CORPORATION
ANALYSIS OF DATA
FROM
THE NEW YORK CITY
TAXI I/M PROGRAM
DRAFT REPORT
Prepared by:
Rob Klausmeier
Nancy Gates
Prepared for:
USEPA
I/M Staff
2565 Plymouth Rd.
Ann Arbor, MI. 48105
February 10, 1984
8501 Mo-Pac Blvd. / P.O. Box 9948 / Austin, Texas 78766 / (512)454-4797
-------
TABLE OF CONTENTS
SECTION PAGE
1 INTRODUCTION 1
1.1 Background 1
1.2 Summary of Findings 2
2 ANALYSIS OF DATA FROM I/M ROADSIDE TESTS 4
2.1 Basic Methodology 5
2.1.2 Quality Assurance of Data Base 7
2.2 Analysis of I/M Data 8
2.2.1 Analysis of Failure Rates 10
2.2.2 Analysis of Emissions Distributions 15
2.2.3 Emissions as a Function of Odometer 30
2.2.4 Emission Reductions 34
2.3 Analysis of Data From Roadside Checks 38
2.4 Results of Meetings With Maintenance
Personnel in Taxi Fleets 43
2.4.1 Fleets Operating Chevrolet/Checker Cabs
(Midland and Metro Cab Co.) 44
2.4.2 Fleets Operating Dodge Cabs
(57th Street Metropolitan Cab Co.) 46
2.4.3 Summary of Meetings 46
3 CONCLUSIONS/RECOMMENDATIONS 48
3.1 Conclusions 48
3.2 Recommendations 49
-------
LIST OF TABLES
TABLE PAGE
2-1 LIST OF KEY DATA FIELDS IN THE DATA BASE 6
2-2 FAILURE RATES 11
2-3 VEHICLES FAILING I/M TEST MORE THAN ONCE 17
2-4 PERCENT OF VEHICLES EXCEEDING IDLE EMISSION LEVELS .... 18
2-5 CROSS TABULATION OF MODERATELY HIGH
EMITTING VEHICLES 20
2-6 CROSS TABULATION OF GROSS EMITTING VEHICLES 21
2-7 EMISSION REDUCTIONS FOR FAILED VEHICLES 35
2-8 CO EMISSION REDUCTIONS FOR VEHICLES
PASSING AND FAILING RETEST 36
2-9 HC EMISSION REDUCTIONS FOR VEHICLES
PASSING AND FAILING RETEST 37
2-10 COMPARISON BETWEEN ROADSIDE CHECK
DATA AND I/M DATA 40
2-11 EMISSION DISTRIBUTIONS IN ROADSIDE CHECKS 42
ii
-------
LIST OF FIGURES
FIGURE PAGE
2-1 ODOMETER DISTRIBUTIONS BY MODEL YEAR ........... 9
2-2 FAILURE RATES VS. ODOMETER READINGS FOR ALL
VEHICLE MAKES AND ALL MODEL YEARS
(1980-1982) ........................ 13
2-3 FAILURE RATES VS. ODOMETER READINGS FOR ALL
VEHICLE MAKES FOR MODEL YEARS 1980, 1981,
and 1982 ......................... 13
2-4 FAILURE RATES FOR MODEL YEAR 1980, 1981, AND
1982 CHEVROLET /CHECKER VEHICLES VS.
ODOMETER READINGS ....................
2-5 FAILURE RATES FOR MODEL YEAR 1980, 1981, AND
1982 DODGE VEHICLES VS. ODOMETER READINGS
2-6 FAILURE RATES FOR MODEL YEAR 1980, 1981, AND
1982 FORD VEHICLES VS. ODOMETER READINGS ......... ]_6
2-7 PERCENTAGE OF VEHICLES (ALL MAKES) WITH
EMISSIONS GREATER THAN 300 ppm HC EMISSIONS
VS. ODOMETER READING ................... 22
2-8 PERCENTAGE OF VEHICLES (ALL MAKES) WITH
EMISSIONS GREATER THAN 700 ppm HC EMISSIONS
VS. ODOMETER READING ................... 22
2-9 PERCENTAGE OF VEHICLES (ALL MAKES) WITH
GREATER THAN 3.0% CO EMISSIONS VS.
ODOMETER READINGS ..................... 23
2-10 PERCENTAGE OF VEHICLES (ALL MAKES) WITH
GREATER THAN 7.0% CO EMISSIONS VS.
ODOMETER READING ..................... 23
2-11 PERCENTAGE OF CHEVROLET /CHECKER VEHICLES
WITH GREATER THAN 300 ppm HC EMISSIONS
VS. ODOMETER READING ................... 25
iii
-------
RADIAN
LIST OF FIGURES con't.
FIGURE PAGE
2-12 PERCENTAGE OF CHEVROLET/CHECKER VEHICLES
WITH GREATER THAN 700 ppm HC EMISSIONS
VS. ODOMETER READING 25
2-13 PERCENTAGE OF CHEVROLET/CHECKER VEHICLES
WITH GREATER THAN 3.0% CO EMISSIONS VS.
ODOMETER READING 26
2-14 PERCENTAGE OF CHEVROLET/CHECKER VEHICLES
WITH GREATER THAN 7.0% EMISSIONS VS.
ODOMETER READING 26
2-15 PERCENTAGE OF FORD VEHICLES WITH
GREATER THAN 300 ppm HC EMISSIONS
VS. ODOMETER READING 28
2-16 PERCENTAGE OF FORD VEHICLES WITH
GREATER THAN 700 ppm HC EMISSIONS
VS. ODOMETER READING 28
2-17 PERCENTAGE OF FORD VEHICLES WITH
GREATER THAN 3.0% CO EMISSIONS VS.
ODOMETER READING 29
2-18 PERCENTAGE OF FORD VEHICLES WITH
GREATER THAN 7.0% CO EMISSIONS VS.
ODOMETER READING 29
2-19 PLOTS OF HC VS. ODOMETER 31
2-20 PLOTS OF CO VS. ODOMETER 32
2-21 AVERAGE EMISSION CONCENTRATION
BY ODOMETER 33
2-22 AVERAGE EMISSION CONCENTRATION BY
ODOMETER - FAILED VEHICLES ONLY 33
2-23 AVERAGE OF REDUCTION AS A FUNCTION
OF ODOMETER 39
iv
-------
1.0 INTRODUCTION
1.1 BACKGROUND
The in-use emission control performance of model year 1980 and newer
vehicles is affected by several factors including misfueling, lack of or
improper maintenance, and emission control component failure. Although ERA
has gathered information on FTP emissions of in-use 1980 and newer vehicles,
only limited information is available on the extent that the above factors
affect emissions. This study was funded for the purpose of gathering and
evaluating Inspection/Maintenance (I/M) emission test data on 1980 and
newer taxis operated in New York City. The objectives of this study were
to: 1) provide insight into the in-use behavior of 1980 and newer vehicles,
especially those with high mileage and 2) learn more about the frequency of
different types of malfunctions in 1980 and newer emission control systems
and the emissions penalties associated with them.
Since it began operation in October 1977, the New York City taxi I/M
program has required all medallion taxis in New York City to pass a short
exhaust emissions test three times a year. The program is jointly adminis-
tered by the NYC Taxi and Limousine Commission and the NYC Department of
Environmental Protection (DEP). It uses a decentralized inspection station
system of 12 fleet stations and 14 other private inspection stations. At
all stations, fleet or private, the inspection is performed by an inspector
certified by the Taxi and Limousine Commission. In addition, the DEP
operates three emission test vans to spot check taxis on the road. These
checks include an idle emission test and a tampering inspection. Taxis
that fail the spot check must be retested within 10 days at the DEP's test
facility in Brooklyn. 1980 and newer model taxis must comply with idle
emission standards of 220 ppm HC and 1.2% CO; no waivers are available.
About 2500 vehicles per month are tested in the New York City taxi I/M
program; over half of these are model year 1980 and newer vehicles. Since
-------
taxis accumulate mileage at a much faster than average rate, the New York
taxi I/M program is a good source of data on high mileage 1980 and newer
vehicles. This I/M program has reported failure rates for new technology
vehicles that are much higher (20%) than those observed in other I/M pro-
grams, suggesting that failure rates may be very sensitive to accumulated
mileage.
1.2 SUMMARY OF FINDINGS
Although the I/M and roadside tests do not provide an indication of
the emission levels as measured by the Federal Test Procedure (FTP), the
results of these tests still provide a picture of the in-use emission control
performance of 1980 and newer model vehicles. Following are the major
findings of the study:
• 1981 and possibly 1982 vehicles with the 229 cubic inch
Chevrolet engines appear to have greater CO and, to a lesser
extent, greater HC emissions than 1980 model year vehicles.
The closed loop fuel metering system may be responsible for
this trend.
• 1981 Dodge vehicles with the 225 cubic inch engines also
appear to have greater CO and HC emissions than comparable
1980 models.
• 1981 Ford vehicles have very similar HC and CO emissions
to the 1980 models which is expected because of similar
fuel metering system designs.
• The failure rate in the I/M test is strongly dependent
on odometer reading.
• The HC and CO emission levels of vehicles that pass the
I/M test are not strongly affected by odometer reading.
• After repair emission levels of failed vehicles are similar
to the emission levels of vehicles that pass the I/M test,
regardless of the mileage.
• 'Although a vehicle c'ould have failed up to four times, most
vehicles only failed once or less. 1981 Dodge models show
the greatest tendency towards repeat failures.
-------
A majority of vehicles emitting moderately high amounts
of CO also emit moderately high amounts of HC; however,
a majority of the moderately high HC emitters do not emit
moderately high amounts of CO. The fact that ignition
defects affect HC but not CO emissions is most likely
responsible for this trend.
For Chevrolet and Ford vehicles, the percentage of gross
CO emitters (greater than 7.0%) does not appear to in-
crease with odometer after approximately 50,000 miles
are accumulated.
The failure rates in the roadside checks are much greater
than the failure rate in the I/M test. However, the road
side checks do not show significantly greater percentages
of gross emitting vehicles (greater than 7.0% CO or 700 ppm
HC).
1982 Chevrolets with the 229 cubic inch engine have had
frequent failures of the evaporative purge values which
have resulted in excessive HC and CO emissions and fuel
consumption. Oxygen sensors, ECMs, and carburetors have
been replaced more frequently than manufacturer's expected
intervals.
-------
2.0 ANALYSIS OF DATA FROM I/M AND ROADSIDE TESTS
Taxis operated in New York City accumulated mileage at a rate of
approximately .50,000 miles per year. As a result, data collected in the
New York Taxi I/M program and during roadside inspections provide a look at
the future performance of certain types of new technology vehicles. This
section presents the results of analysis of data from the New York City Taxi
I/M program.
A majority of the 1980-1982 taxis were Ford, Chevrolet, Checker
or Dodge models. In the analysis, the Checker and Chevy vehicles were
grouped together because they both used engines built by Chevrolet. Accord-
ing to discussions with fleet operators, most of the Chevrolet or Checker
vehicles were equipped with the 229 cubic V6 engine built by the Chevrolet
Motor Division of General Motors. The 1980 version utilized an oxidation
catalyst for emission control and the fuel control system was open loop, i.e.
there was no feedback carburetor. However, the 1980 vehicles did in-
corporate sealed idle mixture adjustment screws. The 1981 Chevrolets are
equipped with a three-way plus oxidation catalyst and a closed loop fuel
control system. Most of the Dodge vehicles were equipped with the 225 cubic
in-line 6 engine. The 1980 and 1981 models that were certified for 49
states (the federal version) used an open-loop fuel control system. In
1980, an oxidation catalyst was used while 1981 models used a 3-way cata-
lyst and an oxidation catalyst. Most likely the carburetors on the 1981
models were set richer than the 1980 models to enhance NO control. Ford
x
taxis were equipped with the 302 cubic inch V8 engine, since this is the
only engine available in a full-sized Ford. Like the Dodge vehicles, the
1980 and 1981 models were open-loop; the 1980 models used an oxidation
catalyst while the 1981 models used a three-way plus an oxidation catalyst.
Most likely the 1981 Ford models also had richer carburetor settings to
enhance NO control.
x
-------
The analysis of the I/M and roadside data attempts to address (
the following questions: j
• What is the reliability of closed-loop fuel control systems? '
• How do disablement/failure rates vary with mileage? '
• Are open-loop systems better for controlling hydrocarbons
(HC) and carbon monoxide (CO) emissions from carbureted vehicles?
• Can high mileage vehicles emit HC and CO at low rates?
• What are the emission reductions from repair? Do some {
I
vehicles show increases in emissions? ,
• How accurate are I/M inspections/do they lower in-use
disablement rates?
• Are vehicles repaired prior to being inspected?
• How many vehicles are repeat failures?
2.1 Basic Methodology
Radian requested and received from the DEP a tape containing taxi
I/M inspection results for the period from January 1, 1982 to May 31, 1983.
These data included 7154 inspection records for 1980 model year vehicles,
8904 for 1981, and 2468 for 1982. Very few records were included for 1983
vehicles and as a result, they were excluded from the analysis. Table 2-1
lists the types of data included on the tape. The data initially were ana-
lyzed on SAS and a clean data tape was generated.
In order to perform more in-depth data analysis, Radian established
additional data sets on the University of Michigan's data base management
system, MICRO. MICRO currently contains almost all of the EPA's mobile source
emission data; it is a very flexible tool for in-depth analysis. With MICRO,
Radian created additional fields that further described the odometer or emis-
sion characteristics of the vehicle. These fields include:
-------
TABLE 2-1 LIST OF KEY DATA FIELDS IN THE DATA BASE
New York Taxi I/M Data
Medallion Number
Month of Test
Year of Test
Make of Vehicle
Model Year of Vehicle
Odometer Reading
HC Emissions in ppm
CO Emissions in %
P/F Status
Retest HC
Retest CO
Retest P/F Status
Technician
Inspection Station
Roadside Test Data
Medallion Number
Make of Vehicle
Model Year of Vehicle
HC Emissions in ppm
CO Emissions in %'
-------
RADIAN
• Odometer group - group describing mileage of vehicle
(0-10,000; 10,000-20,000; etc.)
• HC greater than 300 ppm (yes, no)
• HC greater than 700 ppm (yes, no)
• CO greater than 3.0% (yes, no)
• CO greater than 7.0% (yes, no)
• HC emission reduction (%)
• CO emission reduction (%)
The New York City DEP also provided data sheets containing approxi-
mately 800 roadside emission tests on 1980 and newer model year taxis. Data
from these data sheets were keypunched and entered into a MICRO data set for
analysis. Roadside data were probably more accurate but did not include mile-
age readings or results of repairs.
In addition to collecting data from the inspection program, Radian
conducted personal interviews with maintenance personnel for three of the
major taxi fleets: Midland, Metro and 57th Street Metropolian. These inter-
views were held for the purpose of obtaining information on emission control
system failures and are discussed in Section 2.4.
2.1.2 Quality Assurance of Data Base
Several records in the data base were either modified or removed
to account for the possibility of inaccuracy:
• CO emission levels and
• Odometer readings.
On the inspection form filled out by the inspector, the CO emis-
sions were recorded to the nearest tenth of a percent. However, the emission
analyzers provided a reading to the nearest hundredth of a percent. As a
-------
COdOOOATIOM
result, a few inspectors recorded the data to the nearest hundredth percent
which resulted in the CO emissions being increased by a factor of 10. In
order to correct for these occurrences in the data base, a field named error
was created to identify vehicles that passed the I/M test but had CO emis-
sion levels that would be expected to fail the vehicle. The average error
rate was then calculated for each emission test technician and those that
had a five percent or greater error rate were removed from the data base.
This reduced the total number of records by approximately 20 percent and
made a huge difference in the average CO emissions and the CO frequency
distribution. The percent of vehicles exceeding the 7 percent CO level was
cut in half by these changes and the average emissions were reduced by 25-50
percent.
A more basic approach was used to account for the possibility of
inaccurate odometer readings. Since it is very unlikely that a 1980 model
taxi would have less than 50,000 miles on it, those that showed less than
50,000 miles on the inspection record were not considered, when trends by
odometer were evaluated. Similarly, all 1982 model vehicles with over
50,000 miles also were,removed from the data base. As a result of these
changes, only the 1981 model year vehicles had a full range of odometer
distributions (0-200,000). Figure 2-1 shows a distribution of the vehicle
odometer readings.
2.2 ANALYSIS OF I/M DATA
The analysis of the I/M data targeted on:
• Failure rates,
• Emissions distributions,
• Emissions as a function of odometer, and
• Emission reductions.
-------
yaaa
ODOMETER DISTRIBUTION BY MODEL YEAR
ALL MAKES
N
U
M
B
E
R
O
O
c
c
u
R
N
C
6
Z4oa-
l«aa -
tzaa-
633-
o-io
la~2a 30-40 ca-ea
ODOMETER
Ta-co
toa-zoa
aa-iaa
Figure 2-1. Odometer Distributions by Model Year
-------
2.2.1 Analysis of Failure Rates
Radian's analysis of failure rate data was oriented towards deter-
mining how successful new emission control technologies have been at re-
ducing HC and CO emissions. By examining these data, Radian wanted to
determine how the changes in emission control technologies from 1980 to
1981 affected failure rates and emissions. Failure rates were analyzed
as a function of model year, make, and odometer readings.
Failure Rate by Model Year and Make
Table 2-2 summarizes all of the I/M failure rate data analyzed by
Radian, broken down by model year and make. In this and subsequent tables,
Chevrolets and Checker vehicles are both included as Chevrolets, since they
have identical engines. As shown, the 1981 model year vehicles as a group
did not have lower failure rates than the 1980 vehicles despite the fact
that they had lower mileages. In the case of the Dodge vehicles, the 1981
model showed a failure rate nearly twice that of the 1980 model.
The average HC and CO emissions for passing and failing vehicles
of all makes were generally the same, regardless of model year. The same
emission cutpoints were used for all vehicles in the I/M tests so this
generalization appears reasonable. However, there were some notable excep-
tions. In the case of Chevrolets, the CO emissions from failed vehicles
greatly increased from 1980 to 1981. This trend indicates that I/M failures
in 1981 Chevrolets tend to be the result of rich engine failures, while
1980 models failed largely for HC only problems, such as ignition defects.
Also, 1982 vehicles that passed the test had significantly lower emission
levels.
For all makes, the filed vehicles had higher average odometer
readings than the passing vehicles. This indicates.that, as expected,
mileage affects the I/M failure rate.
10
-------
TABLE 2-2 FAILURE RATES
Model Year
1980
1981
.
1982
Make
All
Chevy*
Dodge
Ford
All
Chevy*
Dodge
Ford
All
Chevy*
Dodge
Ford
Avg.
Odometer
Pass
92K
96K
87K
91K
56K
56K
60K
ASK
18K
18K
21K
12K
Fail
104K
109K
97K
100K
77K
73K
86K
58K
30K
28K
33K
ASK
% Failure
14%
13%
14%
16%
16%
13%
27%
11%
4.5%
4%
7%
4%
Avg. HC (ppm)
Pass Fail
77.3
86.7
63.6
74.2
79.4
82.1
78.7
76.1
60.6
61.5
67.1
47.0
497
574
331
452
441
509
383
404
409
470
275
307
Avg.
Pass
0
0
0
0
0
0
0
0
0
0
0
0
.31
.29
.34
.33
.27
.27
.26
.28
.14
.13
.18
.12
CO (%)
Fail
2
1
4
3
3
2
3
4
2
2
1
3
.44
.09
.04
.48
.11
.77
.16
.72
.30
.25
.90
.77
Count
7154
3454
565
2896
8904
4860
2444
1488
2460
1773
393
283
* Includes Checker
-------
Failure Rate by Odometer Reading
Figure 2-2 shows failure rates for all 1980, 1981, and 1982 model
year vehicles as a function of vehicle odometer readings. As shown, failure
rates increase consistently with increases in odometer readings. Figure 2-3
shows a more in-depth examination of this trend; in this case, failure rate
is shown as a function of odometer reading for each of the model years 1980,
1981, and 1982. This analysis reveals that 1981 model year vehicles have
significantly higher failure rates than 1980 model year vehicles with
similar odometer readings. Odometer readings shown on these figures were
the midpoint of the interval examined. For example, 5,000 relates to the
0-10,000 mile interval.
When the failure rate versus odometer reading is broken down by
vehicle make, other patterns become evident. Figure 2-4 shows failure
rates versus odometer readings for 1980-1982 model year Chevrolet/Checker
vehicles. Here the failure rate for the 1981 vehicles is much higher than
that for the 1980 vehicles for all odometer readings from 50,000 to 150,000.
This indicates that the closed loop fuel control system used in the 1981
vehicles appears to be less reliable at controlling emissions than the
open loop system with sealed idle adjustment screws used on the 1980
Chevrolets.
In Figure 2-5, the Dodge vehicle failure rate is plotted as a
function of odometer reading. Here again, the 1981 vehicles have a con-
sistently much higher failure rate across all odometer readings. The
greater failure rate of the 1981 vehicles may be due to richer or more
sensitive carburetor adjustments required to meet the 1981 NO standard.
However, the 1981 vehicles had sealed idle adjustment screws which should
have caused some drop in their failure rate. These problems may not be
a serious vehicle emission concern since more than 90 percent of Chrysler
vehicles have four cylinder engines with significantly different engine
designs than the 225-6.
12
-------
U FAILURE
ALL I860. 1S«1. ANP 1862 VEHICLES
u
F
A
Z
L
U
R
E
20-
ic-
10-
C-
I
e
I
tc
I
2G
1
SB
I
4C
I
cc
I
ec
I
7C
I
CB
as tea
ALL VEHICLES
ODOMETER
Figure 2-2. Failure Rates vs. Odometer Readings for All Vehicle
Makes and All Model Years (1980-1982)
U FAILURE
ALL MAKES
U
F
A
I
L
U
R
E
40-
32-
24-1
IB
I
B
1
1C
I
2E
I
4C
r
cc
T
ec
T
7C
1
cc
ac ICQ
taea MODEL YEAR
1881 MODEL YEAR
1S«2 MODEL YEAR
ODOMETER
Figure 2-3.
Failure Rates vs. Odometer Readings for All Vehicle
Makes for Model Years 1980, 1981, and 1982
13
-------
F
A
X
L
U
R
E
2C
ZB-
IC-
10-
C-
U FAILURE
CHEVY./CHECKER
1
C
1C
2C
3E
1
4C
1
CC
1
7C
1
«C
1
SC
MOPEL TEAR I SCO
MCJPEL TEAR IS«t
PU3PEL TEAR 1SC2
Figure 2-4. Failure Rates for Model Year 1980, 1981, and 1982
Chevrolet/Checker Vehicles vs. Odometer Readings
U FAILURE
POPfiE
ea
F
A
X
L
U
R
E
30-
24-
12-
1
1C
ZC
1
3S
4B
1
CE
1
6C
1
75
1
CC
MOPEL TEAR tSSO
MOPEL TEAR IS«1
MOPEL TEAR tB«2
Figure 2-5. Failure Rates for Model Year 1980, 1981, and
Dodge Vehicles vs. Odometer Readings
1
8C
1982
1
ICQ
14
-------
RADIAN
The failure rate of Ford vehicles versus odometer reading for
the 1980-1982 model years is plotted in Figure 2-6. In this case, the 1981
failure rate is very similar to the 1980 rate. Most 1981 Fords have open
loop fuel control systems similar to the systems found in the 1980 models,
so this agreement in failure rates is not surprising.
Since the data base covered approximately 16 months, each
vehicle was inspected approximately 4 times and therefore could have
failed 4 times. Table 2-3 shows the percent of vehicles that failed
the I/M test more than once. As shown, most of the vehicles in the data
base had not failed the I/M test previously. When the vehicles that failed
more than once are examined, the 1981 Dodge models appear to be most prone
to repeat failures. This makes sense since the Dodge models are open
loop and more prone towards being readjusted after repair to compensate
for a driveability problem. Also, the 1981 Dodges have the highest failure
rate which increases the chance that they will fail more than once.
2.2.2 Analysis of Emissions Distributions
Additional insight into the cause of I/M failures can be ob-
tained by analyzing the distribution of emission levels. In this analysis,
vehicles were grouped into the following categories:
HC greater than 300 ppm,
HC greater than 700 ppm,
CO greater than 3.0%, and
CO greater than 7.0%.
Emission distribution trends were then analyzed as a function of odometer.
Table 2-4 shows the percent of vehicles in the above categories for each
combination of make and model year. As shown, 1981 Dodge models have the
highest percentage of vehicles with excessive HC or CO emissions. The 1981
15
-------
U FAILURE
FORD
X
F
A
X
u
E
25'
zo-
16-
10-
I
5
I I I I I
1C ZC 35 4C 55
I
65
I
7C
I
65
I
as
I
tea
MODEL TEAR IS8Q
MODEL TEAR 1881
MODEL TEAR 1S«£
ODOMETER
Figure 2-6. Failure Rates for Model Year 1980, 1981, and
1982 Ford Vehicles vs. Odometer Reading
16
-------
TAIJLE 2-3
VEHICLES FAILING I/M TEST MORE THAN ONCE
Model Yr.
1980
, 1981
j
1982
Make
AIL
Chevy
Dodge
Ford
All
Chevy
Dodge
Ford
AL1
Chevy
Dodge
Ford
Total 11 of
Vehicles
7130
3448
565
2896
8879
4856
2444
1488
2454
1772
393
283
Vehicles
654
300
60
294
923
428
376
119
106
75
23
8
Failing Once
9.2%
8.7%
10.6%
10.2%
10.4%
8.8%
15.4%
8.0%
4.3%
4.2%
5.8%
2.8%
Vehicles
276
126
18
132
406
180
196
30
6
—
4
2
Failing Twice Vehicles/3 Times
3.9%
3.7%
3.2%
4.6%
4.6%
3.7%
8.0%
2.0%
0.2%
0%
1.0%
0.7%
54
21
3
30
114
36
69
9
—
—
—
—
0.8%
0.6%
0.5%
1.0%
1.3%
0.7%
2.8%
0.6%
0%
0%
0%
0%
Vehicles/4 Times
16 0.2%
0
0
16 0.6%
20 0.2%
12 0.2%
8 0.3%
0%
0%
0%
0%
0%
-------
TABLE 2-4
PERCENT OF VEHICLES EXCEEDING IDLE EMISSION LEVELS
% Exceeding
HC
Model Year Make > 300
1980 All 9.2%
Chevy * 10.0%
Dodge 6 . 4%
Ford 9.5%
1981 All 9.3%
Chevy * 8.7%
Dodge 12.6%
Ford 6.4%
1982 All 2.6%
Chevy * 2.8%
Dodge 2.3%
Ford 1.8%
Idle
(ppm)
> 700
2.9%
3.9%
1.1%
2.4%
2.5%
2.8%
2.8%
1.1%
0.7%
0.8%
0.5%
0.0%
Emission Levels
CO (%)
> 3.0 > 7.0
4.0% 1.11%
1.5% 0.3%
4.9% 1.9%
7.1% 2.0%
7.3% 1.8%
5.1% 1.7%
11.5% 2.1%
7.0% 2.0%
1.0% 0.30%
1.0% 0.3%
1.5% 0%
.6% .6%
* Including Checker
18
-------
RADIAN
Chevrolet vehicle population has significantly greater percentages of high
CO emitters than their 1980 counterparts, while Ford vehicles show little
change from 1980 to 1981. However, despite the increase, the 1981 Chevro-
lets do not appear to be higher CO or HC emitters than the Ford models.
Table 2-5 presents a cross-tabulation of the high emitting CO
vehicles with the high emitting HC vehicles. As shown, a majority of
the moderately high hydrocarbon emitting vehicles—those greater than
300 ppm—did not have high CO emissions. On the other hand, the majority
of the moderately high CO emitting vehicles did have high HC emissions.
This makes sense since the factors that cause high CO emissions will also
cause high HC emissions, while the reverse is not always true. When the
vehicles with gross emissions, i.e., those over 7% CO or 700 ppm HC are
examined, it appears that both the gross HC and the gross CO emitters occur
independently of each other (see Table 2-6).
Figure 2-7 shows the percentage of all vehicles with moderately
high HC emissions versus odometer readings for model years 1980-1982. In
Figure 2-8, the same relationship is again shown, but for vehicles with
gross HC emissions. Figures 2-9 and 2-10 show the percentage of all vehi-
cles with more than 3.0 percent (moderately high) and more than 7.0 percent
(gross) CO emissions, respectively. Again this percentage is plotted versus
odometer readings for model years 1980-1982.
Although some inconsistencies are present, the overall trend is
towards a greater percentage of high emitters at higher odometer readings.
As shown in Figure 2-7, a greater percentage of 1981 vehicles than 1980
vehicles exceeded 300 ppm HC emissions. Examination of the percentage of
gross HC emitting vehicles by model year shows that a somewhat greater per-
centage of 1981 vehicles than 1980 vehicles have emissions that exceed 700
ppm. However, as shown, the difference between the percent of 1981 and 1980
vehicles exceeding 700 ppm is .not as large as the difference between the
percent of 1981 and 1980 vehicles with HC greater than 300 ppm. It is
19
-------
TABLE 2-5 CROSS TABULATION OF MODERATELY HIGH EMITTING VEHICLES
Vehicle Emission Category
Count
Percent
CO <3.0%, HC <300 ppm
CO <3.0%, HC >300 ppm
CO >3.0%, HC <300 ppm
CO >3.0%, HC >300 ppm
12766
605
321
438
90%
4.3%
2.3%
3.1%
o
-------
TABLE 2-6 CROSS TABULATION OF GROSS EMITTING VEHICLES
Vehicle Emission Category Count Percent
CO <7.0%, HC <700 ppm 13665 97%
CO <7.0%, I1C >700 ppm 269 2.0%
CO >7.0%, HC <700 ppm 140 1.0%
CO >7.0%, HC >700 ppm 56 0.4%
K)
-------
OF VEHICLES GREATER THAN 38O PPM HC
ALL MAKES
* 17.6-
3
a |3-2~
p
P
M «.«-
H
C
4.4-
J
/
J
1
1
/!>
^^ * •
^^ • •
^/ * *
„' ....-••'
1 1 1 1 1 1 1 i 1 1 1
E IE ZE 3E 4E EC CE 7E «C BE 1EB
1 asa (HOPCI. YEAR
1S«1 MODEL. YEAR --_,—,.-,,
i a^« Mnntn •tfcriD aDOMETEK
Figure 2-7. Percentage of Vehicles (All Makes) With Emissions
Greater Than 300 ppm HC Emissions vs. Odometer Reading
X OF VEHICLES GREATER THAN 7OO PPM HC
ALL MAKES
U
7
a
a
p
p
M
H
C
3.6-
2.4-
i.z-
^ .'
1
C
I
IE
I
ZE
1
3E
r
4E
I
CE
1
6C
1
7E
I
8E
SE
I
ICO
MODEL YEAR
last MODEL YEAR
I8CZ MODEL YEAR
ODOMETER
Figure 2-8. Percentage of Vehicles (All Makes) With Emissions
Greater Than 700 ppm HC Emissions vs. Odometer Reading
22
-------
OF VEHICLES GREATER THAN 3.O» CO
ALL MAKES
u
9
U
C
a
IB
12-
«-
4-
T
1C ZB
4S
ES
BE
76 SC
SC
MODEL YEAR tSSB
MODEL YEAR tact
MODEL YEAR 1882
Figure 2-9.
Percentage of Vehicles (All Makes) With Greater Than
3.0% CO Emissions vs. Odometer Readings
U Or VEHICLES CREATES THAN 7.OX CO
ALL MAKES
3.2-
U
7
U
C
o
1 .B-
Q.C-
So
/ \ /
T 1 r—
5 IB ZS
MODEL YEAR 1880
MODEL YEAR 1881
MODEL YEAR 1882
•4S EE BS 75
ODOMETER
as tea
Figure 2-10.
Percentage of Vehicles (All Makes) With Greater Than
7.0% CO Emissions vs. Odometer Reading
23
-------
possible that vehicles exceeding 300 ppm but not 700 ppm HC have fuel-related
problems, while those that exceed 700 ppm HC have ignition-related problems.
The major change from 1980 to 1981 was in the fuel metering system.
As shown in Figure 2-9, a noticeably higher number of 1981 vehi-
cles exceed the 3% CO level than do 1980 vehicles and the number appears
to be very sensitive to odometer reading. Again, this is evidence of the
fuel control problems that appear to exist in 1981 vehicles. The percentage
of vehicles with more than 7 percent CO (Figure 2-10) is also higher for
1981 than for 1980 model year vehicles, although the absolute percentage
is still low, 3 percent.
Trends by Vehicle Make
Chevrolet/Checker
CO emissions were studied for indication of control and full rich
engine failures expected in some of the Chevrolet computer controlled en-
gines. As can be seen in Figures 2-11 through 2-14, the 1981 model year
Chevrolet/Checker vehicles have a greater tendency towards high HC and CO
emissions, indicating the presence of fuel-related problems in this model
year vehicle. For HC, the percent of vehicles exceeding 300 ppm is very
similar for model years 1980 and 1981 until the odometer reading exceeds
65,000. Then the percent of moderately emitting 1981 vehicles sharply
increases, while the 1980 rate only slightly increases. The 1981 vehicles
do not appear to have a greater percent of gross HC emitters (greater than
700 ppm) which could be expected.
In the case of CO, a much greater percentage of model year 1981
vehicles exceed 3.0% and 7.0% than do 1980 vehicles. Also, as shown in
Figure 2-12, the percentage of gross CO emitters (greater than 7.0%) does
not appear to be strongly influenced by mileage after an initial increase
during the first 50,000 miles except for an outlier point at 85,000 miles.
Possibly a disablement of one or more emission related control sensors causes
24
-------
OF VEHICLES GREATER THAN 3OO PPM HC
CHEVY/CHECKER
u
3
a
a
p
p
M
H
C
2O
»e-
•4-
1
c
1
is
1
zs
1
ye.
1
4C
1
ce
1
BE
1
TG
1
«E
1
sc
1
tea
MODEL YEAR
MODEL YEAR 1S«1
MODEL YEAR 18C2
«««M«=^-«,
ODOMETER
Figure 2-11.
Percentage of Chevrolet/Checker Vehicles With Greater
Than 300 ppm HC Emissions vs. Odometer Reading
X OF VEHICLES GREATER THAN 7O8 PPM HC
CHEVY./CHECKER
MODEL YEAR 1S«1
MODEL YEAR I8C2
Figure 2-12. Percentage of Chevrolet/Checker Vehicles With Greater
Than 700 ppm HC Emissions vs. Odometer Reading
25
-------
C
a
tz-
S.B
7.2-
«*.«-
2.4-
OF VEHICLES GREATER THAN 3.a* CO
CHEVY./CHECXER
N /
N J
\ j
V
V7-.
S IS 25
MODEL YEAR 1889
MODEL YEAR 1881
MODEL YEAR I8C2
I
45
I
BE
I
B5
I
76
1
86
I I
85 150
Figure 2-13. Percentage of Chevrolet/Checker Vehicles With Greater
Than 3.0% CO Emissions vs. Odometer Reading
2 OF VEHICLES GREATER THAN 7.03 CO
CHEVY./CHECKER
4-
U
s-
7
U
C
a 2-
1 -
A
''— - ^N / ^^
'? *'• ^^
1 1 1 'i 1 1 i 1 I 1 1
5 16 25 55 45 55 85 75 «5 85 I5Q
MODEL YEAR 1S8H
MODEL YEAR 1881
MODEL YEAR 1882 ODOMETER
Figure 2-14.
Percentage of Chevrolet/Checker Vehicles With Greater
Than 7.0% CO Emissions vs. Odometer Reading
26
-------
the engine to run in the "control rich" mode resulting in moderately, rather
than grossly elevated CO levels. Also, driveability may be impaired or the
owner/operator may have some other indication of a problem when a vehicle
exceeds 7.0% CO.
Fords
Figures 2-15 through 2-18 show the distribution of Ford vehicles
with excess HC and CO emissions. Examination of data on moderately high
and grossly emitting Ford vehicles shows that the 1981 vehicles do not
have higher HC emissions than the 1980 models. Also, it shows that 1981
vehicles have roughly the same percentage of moderately high and gross CO
emitters as the 1980 model year. Since 1981 Fords have similar fuel con-
trol systems to the 1980 models, i.e., open loop, these results are not sur-
prising.
There were an inadequate number of 1980 Dodge vehicles to per-
form a valid comparison of CO and HC emissions distribution between the two
model years.
27
-------
u
s
o
a
p
p
M
H
C
IE
12-
9-
U OF VEHICLES 6REATER THAN 30O PPH HC
R3RP
I I I I I I I I I I 1
E IE 2E 3E 4E EE 6C 7E «E 6E 1EO
tOCO MODEL TEAR
,*<, MOPELTEAR ^^^
Figure 2-15. Percentage of Ford Vehicles With Greater Than 300 ppm
HC Emissions vs. Odometer Reading
X OF VEHICLES BREATER THAN 7OO PPN HC
FORP
3.Z-
7
o
a
p
p
M
H
C
i.e-
o.€-
\ I I 1 T I 1^ I I I I
B 1C ZC 3C 4E CC 6E 7C «E 8C ICO
I8C8 MODEL TEAR
I8C1 MOPEL TEAR
ODOMETER
Figure 2-16. Percentage of Ford Vehicles With Greater Than 700 ppm
HC Emissions vs. Odometer Reading
28
-------
X OF VEHICLES GREATER THAN 3.OH CO
FORD
3
U
C
a
J2
S.B
7.2-
2.4-
S
1C
ZS
35
1
45
55
1
OS
I
7G
1
«C
1
8G
I
(53
MODEL YEAR 1889
MODEL YEAR 1881
Figure 2-17.
Percentage of Ford Vehicles With Greater Than 3.
CO Emissions vs. Odometer Reading
U OF VEHICLES GREATER THAN 7. OX CO
FORD
85 169
MODEL YEAR 1889
MODEL YEAR 1881
ODOMETER
Figure 2-18. Percentage of Ford Vehicles With Greater Than 7.0%
CO Emissions vs. Odometer Reading
29
-------
2.2.3 Emissions As A Function of Odometer
Scatter plots of the HC and CO emission for the 1981 model year
as a function of odometer are shown on Figures 2-19 and 2-20. As shown,
emission levels do not consistently increase as a function of odometer
reading. At the higher odometer reading, there appears to be a few more
vehicles in the high emitting category, but the bulk of the vehicles are
still in the low emitting category. Correlation coefficients were deter-
mined for HC and CO emissions as a function of odometer and the coefficient
was not significantly different from zero.
When the emissions are averaged for different groups of odometer
readings, i.e. 0-10,000 miles, 10,000-20,000 miles, as shown on Figure 2-21,
there does appear to be a trend towards increasing HC and CO emissions. The
reason for this trend can be explained by looking at the emissions as a
function of odometer reading separately for vehicles that pass and fail the
I/M test. As shown on Figure 2-21, the average emissions from vehicles
that pass the I/M test do increase with odometer but generally at a rate
o
much lower than the rate for the total vehicle population. The average HC
emissions varied from approximately 50 ppm at zero miles to approximately
80 ppm at 150,000 miles. On the other hand, the overall average emission
concentrations varied from approximately 50 ppm to over 160 ppm at 150,000
miles. Figure 2-22 shows the average emissions of the vehicles that failed
the I/M test, and although very high, they do not increase with odometer
reading. Consequently, the increase in average emissions as a function of
odometer reading for the overall vehicle population can be explained by the
greater I/M failure rate for the high mileage vehicles. Furthermore, the
data indicate that a properly adjusted or maintained vehicle can continue
to emit low quantities of hydrocarbons or carbon monoxide when it has accu-
mulated high mileage. These trends are consistent through each individual
model year.
30
-------
TAXJ - «:.u "M/SRY PTATS
R= 1981
HT |
I
I
1 A
! A AAB PEAAAA A P?vA A A
2".??+ fl A a A A P a A i
I A
I A n A A A A A « A
! f 1
' .FA
.1 A A
1. <=j .,", * ft f A
I C I: A A3* A A
! * At AA A
I AH fi 1.
I A A" fl A A A A
1 ,'. AAA *BA AAA .1 a,\
in-;i: + APA A A AA APA AAARAA AA
!,1 A A A AA AAA A.^/sr- Rfi f:A A
I AA « CBCAACPFf/'C!: H-'AAAr «
! /ki/.AAAt; ECAPL'C P 8 ft*.
' AA C*CEDSCC-5«f/lACA?a A A A
I AAfC P"BDrJC° jFlfDD^nc n Afl4A.\r"
ML TCEJGFr.] CFALA /".AAA.AA
D'l-K jryHh'IEiCFDAC'A A. A A
!IF?CJTPJKSSyTF*7?F*l--•——-4--———•- — •+•••••• ——»— — «4<-~i
" 5 ^ -1 ? '.I '] •? f1" p " 1 c 9 0 C n 2 0 0 D C 0 ? 5 0 0 0 0 T 9 0 0 G 0
^ i ti o •:n s H i D n r -v
Figure 2-19. Plots of 1IC vs. Odometer
-------
90 r: •»
I
I
P 0 0 +
I
i
TJX I - SUMMARY STATS
LY9= 1981
1 «
I
* /I AA f
' A A « AR n / A
1 /Cf!/»ACrDB3l:.ArPEn AB A
1"C *C.'R C-F:c-FF1"Cr:?rC-''IF3rAGApE.Ac A A R
H5II.LLTIH'?.1jaTGLr)';r!C C
-------
AVERASE EMISSION CONCENTRATION
ALL VEHICLES/PACKED VEHICLES
C
o
N
C
E
N
T
R
A
T
I
0
N
tea-
iza
s
s _
' -• '
\
E
I
IE
I
ZS
I
35
I
45
I
E5
I
0E
I
7B
I
86
8S
I
isa
HC CFPM5/ALL
ca CJi x i aax/AU-
NG CPPN3/PAS6
coca x
Figure 2-21.
ODOMETER
Average Emission Concentrations by Odometer
AVERACE EMISSION CONCENTRATION
FAILED VEHICLES
C
O
N
C
E
N
T
R
A
T
I
a
N
490-
sra
240-
I
E
I
ZS
HC
catu x
I
5E
ODOMETER
I
BB
I
7E
I
SB
I
SB
IBB
Figure 2-22. Average Emission Concentration by Odometer - Failed
Vehicles Only
33
-------
2.2.4 Emission Reductions
The data base contains the results of the retest performed on ve-
hicles that failed the initial inspection along with the final inspection
status, i.e. the results of the retest. Table 2-7 shows the emission re-
duction of the different groups of vehicles. As shown, the average percent
reduction in HC emissions generally exceeded the average percent reduction
in CO emissions. However, percent reduction in average emissions of the
vehicles before and after repairs was greater for CO than HC. The standard
deviation of the percent reduction in CO emissions explains this inconsis-
tency. Standard deviations over 100 percent indicate that some vehicles
showed increases in emissions over 100 percent. The maximum reduction
theoretically is less than 100 percent. Since CO emissions can be extremely
low, i.e. less than .1%, a small absolute increase could equate to a several
hundred percent increase, thereby reducing the average percent reduction
figure and increasing the standard deviation.
Table 2-8 and 2-9 illustrate that the average percent reduction
is influenced by a small percentage of vehicles that show increases in
emissions. Table 2-8 shows the CO emission reductions for vehicles passing
and failing the retest; Table 2-9 shows the HC emission reductions. As
shown when the repair targeted on the pollutant of concern, the effect of
the repair on that pollutant was much more consistent, i.e. there was a
much lower standard deviation. For example, for the cases where CO is re-
duced, the standard deviation is much lower than the overall group that
passed on retest. Similarly, HC reductions are consistently greater and
the standard deviation is consistently lower for the vehicles where HC is
reduced. As shown on Table 2-8, for the group where CO increases, the CO
emissions after repair are 0.61% which is significantly higher than the
rest of the vehicles that passed on retest. For the case where HC increases,
the increase is not excessive (29%), and the average HC emissions fro this
group (130 ppm) is still much lower than the standard of 220 ppm. The in-
crease in CO emissions from repairs appears to be a greater concern than
the increase in HC emissions both because of the magnitude of the increase
34
-------
TABLE 2-7 EMISSION REDUCTIONS FOR FAILED VEHICLES
1980
1981
U)
in
1982
All
Chevy *
Dodge
Ford
All
Chevy *
Dodge
Ford
All
Chevy *
Dodge
Ford
Avg.
* %
Red.
67%
72%
45%
66%
61%
66%
52%
77%
68%
75%
58%
—
IIC
Std.
Dev. of
% Red.
34%
27%
49%
36%
30%
24%
34%
18%
25%
22%
28%
—
% Red.
in Avg.
Emis.
78%
82%
69%
78%
74%
78%
68%
82%
77%
82%
68%
—
Avg.
%
Red.
50%
27%
56%
72%
44%
27%
49%
85%
49%
44%
69%
—
CO
Std.
Dev. of
% Red.
114%
128%
142%
89%
123%
132%
117%
98%
139%
169%
38%
—
% Red.
in Avg
Emis.
91%
86%
91%
93%
88%
87%
85%
97%
88%
91%
81%
—
* Including Checker
-------
TABLE 2-8
UJ
Pass on Retest
CO EMISSION REDUCTIONS FOR VEHICLES PASSING AND FAILING RETEST
educed
C Increases
0 increases
of
Vehicles
98%
87%
2.7%
8%
Avg. CO
before
3.39
3.67
3.52
0.24
Avg . CO
after
0.36
0.34
0.51
0.61
% Re-
duction
in Avg. CO
89%
91%
86%
(-154%)
Avg. %
Reduc-
tion CO
46%
74%
75%
(-259%)
S-I). %
Rcduct ion
in CO
120%
33%
25%
252%
• IIC and CO increases 0% -
Fail on Retest 2% 4.45 2.30 48% (-3.0%) 165%
Sample Size = 1621
S.I). - Standard Deviation
-------
U)
TABLE 2-9
HC EMISSION REDUCTIONS FOR VEHICLES PASSING AND FAILING RETEST
% , Avg. HC
of before
Vehicles (ppm)
Pass on Rctcst 98% 427
• HC and CO are reduced 87% 418
• CO is reduced/IIC increases 2.7% 101
• IIC is reduced/CO increases 8% 625
• IIC and CO increases 0%
F.iil on Rctest 2% 8]0
Avg. HC % Re- Avg. % S.D. %
after duction HC lie- HC Re-
(ppm) in Avg. Emis. duction duction
101 76% 64% 32%
98 77% 66% 24%
130 (-29%) (-48%) 62%
121 81% 71% 22%
_ _ _
569 30% 26% 39%
Sample Size = 1621
S.D. - Standard Deviation
-------
(154% in average CO emissions) and the percent of vehicles affected, 8% CO
vs. 2.7% HC. Note on Table 2-8 and 2-9 that the after repair emission
levels are only slightly higher than emission levels of vehicles that pass
the I/M test (see Table 2-2).
Figure 2-23 shows the average percent reduction for the 1981 and
1982 model year vehicles as a function of odometer. As shown, the percent
reduction appears to be insensitive to odometer, indicating that the emis-
sions from high mileage vehicles can be reduced by the same percent as the
emissions from low mileage vehicles. However, since the idle test does not
measure catalyst activity to a great degree, it is difficult to extrapolate
these results to FTP emissions; i.e. the percent reduction in FTP emissions
may be lower at high mileages. Additional data would be useful to address
this issue.
2.3 ANALYSIS OF DATA FROM ROADSIDE CHECKS
In addition to analyzing data collected during the I/M tests, Ra-
o
dian also analyzed data collected during roadside checks performed by DEP
personnel. The roadside check data is useful for a number of reasons:
• The DEP personnel are likely to perform more accurate in-
spections than the inspections performed by inspectors cer-
tified by the Taxi and Limousine Commission.
• The vehicle owner/operator does not have an opportunity to
prepare the vehicle for the test, consequently, the results
are more representative of vehicles operating on the street.
• The roadside test data were collected much more recently and
therefore are able to give an indication of the performance
of 1983 model vehicles along with a more thorough indication
of the high mileage performance of the 1982 model year ve-
hicles.
The roadside data showed somewhat similar results to the I/M data.
As shown in Table 2-10, the roadside failure rate was much higher than the
38
-------
u
R
E
D
U
c
T
Z
a
N
too
80-
60-
•40-
za-
ENISSION REDUCTION
138! AND 1362 MODEL YEAR VEHICLES
I
1C
i
ZS.
1
3E
I
-4C
I
CC
I
6C
7C
sc tea
AVC HC REDUCTION
AVC CO REDUCTION
ODOMETER
Figure 2-23.
Average of Reduction as a Function of
Odometer
39
-------
TABLE 2-10 COMPARISON BETWEEN ROADSIDE CHECK DATA AND I/M DATA
Roadside Checks Data
Model
Year
1980
1981
1982
1983
Make
All
Chevy
Dodge
Ford
All
Chevy
Dodge
Ford
All
Chevy
Dodge
Ford
All
Chevy
Dodge
Ford
Avg. IIC (ppin)
Pass Fall
71
98
63
68
82
82
84
88
80
86
69
64
46
41
53
—
403
432
540
343
464
498
412
395
501
497
640
240
262
326
188
Avg. CO (%)
Pass Fail
0.24
0.25
0.17
0.27
0.24
0.27
0.24
0.21
0.14
0.15
0.11
0.11
0.11
0.11
0.11
2.16
1.31
0.67
3.56
3.14
2.88
3.92
3.30
3.48
3.62
2.90
0.10
1.39
1.15
1.40
% Failure
28%
31%
17%
31%
36%
39%
39%
25%
24%
30%
11%
4.1%
4%
4%
4%
—
I/M Data
Avg. HC (ppm) Avg. CO (%)
Pass Fail Pass Fail
77.3
86.7
63.6
74.2
79.4
82.1
78.7
76.1
60.6
61.5
67.1
47.0
497
574
331
452
441
509
383
404
409
470
275
307
0.31
0.29
0.34
0.33
0.27
0.27
0.26
0.28
0.14
0.13
0.18
0.12
NO DATA
2.44
1.09
4.04
3.48
3.11
2.77
3.16
4.72
2.30
2.25
1.90
3.77
% Failure
14%
13%
14%
16%
16%
13%
27%
11%
4.5%
4%
7%
4%
-------
RADIAN
I/M failure rate; however, the trends from model year to model year were
similar. For example, the failure rate for 1981 Chevy and Checker vehicles
was higher than the 1980 failure rate. Considering that the roadside data
were collected in the Summer of 1983 vs. 1982 for the I/M data, the higher
failure rate for the 1981 models is consistent with the I/M data that shows
a higher failure rate for the 81 Chevrolets when the failure rate is stan-
dardized by odometer. The roadside data also showed that the emissions re-
ported in the I/M test appear to be reasonable after the obvious errors
are removed. As shown on Table 2-10, there is little difference between
the average emissions of the passed and failed vehicles in either the road-
side test or the I/M test. The roadside test sometimes includes a tampering
inspection but no 1980 and newer vehicles had obvious evidence of tampering.
Table 2-11 shows the emissions distributions in the roadside
checks. Also like the I/M data, the roadside data show a greater percentage
of vehicles with excessively high CO emissions in the 1981 model year. The
roadside test data also indicate that the 1982 Chevrolets show similar
performance to the 1981 model year vehicles; i.e. greater CO emissions than
o
the 1980 model year vehicles. The problems mentioned in Section 2.4 with
the oxygen sensors and the evaporative purge valves could be responsible
for these trends. It is interesting to note that, unlike the failure rates,
the percent of gross CO emitters in the roadside checks were not higher than
in the I/M tests. Possibly, vehicle driveability problems cause the owner/
operator to repair vehicles with extremely high CO emissions.
The comparison between the 1982 model year vehicles in the road-
side checks and the I/M checks is not valid since the I/M data was based on
much newer vehicles than the roadside data. A more valid comparison would
be between the I/M data on the 1982 model year vehicles and the roadside
test data on the 1983 model year vehicles. As shown, there is little
difference between the failure rates and the emissions distributions of
these two groups.
-------
TABLE 2-11
to
Model Yr.
1980
1981
1982
1983
Make
All
Chevy
Dodge
Ford
All
Chevy
Dodge
Ford
All
Chevy
Dodge
Ford
All
Chevy
Dodge
Ford
EMISSION DISTRIBUTIONS
HC
than 300 ppm
17
20
6
17
0
19
26
16
8
13
18
4
0
1
1
0
—
Ill ROADSIDE CHECKS
than 700 ppm
4
3
6
4
6
7
5
3
5
6
4
0
0
1
0
—
CO
3.0%
8
4
0
15
13
14
18
10
11
14
4
0
0
0 .
0
—
7.0%
0%
0
0
0
3
3
5
2
3
5
0
0
0
0
0
—
Count
195
101
18
71
240
137
38
59
173
125
26
22
200
104
82
14
-------
As noted, the roadside test failure rate for the 1981
Chevrolets was three times higher than I/M tests; however, the roadside
failure rate for the 1981 Dodges and Fords was not elevated to
the same degree. This could indicate that the owners are more aware of
problems with the Chevrolets because of check engine lights and are likely
to obtain repairs prior to the I/M test. However, the random nature of the
roadside tests does not give the owner/operator opportunity to repair the
vehicle even if he knows there is a problem, unless driveability is se-
verely affected.
In summary, the roadside data confirmed that the 1981 and 1982
Chevrolet vehicles show inferior CO and possibly HC performance to the 1980
model year vehicles. The closed-loop fuel control system in the 1981 model
year is likely to be responsible for this change. The roadside check data
also confirmed that the 1981 Dodge 225-6 engine apparently has a carburetor
problem as evidenced by its extremely high failure rate in the roadside
check.
The high failure rates in the roadside checks are troublesome,
considering that the taxis are inspected every four months. Are vehicles
being incorrectly passed — either initially or after repairs; are the in-
use disablement rates so high that the failure rate is justifiable; are
vehicles being re-adjusted after repair? Additional data would be useful
in determining the cause of failure for vehicles inspected in the road-
side checks.
2.4 RESULTS OF MEETINGS WITH MAINTENANCE PERSONNEL IN TAXI FLEETS
As part of the analysis of data from the New York Taxi I/M program,
meetings were held with maintenance personnel in three of the major taxi
fleets:
• Midland (operates 1982 Checkers equipped with 229 V6 Chevy
engines)
43
-------
CORPORATION
• Metro (operates 1983 Chevys equipped with 229 V6 Chevy
engines)
• 57th Street Metropolitan (operates 1983 Dodges equipped with
225 cubic inch 6-cylinder engines)
2.4.1 Fleets Operating Chevrolet/Checker Cabs (Midland and Metro Cab Co.)
Midland Cab Company. Midland Cab Co. has experienced frequent
failures of the evaporative purge valve used in the 1982 Checkers with 229
V6 engines. This valve is located in the line between the canister and the
carburetor and is used to purge the canister vapors into the carburetor.
Apparently the valve is failing in the open position which causes the car-
buretor to run in the equivalent of a full rich mode, and in time the rich
operation of the engine fouls the oxygen sensor. The problem is usually
noticed by an operator complaining about poor gas mileage. Fuel economy is
frequently charted and is typically in the area of ten miles per gallon in
city traffic. When the carburetor fails in the rich position, operators
have noticed that the fuel economy drops down to around 7 miles per gallon.
Some operators have noticed that the check engine light has come on, although
many times operators fail to mention the check engine light. (It does not
always remain illuminated.) The maintenance supervisor said that 1981
Checkers with the same basic engine as the 1982 models did not experience
frequent problems with the evaporative purge valve. The only major problem
that they experienced on the 1981 model was water collecting around the
oxygen sensor, causing it to short out.
Midland diagnoses its engines by reading the codes stored in the
computer using the check engine light procedure that GM developed. The
GM diagnostic system has been reliable and is the primary method used to
identify possible sources of the problem. Usually the computer indicates
a defective oxygen sensor. Further investigation has led to the identifica-
tion of the evaporative purge valve as the main culprit.
44
-------
RADIAN
The Checker cabs experience random failures of the carburetor and
the computer. Midland also has replaced the oxygen sensor every 30,000 to
40,000 miles even if the evaporative purge valve is operating correctly.
During the meeting, maintenance logs on four vehicles were reviewed and
every log showed replacement of the oxygen sensor at approximately 30-40,000
mile intervals, and some showed more frequent intervals. Midland also has
encountered fialures of the catalytic converter. In this case, the problem
was either poor fuel economy or no power because of the collapsed converter.
Up to 50,000 miles they have been obtaining new converters from GM; however,
they also had to purchase many new converters because they accumulate mile-
age at a rate of approximately 50,000 per year.
Metro Cab Company. Metro operates a fleet of 1983 Chevys with the 229 cubic
inch V6 engine. Like Midland Cab Co., Metro also has been experiencing fre-
quent failures of the evaporative purge valve. Metro noted that failures
have been experienced on the 1983 model. Similarly, the problem has been
identified by the poor gas economy, poor engine operation or a check engine
light. In order to diagnose the problem in their vehicles, Metro uses a
Mini-Scanner to read the codes stored in the computer. Metro notes that
most of the problems experienced with the 1983 Chevys are related to the
computer system. They encounter few ignition related problems. Like Mid-
land, Metro has to replace the oxygen sensor at periodic intervals and the
average carburetor lasts 40-50,000 miles before needing replacement or an
9
overhaul. In the new fleet of Chevys (1983 models) approximately 150 our
of 200 have had oxygen sensors replaced. Metro has also had to replace
some catalysts. In some cases they experienced catalyst melting and at
other times the catalyst breaks off and plugs the exhaust. Metro has worked
closely with the AC division of GM in diagnosing the problems with their
vehicles.
Metro Cab Co.'s maintenance program operates independently of the
I/M program. Although the I/M program identifies high emitting vehicles,
Metro felt that these vehicles would be identified in the normal course of
45
-------
action because of driveability or fuel economy problems or illuminated
check engine lights.
2.A.2 Fleets Operating Dodge Cabs (57th Street Metropolitan Cab Company)
The 57th Street Cab Company operates a fleet of 1983 Dodges equip-
ped with the 225 cubic inch 6 cylinder engine. Previously, they operated a
fleet of 1981 Dodges. Unlike the other two fleets, there have been no major
mechanical problems such as the evaporative purge valve in the 1981 or 1983
Dodges. Most of the vehicles that failed the I/M test are repaired by a
carburetor adjustment. Although the 1983 225 engine is closed-loop, the
idle circuit is open-loop. Consequently, idle mixture adjustments are per-
formed in a more traditional manner than adjustments on GM vehicles. No
special equipment is needed to diagnose the problems in the Dodges.
The maintenance supervisor for 57th Street noted that occasionally
carburetors are fouled with carbon. The carbon canister has been identified
as being the source of the carbon. To eliminate the problem, an in-line
filter was placed in the "purge line that runs to the carburetor. Sometimes
broken off oxygen snesors have been found along with vacuum leaks and de-
fective vacuum amplifiers. The 57th Street Cab Company services its vehicles
every 21 days which is the equivalent of every 7,000 miles. This service
includes a complete tuneup, brake shoes, and an exhaust emission test. The
service is done in addition to the I/H program.
2.4.3 Summary of Meetings
All three taxi fleets appeared to provide unbiased answers to
questions raised. The problem the 229 V6 Chevrolet engine has with full
rich failures from defective purge valves appears significant and should
be investigated further. This problem could explain the high failure
rate for 1982 Chevrolets in the roadside checks. A much more in-depth
analysis is needed to accurately determine the cause of failure or to
-------
eoi
determine the seriousness of the evaporative purge valve problem. Addi-
tional information is needed to determine if the problem is confined to
the 229 engine families.
-------
3.0 CONCLUSIONS/RECOMMENDATIONS
3.1 CONCLUSIONS
Several conclusions concerning the high mileage emission control
performance of new technology vehicles can be drawn from this analysis:
1981 and possibly 1982 vehicles with the 229 cubic inch
Chevrolet engines appear to have greater CO and to a lesser
extent greater HC emissions than 1980 model year vehicles.
The closed loop fuel metering system may be responsible for
this trend.
1981 Dodge vehicles with the 225 cubic inch engines also
appear to have greater CO and HC emissions than comparable
1980 models.
1981 Ford vehicles have very similar HC and CO emissions
to the 1980 models which is expected because of similar
fuel metering system designs.
The failure rate in the I/M test is strongly dependent
on odometer reading.
The HC and CO emission levels of vehicles that pass the I/M
test are not strongly affected by odometer.
After repair emission levels of failed vehicles are similar
to the emission levels of vehicles that pass the I/M test.
Although a vehicle could have failed up to four times,
most vehicles that failed only failed once. 1981 Dodge
models show the greatest tendency towards repeat failures.
A majority of vehicles emitting moderately high amounts
of CO also emit moderately high amounts of HC; however,
a majority of the moderately high HC emitters do not emit
moderately high amounts of CO. Considering that excess
HC emissions are caused by factors in addition to fuel
enrichment, this finding is logical.
The failure rates in the roadside checks are much greater
than the failure rate in the I/M test. However, the road-
side checks do not show significantly greater percentages
of gross emitting vehicles (greater than 7.0% or 700 ppm HC)
-------
The greater failure rates in roadside checks could be
be the results of:
—improper passing of vehicles in the I/M test
—inadequate repairs
—readjustments after repairs
—excessive disablement/deterioration rates
—greater odometer readings on vehicles tested in
roadside checks.
• For Chevrolet and Ford vehicles, the percentage of gross CO
emitters (greater than 7.0%) does not appear to increase with
odometer after approximately 50,000 miles are accumulated.
• 1982 Chevrolets with the 229 cubic inch engine have had
frequent failures of the evaporative purge valves which have
resulted in excessive HC and CO emissions and fuel consumption.
Oxygen sensors, ECMs and carburetors also have been replaced
more frequently than manufacturer's expected intervals.
3.2 RECOMMENDATIONS
Based upon this analysis, the following recommendations are made
for additional analysis:
• Perform additional surveys of maintenance personnel to
better determine:
—the type of repairs required to bring new technology
vehicles into compliance, and
—frequency of defects/disablements in new technology
vehicles.
• Collect additional data during roadside checks on:
—odometer
—check engine light status
—trouble codes (use a Mini-Scanner to read the codes)
—plumbtesmo/other tampering
—other diagnostic checks.
• Analyze I/M data collected beyond May 1983 to determine
trends for 1982 and 1983 vehicles.
• Perform FTP's on taxis at DEP's Frost St. Lab.
-------
Investigate other GM engine families/evap families to
determine the extent of the evaporative purge valve
problem.
50
------- |