EPA-AA-SDSB-87-10
Technical Report
Effects of Gasoline Volatility on the
Hydrocarbon Exhaust Emissions
From a 1984 Oldsmobile Cutlass
By
Alan E. Schuler
August 19S7
NOTICE
Technical Reports do not necessarily represent final EPA
decisions or positions. They are intended to present
technical analysis of issues using data which are
currently available. The purpose in the release of such
reports is to facilitate the exchange of technical
information and to inform the public of technical
developments which may form the basis for a final EPA
decision, position or regulatory action.
Standards Development and Support Branch
Emission Control Technology Division
Office of Mobile Sources
Office of Air and Radiation
U. S. Environmental Protection Agency
-------
-2-
I. Executive Summary
This report describes the results of a gasoline
volatility/exhaust emission test program done under EPA
contract by ATL Laboratories in Columbus, Ohio. The program
was developed to help explain why HC exhaust emissions
frequently increase when using fuels with high volatility. The
testing involved a 1984 3.8L Olds Cutlass with multi-point fuel
injection. Two different fuels (9.0 and 11.5 psi RVP) were
used, along with three different conditions of the evaporative
canister (no purge, "standard" canister loading, and a loading
beyond breakthrough). In addition, tests were performed with
both fuels with the catalytic converter removed and a standard
canister loading to determine the effect of RVP on engine-out
emissions.
For this vehicle, it appeared that the increase in HC
exhaust emissions with increased fuel RVP is due to the
increase in vapors generated in the fuel tank during vehicle
operation. Canister weight (loading) did not affect exhaust
emissions. Thus, this increase in fuel tank vapors may help
explain the RVP differences seen in EPA's Emission Factor data
base.[l] This means that the temperature of the fuel tank
simulated during FTP testing may be more critical than
previously believed.
This report is divided into four major sections:
Background and Test Format; Test Results; Discussion of
Results; and Conclusions. The fuel summary sheets are included
in the Appendix.
II. Background and Test Format
As previously mentioned, the purpose of this study was to
determine why HC exhaust emissions increase with increased fuel
volatility. Related test programs indicate that unmetered fuel
may be the cause, although direct fuel effects could not be
confidently ruled out.[2,3,4] Most of this unmetered fuel was
thought to have come from the canister purge, with the rest
being generated in the fuel tank during vehicle operation. To
better evaluate the RVP/exhaust emission interaction, a test
program was developed calling for testing with 9.0 and 11.5 RVP
fuel, each with no purge, standard canister loading, and a
canister loading beyond breakthrough condition (hereafter
referred to as saturated though this was not strictly the
case). Under the no purge condition, unmetered fuel to the
engine from the fuel tank was also interrupted.
In addition to the canister variations described above, a
series of tests with the catalyst removed were performed using
a standard canister loading for both fuels. These engine-out
emission values would indicate whether the differences in HC
exhaust emissions occurred in the engine or catalyst. Assuming
-------
-3-
increased purged hydrocarbons (HC) to be the major cause of
increased HC tailpipe emissions, two mechanisms were considered
possible. According to the first mechanism, the increase in HC
emissions could be occurring mostly during combustion. The
catalytic converter - being primarily a proportional reduction
device - would then allow a proportional increase in tailpipe
emissions. According to the second mechanism, formation of the
excess hydrocarbons could be the same as with the smaller
purged HC quantities, but a reduction in excess oxygen could be
reducing catalyst efficiency, resulting in an increase in
tailpipe emissions.
During combustion, the hydrogen atoms in the fuel are
oxidized before the carbon atoms. Therefore, as the oxygen
level starts to decrease, incomplete oxidation will occur first
for the carbon atoms causing an increase in CO levels (with
little or no increase in HC levels) . Because of this greater
sensitivity, CO emissions from all tests were also evaluated
along with HC emissions.
Due to the large number of unknowns and the
correspondingly high test costs for a comprehensive program,
only one vehicle was tested at this stage of the evaluation.
To further reduce costs, hot-start LA-4's were used rather than
full FTPs. A revised version of this program using more
vehicles was left as a future option, depending on these test
results.
To select a vehicle for this test program, an analysis was
made of the test results of 308 cars from EPA's Emission Factor
(EF) data base to determine which, if not all, types of
vehicles have an exhaust emission sensitivity to fuel RVP. The
analysis of these vehicles showed that several groups of
vehicles had a fairly high statistical sensitivity to fuel RVP
(volatility). The most, notable were 3.8 L GM vehicles with
multi-point fuel injection (PFI). -A car was then chosen from
this group (a 1984 Oldsmobile Cutlass Cierra) for testing.
Before starting the actual test program, back-to-back hot start
LA-4's on both 11.5 and 9.0 RVP fuels were run at ATL to verify
the vehicle's continued sensitivity.
Each test condition was to receive a single replicate run,
except for the "baseline" (standard canister loading), which
was to receive two replicates. The second and third baseline
runs (for a given fuel) were made in between the no purge and
saturated canister runs to confirm the vehicle's repeatability.
The adaptive memory software in the vehicle's computer was a
particular concern, since it could be causing the exhaust
emissions of one test to be affected by the previous tests
conditions.
-------
-4-
To ensure consistent canister loadings, the canister was
artificially loaded prior to each test. A baseline canister
weight was obtained" by purging the canister for one hour at one
scfm, and then increasing the weight by 30/46 grams for a
"standard" load (9.0/11.5 RVP, respectively), and 80 grams for
the loading beyond breakthrough (both 9.0 and 11.5 RVP). The
standard load represents the estimated uncontrolled emissions
from one diurnal and one hot soak, plus an additional 10 grams
to account for the estimated residual in a "typical" in-use
canister.
In addition to measuring exhaust emissions and canister
weights, fuel tank, engine block, and engine oil temperatures
were also recorded, along with purge and manifold vacuum. All
tests were performed with the same driver and on the same
dynamometer.
The general testing sequence consisted of:
1. Drain vehicle fuel.
2. Weigh canister (record wet and dry bulb temps).
3. Purge canister for 1 hour at 1 scfm.
4. Weigh canister.
5. Install canister on portable fuel tank.
6. Load canister to desired weight by adding 75°F test
fuel to the portable tank. This pushes fuel vapors
out of the tank into the canister.
7. Weigh canister.
8. Fuel the vehicle to 40 percent of tank capacity
using 75°F test fuel.
9. Vent tank line to air intake of car and plug purge
line. (Canister is not installed on vehicle.) Push
vehicle onto dyno. LA-4 prep to warm-up engine and
catalyst.
10. 10-minute soak. During soak, install canister on
vehicle, connect tank and purge lines.
11. Hot-start LA-4. Measure emissions, traces on
temperatures and vacuums.
12. One hour soak in soak room. (Push vehicle there).
Immediately after soak, remove canister and weigh.
III. Test Results
Summaries of hot-start LA-4 emissions, temperatures, and
canister weights areas are shown in Tables 1, 2, and 3,
respectively. The emission results for bags 1 and 2 of the
LA-4 are shown in Table 4. The tests generally went as
planned. Four additional tests were run, usually to further
replicate questionable data. The retest for 9.0 RVP fuel with
no catalyst was due to a malfunction in the temperature chart
recorder. Since the "no purge" (NP) conditions did not involve
the charcoal canister, the one-hour post-run soaks were not
done.
-------
Table 1
Emission Test Results (g/mi)
1984 Olds Cutlass Cierra (3.8L PFI)
Condition
FTP
EPA Calc.Hot Start
(Bag 2&3)
Qualification
Retest
No Purge
Standard Loading
No Catalyst
Retest
Retest
Fuel
RVP Rep
11.5
9.0
11.5
9.0
11.5
9.0
Standard Loading 11.5
9.0
11.5
9.0
Saturated Loading 11.5
9.0
11.5
9.0
1
1
1
2
3
1
2
3
2
3
1
2
1
2
1
2
1
2
1
2
1
1
1
2
HC
.36
.23
.16
i n
CO
EPA EF
7.56
4.05
5.26
•) OK
NOx
C02
MPG
TEST RESULTS
.76
.64
.65
oq
449.8
461.4
19.15
18.94
ATL TEST RESULTS*
.15
.10
.19
.17
.13
.13
.11
.19
.13
.13
.13
.11
.10
.10
.13
.16
.11
.13
2.30
1.97
1.86
2.19
2.16
2.12
5.60
3.00
5.57
5,93
5.83
2.72
2.60
4.68
3.00
3.90
2.67
2.44
2.25
2.00
4.10
5.51
2.58
3.67
16.59
16.91
16.60
13.78
13.58
14.61
.65
.55
.70
.71
.71
.62
.54
.61
.54
.55
.68
.67
.64
.63
.70
.76
.62
.63
2.05
2.06
2.09
2.21
2.10
2.17
433.9
442.5
442.1
441.6
451.2
429.8
407.9
449.6
443.2
441.8
456.1
454.9
451.8
451.7
438.4
443.6
432.3
444.9
407.3
417.8
405.6
413.4
412.4
417.7
20.01
19.82
19.64
19.64
19.24
20.41
21.51
19.38
19.78
19.78
19.25
19.32
19.46
19.48
19.92
19.59
20.31
19.66
20.12
19.68
20.27
20.07
20.13
19.82
Odom
23512
23550
23512
23550
38431
38461
38490
38505
38610
38639
38713
38728
38742
38757
38520
38535
38654
38669
38580
38595
38683
38698
38831
38846
38860
38772
38787
38802
Run
Order
1
2
7
8
13
14**
13r
14 r
3
4
9
10
5
6
11
12
17
18
17r
15
15r
16
Test
Date
10/08/85
10/11/85
11/19/86
11/20/86
11/25/86
11/25/86
12/01/86
12/02/86
12/04/86
12/04/86
12/05/86
12/05/86
11/26/86
11/26/86
12/02/86
12/02/86
12/01/86
12/01/86
12/03/86
12/03/86
12/09/86
12/09/86
12/10/86
12/08/86
12/08/86
12/08/86
I
ui
LA-4 (2 bag) tests.
Test results excluded from statistical analysis.
-------
Table 2
Temperature Data
1984 Olds Cutlass Cierra
Condition
(jualif ication
Fuel
RVP
11.5
9.0
Standard Loading 11.5
9.0
Retest
No Purge
9.0
11.5
9.0
Saturated Loading 11.5
9.0
Standard Loading
Mo Catalyst
Retest
Retest
11.5
9.0
Rep
1
1
1
2
3
1
2
3
2
3
1
2
1
2
1
2
1
2
1
2
1
1
1
2
TEMPERATURES i
Fuel Tank
Soldered T/C
Test
Start
87
89
88
91
88
90
86
90
86
90
88
91
92
97
85
89
86
90
88
90
88
_
88
91
Test
End
97
98
97
100
98
100
98
102
98
100
98
100
103
107
95
99
98
100
97
98
98
_
96
99
Magnetic T/C
Test
Start/
87
88
87
91
88
89
86
90
85
90
87
90
91
96
84
89
86
90
88
89
88
87
90
Test
End
95
97
96
98
96
99
97
101
97
99
97
99
102
105
93
98
97
99
96
97
97
95
98
(°F)
Engine
Block
Test
Start
196
192
187
189
197
192
189
194
191
194
196
195
193
197
191
195
190
193
195
193
192
191
193
Test
End
185
. 183
181
182
186
187
183
189
184
179
180
182
185
184
186
185
183
183
185
181
183
181
180
Oil
Test
Start
199
197
192
200
197
198
193
198
194
201
195
204
200
205
193
199
194
202
200
202
194
_
201
199
Test
End
243
242
242
242
242
243
242
243
241
244
242
243
243
244
242
243
242
242
242
242
242
_
242
243
Run
Order
A
B
1
2
7
8
13
14
13r
14r
3
4
9
10
5
6
11
12
17
18
17r
15
15r
16
Test
Date
11/19/86
11/20/86
11/25/86
11/25/86
12/01/86
12/02/86
12/04/86
12/04/86
12/05/86
12/05/86
11/26/86
11/26/86
12/02/86
12/02/86
12/01/86
12/01/86
12/03/86
12/03/86
12/09/86
12/09/86
12/10/86
12/08/86
12/08/86
12/08/86
I
(Ti
-------
Table 3
Canister Weights*
1984 Olds Cutlass Cierra
'•.mdition
uilif ication
Purge
.mclard Loading
Catalyst
Retest
Retest
Fuel
RVP
11.5
9.0
•uKlard Loading 11.5
Retest
9.0
11.5
9.0
11.5
9.0
Rep
1
1
1
2
3
1
2
3
2
3
1
2
1
2
indard Loading 11.5 1
2
9,0 1
1
2
1
1
1
2
Fully
Purged
Weight**
791.24
799.39
788.14
788.14
799.03
793.35
788.55
788.55
788.55
788.55
___
793.15
793.15
789.54
789.54
798.42
798.40
798.42
791.68
791.88
791.68
Wet
Bulb
(°F)
61
61
61
60
58
57
59
59
59
59
_._
63
63
58
57
65
65
65
57
57
59
Dry
Bulb
<°F)
76
72
71
71
73
72
72
72
72
72
___
72
72
71
70
75
75
75
71
70
70
Begin
Test
Seq.
802.13
802.83
834.78
834.84
844.64
823.08
819.49
817.89
819.39
819.76
__.
864.44
864.67
850.88
850.54
844.90
844.51
844.34
821.98
821.75
822.00
Wet
Bulb
(°F)
62
62
62
63
58
59
58
58
59
56
56
59
57
56
65
64
61
58
57
58
Dry
Bulb
(°F)
76
72
77
74
71
73
71
72
69
70
...
72
70
70
71
75
74
71
70
70
70
End
Test
Seq.
816.86
801.07
845.70
851.79
851.72
818.14
815.17
819.64
823.17
819.47
_._
867.03
870.06
838.11
839.28
851.11
854.48
857.31
815.09
815.67
825.21
Wet
Bulb
(°P>
61
59
63
63
59
61
58
58
57
56
___
58
59
57
59
66
65
56
56
59
57
Dry
Bulb
(°F)
77
68
74
74
70
72
72
71
69
69
___
70
71
70
71
76
75
68
69
70
69
Weight
Change
+14.73
- 1.76
+10.92
+16.95
+ 7.08
- 4.94
- 4.32
+ 1.75
+ 3.78
- .29
___
+ 2.59
+ 5.39
-12.77
-11.26
+ 6.21
+ 9.97
+12.97
- 6.89
- 6.08
+ 3.21
Run
Order
A
B
1
2
7
8
13
14
13r
14r
3
4
9
10
5
6
11
12
17
18
17r
15
15r
16
All weights are in grams.
The canister was fully purged at the beginning of each day, except for days with retests
-------
Condition
Qualification
No Purge
Standard Loading
No Catalyst
Retest
Retest
Fuel
RVP
11.5
9.0
Standard Loading 11.5
Retest
9.0
11.5
9.0
Saturated Loading 11.5
9.0
11.5
9.0
-8-
Table 4
Emission Results by LA-4 Bag (g/mi)
1984 Olds Cutlass Cierra
Rep
1
1
1
2
3
1
2
3
2
3
1
2
1
2
1
2
1
2
1
2
1
1
1
2
Bag
HC
.095
.074
.121
.106
.108
.111
.083
.134
.105
.109
.117
.093
.097
.095
.081
.108
.077
.102
..956
.829
.808
.978
.945
.902
1
CO
2.
1.
2.
2.
2.
1.
1.
2.
1.
2.
2.
1.
1.
1.
1.
2.
1.
2.
7.
7.
7.
5.
5.
6.
74
46
32
44
94
42
33
59
61
37
05
90
82
64
61
68
19
04
27
30
43
76
67
50
Bag 2
HC
.058
.026
.067
.064
.052
.018
.024
.053
.029
.022
.017
.015
.005
.006
.047
.053
.028
.029
1.34
1.05
1. 14
1.22
1.22
1.22
CO
2.86
1.54
3.25
3.50
2.89
1.30
1.27
2.09
1.40
1.53
.62
.54
. -2
.36
2.49
2.83
1.39
1.62
9.32
9.18
9.61
8.02
7.90
8.11
Run
Order
A
B
1
2
7
8
13
14
13r
14r
3
4
9
10
5
6
11
12
17
18
17r
15
15r
16
Test
Date
11/19/86
11/20/86
11/25/86
11/25/86
12/01/86
12/02/86
12/04/86
12/04/86
12/05/86
12/05/86
11/26/86
11/26/86
12/02/86
12/02/86
12/01/86
12/01/86
12/03/86
12/03/86
12/09/86
12/09/86
12/10/86
12/08/86
12/08/86
12/08/86
-------
-9-
Scatter plots of the HC and CO emissions for the tests run
with the catalytic converter on the vehicle are shown in
Figures 1 and 2, respectively. In both figures, the value for
run 14 - standard loading, 9.0 RVP fuel - is
uncharacteristically high compared to the other indolene (9.0
RVP) values. Using the 9 actual test values for indolene (5
with std. loading, 2 with sat. loading, 2 with no purge), the
mean HC value is 0.125 g/mi with a standard deviation of
0.027. Since the 0.187 HC value for run 14 was greater than
two standard deviations above the mean (.179 g/mi), run 14 was
excluded from the statistical .analysis. Figures 3 through 6
are scatter plots of the HC and CO emissions from each bag of
the LA-4.
Although run 14 was statistically excluded from this
analysis, an investigation was made of the possible reasons for
the higher emissions. No correlation was found with canister
weights. However, run 14 had the highest ending fuel tank
temperature (except for the no purge tests which did not
include a 1-hour soak between runs), which may have caused the
unusually high emission results. (The effect of fuel tank
temperature on the emission results will be fully discussed
later). It is plausible that some change in test cell air.flow
occurred for this test, possibly a slightly misplaced fan — the
engine block temperatures are also high, which could alter the
test conditions. Based on the consistency, trends, and values
of the other results, a change in vehicle (fuel tank) cooling
is the most likely identifiable cause of the unusually high
emissions.
IV. Discussion of Results
The scatter plots shown in Figures 1 through 6 provide a
good visual indication of how the various conditions relate to
each other. The exhaust emissions were further analyzed using
a series of one-way analyses of variance (ANOVAs) with a 10
percent significance level, which compared the variances
between cases to those within cases. However, these statistics
must be considered carefully due to the small amount of data.
Also, ANOVAs assume normal distributions, which cannot be
confirmed with only single replicates at each condition.
A summary of the F-ratios from the one-way ANOVAs are
shown in Tables 5 and 6 for HC and CO emissions, respectively.
Analyses were made of the LA-4 composite emissions, Bag 1
emissions, and Bag 2 emissions. Figures 1 through 6 will be
used simultaneously with Tables 5 and 6 in the following
discussion.
As shown in Figures 1 and 2, the HC and CO emissions are
generally larger with 11.5 RVP fuel than with 9.0 RVP fuel.
Comparisons can also be made for the three canister conditions
-------
-10-
"iGURE 1
LA —4 HC Tailpipe Emissions
.-•-.
s
CD
"~
!-
5
.5
£
S
o
.<
u
0
I
u.^i -
0.21 -
0,20 -
0.19 -
0.18 -
0.1 7 -
0.16 -
0.15 -
0.14 -
0.13 -
0.12 -
0.1 1 -
0,10 -
0,09 -
Qual — Qualification, NP = No Purqe. Std «• Standard Loading, Sat — Saturated Loading
Std Std
a -*- i
i
Std
n
'Sot std
Quai. D
a
NP o Std
0 Sat Std ^ \ '^d
Of ^
NP - , -M
n ""dot JtO
Qual. N^" NP + +
+ • +
d-OS 1 I ! 1 i i 1 ! 1 I I 1 1 i I 1 ! 1
A B 1 2 Ji 4 5 6 7 8 9 10 11 12 13 14 1 3r 1 4r
AIL Run Order
1.5 RVP fuel
+ 9.0 RVP fuel
',00 -
URE 2
L^- — 4 CO Tai
Erni^sions
Qualification, NP — No Purge, *3td =• Standard Loading, Oot =• Sati rated LoadifXJ
5/jo H
4,00 -
Std
0 00
•• nn -
10
1.5 14 1 .5r ' 4r
-t- 90
-------
-11-
FIGURE 3
Bag 1 HC Tailpipe Emissions
1 J . 1 O —
0.15 -
0.14 -
'?
.3 0.13 -
',1
C
0
'3 0.1 2 -
VI
f
t
u
v O.1 1 -
VI
D
3
E
,3 o-io J
0
O.09 -
0.08 -
0.07 -
Qual — Qualification, NP — No Purge, Std — Standard Loading, Sat = Saturated Loading
Std
-f-
Std
n NP
D
Std
Sat Std 4- itd
std a a std *•
a Sat 4-
4-
NP
Q«l. Np ^ NP
S«
a Sat
Uuai, 4-
-f
1 i ! 1 ! 1 1 1 i i I ! 1 1 1 1 ! 1
A B 1 2 o 4 5 6 7 3 9 10 11 12 So 1 4 1 3r 1 4r
ATL Run Orcter
D 1 1,5 RVP fuel
+ 9.0 RVP fuel
5.UU
HGURL 4
Bag, ! CO Tailpipe Emissions
I Dual — Duohfkotion, NP = No Purge, Sid => Standard Loading, Sat = Saturated Loading
4 00 -}
Qual,
D
>.0u -4
! .00
std
a
Std
Oat
c
Std
G
Sat
Qua!.
4- NP
Std
Sid
8 1
ATL Run Order
1 !,5 RVP f
10 i 1 12 13 14 i 5r 1 4r
9.0 RVF1 fue-!
-------
-12-
FIGURE 5
Bag 2 HC Tailpipe Emissions
c'
s
ssions i
fc
UJ
*-»
VI
o
i
O
r
u . i o —
0.15 -
0.14 -
0.13 -
0.12 -
0.1 1 -
0.10 -
0.09 -
0.08 -
O.O7 -
0.06 -
0.05 -
O.04 -
0.03 -
0,02 -
0.01 -
0.00 -
Qual — Qualification, NP = No Purge, Std — Standard Loading, Sat =• Saturated Loading
"a std
Qual. Q
Q Sat -T^-J Std
Sat Da +
,- - Sat Sot ' Std
NP ,jp Std -t- 4-
0 'p "*"
MP NP
-t- -1-
i t t 1 ! ! 1 t 1 t 1 ! 1 1 I 1 1 1
A 3 1 2 o 4 5 6 7 3 9 10 11 12 13 14 13r 1 4r
ATL Run Order
11.5 RVP fuel
+ 9,0 RVP fuel
HGURL 6
8-jO 'j: CO Tailpipe Emissions
Quo. I — Qualification, NP ** No Purae. Std =- Standard Load! no. Sat — Saturated Loadina
id
0
Std
a
.3,00 -
Qual.
Q
(Jual.
Std
Sat
f
Gtd
MP
-------
-13-
Table 5
ANOVA Results of HC Emissions*
(10% Significance Level)
Comparison
Variance in Canister Loadings
(11.5 psi RVP fuel)
LA-4 Values
NP vs. Std. vs. Sat.
Std. vs. Sat.
Purge vs. NP
Bag 1 Values
NP vs. Std. vs. Sat.
Bag 2 Values
NP vs. Std. vs. Sat.
Std. vs. Sat.
Purge vs. NP
Variances in Canister Loadings
(9.0 psi RVP Fuel)
LA-4 Values
NP vs. Std. vs. Sat.
Bag 1 Values
NP vs. Std. vs. Sat.
Bag 2 Values
NP vs. Std. vs. Sat.
Std. vs. Sat.
Purge vs. NP
9.0 vs. 11.5 RVP Fuel
LA-4 Values
NP/Std./Sat. Subset
Std./Sat. Subset
No Purge Subset
No Catalyst Subset
Bag 2 Values
No Catalyst Subset
Sample
Size
7
5
7
7
7
5
7
3
8
8
6
8
15
11
4
6
Deg. of
Freedom**
2,4
1,3
1,5
2,4
2,4
1,3
1,5
2,5
2,5
2,5
1,4
1,6
1,13
1,9
1,2
1,4
F-ratio
5.21
2.97
5.08
1.02
34.2
3.04
41.1
2.21
.60
23.5
2.55
34.1
8.4
13.2
2.3
.80
Fo*** Conclusion
4.32 Differences
5.54 No Difference
4.06 Diff. is Purge
4.32 No Difference
4.32
5.54
4.06
3.78
3.78
3.78
4.54
3.78
3.14
3.36
8.53
4.54
Differences
No Difference
Diff. is Purge
No Difference
No Difference
Differences
No Difference
Diff. is Purge
Differences
Diff. w/fuels
No Difference
No Difference
1,4
.22
4.54 No Difference
*
**
***
Excludes run 14. NP means No Purge.
x,y = the degree of freedom (number of independent
observations) of the F distribution. x equals the degrees
of freedom for the number of cases studied, y equals the
degrees of free'dom for the number of results.
Fo = the theoretical F ratio for the given degrees of
freedom.
-------
-14-
Table 6
ANOVA Results of CO Emissions*
(10% Significance Level)
Comparison
Variance in Canister Loadings
(11.5 psi RVP fuel)
LA-4 Values
NP vs. Std. vs. Sat.
Std. vs. Sat.
Purge vs. NP
Bag 1 Values
NP vs. Std. vs. Sat.
Bag 2 Values
NP vs. Std. vs. Sat.
Std. vs. Sat.
Purge vs. NP
Variances in Canister Loadings
(9.0 psi RVP Fuel)
LA-4 Values
NP vs. Std. vs. Sat.
Bag 1 Values
NP vs. Std. vs. Sat.
Bag 2 Values
NP vs. Std. vs. Sat.
Std. vs. Sat.
Purge vs. NP
9.0 vs. 11.5 RVP Fuel
LA-4 Values
NP/Std./Sat. Subset
Std./Sat. Subset
No Purge Subset
No Catalyst Subset
Bag 2 Values
No Catalyst Subset
Sample
Size
7
5
7
Deg. of
Freedom** F-ratio
2,4
1,3
1,5
2,4
2,4
1,3
1,5
22.8
3.14
25.8
1.17
69.4
4.43
67.2
Fo*** Conclusion
4.32 Differences
5.54 No Difference
4.06 Diff. is Purge
4.32 No Difference
4.32
5.54
4.06
8
8
8
6
8
2,
2,
2,
1,
1,
5
5
5
4
6
2
58
103
.06
.03
.1
.80
3.78
3.78
3.78
4.54
3.78
15
11
4
6
1,13
1,9
1,2
1,4
8.79
34.3
5.38
66.4
3.14
3.36
18.51
4.54
Differences
No Difference
Diff. is Purge
1,4
93.7
4.54
No Difference
No Difference
Differences
No Difference
Diff. is Purge
Differences
Diff. w/ fuels
No Difference
Diff. w/fuels
Diff. w/fuels
* Excludes run 14. NP means No Purge.
** x,y = the degrees of freedom (number of independent
observations) of the F distribution. x equals the degrees
of freedom for the number of cases studied, y equals the
degrees of free'dom for the number of results.
*** Fo - theoretical F ratio for the given degrees of freedom.
-------
-15-
(no purge, "standard" canister loading, "saturated" canister
loading) within a given fuel type. With 9.0 RVP fuel, all
three conditions show basically the same HC exhaust emissions.
CO emissions for all three conditions were also in the same
range. With 11.5 RVP fuel, the HC results are more scattered
than the 9.0 RVP results. Also, the no purge condition (no
purge of the canister or fuel tank vapors) shows a reduction of
the HC emissions to the 9.0 RVP level. The difference in the
11.5 RVP CO emissions between purge and no purge is even more
pronounced. In addition to these comparisons, the F-ratios in
Tables 5 and 6 also show a basic difference in emission levels
between the 9.0 and 11.5 RVP fuels, and a difference between
purge and no purge conditions with 11.5 RVP.
The bag 1 HC and CO results show no difference between the
three conditions, for both 9.0 and 11.5 RVP fuel. However, the
bag 2 emissions shown in Figures 5 and 6 show a very definite
difference between purge and no purge conditions. The
difference between the theoretical and calculated F-ratios for
bag 2 (see Tables 5 and 6) statistically show a large
difference between the purge and no purge conditions. The
amount of canister loading appears to be irrelevant. In some
cases, the lowest of the purged emissions occur with a
saturated canister.
The difference between purge and no purge conditions,
regardless of canister loading, indicates that the fuel tank
vapors generated during vehicle operation are probably the
overriding factor in the emission increases. Table 2 (test
temperatures) shows approximately a 10°F increase in fuel tank
temperature during the LA-4. The temperature traces (not
shown) are approximately linear with run time. No data are
available to confirm that this temperature increase is
representative of in-use conditions for this particular
vehicle. However, both on-road and dynamometer testing of a
1986 Buick and a 1984 Plymouth have shown the current
dynamometer cooling arrangement to be more representative than
alternative arrangements which enhance cooling.[3,4]
As can be seen in Table 2, the starting and ending fuel
tank temperatures always increase by a few degrees when the
replicate run occurs on the same day. Except for the no purge
condition, the CO emissions also increase according to this
same testing pattern. The HC emissions are a little more
variable and do not show this trend.
A rise in fuel tank temperature would also have a larger
effect (generate more vapors) on a high volatility fuel ( i.e.,
11.5 RVP), along with having a larger effect during the latter
and hotter part of the run, (i.e., bag 2). Comparing the
figures for bag 1 and bag 2 (Figures 3 versus 5 and 4 versus
6), the differences in emissions between the two fuels is most
-------
-16-
apparent in bag 2. Once again, increased emissions from the
fuel tank due to the higher RVP fuel appear to be the cause of
the increased exhaust emissions.
The bag 2 tailpipe HC and CO emissions versus the ending
fuel tank temperatures are shown in Figures 7 and 8,
respectively. Run 14 is also included in the plots, since
temperature was suspected to have affected the results in the
first place. The no purge conditions are plotted separately
from the conditions which included purge. It is not known why
the no purge HC values are different for the two fuels.
However, these values are in a range where any uncontrolled
test variation could cause such a difference in HC emissions.
The NP CO emissions are approximately the same for both fuels,
with just a slight inverse relation to fuel tank temperature.
The reason for this apparent trend is not known and may be by
coincidence.
As previously discussed, the HC and CO values which
include purge are larger than the no purge values. With 9.0
RVP fuel, the HC emissions are too scattered to make a definite
correlation with fuel tank temperature. However, the CO
emissions appear to have a slight positive slope. This
CO/temperature correlation may actually be in the form of an
exponential curve, depending on the accuracy of run 14. With
11.5 RVP fuel, both HC and CO appear to have a positive slope
with respect to temperature. In all cases, more data is needed
to draw a strong conclusion.
Table 3 (canister weights) shows that the canister always
gained weight with 11.5 RVP fuel, whereas with 9.0 RVP fuel the
canister usually lost weight. The average change was 9.0 grams
for 11.5 RVP fuel and -3.8 grams for 9.0 RVP fuel. Therefore,
with 11.5 RVP fuel the canister absorbed more vapors from the
fuel tank during the run and the 1-hour post-LA4 soak than were
purged during the LA-4. With 9.0 RVP fuel, the opposite was
true. For the saturated test with 11.5 RVP fuel, the canister
weight gain was small, indicating the canister may have been
close to actually being saturated, or that equilibrium was
obtained between purge and adsorption.
The question still remains as to whether the increase in
hydrocarbon emissions occurred in the engine or catalytic
converter. .The emissions in Table 1 for the no catalyst test
show the engine-out HC levels are virtually equal for both
fuels, while the engine-out CO emissions for 11.5 RVP are
higher than for 9.0 RVP. The F-ratios included in Table 6 show
the CO emissions with 11.5 RVP fuel to be significantly higher
than the 9.0 RVP values. Therefore, the increase in tailpipe
HC emissions appears to be due to a drop in catalyst
efficiency,. since engine-out levels are not changing
significantly. The drop is catalyst efficiency is likely due
to reduced oxygen availability, as evidenced by higher
engine-out CO levels.
-------
a
3
j!
O
1
0,08
0.07 -
0.06 -
0.05 -
0,04 -
0.03 -
0.02 -
0.01 -
94
-17-
FIGURE 7
HC vs. Ending Fuel Tank Temp., Bag 2
NP - No Purge
I
96
I I I I
98
100
I
102
CJ 1 1.5 RVP
Ending Fuel Tank Temperature (F)
9.0 RVP 0- 11.5, NP
104
l
106
A 9,0, NP
FIGURE &
CO vs. Ending Fuel Tank Temp,, Bag 2
t
3.5 -
3 -
a
a 2.5 -
c
0
yi
£ 2 -
y
VI
| 1.5-
u
0
0 1 -
0,5 -
0 -
9
a
n NP - No Purge
a a
a
•f f
I
*
I | | | , | | | | ! 1 1
4 96 98 100 102 104 106
a 11,5 RVP
Ending Fuei Tonk Temperature (F)
9,0 RVP <> 1 1,5," NP
9.0, NP
-------
-18-
Thus, for this vehicle at least, this testing confirms the
assumption made in the Draft Regulatory Inpact Analysis in
support of in-use RVP controls that the effect was occurring in
the catalyst and that no fuel economy credit should accrue with
the elimination of these exhaust emissions.[5]
V. Conclusion
The testing of a 1984 Olds Cutlass conducted in this
program indicates that the increase in exhaust HC emissions
caused by high RVP fuel is occurring in the catalyst and not in
the engine. This was evidenced by the absence of an increase
in engine-out HC emissions. Engine-out CO emissions did
increase, indicating that the engine was running richer and
that oxygen levels in the catalyst were likely lower, thereby
decreasing the efficiency of the catalyst.
The richer running of the engine with higher RVP fuel was
apparently caused by an increase in fuel tank HC emissions
occurring while the engine was running. The effect was not
apparently due to the purge of vapors previously stored in the
evaporative control canister, since changing the loading of the
canister had no effect on exhaust HC emissions. To further
verify this conclusion, the amount of vapors generated in the
fuel tank over the course of a LA-4 could be measured directly.
Further testing of additional vehicles under a procedure
similar to that described in this memo may not be necessary,
since there may not be a need to precisely determine whether
the canister or the fuel tank is the primary cause of the RVP
effect. A control program which lowers in-use RVP will
directly reduce or eliminate both effects. A control program
which raises certification fuel RVP will also eliminate the
exhaust HC excess, regardless of whether the source of the
additional unmetered fuel is the canister or the fuel tank, as
long as the amount of vapors generated in the fuel tank and the
sum of all vapors sent to the canister are representative of
in-use amounts. The costs of the vehicle modifications should
also be similar since they will focus on the amount and control
of purge air during operation. In any event, given that the
fuel tank is definitely implicated by this test program, extra
care should be given to ensure that test cell cooling is
representative of in-use conditions.
-------
-19-
References
1. "Relationship Between Exhaust Emissions and Fuel
Volatility," EPA memo from Thomas L. Darlington to Charles L.
Gray, EPA, QMS, ECTD, June 24, 1985.
2. Final Weekly Report; "Task l - EPA/ATL
Correlation/Temperature Effects," EPA Motor Vehicle Emission
Laboratory, Ann Arbor, MI, June 20, 1986.
3. Letter to API from Exxon Research and Engineering
Company, March 19, 1986.
4. "Effect of Auxiliary Cooling on Fuel Tank
Temperatures," EPA Memorandum from Edward Earth to Robert
Maxwell, February 21, 1986.
5. "Draft Regulatory Impact Analysis; Control of
Gasoline Volatility and Evaporative Hydrocarbon Emissions from
New Motor Vehicles," QMS, OAR, EPA, July 1987.
-------
-20-
Appendix
-------
-21-
ENGINEERING OPERATIONS DIVISION
Fuel Analysis Report
ATL Test Fuel
Supplier: AMOCO
Proposed Use('s): Emission Factors
Quantity: 1 gallon sample Location: ATL AMOCO Indolene
Date placed in service: Nov-86
Date of resupply: End of Program
Analysis by: CORE
Item
RVP (psi)
Distillation
Initial Boiling Point(°F)
10% Evap. Point (°F)
50% Evap. Point (°F)
90% Evap. Point (°F)
End Point (°F)
%Evaporated at 160°F
Blend
Method Specifications
ASTM D 323
ASTM D 86
8.7-9.0
(a)
(a)
(a)
(a)
(a)
(a)
Official
EOD Values
9.0
92
133
218
318
439
20.4
HC Composition
Olefins (vol%)
Aromatics (vol%)
Saturates (vol%)
ASTM D 1319
(a)
(a)
(a)
4.4
24.2
71.4
ASTM D 3343
ASTM D 3338
ASTM D 1298
Weight Fraction Carbon
Net Heat of Combustion
(BTU/lb)
Specific Gravity (60°F/60°F)
Fuel Economy Numerator
(grams carbon/gallon)
Fuel Economy Numerator with R Factor
(grams carbon/gallon)
(a)
(a)
(a)
(a)
(a)
0.8628
18539
0.7428
2421
2414
(a) No requirements or not addressed
Prepared by:
Validated by:
Date:
Date:
-------
-22-
Supplier: AMOCO
Proposed Use('s): Emission Factors
Quantity: 1 gallon sample Location: ATL AMOCO Indolene
Date placed in service: Nov-86
Date of resupply: End of Program
Analysis by: CORE
ITEM -—METHOD-— -----
RVP (PSI) ASTM D 323
Distillation ASTM D 86
initial boiling point
5% evaporated
10% evaporated
20% evaporated
30% evaporated
40% evaporated
50% evaporated
60% evaporated
70% evaporated
80% evaporated
90% evaporated
95% evaporated
end point
evaporated at 160 °F
Sulfur ASTM D 1266
Lead ASTM D 3237
Manganese AA
Phosphorous ASTM D 3231
Water (Wt%) Karl Fischer
Hydrocarbon Composition ASTM D 1319
olefins
aromatics
saturates
Research octane number
Motor octane number
Antiknock index
Sensitivity
Weight fraction carbon
Weight fraction carbon
Weight fraction carbon
Net heat of combustion
Net heat of combustion
API GRAVITY
Specific gravity (60°F/60°F)
Fuel economy numerator (g carbon/gal)
Fuel economy numerator (g carbon/gal) with R Factor
ASTM D 2699
ASTM D 2700
ASTM D 439
RON-MON
ASTM D 2789
ASTM E 191
ASTM D 3343
ASTM D 240
ASTM D 3338
ASTM D 1298
ASTM Official
R R EOD
Values
9.0 9.0 0.55
92
118
133
159
184
204
218
229
241
264
318
364
439
20.4
100
4.4
24.2
71.4
0.0
0.0
0.8628
18539
59
0.7428
2421
2414
9.0
92
118
133
159
184
204
218
229
241
264
318
364
439
20
0
0
4.4
24.2
71.4
0.0
0
0.0
0.0
0.0009 0.8628
20
0.4
18539
59.0
0.7428
2421
2414
-------
-23-
ENGINEERING OPERATIONS DIVISION
Fuel Analysis Report
ATL Test Fuel
Supplier: Chevron
Proposed Use('s): Emission Factors
Quantity: 1 gallon sample Location: ATL Chevron UL7CQ
Date placed in service: Nov-86
Date of resupply: End of Program
Analysis by: CORE
Item
RVP (psi)
Distillation
Initial Boiling Point(°F)
10% Evap. Point (°F)
50% Evap. Point (°F)
90% Evap. Point (°F)
End Point (°F)
%Evaporated at 160°F
Blend
Method Specifications
ASTM D 323
ASTM D 86
11.6-11.9
(a)
(a)
(a)
(a)
(a)
(a)
Official
EOD Values
11.6
74
112
204
343
423
31.7
HC Composition
Olefins (vol%)
Aromatics (vol%)
Saturates (vol%)
ASTM D 1319
(a)
(a)
(a)
5.5
28.1
66.4
ASTM D 3343
ASTM D 3338
ASTM D 1298
Weight Fraction Carbon
Net Heat of Combustion
(BTU/lb)
Specific Gravity (60°F/6p°F)
Fuel Economy Numerator
(grams carbon/gallon)
Fuel Economy Numerator with R Factor
(grams carbon/gallon)
(a)
(a)
(a)
(a)
(a)
0.8649
18484
0.7416
2423
2423
(a) No requirements or not addressed
Prepared by:
Validated by:
Date:
Date:
-------
-24-
Supplier: Chevron
Proposed Use('s): Emission Factors
Quantity: 1 gallon sample Location: ATL Chevron UUCQ
Date placed in service: Nov-86
Date of resupply: End of Program
Analysis by: CORE
ITEM —METHOD-—
RVP (PSI) ASTM D 323
Distillation ASTM D 86
initial boiling point
5% evaporated
10% evaporated
20% evaporated
30% evaporated
40% evaporated
50% evaporated
60% evaporated
70% evaporated
80% evaporated
90% evaporated
95% evaporated
end point
evaporated at 160 °F
Sulfur ASTM D 1266
Lead ASTM D 3237
Manganese AA
Phosphorous ASTM D 3231
Water (Wt%) Karl Fischer
Hydrocarbon Composition ASTM D 1319
olefins
aromatics
saturates
Research octane number
Motor octane number
Antiknock index
Sensitivity
Weight fraction carbon
Weight fraction carbon
Weight fraction carbon
Net heat of combustion
Net heat of combustion
API GRAVITY
Specific gravity (60°F/60°F)
Fuel economy numerator (g carbon/gal)
Fuel economy numerator (g carbon/gal) with R Factor
Chromaspec ASTM
R R
Official
EOD
Values
ASTM D 2699
ASTM D 2700
ASTM D 439
RON-MON
ASTM D 2789
ASTM E 191
ASTM D 3343
ASTM D 240
ASTM D 3338
ASTM D 1298
0
11.6 11.6 0.55
74
102
112
132
156
185
204
235
251
291
343
376
423
31.7
11.6
74
102
112
132
156
185
204
235
251
291
343
376
423
32
0
0
100
5.5 5.5
28.1 28.1
66.4 66.4
0.0
0
0.0 0.0
0.0 0.0
0.8649 0.0009 0.8649
18484 20 18484
59.3 0.4 59.3
0.7416 0.7416
2423 2423
2423 2423
------- |