . 6
3.1.2 Vehicles Passing Both the Pressure
and the Purge Tests 9
3.1.3 Vehicles Failing ONLY the Purge Test . . 11
3.1.4 Summary of Purge and the Pressure
Failure Rates 13
3.1.5 Vehicles Failing Both the Pressure
and the Purge Tests 13
3.2 Modeling "Gross Liquid Leakers" 16
3.3 Combining Purge/Pressure Rates with
Gross Liquid Leaker Rates 18
4.0 Modeling Enhanced EVAP Vehicles Equipped with OBD . . 20
5.0 Comparisons with MOBILES 24
5.1 Comparisons of Weighting Factors 24
5.2 Comparisons of Weighted Diurnal Emissions ... 27
5.3 Comparisons of Diurnal Emissions from Vehicles
Certified to the Enhanced Evaporative Control
Standards 32
6.0 Summary 34
7.0 References 36
11
-------
TABLE OF CONTENTS (Continued)
APPENDICES Page Number
A. Estimates of Purge/Pressure Strata Size by
Vehicle Age for 1995 and Older Model Year
Vehicles for Non-I/M Areas 37
B. Predicted Frequency of Occurrence of "Gross
Liquid Leakers" by Emission Type and Vehicle
Age (for 1995 and Older Model Year Vehicles) .... 38
C. Estimates of Purge/Pressure Strata Size by
Vehicle Age for 1999 and Newer Model Year
Vehicles for I/M Areas 39
D. Estimates of Purge/Pressure Strata Size by
Vehicle Age for 1999 and Newer Model Year
Vehicles for Non-I/M Areas 40
E. Predicted Frequency of Occurrence of "Gross
Liquid Leakers" by Emission Type and Vehicle
Age (for 1999 and Newer Model Year Vehicles) .... 41
111
-------
*** DRAFT ***
Estimating Weighting Factors
for Evaporative Emissions in MOBILE6
Report Number M6.EVP.006
Larry C. Landman
U.S. EPA Assessment and Modeling Division
1 .0 Introduction
In three recently released draft reports [1,2,3]*, the US
Environmental Protection Agency (EPA) proposed methods of
estimating the resting loss and diurnal emissions from results of
real-time diurnal (RTD) tests of in-use vehicles in which the
ambient temperature cycled over a 24 degree Fahrenheit range to
simulate in real-time the daily heating and cooling that parked
vehicles experience over a 24 hour period. For many of the
vehicles used in these studies, the recruitment method was
designed to recruit a relatively large number of vehicles that had
problems with their evaporative control systems. Specifically, two
tests of the integrity of each vehicle's evaporative control
system (a pressure test** and a purge test) were used to screen
the candidate vehicles. This recruitment bias did not affect the
analysis of these data as described in earlier reports; since
those analyses were performed within each purge/pressure grouping,
the selection was random within the purge/pressure and model year
groups. However, to correctly represent the entire in-use fleet
the results must be weighted. In this report, EPA proposes
weightings for each stratum to estimate the emissions of the
entire in-use fleet. EPA will use these factors to weight
together the results of the RTD tests, as well as the results of
hot soak tests and running loss tests which were also derived from
measurements of a stratified sample.
For each of the earlier analyses of resting loss and diurnal
data, the sample of test vehicles was divided into four strata.
The first of these strata consisted of several vehicles having
* The numbers in brackets refer to the references in Section 7 (page 36).
* * This pressure test was performed by disconnecting the vapor line at the
canister and then pressurizing the tank from that position with the gas
cap in its normal position. This procedure differs from the method
currently being used in Inspection and Maintenance (I/M) lanes.
-------
-2- DRAFT
substantial leaks of liquid gasoline (as opposed to simply vapor
leaks); these vehicles were labeled "gross liquid leakers." EPA
proposed [4] using the following three definitions (based on the
evaporative emissions test used) for such vehicles with:
resting loss emissions (i.e., the mean emissions during the
last six hours of the 24-hour RTD test) were at least 2.00
grams per hour (see also reference [1]) , or
hot soak test emissions were at least 10.00 grams per test
(see also reference [5]) , or
running loss test emissions were at least 7.00 grams per
mile (see also reference [6]) .
These three different definitions will identify potentially
different sets of vehicles as being "gross liquid leakers" (see
Section 3.2). For the remaining three strata, we used the results
of the purge and pressure tests to match the stratification of the
recruitment process. This approach produces the following three
additional strata:
1) vehicles that pass both the purge and pressure tests,
2) vehicles that fail the pressure test (regardless of their
performance on the purge test),* and
3) vehicles that fail only the purge test.
This document reports on EPA's proposal to weight those four
strata together to obtain estimates (of running loss, hot soak,
resting loss, and diurnal emissions) for the entire in-use fleet.
2.0 Data Sources
To develop the appropriate weighting factor for the stratum
of vehicles identified as "gross liquid leakers," EPA relied on
five groups of data to estimate the frequency of the occurrence of
these vehicles (see also reference [4]) :
For the "gross liquid leakers" identified by the RTD test,
EPA used a sample consisting of 151 vehicles tested by the
Coordinating Research Council (CRC) during 1996 as part of
its real-time diurnal testing program (Program E-9)
combined with 119 vehicles tested by EPA. [1]
For the "gross liquid leakers" identified by the hot soak
test, EPA used a sample consisting of 300 vehicles tested
* Vehicles failing both purge and pressure are discussed in Section 3.1.4.
-------
-3- DRAFT
by Auto/Oil during 1993 as part of its real world hot soak
testing program combined with 197 vehicles tested by EPA.
[5]
For the "gross liquid leakers" identified by the running
loss test, EPA used a sample consisting of 150 vehicles
tested by the CRC during 1997 as part of its running loss
testing program (program E-35) . [6]
The CRC also tested 50 late-model year vehicles during 1998
as part of its combined hot soak, real-time diurnal,
running loss testing program (E-41) . [7] (These results are
used in reference [4] to test the predictions of the
occurrence of "gross liquid leakers" among newer vehicles.)
A fifth source of data consisted of the results of a
testing program run jointly by the CRC and the American
Petroleum Institute (API) . [8] This program was designed to
determine the frequency of vehicles with liquid leaks.
Actual measurements of evaporative emissions were not
performed in this program; therefore, we cannot determine
which of those vehicles identified as having liquid leaks
would have actually met any of our definitions of a "gross
liquid leaker."
To develop the appropriate weighting factors based on the
performance on the purge and pressure tests, EPA used data from an
EPA testing contractor, Automotive Testing Laboratories (ATL),
which performed purge and pressure tests on a random sample of
13,425 vehicles at its Inspection and Maintenance (I/M) lanes in
Indiana and Arizona between the years 1990 and 1995. Since the
testing protocols were changing in the early months of the
program, we omitted the first nine months of data. We then
identified the initial test of each of the test vehicles and
calculated, by vehicle age, the number of pre-1996* model year
vehicles in each of the three purge/pressure categories.
We combined the results for the I/M lane testing into a
single table (Table 1 on page 5). Omitted from all of the columns
in Table 1 are the results on approximately fifteen percent of the
vehicles for which the purge or pressure tests were not performed.
The reasons that testing was not performed varied, and included
both periodic problems with the testing equipment as well as
problems related to the vehicle (e.g., presence of check-valves or
difficulty accessing the necessary lines). All of the subsequent
analyses were performed on the sample of vehicles for which the
purge/pressure classification could be made. Since all of the
subsequent analyses are based on ratios from Table 1 (e.g., the
* Limiting the analysis to pre-1996 model year vehicles is related to the
phase-in of the enhanced evaporative control vehicles (see Section 4)..
-------
-4- DRAFT
number of classified pressure failures divided by the total number
of vehicles that were classifiable), EPA proposes to treat the
results of those analyses as if they applied to the entire in-use
fleet. This proposal is equivalent to assuming that those 15
percent of unclassifiable vehicles were distributed
proportionately among the three purge/pressure strata.
In examining the data in Table 1, we note that there were
relatively few vehicles more than 20 years of age. Since small
sample sizes tend to result in low statistical confidence in the
calculated ratios (i.e., the percent of vehicles at each age that
fall into each of the purge/pressure strata), those small sample
sizes are an obvious weakness of this analysis. We will address
that weakness by using the calculated variances in the ratios to
weight the analyses. (That is, the ratios from the model years
containing the most vehicles will be counted more heavily than the
ratios from the more sparsely sampled model years.)
An alternative approach (not being used) is to smooth the
data from the older vehicles by averaging the results from the 66
vehicles over the age of 20 years (all of which were from the
industry programs) to obtain a sample with:
a mean age of 23.23 years,
19 vehicles (28.8 percent) passing both the pressure test
and the purge test,
38 vehicles (57.6 percent) failing the pressure test, and
9 vehicles (13.6 percent) failing only the purge test.
That averaged failure rate on the pressure test of almost 60
percent among the vehicles over 20 years of age suggests a
substantially higher failure rate among these vehicles than was
predicted in MOBILES (i.e., under 35 percent). This is due
entirely to data recently obtained in the CRC testing programs.
-------
-5-
DRAFT
Table 1
Distribution of 14,061 1971-95 Model Year Vehicles
Vehicle
Age*
0
1
2
3
4
5
6
7
8
9
1 0
1 1
1 2
1 3
1 4
1 5
1 6
1 7
1 8
1 9
20
21
22
23
24
25
26
Performance on Purge and
Pressure Tests
Fail
Pressure
5
48
42
61
81
94
91
94
88
68
46
41
64
49
29
1 9
1 3
7
4
1 2
1 2
3
7
1 0
6
6
6
Fail Only
Purge
9
29
33
33
42
50
76
74
46
68
44
24
23
20
5
6
3
1
0
1
3
2
0
2
2
3
0
Passing
Both
228
1 ,448
1 ,302
1 ,494
1 ,308
1 ,475
1 ,403
1 ,261
888
682
369
192
152
102
62
34
1 7
7
2
4
7
7
5
3
2
1
1
Total
242
1,525
1,377
1,588
1,431
1,619
1,570
1,429
1,022
818
459
257
239
171
96
59
33
1 5
6
1 7
22
1 2
1 2
1 5
1 0
1 0
7
The quantity "Vehicle Age" is the whole number calculated by subtracting
model year from test year and then changing all negative results to zero.
-------
-6- DRAFT
3.0 Analysis
3.1 Modeling Strata Based on Purge and Pressure Tests
Using the data from Table 1, we calculated the rate at which
vehicles were present (by age) in the following three categories
determined by the results on the pressure test and the purge test.
vehicles passing both the pressure test and the purge test,
vehicles failing the pressure test, and
vehicles failing only the purge test.
These three categories are not independent. Given the results
from any two would permit the size of remaining category to be
determined. EPA chose to model the rates at which vehicles were
present in the first two of those categories. These rates by
vehicle age (in years) along with the corresponding 90 percent
confidence intervals are given in Tables 2 and 3. The confidence
intervals were calculated separately for each vehicle age rather
than having an overall calculation for the entire sample.
Calculating confidence intervals independently (as if the failure
rate at one age were not related to the failure rates of
neighboring ages) emphasizes the disparity in the sizes of some of
the samples by age, as the size of the confidence interval is
substantially controlled by the sample size.
3.1.1 Vehicles Failing the Pressure Test
Calculating (from Table 1) the rates at which vehicles failed
the pressure test (regardless of the performance on the purge
test) produces the data given in Table 2.
As previously stated, since the 90 percent confidence
intervals in Table 2 were calculated separately for each age that
was sampled, the confidence intervals are most representative of
the relative sample sizes. Rather than immediately attempting to
use a regression analysis to obtain an equation relating the rate
of vehicle's failing the pressure test to the vehicle's age, we
first examined the calculated 90 percent confidence intervals in
Table 2.
-------
-7-
DRAFT
Table 2
Estimating Rate of Failing the Pressure Test
For 14,061 1971-95 Model Year Vehicles
With 90 Percent Confidence Intervals
Vehicle
Age
0
1
2
3
4
5
6
7
8
9
1 0
1 1
1 2
1 3
1 4
1 5
1 6
1 7
1 8
1 9
20
21
22
23
24
25
26
Sample
Size
242
1 ,525
1 ,377
1 ,588
1 ,431
1 ,619
1 ,570
1 ,429
1 ,022
818
459
257
239
171
96
59
33
1 5
6
1 7
22
1 2
1 2
1 5
1 0
1 0
7
Failure
Rate
2.1%
3.1%
3.1%
3.8%
5.7%
5.8%
5.8%
6.6%
8.6%
8.3%
10.0%
16.0%
26.8%
28.7%
30.2%
32.2%
39.4%
46.7%
66.7%
70.6%
54.5%
25.0%
58.3%
66.7%
60.0%
60.0%
85.7%
90 Percent
Confidence Interval
0.6% - 3.6%
2.4% - 3.9%
2.3% - 3.8%
3.0% - 4.6%
4.7% - 6.7%
4.8% - 6.8%
4.8% - 6.8%
5.5% - 7.7%
7.2% - 10.1%
6.7% - 9.9%
7.7% - 12.3%
12.2% - 19.7%
22.1% - 31.5%
23.0% - 34.3%
22.5% - 37.9%
22.2% - 42.2%
25.4% - 53.4%
25.5% - 67.9%
35.0% - 98.3%
52.4% - 88.8%
37.1% - 72.0%
4.4% - 45.6%
34.9% - 81.7%
46.6% - 86.7%
34.5% - 85.5%
34.5% - 85.5%
64.0% - 100.0%
Examining the confidence intervals in Table 2, we found that
some of those confidence intervals are so large as to be almost
useless. (For example, knowing that seven of 15 of the vehicles
17 years of age failed the pressure test indicates that the actual
failure is most likely between 25 percent and 68 percent. A range
that large is not helpful in predicting the true failure rate.)
However, using both the sample failure rates and the confidence
-------
DRAFT
intervals, we were able to make the following four observations
that were then used to select an appropriate mathematical model:
The pressure failure rate appears to start (i.e., for new
vehicles) between two and four percent.
The pressure failure rate increases gradually for the first
seven years of the vehicle's life.
The pressure failure rate then increases more rapidly for
the next ten years.
The pressure failure rate then begins to level off
(approaching 60 percent, according to the data from the
industry programs) (see third "bullet" on page 4).
This type of behavior is typical of a logistic growth function.
Prior to constructing an appropriate logistic growth
function, we first "adjusted" the variable "AGE" (in Tables 1, 2,
and 3) which is based on the test date (since it is the integer
calculated by subtracting the model year from the test year).
However, since the typical test was performed in early July (mean
date of July 3 and median date of July 10), we modified that
variable by adding 0.5 so that the predicted rate of vehicles
failing the pressure test would be based on the age of the
vehicles as of the first of January which coincides with the date
used in MOBILE.
To account for differences in the size of the confidence
intervals (or, equivalently the sample size) the data were
weighted by the reciprocal of the variance. The "logistic growth"
function that best models the weighted pressure test failure rates
from Table 2 is given by the following equation:
_ ., _ 4 0.6045
Pressure Fa.lure Rate = 1 + 1 7.733.exp[_0.01362*(AG EA2)] (1 )
Estimates based on this equation of failure rates on the pressure
test are given in Appendix A. These estimates must be adjusted
for the "gross liquid leakers" (see Section 3.3).
In Figure 1 (on the following page), we plotted both the
measured failure rate on the pressure test (from Table 2, shifted
by six months to compensate for the July testing) and the curve
described by equation (1). That graph suggests that equation (1)
is a very good fit for the measured failure rates except at ages
18, 19 and 21 years (for which fewer than 20 vehicles were
recruited at each age). Also, for vehicles over 20 years of age,
the predicted failure rate on the pressure test is close to 60
percent which closely approximates the results of those 66 tests
from industry programs (see third "bullet" on page 4).
-------
-9-
DRAFT
Figure 1
Comparison of Measured and Predicted Rates
For Vehicles Failing the Pressure Test
(by vehicle age)
90%
0)
4-*
TO
(A
0)
0)
3
(A
(1)
60% --
30% --
0%
1 0 20
Vehicle Age (years)
30
3.1.2 Vehicles Passing Both the Pressure and the Purge Tests
As described in the preceding section, we first calculated
from Table I the rates at which vehicles passed both the pressure
and the purge tests, yielding the results given in Table 3.
As in the case of the failure rate on the pressure test, we
were able to make the following four observations from Table 3:
The rate at which vehicles passed both the pressure test
and the purge test starts (i.e., for new vehicles) between
92 and 96 percent.
The rate at which vehicles passed both the pressure test
and the purge test decreases gradually for the first seven
years of the vehicle's life.
The rate at which vehicles passed both the pressure test
and the purge test then decreases more rapidly for the next
ten years.
The rate at which vehicles passed both the pressure test
and the purge test then begins to level off (approaching 20
to 40 percent, according to the data from the industry
programs) (see second "bullet" on page 4).
-------
-10-
DRAFT
Table 3
Estimating Rate of Passing Both the Pressure and Purge Tests
For 14,061 1971-95 Model Year Vehicles
With 90 Percent Confidence Intervals
Vehicle
Age
0
1
2
3
4
5
6
7
8
9
1 0
1 1
1 2
1 3
1 4
1 5
1 6
1 7
1 8
1 9
20
21
22
23
24
25
26
Sample
Size
242
1 ,525
1 ,377
1 ,588
1 ,431
1 ,619
1 ,570
1 ,429
1 ,022
818
459
257
239
171
96
59
33
1 5
6
1 7
22
1 2
1 2
1 5
1 0
1 0
7
Rate
94.2%
95.0%
94.6%
94.1%
91.4%
91.1%
89.4%
88.2%
86.9%
83.4%
80.4%
74.7%
63.6%
59.6%
64.6%
57.6%
51.5%
46.7%
33.3%
23.5%
31.8%
58.3%
41.7%
20.0%
20.0%
10.0%
14.3%
90 Percent
Confidence Interval
91.7% - 96.7%
94.0% - 95.9%
93.5% - 95.6%
93.1% - 95.1%
90.2% - 92.6%
89.9% - 92.3%
88.1% - 90.6%
86.8% - 89.6%
85.2% - 88.6%
81.2% - 85.5%
77.3% - 83.4%
70.2% - 79.2%
58.5% - 68.7%
53.5% - 65.8%
56.6% - 72.6%
47.0% - 68.2%
37.2% - 65.8%
25.5% - 67.9%
1.7% - 65.0%
6.6% - 40.5%
15.5% - 48.2%
34.9% - 81.7%
18.3% - 65.1%
3.0% - 37.0%
0.0% - 40.8%
0.0% - 25.6%
0.0% - 36.0%
As before, the "logistic growth" function appeared to be the
best choice for modeling the rates at which vehicles passed both
the pressure and the purge tests. After adjusting for age, the
resulting equation is given below as equation (2):
Rate of Passing Both = 1 -
0.7200
13.40*exp[-0.0145*(AGEA2)]
(2)
-------
-11-
DRAFT
Estimates based on equation (2) of the rates of vehicles
passing both the purge and pressure tests are given in Appendix A.
These estimates must be adjusted for the "gross liquid leakers"
(see Section 3.3).
In Figure 2, for vehicles passing both the purge and pressure
tests, we plotted both the measured rates (from Table 2, with age
shifted by six months to compensate for the July testing) and the
curve described by equation (2). That graph suggests that
equation (2) is a very good fit for the measured rates for ages at
which at least 20 vehicles were sampled. Also, for vehicles over
20 years of age, the predicted rate of vehicles passing both the
purge and pressure tests is close to 29 percent which closely
approximates the results of those 66 tests from industry programs
(see second "bullet" on page 4).
Figure 2
Comparison of Measured and Predicted Rates
For Vehicles Passing Both the Purge and Pressure Tests
(by vehicle age)
100%
(A
0)
o
CD
O)
c
'in
(A
TO
D.
50% --
0%
1 0 20
Vehicle Age (years)
30
3.1.3 Vehicles Failing ONLY the Purge Test
The third (and final) stratum based on vehicles' performance
on the purge and pressure tests is that containing vehicles that
failed only the purge test. EPA proposes to estimate that stratum
by subtracting from one hundred percent the total of equation (1)
plus equation (2) (prior to adjusting for the gross liquid
leakers, as discussed in Section 3.3). Since, for some vehicle
ages, the rate of decline in equation (2) is greater that the rate
of growth in equation (1), this approach has the effect of
predicting a decrease in the rate of purge only failures for
-------
-12-
DRAFT
vehicles older than 16 years of age. (This effect suggests that
some of the vehicles that had failed only the purge test would
begin to also fail the pressure test, thus migrating into the
pressure failure stratum.)
This effect is illustrated in Figure 3. If we combine this
estimate of the incidence of vehicles' failing only the purge test
with the estimate (from Section 3.1.4) of failing both the purge
and pressure tests, we obtain a predicted failure rate on the
purge test (regardless of any pressure test result). This purge
test failure rate does not exhibit that quirk of a decrease in
failure rate with increasing age.
The predicted size (in percent) of the stratum of vehicles
that failed only the purge test is given in Appendix A. These
estimates must be adjusted for the "gross liquid leakers" (see
Section 3.3).
In Figure 3, for vehicles failing only the purge test, we
plotted both the measured rates (calculated from Table 1 and
shifted by six months to compensate for July testing) and the
curve described by subtracting from one hundred percent the total
of equation (1) plus equation (2).
Figure 3
Comparison of Measured and Predicted Rates
For Vehicles Failing ONLY the Purge Test
(by vehicle age)
30%
20% --
TO
o:
10%--
(A
0)
0)
O)
D
Q_
0%
1 0 20
Vehicle Age (years)
30
The preceding graph indicates that the combination of equations
(1) and (2) is a very good fit for the measured rates for ages at
which at least 100 vehicles were sampled (i.e., through age 13).
-------
-13-
DRAFT
Also, for vehicles over 20 years of age, the predicted rate of
vehicles failing only the purge test is close to 12 percent which
closely approximates the results of those 66 tests from industry
programs (see fourth "bullet" on page 4).
3.1.4 Summary of Purge and the Pressure Failure Rates
Combining the predicted rates from Figures 1, 2, and 3 (or
Appendix A) into a single area graph produces the following:
Figure 4
Predicted Pressure and Purge Rates
(by vehicle age)
1 0 0 %
7 5 %
> 5 0 %
c
0)
3
. 25 % H
0 %
Fail ONLY Purge
Pass Both
Purge and Pressure
10 15
Vehicle Age (years)
2 0
2 5
3.1.5 Vehicles Failing Both the Purge and the Pressure Tests
When EPA analyzed the RTD data, it was determined that there
were insufficient test results to distinguish between the diurnal
emissions of vehicles that failed both the purge and pressure
tests from those that failed only the pressure test. Therefore,
those vehicles were combined into the single stratum of vehicles
that failed the pressure test (regardless of their performance on
the purge test). Since the purpose of this study is to develop
factors to weight together the estimates of the individual stratum
to predict the in-use fleet emissions, it was not necessary to
model frequency of the stratum of vehicles that failed both the
purge and pressure tests.
As a service to possible future researchers and modelers who
may require an estimate of the number of vehicles failing both the
-------
-14-
DRAFT
purge and pressure tests (and based on the same sample that
produced equations (1) and (2)) , EPA performed the following
analysis.
The approach was similar to the one used in Sections 3.1.1
and 3.1.2 in which a table containing the frequencies with the
corresponding ninety percent confidence intervals was created (see
Table 4).
Table 4
Estimating Rate of Failing BOTH the Pressure and Purge Tests
For 1971-95 Model Year Vehicles
With 90 Percent Confidence Intervals
Vehicle
Age
0
1
2
3
4
5
6
7
8
9
1 0
1 1
1 2
1 3
1 4
1 5
1 6
1 7
1 8
1 9
20
21
22
23
24
25
26
Sample
Size
242
1 ,522
1 ,377
1 ,587
1 ,430
1 ,619
1 ,568
1 ,428
1 ,020
814
458
254
235
169
94
58
33
1 5
6
1 7
22
1 2
1 2
1 5
1 0
1 0
7
Rate
0.0%
0.2%
0.0%
0.3%
0.8%
0.4%
0.6%
0.8%
1.3%
1.5%
2.6%
2.8%
6.0%
7.1%
7.4%
6.9%
15.2%
26.7%
0.0%
23.5%
4.5%
0.0%
16.7%
53.3%
30.0%
30.0%
57.1%
90 Percent
Confidence Interval
0.0% - 0.5%
0.0% - 0.4%
0.0% - 0.2%
0.0% - 0.5%
0.4% - 1.1%
0.1% - 0.6%
0.3% - 0.9%
0.4% - 1.2%
0.7% - 1.9%
0.8% - 2.2%
1.4% - 3.8%
1.1% - 4.4%
3.4% - 8.5%
3.9% - 10.4%
3.0% - 11.9%
1.4% - 12.4%
4.9% - 25.4%
7.9% - 45.4%
0.0% - 28.5%
6.6% - 40.5%
0.0% - 11.9%
0.0% - 17.7%
0.0% - 34.4%
32.1% - 74.5%
6.2% - 53.8%
6.2% - 53.8%
26.4% - 87.9%
-------
-15-
DRAFT
As the reader may note, some of the sample sizes in Table 4
do not match the supposedly same samples in the first three
tables. In the first three tables, vehicles that failed the
pressure test but did not have a successful purge test were
included in the stratum "fail pressure" and, thus, included in the
total as well. However, those same vehicles would not be included
in Table 4.
After adjusting for age, the "logistic growth" function that
best models the frequency of vehicle's failing both the pressure
and purge tests from Table 4 is given by the following equation:
Rate of Failing Both =
0.3536
1 + 414.96*exp[-0.32955*AGE]
In Figure 5, for vehicles failing both the purge and pressure
tests, we plotted both the measured rates (from Table 4, shifted
by six months to compensate for the July testing) and the curve
described by equation (3).
Figure 5
Comparison of Measured and Predicted Rates
For Vehicles Failing Both the Pressure and Purge Tests
(by vehicle age)
0)
O)
D
Q_
T3
c
TO
D
(A
(A
0)
O
CD
60%
40% --
20% --
0%
1 0 20
Vehicle Age (years)
30
Figure 5 suggests that equation (3) is a very good fit for the
measured rates for ages at which at least 30 vehicles were sampled
(i.e., through age 16). Also, for the vehicles over 20 years of
age, the data in Table 4 indicates that 20 out of 66 (30.3%) of
those vehicles over 20 years of age (with a mean age of 23.2
-------
-16- DRAFT
years) failed both the purge and pressure tests, and equation (3)
predicts that the failure rate would be 29.5 percent. Thus,
equation (3) is also a very good fit for the measured rates for
the older vehicles.
3.2 Modeling the Stratum of "Gross Liquid Leakers"
The set of vehicles identified as "gross liquid leakers"
varies depending upon which type of evaporative emission is being
considered. Earlier (see "bulletted" points on page 2), we
presented three definitions each based on one type of test (i.e.,
RTD test, hot soak test, and running loss test). These
definitions were developed in a recent report devoted exclusively
to the subject of "gross liquid leakers" [4]. In that report, EPA
produced the following two equations to predict the frequency of
"gross liquid leakers" occurring on the RTD and on the running
loss tests for evaporative emissions:
Rate of Gross Liquid Leakers
Based on RTD Testing = 1 + ^ 4^ 3,°e^l°Q236Q4 . A G E j ( 4 )
Rate of Gross Liquid Leakers
Based on Running Loss Testing = 1 + 120 * exp^O .4 * A G E ] ( 5 >
See reference [4] for the details of how these equations were
derived.
In that same report, EPA proposed that the vehicles
classified as "gross liquid leakers" on the hot soak test are the
same vehicles identified as gross liquid leakers on either the
running loss or RTD tests. (That is, the set of vehicles
classified as gross liquid leakers on the hot soak test is the
union of the set of vehicles classified as gross liquid leakers on
the RTD test with the set of vehicles classified as gross liquid
leakers on the running loss test.) Therefore, the rate of gross
liquid leakers as identified on the hot soak test would be the sum
of the two rates for the RTD testing and the running loss testing
minus the number of double counted vehicles (i.e., the product of
those two rates assuming these two categories are independent of
each other) . Using equations (4) and (5) plus the preceding
assumption, the predicted rates of "gross liquid leakers" were
calculated (for each of the three test types) and are plotted in
Figure 6 (on the following page) and appear in Appendix B.
-------
-17-
DRAFT
Figure 6
Predicted Occurrences of "Gross Liquid Leakers"
On Each of Three Tests of Evaporative Emissions
(by vehicle age)
Running Loss Test
1 0 20
Vehicle Age (years)
30
It is important to note that this model of the frequency of
gross liquid leakers is based on the assumption that modern
technology vehicles will show the same tendency toward gross
liquid leaks as do the older technology vehicles at the same age.
However, if the modern technology vehicles were to exhibit a lower
tendency to leak (due to the more stringent demands imposed by the
new evaporative emissions certification procedure as well as
heightened attention to safety, e.g., fuel tank protection and
elimination of fuel line leaks), the effect would be to replace
each of the three logistic growth functions with two or three
curves specific to different model year groups.
Since EPA has no data to indicate model-year specific rates,
EPA proposes to use the model illustrated in Figure 6, to estimate
the occurrence in the in-use fleet of these vehicles that have
substantial leaks of liquid gasoline (i.e., "gross liquid
leakers") for vehicles that were not designed to meet the new
enhanced evaporative test procedure (i.e., vehicles up through the
1996 model year along with some of the 1997 and 1998 model years).
For the vehicles designed to meet the new enhanced evaporative
test procedure, EPA proposes to modify that equation (see Section
4.0) .
-------
-18-
DRAFT
3.3 Combining Purge/Pressure Rates with Gross Liquid Leaker Rates
In Section 3.1 we characterized the three strata resulting
from the individual vehicle's performance on the purge and
pressure tests. In Section 3.2, we characterized the additional
stratum created for the "gross liquid leakers." In order to make
these strata non-overlapping (i.e., mutually exclusive), we must
remove the "gross liquid leakers" from the other three strata.
To determine the distribution of the gross liquid leakers
among the other three strata, we examined the 270 vehicles in the
combined EPA/CRC RTD testing programs. Seven vehicles were
identified as "gross liquid leakers" out of those 270 vehicles
that were tested. Of these seven gross liquid leakers:
four had failed both the purge and pressure tests,
two had failed only the pressure test, and
one had passed both the purge and pressure tests.
This distribution of seven gross liquid leakers proves that gross
liquid leakers can and do occur within the all three of the
purge/pressure strata. EPA proposes to first estimate the number
of gross liquid leakers within each of the purge/pressure strata
and then to remove them to form a fourth stratum consisting of
only the gross liquid leakers. Rather than attempting to estimate
the distribution of all gross liquid leakers based on a sample of
only seven vehicles, EPA proposes that MOBILE6 will simply
distribute the liquid leakers proportionately among the three
purge/pressure categories.
For example, if we were to take a hypothetical fleet of
10,000,000 vehicles 10 years of age, then Appendix B predicts that
780,000 (7.8 percent) of them will be "gross liquid leakers" on
the RTD test. Similarly, Appendix A indicates that 10,910,000
(10.91 percent) will fail the pressure test, 647,000 (6.47
percent) will fail only the purge test, and the remaining
8,262,000 (82.62 percent) will pass both tests. Distributing
those 780,000 "gross liquid leakers" proportionately among the
three purge/pressure strata predicts the following distribution:
Table 5
Predicted Distribution on RTD Test
Purge/Pressure
Strata
Fail Pressure
Fail ONLY Purge
Pass Both
TOTALS
"Gross Liquid
Leakers"
85,098
50,466
644,436
780,000
Not "Gross
Liquid Leakers"
1,005,902
596,534
7,617,564
9,220,000
TOTALS
1,091,000
647,000
8,262,000
10,000,000
-------
-19-
DRAFT
Thus, Table 5 indicates that for 10 year old vehicles on the RTD
test:
7.80 percent of those vehicles will be "gross liquid
leakers" on a RTD test,
10.06 percent of those vehicles will fail the pressure
test, but will not be "gross liquid leakers,"
5.97 percent of those vehicles will fail only the purge
test, but will not be "gross liquid leakers," and
76.18 percent of those vehicles will pass both the pressure
and purge tests, and will not be "gross liquid leakers."
Repeating this process for each vehicle age in Appendices A and B
produces the estimated size of each of the four strata for the RTD
test for each age. The results are plotted below:
Figure 7
Predicted Strata Sizes on RTD Test
(by vehicle age)
1 0 0 %
80% --
60% -
40% -
20% -
0 % -I
Pass Both
Purge and
Pressure
10 20
Vehicle Age (years)
3 0
-------
-20-
DRAFT
Repeating this process using the "running loss" and "hot
soak" columns from Appendix B will produce the estimates of the
size of the four strata for use with each of those two types of
evaporat ive emi s s ions.
4.0
Modeling Enhanced EVAP Vehicles Equipped with OBD
Beginning with the 1996 model year, manufacturers were
required to certify twenty percent of their vehicles using a new
"enhanced" evaporative testing procedure and to equip those
vehicles with an on-board diagnostic (OBD) system that would
detect pressure leaks and malfunctions in the purge system; that
percentage is scheduled to increase to one hundred percent by the
1999 model year. The phase-in percentages are prescribed in
40 CFR 86.096-8 and shown below in Table 6.
Table 6
Phase-In of Vehicles with
Enhanced Evaporative Controls
Model Year
1995
1996
1997
1998
1999
Percentage
0%
20%
40%
90%
100%
To predict the performance of these 1996 and newer vehicles
on the purge and pressure tests, the effects of two factors must
be considered:
1 ) A change in the regulations requires a doubling of the
period during which these vehicles must meet the
evaporative emissions standards (increased from 50,000
to 100,000 miles for light-duty vehicles).
2) The OBD system is expected to alert each vehicle's owner
(or driver) to most problems with the evaporative
control system, thus permitting the owner to decide
whether to repair the problem.
In order to meet the increased durability and more stringent
evaporative standards, manufacturers have implemented a number of
changes, including (but not limited to):
-------
-21- DRAFT
"quick connects" that reduce the possibility of improper
assembly when the vehicle is serviced,
advanced materials that are less permeable, less
susceptible to puncture, and more durable (i.e.,
elastomeric materials used in hoses and connectors),
improvements made to the purge system (to enable the
vehicles to pass both the running loss test and the
multi-day diurnal test),
tethered gas caps, and
improved fractional-turn gas caps.
Since these changes are expected to result in improved control of
evaporative emissions, EPA proposed in a separate report
(M6.EVP.005) to use a separate set of estimates of both resting
loss and diurnal emissions for these vehicles. However, EPA does
not have actual data on the effects of these changes in durability
that translates into changes in the purge and pressure failure
rates estimated in Section 3.1 for the pre-1996 model year
vehicles.
EPA, therefore, proposes to use the doubling in the
durability requirement to modify equations (1), (2), (4), and (5)
(from Sections 3.1.1, 3.1.2, and 3.2) by replacing the variable
"AGE" with "AGE/2" resulting in:
Pressure Failure Rate of Enhanced Evaporative Control Vehicles
0.6045
1 + 17.733*exp[-0.003405*(AGEA2)]
Rate of Passing Both for Enhanced Evaporative Control Vehicles
0.7200
(6)
13.40*exp[-0.003625*(AGEA2)]
(7)
Rate of Gross Liquid Leakers on the RTD Test for the Enhanced Evaporative
Control Vehicles
0.08902 (8)
1 + 414.613*exp(-0.1842*AGE)
Rate of Gross Liquid Leakers on the Running Loss Test for the Enhanced
Evaporative Control Vehicles
= P-06
1 + 1 20*exp[-0.2 * A G E ] v
-------
-22- DRAFT
Using these equations, we generated the estimated failure
rates in Appendix C for those (1996 and newer) vehicles certified
to the enhanced evaporative control standards. These estimates
assume only the benefits of changes in durability, with no
estimation of the effect from the OBD system.
Since the OBD system is designed to alert each vehicle's
owner to problems with the evaporative control system. The
appearance of a warning light should result in at least some
owners having their malfunctioning vehicles repaired. Thus, the
OBD system has the potential to affect the relative sizes of the
purge/pressure strata (in Appendix C), depending both on its
ability to identify problems in the evaporative control systems
and on the owners inclination to repair such problems.
In an separate analysis (M6.IM.001) of exhaust (not
evaporative) emissions, EPA proposed that the effect of an OBD
system will vary, based on the vehicle's warranty and the presence
of an I/M program. That proposal stated that:
The vehicle's malfunction indicator light (MIL) would
detect/identify 85 percent of the instances of the
vehicle's exceeding twice the exhaust emission standard.
While the vehicle is under its full ("bumper to bumper")
warranty (i.e., up through 36,000 miles), 90 percent of
the owners will have the vehicle repaired when the MIL
indicates a problem.
When the warranty covers only the electronic control
module and the catalytic converter (i.e., from 36,000
through 80,000 miles), only 10 percent of the owners
will have the vehicle repaired when the MIL indicates a
problem. That percentage would increase from 10 to 90
percent if that geographic area has an I/M program that
requires the MIL to indicate that there are no problems.
When the vehicle is no longer covered by a
manufacturer's warranty (i.e., over 80,000 miles), none
(i.e., zero percent) of the owners will have the vehicle
repaired when the MIL indicates a problem. That
percentage will increase from zero to 90 percent if that
geographic area has an I/M program that requires the MIL
to indicate that there are no problems.
EPA proposes to adapt those assumptions (modified slightly)
for evaporative emissions. Specifically:
The vehicle's malfunction indicator light (MIL) would
detect/identify 85 percent of the instances of the
vehicle's failing the purge test or failing the pressure
test. (It is assumed that the OBD system would not
detect the presence of a gross liquid leak.)
-------
-23- DRAFT
While the vehicle is within its full warranty period
(i.e., approximately through the age of three years), 90
percent of the owners will have the vehicle repaired
when the MIL indicates a problem. (These first two
assumptions suggest that the OBD system combined with
the manufacturer's warranty program will reduce the
incidence of vehicles failing either the pressure or
purge tests by 76.5 percent.)
While the vehicle is under its partial warranty period
(i.e., approximately ages four through six years), only
10 percent of the owners will have the vehicle repaired
when the MIL indicates a problem. (This assumption
suggests that the OBD system combined with the
manufacturer's warranty program will reduce the
incidence of vehicles that in their fourth, fifth, or
sixth years newly fail either the purge or pressure
tests.) That percentage would increase from 10 to 90
percent if that geographic area were to have an I/M
program requiring the MIL to indicate that there are no
problems.
When the vehicle's evaporative control system is no
longer covered by a manufacturer's warranty (i.e.,
beyond about six years of age), none (i.e., zero
percent) of the owners will have the vehicle repaired
when the MIL indicates a problem. That percentage would
increase from zero to 90 percent if that geographic area
were to have an I/M program requiring the MIL to
indicate that there are no problems.
Note: EPA is currently studying in-use OBD systems. The
assumptions listed here are working assumptions and
subject to change.
Applying these assumptions to the strata sizes in Appendix C
produces the tables in Appendices D and E that cover the 1999 and
later model year vehicles, for the I/M and non-I/M areas
respectively.
For the vehicles produced during the phase-in years of the
enhanced evaporative standard (i.e., model years 1996 through
1998), EPA proposes to use a three-step approach:
First, the fraction of the fleet not manufactured to the
new enhanced evaporative standard will be treated the same
as the 1986-95 model year vehicles. That is, the emissions
for each of the four purge/pressure/leaker strata will be
calculated and then weighted together (using Appendix A)
for each of the three years (1996-1998) .
Then, for the vehicles manufactured to the new enhanced
evaporative standard, we will calculate the predicted
-------
-24- DRAFT
emissions for each of the four purge/pressure/leaker
strata, and then weight them together using the rates in
either Appendix D (I/M area) or Appendix E (non-I/M area).
Finally, we will weight those two sets of results together
using the phase-in percentages in Table 6.
5.0 Comparisons with MOBILES
5.1 Comparisons of Weighting Factors
Both the MOBILES model and this proposal weight the estimated
evaporative emissions based on each vehicle's performance on purge
and pressure tests; however there are several structural
differences:
The weighting factors in MOBILES are each functions of a
continuous variable (i.e., each model year's average
odometer) while the weighting factors in this proposal are
functions of a discrete variable (i.e., each model year's
estimated age).
The weighting factors in this proposal are smooth functions
(i.e., exponential) of the estimated age. Although the
weighting factors in MOBILES are continuous, they are not
smooth functions (they are piece-wise linear).
MOBILES does not have a separate stratum for the vehicles
classified as "gross liquid leakers" while this proposal
does. We will, therefore, compare the MOBILES weighting
factors with the unadjusted factors from Section 3.1 of
this report.
The comparisons between the MOBILES weighting factors for
light-duty vehicles for each of the three purge/pressure strata
and the weighting factors in this proposal are illustrated in the
following three figures:
Figure 8 compares the estimates of vehicles passing both
the purge test and the pressure test. (Subtracting each of
those estimates from 100 percent will yield the associated
rates of vehicles failing either the purge test, or the
pressure test, or both tests.)
Figure 9 compares the estimates of vehicles failing the
pressure test (regardless of the performance on the purge
test).
Figure 10 compares the estimates of vehicles failing only
the purge test.
-------
-25-
DRAFT
Figure 8
Comparison of Predictions for Frequency of Vehicles
Passing Both the Purge and Pressure Tests
(by vehicle age)
1 0 0 %
> 5 0 %
c
0)
3
of 25%
0 %
'MOBILES
Proposal
10 15
Vehicle Age (years)
2 0
2 5
Figure 9
Comparison of Predictions for Frequency of Vehicles
Failing the Pressure Test
(by vehicle age)
60%
=^ 4 0 % - -
0 %
MOBILES
Proposal
10 15
Vehicle Age (years)
2 0
2 5
-------
-26-
DRAFT
Figure 10
Comparison of Predictions for Frequency of Vehicles
Failing ONLY the Purge Test
(by vehicle age)
40%
2 0 % - -
0)
3
0)
0 %
MOBILES
~ ~ ~ Proposal
10 15
Vehicle Age (years)
2 0
2 5
Based on examinations of these three graphs, we can make the
following observations:
The most obvious difference between the weighting factors
used in MOBILES and those being proposed in this report is
that the factors proposed in this report are capped (around
age 20) while the MOBILES factors are not.
Despite the structural differences (i.e., smooth function
of a discrete variable versus piece-wise linear function of
a continuous variable) between these two sets of weighting
factors, they produce similar results for vehicles up to
about 12 to 13 years of age.
The estimates being proposed in this report predict
substantially a higher proportion of vehicles failing the
pressure test for vehicles older than 13 years of age than
does the MOBILES model.
-------
-27- DRAFT
5.2 Comparisons of Weighted Diurnal Emissions
By combining the information in this report with the
information in earlier reports, we can estimate the diurnal
emissions for the in-use fleet.* To compare these proposed
estimates with those predicted by the MOBILES model, the MOBILES
model was run for a fleet with a national distribution of model
years (as of January 1, 1995) with two likely combinations of
temperature cycle and fuel RVP (assuming no weathering of the
fuel):
daily temperatures cycling between 60 and 84 degrees
Fahrenheit using fuel with a 9.0 RVP (see Figure 11) and
daily temperatures cycling between 82 and 106 degrees
Fahrenheit using fuel with a 7.0 RVP (see Figure 12).
Each MOBILES run generates estimated diurnal emissions for the 25
most recent model years, which for these runs were the 1971
through 1995 model years.
In Figures 11 and 12, the typical (i.e., mean) 24-hour
diurnal emissions for each vehicle are plotted against model year.
Visual inspections of Figures 11 and 12 suggest that the estimates
of diurnal emissions resulting from this new proposal (for each of
those two combinations of fuel RVP and temperature cycle) for each
model year are:
close (within a gram per vehicle per day) to MOBILES
estimates for the 1986 to 1995 model year vehicles,
typically 25 to 75 percent higher (for 9.0 and 7.0 RVP
fuels, respectively) than MOBILES estimates for 1980 to
1985 model year vehicles, and
typically 30 to 40 percent higher than MOBILES estimates
for 1972 to 1979 model year vehicles.
In these two examples the variable MODEL YEAR can be transformed
into AGE using the equation: AGE = 95 - (MODEL YEAR).
It is important to note that these (following) proposed estimates are only
of the full one-day diurnal emissions. They do include diurnal emissions
from "gross liquid leakers." But, they do not include evaporative
emissions from interrupted (i.e., partial-day) diurnals, nor from multi-
day diurnals, nor from running loss, nor from hot soaks. Estimates of
these excluded sources are being developed. A true comparison between
this proposal and MOBILES will require using activity data to weight
together all of these individual components of evaporative emissions.
-------
-28-
DRAFT
Figure 11
Comparison of Predictions for In-Use Diurnal Emissions by Model Year
60° to 84° F Cycle -- Using 9.0 RVP Fuel
(As of January 1, 1995)
4 0
in
E
o>
O
I
re
c
20--
MOBILE5
~ ~ ~ Proposed
7 0
7 5
80 85
Model Year
9 0
9 5
Figure 12
Comparison of Predictions for In-Use Diurnal Emissions by Model Year
82° to 106° F Cycle -- Using 7.0 RVP Fuel
(As of January 1, 1995)
4 0
in
E
re
o>
O
X
c
20--
V
7 0
7 5
80 85
Model Year
MOBILES
Proposed
9 0
9 5
-------
-29-
DRAFT
A similar comparison for the vehicles certified to the new
enhanced evaporative emission requirements (i.e., 1999 and newer
vehicles, along with some 1996-98 model year vehicles) is provided
in Section 5.3.
In Figures 11 and 12, we weighted each model year's estimated
diurnal emissions by the relative number of vehicles in the fleet,
we obtain the estimate of the mean daily diurnal emissions for an
average in-use vehicle subject to a full day's diurnal:
Temperature Cycle and Fuel
60° to 84° F with 9.0 RVP
82° to 106° F with 7.0 RVP
MOBILES
Estimate
4.86
7.61
New
Proposal
5.42
10.86
We repeated these calculations for several dozen combinations
of fuel RVP and temperature cycles and then graphed the resulting
averages in Figures 13 through 16. The first three figures
(Figures 13 through 15) are based on January 1, 1995 (thus
covering model years 1971 through 1995) . Figure 16 is based at
January 1, 1985 (thus, covering model years 1961 through 1985) .
Therefore, Figures 15 and 16 differ only by the model years
covered.
Since both the horizontal and vertical scales vary among
these four graphs, care should be taken in making comparisons
between these figures.
Figure 13
Comparison of Predictions for Mean 24-Hour Diurnal Emissions
60° to 84° F Cycle -- Per In-Use Vehicle
(As of January 1, 1995)
1 5
(A
| 1 0 -'
O)
O
X
- 5 4-
re
c
3
Q
MOBILES
Proposed
8 1 0
Fuel RVP (psi)
1 2
-------
-30-
DRAFT
Figure 14
Comparison of Predictions for Mean 24-Hour Diurnal Emissions
72° to 96° F Cycle -- Per In-Use Vehicle
(As of January 1, 1995)
3 0
(A
I*
O)
O
I
"re
c
3
Q
0 --
0 --
MOBILES
Proposed
8
Fuel RVP (psi)
1 0
Figure 15
Comparison of Predictions for Mean 24-Hour Diurnal Emissions
82° to 106° F Cycle -- Per In-Use Vehicle
(As of January 1, 1995)
4 0
(A
E
re
o>
O
I
"re
20--
8
1 0
Fuel RVP (psi)
-------
-31-
DRAFT
Figure 16
Comparison of Predictions for Mean 24-Hour Diurnal Emissions
82° to 106° F Cycle -- Per In-Use Vehicle
(As of January 1, 1985)
6 0
(A
re 4
o>
O
I
- 2
re
c
3
Q
0 --
0 --
MOBILES
Proposed
8
Fuel RVP (psi)
1 0
Based on examinations of these four graphs, we made the
following observations:
For the lowest temperature cycle (i.e., daily temperatures
cycling between 60° and 84°F), the newly proposed methods
of predicting fleet diurnal emissions produce results quite
similar to MOBILES for the full range of fuel RVPs.
(Figure 13.)
For the two higher temperature cycles, these newly proposed
methods of predicting diurnal emissions produce results
similar to MOBILES for fuel RVPs up to 10 psi, which is a
reasonable upper bound for in-use fuel RVP at those
temperatures. (Figures 14 through 16.)
The new approach (compared to MOBILES) predicts slightly
higher diurnal emissions for the lower RVP fuels and the
reverse (i.e., lower diurnal emissions) for the higher RVP
fuels. The range of RVPs for which the new estimates are
higher than the MOBILES estimates varies with the
temperature cycle:
For the low temperature cycle (i.e., 60° to 84°F), the
new approach predicts slightly higher diurnal emissions
for fuel RVPs up through 11 psi. For fuel RVPs near 7
-------
-32- DRAFT
psi, the new approach predicts less than 0.9 grams of
HC (above the MOBILES estimates) per day per vehicle
undergoing a full (24-hour) diurnal. This difference
gradually shrinks to zero as the fuel RVP nears 11 psi.
For fuel RVPs above 11 psi, the new estimates are
slightly lower than the MOBILES estimates.
For the moderate temperature cycle (i.e., 72° to 96°F),
the new approach predicts slightly higher diurnal
emissions for fuel RVPs up through 10 psi. For fuel
RVPs above 10 psi, the new estimates are lower than the
MOBILES estimates. For RVPs above 11 psi (an unlikely
occurrence with this temperature cycle), the two models
move farther apart.
For the high temperature cycle (i.e., 82° to 106°F),
the new approach predicts slightly higher diurnal
emissions for fuel RVPs up through 8.5 psi for the in-
use fleet as of January 1995 (Figure 15). For fuel
RVPs above 8.5 psi, the new estimates are lower than
the MOBILES estimates. For RVPs above 10 psi (an
unlikely occurrence with this temperature cycle), the
two models move farther apart. A snapshot of the in-
use fleet as of January 1985 (Figure 16) yields similar
results.
5.3 Comparisons of Diurnal Emissions from Vehicles
Certified to the Enhanced Evaporative Control Standards
In the preceding section, we compared, for 1995 and older
model year vehicles, these proposed estimates with those predicted
by the MOBILES model. In this section, we perform a similar
comparison of the two estimates for the vehicles certified to the
enhanced evaporative standard (i.e., the 1999 and newer along with
some 1996 through 1998 model year vehicles).
Repeating the process used to create Figures 11 and 12 but
with January 1, 2020 as the base produced the two following
figures (Figures 17 and 18). Visual inspections of Figures 17 and
18 leads to the following conclusions:
This proposal predicts substantially lower diurnal
emissions for these new vehicles than does MOBILES.
For vehicles between 10 and 25 years of age, this proposal
predicts a substantial rise in the diurnal emissions of the
in-use fleet (on a per vehicle basis).
This increase in emissions is driven primarily by the
increasing numbers of gross liquid leakers. With
improvements in the durability of evaporative emission
control systems, discussed in Section 4.0, the high
-------
-33-
DRAFT
emissions associated with "gross liquid leakers" eventually
drive the overall fleet average diurnal emissions curve.
This is despite the fact that their predicted rate of
occurrence among vehicles certified to the enhanced
evaporative standards is very low (less than two percent at
age 25).
Thus, modifying assumptions on the frequency of gross
liquid leakers as a function of age among these vehicles
would significantly affect the graphs for "Proposal" in
Figures 17 and 18, with that curve remaining much flatter
throughout. As we have no information of the impact of the
new enhanced evaporative control requirements on the rate
of occurrence of "gross liquid leakers," we are especially
interested in comments in this area.
Figure 17
For Enhanced Evaporative Control Vehicles
Comparison of Predictions for In-Use Diurnal Emissions by Age
60° to 84° F Cycle -- Using 9.0 RVP Fuel
(Non-l/M Area)
n
E
TO
^
O)
o
TO
c
D
b
6 --
3 --
10 15
Vehicle Age (years)
20
2 5
-------
-34-
DRAFT
Figure 18
For Enhanced Evaporative Control Vehicles
Comparison of Predictions for In-Use Diurnal Emissions by Age
82° to 106° F Cycle -- Using 7.0 RVP Fuel
(Non-l/M Area)
n
E
TO
^
O)
o
TO
c
D
b
6 --
3 --
10 15
Vehicle Age (years)
20
2 5
6.0
Summary
Estimates of evaporative emissions in MOBILE6 will be modeled
based on their type:
resting loss emissions,
running loss emissions,
hot soak emissions, and
diurnal emissions.
Each of these types will be calculated based on whether the
individual vehicles:
are gross liquid leakers,
failed the pressure test,
failed only the purge test, or
passed both the purge and pressure tests.
Once the estimated evaporative emissions of each sub-strata is
calculated, they will be weighted together using the equations
developed in this report.
-------
-35- DRAFT
Very preliminary analyses suggest that for the full-day
diurnal (see the footnote on page 27), the proposals for MOBILE6
would:
predict lower diurnal emissions than does MOBILES for the
1999 and newer vehicles,
predict similar diurnal emissions than does MOBILES for the
1986-1995 model year vehicles, and
predict higher diurnal emissions than does MOBILES for the
1985 and older vehicles.
Based on this proposal, we can perform a simplified analysis
to estimate the effect (benefit) on diurnal emissions of reducing
fuel RVP:
In each of the four figures (Figure 13 through 16), the
MOBILES graph is steeper than the graph of the new
approach. The lower slopes associated with the MOBILE6
proposal will result in smaller decreases in predicted
diurnal emissions associated with a change in fuel RVP than
the corresponding changes predicted by MOBILES.
Analyses performed to calculate the effect (e.g., either
the cost per ton or the benefit) will make RVP control
programs slightly less attractive.
The difference in the predicted effect of lowering fuel RVP
is small for lower RVP fuels. Thus, estimating the effects
of reducing the fuel RVP from 8 psi (or from a lower value)
will be similar under both methods.
-------
-36- DRAFT
7.0 References
1 ) Larry Landman, "Evaluating Resting Loss and Diurnal Evaporative
Emissions Using RTD Tests," Report numbered M6.EVP.001, July
1999.
2 ) Larry Landman, "Modeling Hourly Diurnal Emissions and
Interrupted Diurnal Emissions Based on Real-Time Diurnal Data,"
Report numbered M6.EVP.002, July 1999.
3 ) Larry Landman, "Modeling Diurnal and Resting Loss Emission from
Vehicles Certified to the Enhanced Evaporative Standards,"
Report numbered M6.EVP.005, November 1998.
4 ) Larry Landman, "Evaporative Emissions of Gross Liquid Leakers
in MOBILE6," Report numbered M6.EVP.009, June 1999.
5 ) Louis Browning, "Update of Hot Soak Emissions Analysis"
prepared by Louis Browning of ARCADIS Geraghty & Miller, Inc.
for EPA, Report numbered M6.EVP.004, September 1998
6 ) Larry Landman, "Estimating Running Loss Evaporative Emissions
in MOBILE6," Report numbered M6.EVP.008, June 1999.
7 ) D. McClement, "Real World Evaporative Testing of Late Model In-
Use Vehicles, CRC Project E-41", Prepared for the Coordinating
Research Council, Inc. by Automotive Testing Laboratories,
Inc., December 17, 1998.
8 ) D. McClement, "Raw Fuel Survey in I/M Lanes", Prepared for the
American Petroleum Institute and the Coordinating Research
Council, Inc. by Automotive Testing Laboratories, Inc., June
10, 1998.
-------
-37-
DRAFT
Appendix A
Estimates of Purge/Pressure Strata Size by Vehicle Age
For 1995 and Older Model Years Vehicles
(For Non-l/M Areas)
Vehicle
Age
(years)
0
1
2
3
4
5
6
7
8
9
1 0
1 1
1 2
1 3
1 4
1 5
1 6
1 7
1 8
1 9
20
21
22
23
24
25
Failing
Pressure
Test
3.23%
3.27%
3.40%
3.62%
3.96%
4.44%
5.10%
5.99%
7.18%
8.78%
10.91%
13.70%
17.30%
21.79%
27.12%
33.07%
39.19%
44.90%
49.76%
53.50%
56.16%
57.92%
59.02%
59.66%
60.03%
60.24%
Failing
Only
Purge
Test
1.77%
1.80%
1.88%
2.02%
2.23%
2.53%
2.95%
3.51%
4.25%
5.23%
6.47%
8.00%
9.76%
11.61%
13.29%
14.51%
15.06%
14.95%
14.41%
13.70%
13.03%
12.50%
12.13%
11.89%
11.74%
11.65%
Passing
Both
Purge and
Pressure
Tests
95.00%
94.93%
94.72%
94.36%
93.81%
93.03%
91.96%
90.51%
88.57%
85.99%
82.62%
78.30%
72.94%
66.60%
59.58%
52.42%
45.76%
40.14%
35.84%
32.80%
30.81%
29.58%
28.85%
28.45%
28.23%
28.11%
-------
-38-
DRAFT
Appendix B
Predicted Frequency of Occurrence of "Gross Liquid Leakers"
by Emission Type and Vehicle Age
For 1995 and Older Model Years Vehicles
(Reproduced from Report Number: M6.EVP.009 [4])
Vehicle
Age
(years)
0
1
2
3
4
5
6
7
8
9
10
1 1
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Resting
Loss /
Diurnal
0.02%
0.03%
0.04%
0.06%
0.09%
0.13%
0.19%
0.27%
0.39%
0.55%
0.78%
1 .08%
1 .49%
2.00%
2.63%
3.36%
4.15%
4.97%
5.75%
6.46%
7.05%
7.54%
7.91%
8.19%
8.40%
8.55%
Running
Loss
0.05%
0.07%
0.11%
0.16%
0.24%
0.35%
0.50%
0.72%
1.02%
1 .40%
1.88%
2.43%
3.02%
3.61%
4.16%
4.62%
5.00%
5.29%
5.51%
5.66%
5.77%
5.84%
5.89%
5.93%
5.95%
5.97%
Hot
Soak
0.07%
0.10%
0.15%
0.23%
0.33%
0.48%
0.70%
1.00%
1 .41 %
1.95%
2.64%
3.48%
4.46%
5.54%
6.67%
7.83%
8.95%
10.00%
10.94%
1 1 .75%
12.42%
12.94%
13.34%
13.63%
13.85%
14.00%
-------
-39-
DRAFT
Appendix C
Estimates of Purge/Pressure Strata Size by Vehicle Age
For 1999 and Later Model Years Vehicles
(For I/M Areas*)
Vehicle
Age
(years)
0
1
2
3
4
5
6
7
8
9
1 0
1 1
1 2
1 3
1 4
1 5
1 6
1 7
1 8
1 9
20
21
22
23
24
25
Failing
Pressure
Test
0.76%
0.76%
0.77%
0.78%
0.80%
0.82%
0.85%
0.89%
0.93%
0.98%
1.04%
1.11%
1.20%
1.29%
1.41%
1.54%
1.69%
1.86%
2.06%
2.30%
2.56%
2.87%
3.22%
3.62%
4.07%
4.57%
Failing
Only
Purge
Test
0.42%
0.42%
0.42%
0.43%
0.44%
0.46%
0.47%
0.50%
0.52%
0.56%
0.60%
0.64%
0.69%
0.75%
0.82%
0.91%
1.00%
1.11%
1.23%
1.37%
1.52%
1.69%
1.88%
2.08%
2.29%
2.51%
Passing
Both
Purge and
Pressure
Tests
98.83%
98.82%
98.81%
98.79%
98.76%
98.72%
98.67%
98.62%
98.54%
98.46%
98.36%
98.25%
98.11%
97.95%
97.77%
97.56%
97.31%
97.03%
96.71%
96.34%
95.92%
95.44%
94.90%
94.30%
93.64%
92.92%
This assumes that the I/M program requires repairs to vehicles with a MIL
that indicates that there is a problem.
-------
-40-
DRAFT
Appendix D
Estimates of Strata Size by Vehicle Age
For 1999 and Later Model Years Vehicles
(For Non-l/M Areas*)
Vehicle
Age
(years)
0
1
2
3
4
5
6
7
8
9
1 0
1 1
1 2
1 3
1 4
1 5
1 6
1 7
1 8
1 9
20
21
22
23
24
25
Failing
Pressure
Test
0.76%
0.76%
0.77%
0.78%
0.85%
0.94%
1.06%
1.21%
1.39%
1.61%
1.87%
2.18%
2.53%
2.94%
3.42%
3.97%
4.62%
5.36%
6.21%
7.20%
8.34%
9.64%
11.13%
12.82%
14.73%
16.86%
Failing
Only
Purge
Test
0.42%
0.42%
0.42%
0.43%
0.47%
0.53%
0.60%
0.70%
0.81%
0.95%
1.12%
1.31%
1.53%
1.79%
2.09%
2.44%
2.83%
3.29%
3.81%
4.40%
5.05%
5.78%
6.58%
7.44%
8.34%
9.27%
Passing
Both
Purge and
Pressure
Tests
98.83%
98.82%
98.81%
98.79%
98.68%
98.53%
98.34%
98.09%
97.79%
97.43%
97.01%
96.52%
95.94%
95.27%
94.49%
93.59%
92.55%
91.35%
89.98%
88.40%
86.61%
84.57%
82.29%
79.74%
76.92%
73.86%
Up through the age of three (3) years,
identical to those in Appendix C.
the values are
This assumes that either the geographic area has no I/M program or that
the existing I/M program does not include a check of the OBD MIL.
-------
-41-
DRAFT
Appendix E
Predicted Frequency of Occurrence of "Gross Liquid Leakers"
by Emission Type and Vehicle Age
For 1999 and Newer Model Years Vehicles
Vehicle
Age
(years)
0
1
2
3
4
5
6
7
8
9
10
1 1
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Resting
Loss /
Diurnal
0.02%
0.03%
0.03%
0.04%
0.04%
0.05%
0.06%
0.08%
0.09%
0.11%
0.13%
0.16%
0.19%
0.23%
0.27%
0.33%
0.39%
0.47%
0.55%
0.66%
0.78%
0.92%
1 .08%
1 .27%
1 .49%
1 .73%
Running
Loss
0.05%
0.06%
0.07%
0.09%
0.11%
0.13%
0.16%
0.20%
0.24%
0.29%
0.35%
0.42%
0.50%
0.61%
0.72%
0.86%
1.02%
1.20%
1 .40%
1.63%
1.88%
2.14%
2.43%
2.72%
3.02%
3.32%
Hot
Soak
0.07%
0.09%
0.10%
0.13%
0.15%
0.19%
0.23%
0.27%
0.33%
0.40%
0.48%
0.58%
0.70%
0.83%
1 .00%
1.19%
1.41%
1 .66%
1 .95%
2.28%
2.64%
3.04%
3.48%
3.96%
4.46%
4.99%
------- |