EPA-AA-IMS/PS-80-3
Analysis of Propane Gain
Used as a Screen on the
Portland Element III Sample
March, 1980
NOTICE
Technical Reports do not necessarily represent final EPA decisions or posi-
tions. 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.
David Brzezinski
Inspection/Maintenance Staff
Emission Control Technology Division
Office of Air, Noise, and Radiation
U.S. Environmental Protection Agency
-------
Propane gain values have been collected from 292 Portland Element III
(1975-77 model year) vehicles along with as-received FTP and idle test
emission measurements. This data is used here to compare the expected
failure rate, errors of commission, and the amount of FTP excess emissions
identified if one of two sets of idle values or one of four propane gain
values had been used as a cutpoint to screen the vehicles in the sample with
high emissions. Table 1 contains the results when all 292 vehicles are
combined into a single group and the cutpoints applied. Table 2 applies the
cutpoints to only those 127 vehicles (43.5%) which are equipped with air
injection systems and Table 3 applies the cutpoints to the remaining 165
vehicles (56.5%), which are without air injection systems.
The Portland Element III sample was chosen to represent a wide range of
emission control technology types existing in the Portland vehicle popula-
tion. This means that in some cases low selling vehicles with a unique
emission control technology type may be somewhat over-represented in the
Element III sample. A complex weighting scheme allows this sample to be
used to represent a Portland or National vehicle mix when desired. The
values calculated in this analysis, however, have not been weighted to
represent any known vehicle mix. The uncertainty introduced by not weight-
ing the vehicle sample is not expected to change the observed trends or
conclusions reached in this analysis.
The idle cutpoints referred to in this analysis as "Houston" idle cutpoints
are the proposed cutpoint values to be used in the Houston pilot I/M pro-
gram. The "Portland" idle cutpoints are the most typical cutpoint values
used in the current Portland, Oregon Department of Environmental Quality
(DEO) I/M program. Neither of these sets of idle cutpoints represent sug-
gestions by EPA, but only represent points for discussion of the relative
merits of idle and propane test procedures.
Vehicles whose measured idle exhaust emissions are greater than either the
idle hydrocarbon (HC) or idle carbon monoxide (CO) cutpoint or both fail the
idle test. Vehicles whose measured propane gain values in neutral are less
than the propane gain cutpoint fail the propane test.
Errors of commission are traditionally defined as those vehicles which pass
their FTP test standards, both for hydrocarbons (HC) and carbon monoxide
(CO), but which are failed by a particular I/M short test. For example, in
Table 1, 5.1% (15 vehicles) of the sample had measured HC emissions of 1.5
gm/mi or less and CO emissions of 15 gm/mi or less, but had either a mea-
sured idle exhaust HC value more than 225 ppm or a measured idle exhaust CO
value more than 1.0%. This value (5.1%) represents the errors of commission
(as traditionally defined) in this sample using the "Portland" idle cut-
points. It is recorded in Table 1 under the Errors of Commission heading in
the "FTP HC & CO" column. The "FTP HC" and the "FTP CO" columns under the
Errors of Commission heading in Tables 1-3 represent a slightly different
view of errors of commission. The rates in the "FTP HC" column represent
those vehicles which pass the HC portion of the FTP test but fail a short
-------
test. The rates In the "FTP CO" column represent those vehicles which pass
the CO portion of the FTP test but fail a short test. Some of these vehi-
cles which pass the FTP for one pollutant may have failed the FTP for the
other pollutant and therefore would have failed the overall FTP standards.
As a result, these rates will be higher than the rates of error of commis-
sion involving vehicles which pass both the HC and CO portions of the FTP
test.
FTP excess emissions are the total gm/mi emissions from the sample fleet in
excess of their FTP standards. The FTP excess emissions identified by a
short test is the sum of the FTP excess emissions of vehicles which fail the
short test. The percent of FTP excess emissions identified is one measure
of its effectiveness as a screen to identify vehicles which need emission
related repair and/or adjustment. The more accurately a short test selec-
tively fails the dirtiest cars, the higher will be the percent excess emis-
sions identified.
Using a single propane test cutpoint of 10 rpm on all vehicles based on this
sample, the failure rate would be about 30% (Table 1). Using an idle test
and the "Houston" idle cutpoints for all 1975-77 vehicles the failure rate
would drop to about 22%, yet the excess emissions identified would increase
and the errors of commission would reduce significantly. This shows that,
when applied to all 1975-77 vehicles as a group, the "Houston" idle test
might prove to be a more effective screen than the propane test. However,
if we divide the sample into two groups, with and without air injection
systems (Tables 2 and 3), it is clear that the problems with the propane
test primarily lie in its inability to screen vehicles with high emissions
which have air injection systems. Table 4 compares the average FTP results,
FTP failure rates, and total excess FTP emissions of the vehicles in the
sample to those vehicles in the sample with and without air injection sys-
tems.
The error of commission rates for the propane test are higher for vehicles
with air pumps (Table 2) than for vehicles without air pumps (Table 3). In
Table 3, if we compare the propane cutpoint of 20 rpm with the "Houston"
idle cutpoints on vehicles without air pumps; we can see that the propane
test as a screen for high emitters is in this case very comparable to the
idle test. Both have low error of commission rates and have fairly equal
failure rates (about 30%). Even the FTP excess emissions identified are
fairly close. The "Houston" idle test appears to have a tendency to iden-
tify vehicles with excess HC emissions, while the propane test, which only
checks carburetor adjustment, appears to have a tendency to screen more
vehicles with excess CO emissions than excess HC emissions.
Vehicles with air injection systems (Table 2) present problems for both the
idle test and the propane test. For example, the data indicate that, in
order to identify 40 percent of the excess HC d-iissions from cars with air
pumps using the propane test, over 40% of then would have to be failed,
using a propane gain cutpoint of 30 rpm. , (Passing and failing cars at the
random would on average identify the same percent excess emissions at the
same failure rate). The error of commission rate (18.1%) at that propane
gain cutpoint is quite high. The error of commission rate (0.0%) for the
-------
"Houston" idle outpoints Is small, however the percentage of excess HC
emissions identified is also small. This indicates that the "Houston" idle
cutpoints, while avoiding errors of commission much more successfully than
the propane test, have failed to screen out as many of the air pump vehicles
which exceed FTP standards no doubt in large part because of the lower
failure rate (9.4%). More vehicles would need to be failed to allow a
substantial reduction in emission from those vehicles. The "Portland" idle
cutpoints fail more cars (35.4%) and are more effective for these vehicles
in terms of excess emissions; the error of commission rate (5.5%) is higher
for this idle test but still less than one half the error of commission rate
for any of the propane tests.
A possible reason for the higher error of commission rates using the propane
test on vehicles equipped with air injection systems could be the combined
effect of air injection in catalyst-equipped vehicles and the typical car-
buretor adjustment specifications of these vehicles. Air injection causes
the catalyst to work more efficiently. The manufacturers who install air
pumps can often readjust their carburetor specifications richer to improve
driver-perceived performance and still pass FTP standards. These richer
carburetor specifications result in small or zero propane gain values.
Therefore vehicles with air pumps are over-represented in the group of
vehicles which when adjusted to manufacturers specification have measured
propane gain values which are below a given propane test cutpoint.
A propane I/M program testing vehicles with air injection systems might
cause owners of many air injection vehicles to adjust their carburetors
leaner than recommended by the manufacturer in order to pass the propane
gain cutpoint. The leaner carburetor setting probably will result in some
additional reduction in emissions from those cars and some fuel economy
gains, possibly without causing any significant performance problems. If
this is the case, no serious problems would arise. If serious problems
(e.g., a group of vehicles is consistently unable to pass the propane cut-
point with acceptable driveability, leading to public dissatisfaction and
ad just-readjust behavior), do arise because of propane testing of vehicles
with air pumps, perhaps those vehicles could be exempted from propane test-
ing altogether. Vehicles with air injection systems tend to have lower
emissions than vehicles without air pumps (Table 4), even when maladjusted.
It is estimated that vehicles with air pumps account for less than 20% of
FTP excess emissions from all vehicles. The necessary reductions in overall
automotive pollutants could be gained from other vehicles which, without air
pumps, tend to have much higher emissions when maladjusted. Another approach
to vehicles which have air injection systems might be some combination of
propane testing, to determine the carburetor adjustment, and an idle test to
gauge the overall vehicle's exhaust emissions.
The trends and values here were all based on the observations of the 292
Portland Element III vehicles. It should be noted that the Portland cars
may be lower emitting than vehicles in areas without I/M programs and that
consequently, the failure rates presented here may be lower than would occur
in a non-I/M area. Nevertheless, these results are currently EPA's best
estimate of how an actual propane test would fare when applied to a much
larger sample of vehicles. New observations may surface in the existing or
new data which may affect any conclusions presented here.
-------
Table 1
Portland Element III
1975-77 Model Years
(292 Vehicles)
Short Test
FTP Excess Emissions Identified
Errors of Commission (%)
Failures3 (%)
Idle
Cutpoint
"Houston" . 21.6
9
"Portland" 43.5
Propane Gain
Cutpoint
10 29.5
20 31.8
30 ' 37.7
40 40.8
1: Idle CO cutpoint = 3.5%;
2: Idle CO cutpoint = 1.0%;
3: Idle exhaust measurement
HC(%) C0(%) FTP
65.0 62.2
82.5 82.5
55.3 58.2
56.2 61.5
62.1 67.1
64.7 70.7
Idle HC cutpoint = 400 ppm.
Idle HC cutpoint = 225 ppm.
HC
0.
5.
6.
6.
8.
9.
greater than either the Idle CO or Idle HC cutpoint
Test; Propane Gain measurement less than the Propane Gain cutpoint fails
4: Vehicles which pass both
5: Vehicles which pass the
6: Vehicles which pass the
the FTP HC and CO standards but fail the short
CO portion of the FTP but fail the short test.
HC portion of the FTP but fail the short test.
the
test
& CO4 FTP CO5 FTP HC6
0 0.0 1.4
1 7.5 8.9
5 8.2 9.9
5 8.2 11.6
9 10.6 14.7
6 11.3 15.8
fails the Idle
Propane Test.
•
Corrected February 20, 1980
-------
Table 2
Portland Element III
1975-77 Model Years With Air Pumps
(127 Vehicles)
Idle
Cutpoint 1
"Houston" ,
"Portland""
Short Test
3
Failures
9.4
35.4
FTP Excess Emissions Identified
HC(%)
30.8
74.0
Errors of Commission (%)
35.5
78.4
FTP HC & CO
0.0
5.5
FTP CO
0.0
9.4
FTP HC
2.4
10.2
Portland Gain
Cutpoint
10 33.9
20 36.2
30 41.7
40 44.9
30.1
32.3
41.5
48.2
42.7
46.0
49.4
54.9
15.0
15.0
18.1
18.9
18.9
18.9
22.0
22.8
22.0
23.6
26.8
27.6
1: Idle CO cutpoint = 3.5%; Idle HC cutpoint = 400 ppm.
2: - Idle CO cutpoint = 1.0%; Idle HC cutpoint = 225 ppm.
3: Idle exhaust measurement greater than either the Idle CO or Idle HC cutpoint fails the Idle
Test; Propane Gain measurement less than the Propane Gain .cutpoint fails the Propane Test.
4: Vehicles which pass both the FTP HC and CO standards but fail the short test.
5: Vehicles which pass the CO portion of the FTP but fail the short test.
6: Vehicles which pass the HC portion of the FTP but fail the short test.
Corrected February 20, 1980
-------
Table 3
Portland Element III
1975-77 Model Years Without Air Pumps
(165 Vehicles)
Idle
Outpoint .
"Houston" ,
"Portland""
Short Test
Failures (%)
30.9
49.7
FTP Excess Emissions Identified
Errors of Commission (%)
HC(%)
75.8
89.1
C0(%)
70.7
85.6
FTP HC & CO
0.0
4.8
FTP CO"
0.0
6.1
FTP HC
0-.6
7.9
Propane Cain
Cutpoint
10
20
30
40
26.1
28.5
34.5
37.6
62.3
63.7
68.7
69.8
63.0
66.4
72.6
75.8
0.0
0.0
1.8
2.4
0.0
0.0
1.8
2.4
0.6
2.4
5.5
6.7
1: Idle CO cutpoint = 3.5%; Idle HC cutpoint = 400 ppm.
2: Idle CO cutpoint = 1.0%; Idle HC cutpoint = 225 ppm.
3: Idle exhaust measurement greater than either the Idle CO or Idle HC cutpoint fails the Idle
Test; Propane Gain measurement less than the Propane Gain cutpoint fails the Propane Test.
4: Vehicles which pass both the FTP HC and CO standards but fail the short test.
5: Vehicles which pass the CO portion of the FTP but fail the short test.
6: Vehicles which pass the HC portion of the FTP but fail the short test.
Corrected February 20, 1980
-------
c
in
m
a
s
m
5
2
z
-«
1
o
11
•n
o
m
(•^
00
(T>
Ol
1
t-^
N>
C
Ul
With
Air Pump
Without
Air Pump
Total
Sample
Percent of
Sample
43.5
56.5
100.0
Average FTP
H£
1.52
2.15
1.88
Portlai
1975-
(29
\ £~ J
(gm/mi)
CO
18.6
30.4
25.2
Table 4
(292 Vehicles)
FTP Failure Rate(%)
HC or CO HC CO
52.8
64.2
38.6 42.5
54.5 61.8
59.2 47.6 53.4
FTP Excess Emissions(%)*
HC CO
23.9
76.1
100.0
24.4
75.6
100.0
oo
* Percent of all FTP excess emissions from the total sample
Corrected February 20, 1980
------- |