United States       Air and Radiation       EPA420-R-01-002
         Environmental Protection                January 2001
         Agency
&EPA    Final Technical Support
          Document for
          "Amendments to Vehicle
          Inspection Maintenance
          Program Requirements
          Incorporating the
          Onboard Diagnostic
          Check"
                               > Printed on Recycled Paper

-------
                                                     EPA420-R-01-002
                                                         January 2001
        Final Technical Support Document for
 "Amendments to Vehicle Inspection Maintenance
       Program Requirements Incorporating the
              Onboard Diagnostic Check"
               Edward Gardetto, Ted Trimble, Martin Reineman
                               and
                       David Sosnowski (Editor)
               Transportation and Regional Programs Division
                  Office of Transportation and Air Quality
                  U.S. Environmental Protection Agency
                             NOTICE

  This technical report does not necessarily represent final EPA decisions or positions.
It is intended to present technical analysis of issues using data that 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.

-------
                                TABLE OF CONTENTS




1.0          Overview




2.0          OBD-I/M Pilot 1: OBD Checks and Tailpipe Testing




2.1          Summary of Goals and Conclusions




2.2          Background




2.3          Vehicle Sampling




2.3.1         Methodology




2.3.2         Results




2.4          Vehicle Testing




2.4.1         Methodology




2.4.2         Results




2.4.2.1       Emission Reductions




2.4.2.2       OBD and Preventative Maintenance




2.4.2.3       OBD and Errors of Omission ("False Passes")




2.4.2.4       OBD and Errors of Commission ("False Failures")




2.5          Conclusions




3.0          OBD-I/M Pilot 2: OBD Checks and Evaporative Emission Testing




3.1          Summary of Goals and Conclusions




3.2          Background




3.3          Vehicle Sampling




3.3.1         Methodology




3.3.2         Results

-------
3.4
3.4.1
3.4.2
3.5
4.0
4.1
4.2
4.2.1
4.2.2
4.2.3
4.3
4.3.1
4.3.2
Vehicle Testing
Methodology
Results
Conclusions
OBD-I/M Pilot 3: Analyzing the Wisconsin OBD-I/M Program Experience
Summary of Goals and Conclusions
DLC Location
Background
Results
Conclusions
Vehicle Readiness
Background
Results
4.3.2.1       Readiness




4.3.2.2       MIL-on and IM240 Failure Rates




4.3.3         Conclusions




4.4          Gas Cap Testing vs. OBD-I/M




4.4.1         Background




4.4.2         Results




4.4.3         Conclusions

-------
1.0    Overview

       On September 20, 2000, EPA proposed to revise existing Motor Vehicle
Inspection/Maintenance (I/M) requirements related to the incorporation of Onboard Diagnostic
(OBD) checks into such programs.  Among other things, the proposed regulatory revisions —
once adopted - will accomplish the following:

1)                   Allow the OBD-I/M check to replace tailpipe and evaporative system
              testing on OBD-equipped vehicles (with the exception of the gas cap evaporative
              system test); and

1)                   Revise the failure and rejection criteria for the OBD-I/M check.

       This Technical Support Document (TSD) provides EPA's technical justification for these
amendments, based upon the Agency's findings gathered during three separate OBD-I/M pilot
studies. These three pilot studies focused on the following aspects of OBD-I/M testing: 1) OBD-
I/M's effectiveness as compared to existing exhaust emission testing; 2) OBD-I/M's
effectiveness in identifying faults in the evaporative system;  and 3) the unique implementation
issues associated with incorporating checks of the OBD system into a traditional I/M setting.
The results of EPA's pilot testing were shared while still in progress with members of the  OBD
workgroup of the Mobile Source Technical Review Subcommittee established under the Federal
Advisory Committee Act (FACA).  The OBD workgroup's membership includes representatives
from the testing and repair industries, vehicle manufacturers, the states, EPA, scan tool
manufacturers, the academic community, private consultants, and providers of OBD technician
training. Feedback from the workgroup was used to help guide the progress of the pilots, to
interpret the results along the way, and to develop the proposed amendments to the rule's
existing OBD-I/M requirements.

       This TSD is divided into four main sections. Following the overview section are three
main sections that coincide with the three pilots identified above.
2.0    OBD-I/M Pilot 1: OBD Checks and Tailpipe Testing

2.1    Summary of Goals and Conclusions

       Between September 1997 and October 1999, EPA recruited 201 in-use MY 1996 and
newer OBD-equipped vehicles and performed an EVI240 transient test, an OBD-I/M inspection,
and an abbreviated version of the vehicle certification test known as the Federal Test Procedure
(FTP) on each vehicle1. Vehicles identified as needing repairs were repaired after this initial test
        Because the focus of this pilot was comparing OBD-I/M checks to more traditional, tailpipe-based I/M tests, only the
tailpipe portion of the FTP was performed for this pilot study. The evaporative emission portion of the FTP was performed on a
smaller sample of vehicles included in the separate, OBD evaporative pilot discussed in section 3 of this TSD.

-------
sequence and then subjected to the same sequence again, after repairs. The goal of this test
program was to answer the following question:

       Is it necessary to conduct both an OBD-I/M check and the IM240 (or some other
       tailpipe test) on OBD-equipped vehicles?

To answer this question, EPA had to determine if the OBD-I/M check: 1) failed vehicles it
should have passed; 2) passed vehicles it should have failed; and 3) whether the rate at which it
did either of these was higher than, lower than, or the same as the EVI240.  Since it is widely
considered the most accurate tailpipe-based I/M test available, the EVI240 was used to represent
the "best case" scenario with regard to tailpipe testing in general.

       Based upon the test results detailed in this TSD, EPA concluded that it should not require
both an OBD-I/M check and the EVI240 (or other tailpipe test) on MY 1996 and newer OBD-
equipped vehicles2.  Specifically, EPA found that while the OBD-I/M check did falsely pass and
falsely fail some vehicles, in both cases the percentage of vehicles impacted was smaller than
would be the case with the EVI240 and other tailpipe tests. Furthermore, even though the EVI240
caught some of the very few vehicles missed by the OBD-I/M check, the additional cost that
would result from subjecting OBD-equipped vehicles to two tests instead of one will likely
outweigh any additional benefit that may be achieved.  Lastly, the emission reductions available
from basing repairs on the OBD-I/M check appear to be at least as large as the emission
reductions obtained from EVI240-triggered repairs on OBD-equipped vehicles.
2.2    Background

       Under the Clean Air Act as amended in 1990 (CAA), EPA was required to promulgate
two categories of regulations related to OBD. The first regulated vehicle manufacturers and
required the installation of the OBD system on all new light-duty vehicles and light-duty trucks.
The second regulated state I/M programs and required that all such programs - whether basic or
enhanced - include an inspection of the OBD computer for vehicles so equipped. In 1992, when
EPA published its original I/M rule, federal OBD certification requirements were still being
developed, and so sections were reserved in the I/M rule to address the OBD-I/M testing
requirement at a later date. Since the 1992 I/M rule was published, EPA has amended it twice to
address OBD-I/M testing requirements - first, on August 6,  1996, and again on May 4, 1998.

       In the 1996 amendments, EPA described  how OBD was to be addressed as part of the
basic and enhanced I/M performance standards and established OBD-I/M SIP requirements. The
1996 amendments also specified data collection,  analysis, and summary reporting requirements
for the OBD-I/M testing element; established OBD test equipment requirements and the OBD
test result reporting format; and identified those conditions that would result in either an OBD-
        Although EPA does not intend to require dual testing of OBD-equipped vehicles for reasons detailed in this TSD,
states wishing to dual test MY 1996 and newer vehicles may do so. EPA will work with individual states to determine whether
or not additional credit is warranted on a case-by-case basis.

-------
I/M failure or rejection. Lastly, the 1996 amendments established January 1, 1998 as the
deadline by which most I/M programs were to begin OBD-I/M checks, though vehicles were not
required to be failed based upon the OBD-I/M check until January  1, 20003. The data gathered
by state programs between January 1, 1998 through December 31,  1999 was to be used to assess
the effectiveness of the OBD-I/M check relative to the EVI240.

       Subsequent to the 1996 amendments, the I/M test environment changed significantly,
with the result that use of the EVI240 was not as prevalent as had once been expected. In the
same time frame, EPA discovered that the EVI240 test as originally designed might lead to false
failures for some vehicles due to insufficient preconditioning4, and as a result the test itself might
not be as effective as once thought. This latter issue suggested that evaluating the OBD-I/M
check based upon a comparison to the EVI240 could unfairly penalize the OBD-I/M check.
Members of the OBD workgroup5 (established under the Federal Advisory Committee Act)
raised  similar concerns regarding the appropriateness of comparing the OBD-I/M check to a "hot
start" test like the EVI240 as opposed to the FTP, which is a "cold start" test and the standard to
which new cars are certified. Furthermore, OBD design requirements are based in part on
detecting emission failures which are directly related to the FTP6.

       As a result of these changing conditions and concerns, EPA revisited its original plans for
evaluating the effectiveness of OBD-I/M testing by comparing it to state-gathered EVI240
inspection lane data. Instead, EPA decided to pursue the test program described here, in order to
alleviate the need for states to run dual tests (tailpipe and OBD) in their I/M lanes merely as  a
form of data gathering7.  The May 4, 1998 amendments to the I/M rule addressed this change by
delaying the date by which I/M programs were to begin OBD testing to no later than January 1,
     o   	
2001  .  To generate the necessary data for comparison, EPA and its research partners conducted
sample testing at four different labs: the National Vehicles and Fuels Emissions Laboratory
(NVFEL) in Ann Arbor, Michigan; the Automotive Testing Laboratory (ATL) in Mesa, Arizona;
the Colorado Department of Health Laboratory (CDH) in Aurora, Colorado; and the  California
Air Resources Board (CARB) test facilities in El Monte, California.
        Programs qualifying for the Ozone Transport Region (OTR) low enhanced performance standard were allowed to
postpone mandatory OBD-I/M testing until January 1, 1999.

       4
        SAE paper 962091, "Preconditioning Effects on I/M Test Results Using IM240 and ASM Procedures," Heirigs,
Philip; Gordon, Jay.

        The OBD workgroup is a subgroup of the Mobile Sources Review Subcommittee, which was itself formed under the
Federal Advisory Committee Act  (FACA) in order to advise the Agency on technical matters.

        The exhaust standards for OBD require that a dashboard light be illuminated under circumstances which could lead
the vehicle to exceed its certification standards by 1.5 times the standard.

       7 Federal Register Volume 61, No. 152; August 6,1996; page 40940.

       o
        In its September 20, 2000 notice of proposed rulemaking, EPA proposed to extend this deadline to January 1, 2002 in
addition to the other revisions discussed in this TSD — in part due to the proximity of the current deadline to the release of these
findings and the proposed amendments.

-------
2.3    Vehicle Sampling
2.3.1   Methodology

     Based upon advice from the Mobile Source Review Subcommittee, EPA decided to
conduct an FTP-based test program with a minimum of 200 vehicles9.  The goal of the test
program was to determine how well OBD-I/M testing compared to tailpipe I/M testing. Because
the IM240 is generally accepted as the most accurate tailpipe test available for use by I/M
programs, EPA decided that a comparison to the IM240 would be considered a "best case"
comparison for establishing relative tailpipe test effectiveness
10
       The recruitment of vehicles for pilot testing was controlled by the need to answer two
basic questions concerning the effectiveness of OBD-I/M testing relative to traditional tailpipe
tests: 1) Do vehicles identified by the OBD-I/M check actually need repair, and 2) Does the
OBD-I/M check miss high emitters that would be caught by traditional tailpipe testing? To
address the first question, EPA recruited vehicles identified by OBD  as possible high emitters in
need of repair (i.e., vehicles with the malfunction indicator light — or MIL — illuminated).  To
address the second question, EPA focused on those vehicles that failed a properly preconditioned
EVI240, but for which no MIL was illuminated.

       Concern about the relatively small sample size and the degree to which it would represent
the fleet at large led EPA to weight its sample based upon manufacturer production for the six
largest producers. The remaining manufacturers represent a small percentage (<10%) of the
entire fleet and  are represented by the category "other."  The sample was also weighted to
account for the  growing fraction of light-duty trucks (LDTs) in the fleet.  Table 1 below was
developed to act as a target for the 200 vehicle sample based on 1997 vehicle sales

                      Table 1: Procurement  Goals Based on Production
                   11
MFR
LDV
LDT
Total
GM
35
27
62
Ford
21
29
50
Daimler-
Chrysler
10
20
30
Toyota
11
5
16
Honda
11
1
12
Nissan
7
3
10
Other
10
10
20
Total
105
95
200
        Mobile Source Review Subcommittee meeting of 7716/97.

         Sierra Research Report under EPA Contract 68-C4-0056; WA 2-03; "Development of a Proposed Procedure for
Determining the Equivalency of Alternative Inspection and Maintenance Programs," page 7.
       11
         Automotive Industries: February, 1998, page 17.

-------
       For vehicles with the MIL illuminated, any vehicle with a non-evaporative, emissions-
related trouble code commanding the MIL on was accepted into the program12. These vehicles
were selected without knowledge of their tailpipe emissions.  However, because misfire codes
are relatively common among the MILs observed in the field, an upper limit of 25% was
established for misfire codes per manufacturer represented in the overall sample.  This 25% limit
was  derived from a fleet survey of over 160,000 vehicles in Wisconsin and represents the relative
occurrence of misfire diagnostic trouble codes (DTCs) seen in the I/M lane13.

       The pilot study also called for the recruitment of vehicles with (potentially) high
emissions and no MIL illumination. These no-MIL/high-emitter vehicles were identified and
recruited based upon two primary criteria: 1) High LANE24014 test results or 2) other
characteristics which  experience suggested would result in high emissions (i.e., high mileage,
and/or driveability problems).  Using the first criteria, the most stringent EVI240 standards15 were
applied even though the actual state I/M program did not fail vehicles based on those values.

       Additional,  potential high emitters were recruited based upon very high mileage, or a
mechanic's report that a particular vehicle was running poorly.  Because NVFEL  is not located
near an operating I/M program, the Ann Arbor lab used this secondary method as its primary
means for identifying and recruiting no-MIL/high-emitter vehicles for subsequent testing.  ATL
and CDH also attempted to find additional no-MIL/high-emitter vehicles based upon these more
qualitative (as opposed to quantitative) criteria. On the vehicles with suspected high emissions
but no lane-based tailpipe data, the LAB240 was again used to screen which vehicles were kept
in the sample and which were released from further participation in the pilot study.
2.3.2  Results

       201 vehicle test slots were filled during the program versus the 200-vehicle target (2
vehicle slots were filled by the same vehicle, which was recruited twice, six months apart, with
different problems each time).  Table 2 represents the breakdown of this sample by
manufacturer and is also segregated into cars (LDVs) versus trucks (LDTs). The category of
"other" is made up of the following manufacturers followed by the number of sample vehicles
         "Recommended Practice for Diagnostic Trouble Code Definitions," SAE J2012, Society of Automotive Engineers,
Inc., revised March 1999.

         "Analyses of the OBDII Data Collected from the Wisconsin I/M Lanes," Trimble, Ted, Environmental Engineer,
U.S. EPA, August 2000.

         EPA distinguishes between "LANE240" tests (i.e., those conducted by a commercial testing contractor as part of the
routine operation of an existing program) and "LAB240" tests (i.e., those conducted under controlled, laboratory conditions for
test type comparison and evaluation purposes). More information concerning the differences between "LANE" and "LAB"
IM240s is available in Appendix 3.

         "EPA I/M Briefing Book: Everything You Ever Wanted to Know About Inspection and Maintenance," EPA-AA-
ESPD-IM-94-1226, Section 4, page 10. U.S. EPA, Office of Air and Radiation, February 1995.

-------
from each: Mazda (2), Volkswagen (3), Isuzu (2), Hyundai (3), Kia (1), Saab (1), Volvo (1), and
Suzuki (3). Of the 201 vehicles in the sample, 193 were recruited as MIL-on vehicles, while the
remaining 8 were recruited as no-MIL/high-emitter vehicles.
Table 2: Descrii
MFR
LDV
LDT
Total
GM
*45
"(128%)
18
(66%)
63
(102%)
Ford
31 (148%)
28
(96%)
59(116%)
Dtion of Sample by Manufacturer and Type
Daimler-
Chrysler
22
(220%)
16
(80%)
38
(127%)
Toyota
5
(45%)
1
(20%)
6
(38%)
Honda
8
(73%)
0
(0%)
8
(67%)
Nissan
7
(100%)
4
(133%)
11 (110%)
Other
14 (140%)
2
(20%)
16
(80%)
Total
132
(126%)
69
(73%)
201
(100%)
* = number procured ** = percent of goal

       The sample breakout by model year and by minimum, maximum, and average odometer
readings are listed in tables 2a and 2b, respectively.

                            Table 2a: Breakout by model year

LDV
LDT
1996
28
27
1997
33
22
1998
38
14
1999
32
6
2000
1
0
                              Table 2b: Odometer readings


MINIMUM
AVERAGE
MAXIMUM
LDV

29
26,440
93,575
LDT

3,981
54,505
245,000
       As the results of these procurement efforts are considered, it is important to keep in mind
the relatively low age of the fleet of vehicles being evaluated (i.e., MY 1996 and newer).  The
relatively low age (four years old or newer) and mileage (average = 37,000 miles) of the vehicle
population targeted by the study led to fewer MIL illuminations in the general fleet than would
be expected. We do not believe, however, that the relative newness of the vehicles that
participated in this pilot will change the direction of the conclusions drawn from this study.
Specifically, we do not believe that OBD systems will prove somehow less effective at
identifying vehicles in need of repair as the  OBD-equipped population ages.  This is because the
OBD system itself (as opposed to the hardware the OBD system monitors) is primarily a self-
contained, software-based system and not likely to be subject to substantial degradation due to
aging.  The practical impact of the  newness of these vehicles was the limited  exposure of the
hardware being monitored to the real world effects of heat, cold, water, salt, etc. However,
because the possibility exists that multiple-component aging may have a negative, synergistic

-------
effect on OBD's ability to detect vehicles which have high emissions in the future, continued
study of the  OBD-equipped fleet as it ages and accumulates mileage seems warranted16.
2.4    Vehicle Testing

2.4.1  Methodology

       During the two-year period from September 1997 through October 1999, EPA and its
research partners conducted sample testing at four different laboratories across the country: the
National Vehicles and Fuels Emissions Laboratory (NVFEL) in Ann Arbor, Michigan; the
Automotive Testing Laboratory (ATL) in Mesa, Arizona; the Colorado Department of Health
Laboratory (CDH) in Aurora, Colorado; and the California Air Resources Board (CARB) test
facilities in El Monte, California.  FTP testing was performed using the methods described in
CFR 86.130-96 with the exception that no diurnal heat build or SHED testing was conducted17.
EVI240 testing was done in accordance with EPA Technical Guidance EPA-AA-RSPD-EVI-98-1.
OBD information was gathered using scan tools from various manufacturers complying with the
standards established by SAE 1978. Maintenance was performed at either the original
manufacturer's dealership or by mechanics following the manufacturer's available service
information.

       MIL-on vehicles were inspected when they first arrived at the lab using the LAB240
procedure and the fuel that was already in the vehicle's tank.  The tanks were then drained of in-
use fuel and refilled with indolene test fuel.  The vehicles then received a standard FTP and a
second LAB240. This provided the "As-Received" emissions profile of the vehicle. The FTP
was the standard for comparing actual emissions reductions; the EVI240 and the OBD-I/M test
results were only used to identify vehicles as either "pass" or "fail," relative to the respective I/M
test type. Vehicles identified as "failures" based upon either their tailpipe or OBD-I/M results
were then sent for repairs, after which they were again tested on the FTP to determine their "after
repair" emission levels.  Any difference measured between the two FTPs represented the air
quality improvement18 or emission benefit. Most maintenance was performed following original
equipment manufacturer (OEM) published procedures, while in some cases, this information was
supplemented through consultation with OEM engineers.  In cases where DTCs were present but
          In recognition of the potential impact of high mileage on OBD effectiveness, EPA recently completed testing and
has begun analyzing the results from a study of 43 OBD-equipped vehicles with mileages of approximately 100,000 miles to as
high as 273,000 miles. Early indications suggest that high mileage does not have a noticeable impact on the effectiveness of the
OBD system to detect needed repairs. With regard to the impact of simple aging, EPA recognizes the value of gathering
additional information on the durability of OBD systems as they age, and stands ready to revise the OBD-I/M requirements
should future study suggest such is warranted.

          Sealed Housing for Evaporative Determination (SHED) testing was conducted on a different subset of vehicles, as
part of the evaporative system pilot, which is addressed in section 3.0 of this TSD.

       1 &
          Two vehicles were too dirty and/or running too poorly to test on the FTP. Since it was not possible to establish an
accurate "before repair" emission measurement for these vehicles, no air quality benefit was assigned to them. (See discussion in
Table 10, and Appendix 4).

                                             10

-------
the mechanics could find nothing in need of repair or replacement, the OBD system was allowed
to reset and was then monitored to verify the absence of any OBD- or emission-related problem.
(See Appendix 1 for additional test sequence details.)
2.4.2  Results
              Table 3 shows the number of MIL-on vehicles with initial FTP readings
exceeding 1 and 1.5 times the applicable certification standard (the latter being a subset of the
former).  One and a half times the standard was chosen as a criterion for comparison because the
certification requirements for OBD specify that the MIL is to be illuminated if a problem is
detected that could  result in emissions exceeding 1.5 times the exhaust certification standard.
Table 3 also shows the subset of MIL-on vehicles for which the MIL cleared on its own, after
being recruited but  before any repairs could be attempted
19
                    Table 3: MIL-On Vehicles vs. FTP and 1.5 Times FTP

LDV
LDT
Total
MIL on
128
66
194
MIL self-cleared
5
6
11
FTP > 1 times cert.

40
18
58
FTP> 1.5 times cert.

21
10
31
       As noted in footnote 18, Table 3 includes two vehicles which are assumed to have failed
their as-received FTP at over 1.5 times the applicable tailpipe standards. These vehicles could
not be driven on the FTP trace and therefore no tailpipe readings are available. A description of
these two vehicles is included in Appendix 4.

       EPA also recruited vehicles with suspected high emissions but no MIL illumination.
Using the screening methods discussed earlier, 8 vehicles qualified to represent this category.
Table 4 shows the number of MIL-off vehicles with initial FTP readings exceeding 1 and 1.5
times the applicable certification standard (the latter being a subset of the former).

        Table 4: MIL-Off Vehicles vs. Certification and 1.5 Times Certification Standard


LDV
LDT2°
Total
MIL-off

4
4
8
FTP > 1 times cert.

2
3
5
FTP > 1.5 times cert.

1
3
4
       19
         MIL self-clearing is a design feature of OBD systems, and is the way the system accounts for intermittent problems
(like misfire) that may occur once under atypical vehicle operation, but then seem to disappear during more normal driving. This
aspect of OBD is discussed in more detail in section 2.4.2.2, "OBD and Preventative Maintenance."
       20
         All three figures in this row include vehicle CDH04, which was recruited with no MIL but is considered an accurate
                                 OBD identification (see Appendix 4).
                                             11

-------
       The ability of OBD to correctly identify vehicles which are emitting at levels
significantly over their applicable certification standard (2 times or higher) was also investigated.
The subgroup of vehicles making up this sample is listed in Table 5 below.

       Note that OBD-I/M missed two of a total of 21 vehicles identified as gross emitters with
FTP scores of two or more times their certification standards (i.e., OBD successfully identified
90% of these gross emitters). Note further that the LAB240 missed four times as many vehicles
from this category, and identified only 62% of the grossest emitters.  The two vehicles missed by
OBD-I/M - one LDV (CDH03) and one LOT (CDH33) -- were correctly failed by the LAB240,
while  the eight LDVs missed by the LAB240 were correctly failed by OBD-I/M.

         Table 5:  Vehicles with FTP Results Over 2 Times the Certification Standard

LDV
LDT
As measured by FTP
15
6
Identified by OBD
14
5
Identified by LAB240
7
6
       EPA also collected data on the degree to which repairing vehicles solely to turn the MIL
off resulted in emission reductions that changed FTP-failing vehicles into FTP-passing vehicles.
Of the 15 LDVs with emissions over twice their standard (>2xFTP), 12 (or 80%) were repaired
to below certification levels by targeting repairs solely at correcting the conditions that led to the
MIL being on. All 14 of the >2xFTP LDVs identified by OBD-I/M that were repaired based
upon OBD-targeted repairs tested below 1.5 times the applicable standard after those repairs
(i.e., below the minimum required detection threshold for OBD). The two vehicles that remained
above the FTP standard (but below 1.5 times that standard) after repairs to turn off the MIL are
discussed in section 2.4.2.3 of this TSD (Table 9).

2.4.2.1       Emission Reductions

       The emissions reductions attributable to OBD- and LAB240-triggered repairs performed
as part of this pilot are presented in Table 6 below. Vehicles which failed both the LAB240 and
OBD-I/M tests are included in both the EVI240 and OBD categories of repair data presented
below. Nevertheless, vehicles that failed both the LAB240 and OBD-I/M tests were repaired
based mainly on the OBD codes and therefore the EVI240 repair data are not completely
independent of OBD effects.  Wholly separate from its use in the I/M arena, OBD is a powerful
tool for diagnosing and  repairing vehicles in the real world, and more and more repair
technicians are using the OBD scan as their starting point for diagnosing vehicles prior to repair.
Although EPA could have required technicians to ignore OBD when attempting to fix vehicles
identified by EVI240 for repair, we believe that such a restriction would be artificial and
unnecessarily limiting.

       The varying sample sizes listed above for non-methane hydrocarbon (NMHC), total
hydrocarbon (THC), and CO2 are due to the fact that while NVFEL and ATL measured all five
pollutant categories for the pilot vehicles they tested, the CDH did not measure NMHC and
CARS did not measure  THC or CO2 for their respective vehicles. The THC and NMHC

                                           12

-------
averages quoted below are therefore based upon the subset of vehicles for which those emissions
were measured. LDV and LDT data are presented separately because of significant differences
in their certification standards and control strategies.

  Table 6: Average Reductions and Fuel Economy Improvement from OBD vs. EVI240 Repairs
LDV
Avg. OBD
n=126
Avg. IM240
n=7
LDT
Avg. OBD
n=65
Avg. IM240
n=7
THC
(gpm)
0.138
n=114
1.04
n=7

0.11
n=65
0.84
n=7
NMHC
(gpm)
0.1
n=lll
0.9
n=5

0.05
n=49
0.37
n=5
CO
(gpm)
2.4
n=126
15.4
n=7

1.56
n=65
10.47
n=7
NOx
(gpm)
0.1
n=126
0.6
n=7

0.13
n=65
0.60
n=7
C02
(gpm)
6.47
n=114
14.71
n=7

-2.66
n=64
8.27
n=7
FE Increase
(mpg)
0.53
n=114
2.36
n=7

0.03
n=64
0.79
n=7
       Another way to look at the same repair reductions is to quantify the total grams per mile
(gpm) reduced over the course of the study as opposed to average gpm reductions. Table 7
below quantifies the total gpm reductions attributable to repairs triggered by either OBD or
EVI240, per pollutant category and segregated by LDVs and LDTs. If a vehicle failed both the
LAB240 and the OBD tests, the gpm reductions resulting from repairs were counted under both
categories.  Note that with the exception  of fuel economy improvement on LDTs, OBD-triggered
repairs consistently produced more total reductions and fuel economy improvement than did the
IM240-triggered repairs.

     Table 7:  Summation of Reductions Associated with OBD vs. IM240 Triggered Repairs
LDV
OBD reductions
IM240 reductions
LDT
OBD reductions
IM240 reductions
THC
(gpm)
15.7
10.0

7.5
6.4
NMHC
(gpm)
11.1
4.9

2.6
1.9
CO
(gpm)
298
111

101
90
NOx
(gpm)
12.1
5.4

8.2
7.1
C02
(gpm)
737
25

43
61
FE Increase
(mpg)
60
27

2
6
       It should be pointed out that in its comparison of the emission reductions attributable to
the OBD-I/M check versus the EVI240, the OBD tailpipe study was biased in favor of the EVI240
to ensure that the conclusions drawn regarding the OBD-I/M check's relative effectiveness were
conservative. Specifically, when a vehicle was identified as a likely EVI240 false failure based
upon a comparison of LANE240 and LAB240 test results, that vehicle was then dismissed from
further participation in the study. As a result, the gpm emission reductions attributed to EVI240
were not "watered down" down by the false failures noted between the LANE- and LAB240s.
Conversely, potential OBD-I/M check false failures were included in the sample and were
                                          13

-------
actively recruited.  Therefore, the gpm reductions attributed to either test based upon this pilot
really do represent the "best case" scenario for the IM240 and the "worst case" scenario for the
OBD-I/M check.
2.4.2.2
OBD and Preventative Maintenance
       As a matter of design, OBD should be able to identify the need for repairs and/or
maintenance prior to actual increased emissions. This is because OBD monitors the performance
of individual emission control components, several of which may need to fail in sequence, or
over a period of time before the problem shows up at the tailpipe. For example, a periodic
misfire might not lead to immediate increases in emissions, but eventually can destroy the
catalyst, at which time tailpipe emissions will increase substantially (as will the likely cost of
repairs).  Traditional tailpipe tests are less capable of identifying this kind of preventative repair,
because such tests rely exclusively upon measurement of post-catalyst tailpipe emissions.
Therefore, with traditional tailpipe tests, a relatively inexpensive problem to begin with may
become critical before it can be detected.

       The tailpipe pilot evaluated this aspect of OBD, and Table 8 lists the results of
maintenance performed on vehicles with tailpipe emissions below the applicable certification
standards. Vehicles for which EPA was unable to identify or reproduce the condition that led to
the original MIL illumination are designated below as MNR  (for "Malfunction Not
Reproduced").  Vehicles for which the MIL went out on its own after procurement but prior to
attempted repair are designated below as "MIL  self-cleared." See Appendix 2 for a list of the
parts replaced on these vehicles OBD identified as needing maintenance.
Table 8: Maintenance asi

LDV
LDT
Total
MIL on/FTP pass
88
48
136
Dect of OBD MIL illumination identification
Broken paiKs) found
63
34
97
MNR
25
14
39
MIL self-cleared
3
6
9
       In considering these results, it is important to understand that a MIL going out on its own
is considered a part of the normal operation of the system; it is not necessarily an indication that
the OBD system itself is having a problem. Every mechanic knows that vehicles are complex
systems and can experience intermittent or transient problems that seem to go away on their own.
And most motorists have had the experience of having a problem that mysteriously "goes away"
the second they take it into the shop. Perhaps the vehicle has been put under an unusually high
load, the fuel quality is below par,  or the vehicle is otherwise being operated under atypical
conditions. The OBD system is designed to account for intermittent problems by setting a code
when a problem is first detected (for example, a misfire) and then to monitor the vehicle to see if
the problem recurs after a certain number of key-on/key-off cycles. If the problem does not
recur, the system is allowed to extinguish the MIL, though a record of the problem is recorded
and retained by the OBD system for a certain period of time after the MIL is turned off,
                                           14

-------
depending upon the nature of the code.  The last column of Table 8 above - "MIL self-cleared" -
represents this particular subset of MNR vehicles recruited as part of the tailpipe pilot. Given the
difficulty EPA had in finding MY 1996+ vehicles with the MIL on for recruitment, we may have
tended to procure vehicles as soon as the MIL turned on (and before the OBD system had a
chance to clear itself if the problem detected was an intermittent or transient condition), which
does not reflect the anticipated experience of an OBD-I/M program. As a result, EPA believes
its sample may have been biased in the direction of recruiting vehicles with MILs lit for
intermittent problems.
2.4.2.3       OBD and Errors of Omission ("False Passes")

       During the pilot program, 4 vehicles were found with no MILs illuminated or DTCs set,
but which nevertheless had tailpipe emissions exceeding both their certification standards and the
1.5 times the certification level at which OBD is supposed to command the MIL to illuminate
and set relevant DTCs.  A fifth vehicle was also found to have high emissions and no visible
MIL illumination, though when this vehicle was scanned, it was found that the MIL was, indeed,
commanded on, but had not illuminated due to a short in the system.  All 5 of these vehicles are
listed in Table 9 below along with a brief summary of the cause(s) of their high emissions.
       The first two vehicles (CDH03 and CDH33) had high emissions  and no MIL or pending
DTCs prior to repair, while the next two vehicles (ATL120 and ATL130) arrived with the MIL
on. In the case of ATL120, the MIL was extinguished by repairs, but the vehicle still produced
high emissions after these repairs. In the case of ATL130, a diagnostic scan revealed nothing to
fix and the MIL did not re-light after being cleared by the scanner, even  though the vehicle's
emissions were still high after the MIL was cleared. And the last vehicle (CDH04) could not be
driven on the FTP because it stalled in third gear, but was assumed to be a high emitter due to the
fact that it produced a plume of black smoke when tested on the LAB240.

       CDH03 is considered an OBD error of omission due to its emission levels and the lack of
MIL illumination and DTCs.  The repair of the oxygen sensor returned this vehicle to acceptable
emissions level.  Further investigation of this problem by Daimler-Chrysler engineers found an
unanticipated failure mode of the rear oxygen sensor. Daimler-Chrysler found that this failure
mode would be detected by all later OBD systems in their product line.  No additional examples
of this type of oxygen sensor failure mode were located in this test program.
                                          15

-------
                           Table 9: Discussion of Specific Vehicles
Vehicle
CDH03: 1996 Chrysler Neon;
86,236 miles; LANE240 failure
CDH33: 1 997 Daimler-Chrysler
1500 Pick-up truck, odometer
1 13,543; LAB IM240 failure
ATL120: 1 997 GM Grand Am;
47,173 miles; MIL extinguished
after diagnostics.
ATL130: 1996 IsuzuHombre
(GM system); 235,000 miles; MIL
extinguished after diagnostics.
CDH04: 1996 GM S10 Pickup
Truck; 27,063 miles; LANE240
failure
FTP Emissions
(As received)
THC/NMHC/CO /NOx
FTP: 1.73/XX/52.0/0.25
(As received)
FTP: 0.55/xx/12.8/2.9
Standard: xx/0.4/5.5/0.97

(Post diagnostics; no repair)
FTP: 0.14/0.12/1.6/0.97
Standard: xx/0 .25/3.4/0.4

(Post repair)
FTP: 0.5/0.39/17.1/0.6
Standard: xx/0. 31/4.2/0.6

Could not be driven on FTP;
projected failure (See Appendix 4)
Problem found
OBD error of omission; unanticipated oxygen
sensor failure; problem fixed for later model years.
THC was < 1 .5 times standard (NMHC unknown)
but CO and NOx > 1 . 5 times. See catalyst monitor
discussion below.
No problem found during diagnostics and MIL did
not reset after clearing, even though NOx was
above OBD trigger level. (HC and CO remained
below the OBD trigger level.)
On post-repair FTP with MIL off, CO was still >
1.5 times standard, while HC fell below the OBD
trigger level. See catalyst monitor discussion
below.
MIL commanded on, but electrical short prevented
illumination; would be caught by OBD-I/M scan.
       The issue with CDH33, ATL130, and ATL120 seems to be a matter of timing and the
way that catalyst efficiency losses are monitored by OBD21.  Currently, catalyst monitors only
target HC to establish catalyst efficiency22 based on the fact that the vast majority of possible
failure modes leading to increased CO and NOx emissions from the catalyst will also eventually
lead to increased HC emissions — at which time a DTC will be set and the MIL illuminated.
While CDH33, ATL130, and ATL120 showed high emissions for NOx and/or CO, the
malfunction in question simply had not had time to lead to excessive HC emissions.

       Lastly, CDH04 is not considered an OBD error of omission because the computer was
commanding the MIL on, but the nature of the problem (a short in the electrical system)
prevented the MIL from illuminating. This type of problem would be identified as a failure in an
OBD-I/M program and would be required to be repaired.  This vehicle helps illustrate why a
simple pass-fail visual check for MIL illumination is not enough on which to base an I/M
inspection; a scan of the onboard computer is also needed to help determine if there is a problem
with vehicle readiness, a malfunctioning MIL,  a short in the wiring, et cetera.
         EPA did not perform a detailed analysis of the entire emissions systems on these vehicles. Therefore, we cannot say
for certain that these CO and NOx problems are exclusively due to loss of catalyst efficiency, though it is our engineering
judgment that catalyst efficiency is a significant, contributory cause of the results observed.
1.2.4).
         California Air Resources Board Regulation, "Malfunction and Diagnostic System Requirements, 1968.1(b)(1.2.1-
                                             16

-------
2.4.2.4        OBD and Errors of Commission ("False Failures")

       Of the 194 vehicles that were accepted into the program with the MIL on, 43 or 22%
were sent home without a repair identified because repair technicians were unable to replicate the
cause of MIL illumination. Ten of these vehicles were sent home because the light went out
before initial testing was completed; this was attributed to OBD's ability to self-clear non-
recurring, intermittent fault codes (as discussed above). Because the repair goal of the tailpipe
pilot was to extinguish the MIL, no repairs were attempted on these self-clearing vehicles and
they were dismissed without further testing.

       Of the 33 remaining vehicles, 30 had FTP measurements at or below the applicable
certification standards, 2 (ATL120 and ATL94) had FTP scores below the OBD threshold of 1.5
times the certification standard, and 1 (ATL120) was a gross emitter which EPA was unable to
fix based upon OBD diagnostics.  Based upon engineering judgment, EPA believes that the
majority of these vehicles had intermittent problems that for one reason or another were not
manifesting themselves at the time repairs were being attempted.  Almost half (15) had misfire
codes, while an additional 11 had fuel trim OBD codes for which  OEM diagnostics failed to
identify a specific cause. Misfires, it should be noted,  are notoriously intermittent.  In some of
these cases the repair technicians were able to reproduce the misfire by spraying the engine
compartment with water.  In at least one case, however, the repair technicians were unsuccessful
with this technique even though they could plainly see where the misfire was occurring from a
plug wire that was not routed correctly. In one case where the repair technicians were unable to
reproduce the misfire on their own, the owner took technicians out on the road to demonstrate
when the misfire occurred (e.g., at high rpm  and load, off the  FTP cycle). This was a case where
EPA was able to convert a vehicle that seemed like a potential false failure into one where the
OBD system successfully identified a vehicle in need of repair. It is possible that other pilot
vehicles identified as potential OBD false failures based upon the repair technicians' inability to
identify a fixable problem could have been successfully repaired if EPA  had access to the vehicle
owners. Such access was the exception rather than the rule during the tailpipe pilot study,
because in many cases the "owners" of the vehicles were actually car rental agencies or
dealerships without practical knowledge of the individual vehicles in their fleets. In a real world
I/M program, the repair technician would be able to consult with the vehicle owner concerning
the conditions under which the MIL was illuminated, and as a result, the incident of unfixable
OBD failures should be lower than suggested by the pilot sample.
2.5    Conclusions

       Based upon the results of the OBD tailpipe pilot, EPA concluded that OBD scanning and
repair is a viable basis for I/M testing for MY 1996 and newer, OBD-equipped vehicles. The
emission reductions attributable to OBD-triggered repairs appear to be at least as large as those
attributable to repairs triggered by the most accurate, traditional I/M tailpipe test (i.e., the
EVI240).  In direct comparison to the EVI240, OBD-I/M checks identified more vehicles with
tailpipe emissions exceeding their certification standards.  With few exceptions, OBD-I/M
checks identified the same true failures as did the EVI240, while also providing diagnostic

                                           17

-------
information for repairing those vehicles.  Furthermore, OBD-I/M checks: 1) identified vehicles
that were falsely failed on the EVI240 as clean; 2) identified high emitting vehicles missed by the
EVI240; and 3) identified vehicles with broken or worn components that needed replacement or
repair prior to the actual development of emissions problems (thereby providing additional air
quality benefits in the form of pollution prevention).  Additionally, EPA found that OBD-
triggered repairs effectively returned vehicles to their proper operating conditions and that
tailpipe emissions returned to below certification levels in the majority of cases.
3.0    OBD-I/M Pilot 2: OBD-I/M Checks and Evaporative Emission Testing

3.1    Summary of Goals and Conclusions

       From April 1999 through May 2000, Automotive Testing Laboratories, Inc. (ATL) in
Mesa, Arizona conducted pre- and post-repair evaporative system emission testing on 30 OBD-
equipped vehicles under contract to EPA.  Unlike the tailpipe pilot described in section 2, the
evaporative pilot did not use vehicles with naturally occurring system failures.  Instead, specific
purge system malfunctions and evaporative system leaks were induced to see whether or not the
vehicles' OBD systems were capable of detecting a range of evaporative system problems. After
the failures were induced, the vehicles were then tested using the evaporative portion of the FTP.
Once OBD's ability to detect these induced evaporative system failures was established along
with the vehicles' pre-repair FTP scores, ATL technicians then repaired the vehicles, and a
second round of FTP testing was conducted.  The goal of this pilot test program was to answer
the following question:

       Can the  OBD-I/M check accurately detect evaporative system purge malfunctions and
       leaks on  OBD-equipped vehicles and, once these failures are repaired and the codes
       cleared, does the OBD system respond correctly by leaving the MIL extinguished?

       Unlike the tailpipe pilot, the OBD evaporative pilot did not focus on comparing the
OBD-I/M check to traditional I/M tests like the purge and pressure tests.  The reason for this is
because OBD-equipped vehicles with enhanced evaporative system monitoring are largely
considered untestable using traditional I/M program evaporative system tests23. In many cases,
the intrusive nature of the traditional I/M evaporative system tests could easily  compromise the
evaporative control systems on these vehicles, which, for example, frequently have hard lines
that cannot be crimped without causing damage to the vehicle.  In other cases, the lines are
simply inaccessible, or cannot be disconnected as required by some of the traditional evaporative
system test procedures.

       Based upon the test results detailed in this TSD  — and given the impracticality of using
traditional purge and pressure checks on most OBD-equipped vehicles — EPA concluded that
       23
         A notable exception is the gas cap pressure test (see section 4). Another exception is OBD-equipped vehicles that
have been built with special evaporative system service ports. However, because such service ports are not required on these
vehicles — and EPA has no reliable data on how many vehicles are so equipped — it is difficult to imagine basing a program of
evaporative system testing upon the presumption that such ports will be generally available to facilitate testing.

                                            18

-------
OBD-based evaporative emissions checks are a suitable replacement for traditional evaporative
emission I/M tests on OBD-equipped vehicles. This conclusion is based upon the following
observations made during the OBD evaporative system pilot:

       1) OBD evaporative system monitors appear to operate within their design specifications
       in the majority of cases. When evaporative system failures of the type found in traditional
       I/M test programs were induced, in most cases the OBD system responded correctly by
       lighting a MIL and setting an evaporative system DTC. Furthermore, once these failures
       were corrected and the codes cleared, the OBD system again responded correctly by not
       resetting the DTCs and re-lighting the MIL.

       2) The emission reductions associated with performing repairs triggered by OBD-based
       evaporative system testing appear to be substantial. In general, vehicles with evaporative
       emission DTCs and lit MILs were found to substantially exceed their FTP evaporative
       emission standards, while repaired vehicles fell well below those standards.
3.2    Background

       In addition to monitoring components the failure of which could lead to exhaust
emissions exceeding their FTP standards by 1.5 times the standard, OBD certification
requirements also phase-in separate standards for monitoring the evaporative control systems on
OBD-equipped vehicles. The first of these OBD evaporative system standards is phased in with
the 1996 through 1999 model year and applies to those vehicles which meet the enhanced
evaporative emission certification standards.  Vehicles built to meet the enhanced evaporative
emission standard must limit running losses to less than 0.05 gpm and hot soak/diurnal losses on
the FTP to no more than 2.0 grams. Beginning with MY 1998, and phasing in through MY
2006, OBD-equipped vehicles must also meet the Onboard Refueling Vapor Recovery (ORVR)
standards, which prohibit vehicles from emitting more than 0.02 grams per gallon of fuel
dispensed during vehicle refueling.

       To determine whether vehicles built to meet these requirements were operating as
required in the field, EPA contracted with ATL to conduct a pilot study of OBD evaporative
monitor effectiveness. As previously stated, the pilot testing ran from April 1999 to May 2000
and included a mix of 30 LDVs and LDTs from MY 1996 through 200024.  The goal of this pilot
was to determine whether OBD reacted correctly to evaporative system malfunctions and to the
repair of those malfunctions, as well as to determine the degree to which either condition (i.e.,
malfunctioning vs. repaired) affected FTP evaporative emissions. This pilot did not examine the
issue of OBD evaporative emission readiness under in-use driving conditions, nor did it address
whether the gas cap test is a suitable supplement to OBD-I/M evaporative system testing.  Those
issues were addressed as part of EPA's analysis of OBD test results from the Wisconsin I/M
program and will be discussed in section 4 of this  TSD.
       24 Data gathered for EPA under Work Assignments 3-12 and 0-4, SHED Tests on OBD II Evap Vehicles, EPA
Contract No. 68-C99-241 - Automotive Testing Laboratories; 1999-2000.

                                           19

-------
3.3    Vehicle Sampling

3.3.1   Methodology

       Unlike the OBD tailpipe pilot, the vehicles used in the OBD evaporative pilot were not
recruited from operating I/M lanes, and recruitment was not based upon naturally occurring, pre-
existing evaporative system failures. This decision was based upon prior attempts to recruit
natural OBD-I/M failures which showed that the majority of such failures with evaporative
system DTCs set were the result of gas caps that had not been tightened properly (this issue is
discussed further under section 3.4.1, which addresses vehicle testing methodology). Instead, the
majority of vehicles for the OBD evaporative pilot were recruited from fleet owners in the Mesa,
Arizona area, including both commercial rental agencies and local auto dealerships with which
ATL had a standing arrangement for procuring test vehicles.  Only one vehicle involved in the
pilot was a privately owned vehicle recruited from an ATL employee. Though every effort was
made to recruit as diverse a sample as possible, no attempt was made to weight the vehicle
sample by vehicle type and manufacturer.  This was due largely to limitations resulting from the
small sample size (less than one-sixth the size of the OBD tailpipe sample) which, in turn, was
the result of the high cost and time requirements associated with FTP evaporative system testing,
where a single test can take several days to complete.
3.3.2   Results

       A complete description of the 30 vehicles participating in the OBD evaporative pilot can
be found in Table A-l of Appendix 6. The descriptive details identified include vehicle make,
model, model year, mileage, engine family, evaporative system family, chassis dynamometer
testing parameters, and whether the vehicle was designed to comply with ORVR and/or
enhanced evaporative control standards.  A snapshot of the 30-vehicle OBD evaporative test
sample is provided below:

Manufacturers represented: 8
       Ford (7), GM (7), Honda (3), Isuzu (1), Mazda (2), Mitsubishi (1), Nissan (4), Toyota (5)

Model years represented: 5
       1996 (2), 1997 (1), 1998 (9), 1999 (16), 2000 (2)

       Lowest mileage:  5,259 miles       Highest mileage: 116,730 miles

       Light-duty vehicles: 20            Light-duty trucks: 10

       Enhanced  evap system only: 14     Enhanced evap and ORVR:  16

       Fleet vehicles: 29                 Privately owned vehicles: 1

                                           20

-------
       As discussed in section 3.3.1, the 30-vehicle sample was not sales weighted among
manufacturers; neither was it weighted based upon car versus truck sales. Most of the vehicles
in the sample were of relatively low mileage, with only 5 exceeding 50,000 miles, while the
sample average was just over 31,000 miles.  As can be seen from the above manufacturer
summary, Chrysler vehicles are not represented in the 30-vehicle sample. This is because
Daimler-Chrysler used an alternative Federal OBD certification provision available for MY
1996-99 vehicles which allowed manufacturers to postpone use of OBD evaporative emission
monitors in their Federally certified vehicles until MY 2000.

3.4    Vehicle Testing

3.4.1   Methodology

       Prior to being accepted into the OBD evaporative pilot study, candidate vehicles  were
evaluated to ensure that there were no driveability, braking, or exhaust leak problems. Once a
vehicle was accepted, its OBD emission control system was then checked for readiness status
and the presence of a lit MIL and/or DTCs. Unlike the OBD tailpipe pilot, the pre-existence of
naturally occurring system failures was not one of the criteria for participation in the OBD
evaporative pilot.  Instead, EPA opted to use induced failures.

       Induced failures were used due to the difficulty EPA had in finding MY 1996 and newer
OBD-equipped vehicles with naturally occurring evaporative system problems, which, in turn,
was due to the relative newness of the vehicles in question.  Unlike tailpipe problems which are
largely a function of mileage accumulation and general wear-and-tear,  evaporative system
problems tend to be a function of vehicle age, as the components of the system lose elasticity and
become brittle and more leak-prone. Furthermore, when naturally occurring evaporative system
DTCs were found, the vast majority turned out to be due to gas caps that had not been properly
tightened after refueling. EPA decided to use induced evaporative system failures to more
thoroughly investigate the effectiveness of OBD systems in detecting a wide variety of potential
in-use failures, above and beyond loose gas caps.   Table A-2 in Appendix 6 provides a vehicle-
by-vehicle account of the induced failures that were used in the 30-vehicle sample, the resulting
DTCs, and the drive cycles required to satisfy the readiness criteria for both "failure" and
"repair" sequences. Table A-2 also includes a comment field for more detail on specific vehicle
test issues. A summary of the induced failures used in the pilot is provided in Table 10 below:

       Only one failure was induced per vehicle.  The failures used were not meant to represent
the variety of real  world failures, nor were they necessarily representative of the range of excess
emissions which results from real failures.  Rather, the failures used were selected because they
are reproducible, easy to repair, and are the sorts of failures properly functioning OBD
evaporative system monitors should detect. Following the induced failures, vehicles were then
given the evaporative portion of the FTP to help estimate the mass of excess evaporative
emissions associated with these failures.

                 Table 10: Summary  of Induced Evaporative System  Failures

                                           21

-------
Type of Failure
Missing gas caps
Loose gas caps
0.040 inch leaks in gas caps, vent lines (initial OBD leak detection threshold)
Disabled canister fresh air inlet
Disconnected purge lines
0.020 inch leaks in gas caps (stricter OBD leak threshold, begins phasing-in MY 2000)
Number of Instances in Sample
3
2
11
1
8
5
       Two varieties of small orifice leaks were induced under the OBD evaporative pilot - a
0.040 inch leak and a more stringent 0.020 inch leak. Under California and Federal OBD
requirements, MY 1996 and newer vehicles equipped with OBD II evaporative system monitors
are required to detect leaks from a hole 0.040 inches in diameter or larger, and must also detect
and identify a malfunctioning purge system25. Beginning with MY 2000, LDVs and LDTs must
begin phasing in a more stringent leak detection threshold of  0.020 inches in diameter. Under
the OBD evaporative pilot, 5 vehicles were tested with 0.020  inch diameter leaks to examine the
robustness of the current systems (i.e., whether or not they can detect leaks below the level
minimally required), as well as to estimate the incremental emission impact from identifying
vehicles which pass the 0.040 inch limit while failing the more stringent 0.020 inch limit.  Gas
caps with 0.040 and 0.020 inch diameter leaks were supplied by  Stant Manufacturing
Corporation and were built with  flow tested, precision machined, square-edged orifices.

       Once a failure was induced, the impact on the vehicle's evaporative emission system was
verified by performing functional "pressure" and "purge" tests on the vehicle in question. These
traditional evaporative system tests were conducted by experienced ATL laboratory technicians26
who were not under the time constraints that make such testing impractical in most high volume
I/M test lanes. For vehicles with service ports, the tests were  conducted by measuring pressure
loss and purge system vacuum through the service port. For vehicles without service ports, the
ATL technicians used test procedures designed for pre-OBD-equipped vehicles.  These pre-OBD
test procedures consisted of measuring pressure loss by pressurizing the evaporative system from
the fill-pipe  and then monitoring the loss of pressure with time.  Purge system failures were
verified by using a roto-meter to check for the presence (or lack) of purge flow.

       After a failure condition was induced, the vehicle's OBD computer was then reset to
clear codes so that all monitors registered  as "not ready." The vehicles were then operated in a
manner that would exercise the monitors and — if the OBD system was functioning properly — a
DTC would be set and the MIL illuminated.  Typically, "exercising the monitors" meant driving
       25
         Not all MY 1996-99 vehicles are equipped with OBD evaporative system monitors. Manufacturers were allowed to
phase-in the use of such monitors from MY 1996 through MY 1999.

         We stress that the repair technicians were "experienced" because many manufacturers have opposed the intrusive
nature of EPA's original evaporative system tests, particularly in high volume I/M lanes. Particular care was taken during this
pilot to conduct these tests in a manner that did not adversely influence the evaporative emission results.

                                            22

-------
the vehicle on a dynamometer prior to formal FTP testing. The only exception to this practice
was when a dynamometer was not available, at which time "exercising the monitor" was
achieved by operating the vehicle over a local surface street, following a route approximating the
speed-time relation of the drive cycle used in the FTP (also known as the LA-4).  Following FTP
testing of vehicles with OBD-confirmed, induced failures, the vehicles were then repaired by
ATL technicians.  After repairs, the OBD system was again reset to clear the fault code and
return the readiness status to "not ready." The vehicle was then driven to exercise the monitors
to determine if the OBD system responded correctly (i.e., by not setting a DTC or commanding
the MIL to light).

       The FTP evaporative emission test selected for this study was the 3-day diurnal
procedure with running loss test. An abbreviated flowchart of the test procedure  for the FTP
evaporative test is presented in Figure A-l in Appendix 6. In general, tests were  conducted in
accordance with 40 CFR Part 86, Subpart B, as revised July 1,  199827. The test fuel used in this
pilot was indolene, as specified by the FTP.
3.4.2  Results

       The 30-vehicle sample was divided into two groups, based upon detection threshold.  The
first is a group of 25 vehicles, 9 with purge system failures, and 16 with leaks greater than or
equal to 0.040 inches in diameter (i.e., vehicles with induced failures within the required
detection range of current OBD evaporative system standards).  The second group consists of 5
vehicles with induced leaks produced by a 0.020 inch diameter hole in the gas cap (i.e., vehicles
with induced failures falling below the currently required detection threshold for OBD
evaporative systems).  The two groups were separated so as not to "penalize"  the overall sample
for vehicles in the second group which failed to find leaks more stringent than their design
requirements. Table 11 below looks at each subset separately.

                   Table 11: DTC and MIL-on Rates After Induced Failures

25 Vehicle Sample (purge, 0.040 leaks)
5 Vehicle Sample (0.020 leaks)
DTC Registered

22
3
MIL Illuminated

The same 22
The same 3
       Emission results for 22 vehicles repaired as part of the OBD evaporative pilot are
summarized in Table 12 below.  Only 22 of the 30-vehicle sample are included in the repair
results summary because not all vehicles registered DTCs, and valid "repair" results were not
accomplished for all vehicles. Vehicles without a complete set of "fail" vs. "repair" data were
excluded from the analysis used to prepare Table 12.
         Some minor deviations from Subpart B were allowed during pilot testing, including the use of: 1) external fuel tank
temperature measurement (on steel fuel tanks) as a surrogate for installing internal thermocouples, 2) the vehicle's fuel pump to
drain the tank instead of installing a drain(s) at the lowest point in the tank, and 3) a greater than +/- 3 degree F disagreement
between measured and target temperatures on the running loss test for a small number of vehicles. Table A-5 in Appendix 6 lists
all target vs. actual temperature differences observed during pilot testing (see Appendix 7 for further discussion).
                                             23

-------
       The summary results presented in Table 12 are stratified as a function of evaporative
emission control design (i.e., whether enhanced evap or ORVR designs). Of the 22 vehicles
included in the sample, 11 were certified to the enhanced evaporative standard and 11 were
designed to comply with ORVR requirements. Table 12 divides the evaporative emission results
into these strata because the design of ORVR systems (larger canisters, larger vapor lines, other
unique components to control  refueling loss) may lead to lower overall evaporative emission
losses in the case of a leak or other malfunction.  Because ORVR designs are manufacturer and
vehicle design specific and the sample size used in the pilot was too small to be representative
across manufacturers and models, an investigation into how and why ORVR compliant vehicles
seem to have inherently low evaporative emissions was not performed as part of the OBD
evaporative pilot. Nevertheless, the data in Table 12 suggest that when compared to vehicles
designed to only meet the enhanced evaporative emission requirements, ORVR-controlled
vehicles have significantly lower evaporative emissions, even when leaks or other malfunctions
have been introduced into the  system.

       To get an idea of the impact on the mean and standard deviations when the subsets of
enhanced evap and ORVR vehicles are averaged together, see Table 13 below. Complete
evaporative emission results for all 30 vehicles are presented in Table A-3 in Appendix 6, while
FTP exhaust results for these same vehicles are summarized on a bag-by-bag basis in Table A-4.

         Table  12: Summary Statistics for 11 Enhanced Evap and 11 ORVR Vehicles

Running Losses (gpm)

1 hour Hot Soak Loss (grams)

High 24 hour Diurnal Loss (grams)

Enhanced
evap fails
x=7.86
s = 7.89
x= 10.74
s=16.12
x= 20.83
s=17.77
Enhanced
evap repairs
x=0.02
8 = 0.01
x = 0.13
s = 0.08
x=0.95(N=10)28
s = 0.87
ORVR fails
x = 4.51
s = 5.29
x = 2.89
s = 3.20
x= 12.31
s= 12.00
ORVR repairs
x = 0.02
8 = 0.01
x=0.14
s = 0.07
x = 0.87
s = 0.51
x = mean; s = standard deviation
        Table 13: Average Emission Reductions From Sample of 22 Repaired Vehicles
Running Losses (gpm)
1 hour Hot Soak Loss (grams)
High 24 hour Diurnal Loss (grams)
x = 6.17
x = 6.68
x=14.18(N=21)29
s = 6.78
s=12.04
s=14.54
       In addition to the summary results presented in Tables 12 - 13 above, EPA wishes to
highlight the following findings made as a result of the OBD evaporative pilot study:
       28
         One of the 11 enhanced evap vehicles had to be returned to its owner prior to post-repair evaporative system testing.

                                  29
                                    See explanation in footnote 28 above.
                                           24

-------
       1) 88% of OBD-equipped vehicles (22 of 25) set DTCs and lit MILs when
       evaporative system failure conditions were induced and subsequently showed no
       DTCs and MILs when the vehicles were repaired. This suggests that OBD
       evaporative system monitors are working as designed in the vast majority of
       cases. EPA considers these results impressive, compared to the existing purge
       and fill-neck pressure tests, which both suffer from relatively low testability rates
       — approximately 70% for pre-OBD-equipped vehicles and less than 15% for
       OBD-equipped vehicles.

       2) Of the 3 vehicles that did not light a MIL or set a DTC during ATL testing, 2 were
       Mazda 626s, which represents 100% of that manufacturer's fraction of the 30-vehicle
       sample.  To see whether there was a possible design problem with this particular make
       and model, EPA procured two "sister" vehicles in Ann Arbor, and found them to be
       functioning properly. EPA is pursuing a resolution regarding the third vehicle's results
       with the manufacturer.

       3) 60% of OBD-equipped vehicles tested (3  of 5) identified a 0.020 inch diameter leak
       (i.e., a leak below the required leak detection threshold for the OBD-equipped vehicles in
       the sample) by setting a DTC and lighting the MIL. This suggests that the majority of
       OBD systems are quite robust and have leak detection capability well below the
       minimum requirement.

       4) Three OBD-equipped vehicles which set MILs for evaporative system problems
       (different from the 3 of 5 listed in item # 3 above) produced FTP evaporative emissions
       less than half the levels of the enhanced evaporative emission standards. This suggests
       that "maintenance" problems are being identified by OBD even though they result in
       emission levels below FTP standards.

       5) 95% of repaired OBD-equipped vehicles (21 of 22) had FTP-measured running loss
       emissions that were actually below the certification standard for enhanced running losses.
       95% of repaired OBD-equipped vehicles (20 of 21) had FTP-measured diurnal  plus hot
       soak emissions that were below the certification standards for those categories of
       evaporative emissions.  The running loss and diurnal plus hot soak emissions for ORVR
       vehicles with induced failures averaged approximately half the levels measured for
       comparable, enhanced evap-only vehicles.

       6)  The average emission reductions for repairing OBD-identified DTCs is substantial:
       6.17 gpm for the running loss test, 6.7 g for the hot soak test,  and 14.2 g for the high 24
       hour result for the diurnal loss test.

3.5    Conclusions

       In the majority of cases, OBD evaporative emission monitors appear to be operating as
designed, and, in some cases, better than required. This  conclusion is based upon an admittedly

                                          25

-------
small sample of OBD-equipped vehicles with induced failures specifically aimed at triggering
the evaporative system monitors. Nevertheless, the majority of OBD-equipped vehicles in the
test sample correctly set evaporative DTCs and lit the MIL when evaporative system failures
were induced, while also successfully showing no MILs or DTCs when those failures were
removed. An analysis of the FTP mass emissions data before and after these induced failures
suggests that the emission reductions attributable to OBD-triggered evaporative system repairs is
substantial, with pre-repair vehicles registering evaporative emissions well above the applicable
FTP standards and post-repair vehicles having evaporative emissions well below those standards.
Based upon these observations — and given the impracticality of using pre-OBD-style purge and
pressure checks on most  OBD-equipped vehicles — EPA believes that OBD-I/M evaporative
emissions checks are a suitable replacement for the traditional purge and fill-neck pressure tests
for MY 1996 and newer, OBD-equipped vehicles.

       Neverthless, EPA still recommends that states continue to perform gas cap pressure tests
on OBD-equipped vehicles.  Unlike other, traditional evaporative system tests, the gas cap test
does not suffer from the material composition  and accessibility problems that make many OBD-
equipped vehicles untestable using the purge and fill-neck pressure tests.  Furthermore, the
failure threshold for the gas cap pressure test is more stringent than even the most stringent
OBD-based evaporative emission standards. As will be shown in section 4, which details EPA's
analysis of Wisconsin's operating OBD-I/M program data, EPA believes that there is real-world
data to suggest that additional evaporative system failures can be identified by performing a
separate gas cap pressure test in conjunction with the OBD-I/M check (see Table  19, "Gas Cap
vs.  OBD Evaporative System Failure Rates" later in this document).
4.0    OBD-I/M Pilot 3: Analyzing the Wisconsin OBD-I/M Program Experience

4.1    Summary of Goals and Conclusions

       The last of the three OBD-I/M pilot studies was aimed at identifying the real-world
implementation issues associated with OBD-I/M testing and was conducted using data gathered
from the Wisconsin enhanced I/M test lanes, where OBD checks are being implemented
voluntarily by the state. The analysis of Wisconsin's operating program data for OBD-equipped
vehicles was conducted in two stages, the first performed by Sierra Research in 1998 under
contract to EPA and the second in 1999-2000, performed by EPA staff from the Office of
Transportation and Air Quality (OTAQ) in Ann Arbor, Michigan.

       Although the original focus of the Sierra study was intended to be broader, flaws in
Wisconsin's I/M  contractor's OBD hardware and software limited the scope of the analysis to
identifying physical aspects of the OBD-I/M testing process that could lead to implementation
difficulties.  Specifically, the Sierra study provided an estimate of the time needed to perform a
typical OBD-I/M inspection (on average, about 31 seconds) and also identified atypical data link
connector (DLC) location as a potential bottleneck in high-volume I/M test lanes. In response to
this latter issue, EPA has developed a database of DLC locations based upon the Wisconsin data
and manufacturer-supplied information. Electronic copies of this database are available on

                                          26

-------
EPA's web site at http:\\www.epa.gov\OMSWWW. EPA has found that the development of this
database and increased inspector experience has eliminated DLC location as a problem area in
the Wisconsin program.  While a period of introductory learning will be necessary, we do not
believe that DLC location will be a significant problem for future, OBD-based I/M efforts.

       Separate from the Sierra Research analysis, EPA looked at data from Wisconsin's I/M
program for the last eight months of 1999, by which time the OBD software and hardware
problems mentioned above had been corrected.  The program data EPA analyzed included
EVI240, gas cap, and OBD MIL illumination and readiness data for over 116,000 MY 1996 and
newer vehicles.

       Using the above database, EPA compared evaporative system failure rates for vehicles
based upon the OBD evaporative system test and the separate gas cap check and found that the
gas cap test identified significantly more evaporative system leaks than were identified based
upon the OBD evaporative system monitors alone.  EPA believes this demonstrates the
complementary nature of these two tests — not an unanticipated conclusion, given the different
standards and stringencies involved. We believe that these findings support our recommendation
that states continue gas cap evaporative system testing on OBD-equipped vehicles in conjunction
with OBD-I/M testing (as mentioned earlier, in section 3 of this TSD).

       In analyzing Wisconsin's OBD-I/M data, EPA also looked at MIL illumination and
monitor readiness results and concluded that there is a small fraction of vehicles that arrive for
testing with  one or more of their OBD readiness codes unset, although the problem seems largely
limited to the earliest of the OBD-equipped vehicles (i.e., MY 1996). Looking at the raw data,
EPA found a 5.8% not-ready rate among MY 1996 vehicles.  However, when we excluded
vehicles for  which corrective measures are being taken by the manufacturers, the not-ready rate
for MY 1996 dropped to roughly 3%.  By MY 1998, the OBD not-ready rate dropped even
further — to  below 1%. EPA believes that offering states the ability to waive vehicles with one
or two unset readiness codes instead of rejecting them  (as currently required) will go a long way
toward eliminating vehicle readiness as an obstacle to smooth implementation.

       Because the two-staged analysis of Wisconsin's OBD-I/M data has three separate points
of focus - DLC location, vehicle readiness, and the relative effectiveness of the gas cap test -
each will be dealt with individually below, identified by focus.
4.2    DLC Location
4.2.1   Background

       In 1998, EPA contracted with Sierra Research to gather information on approximately
2,500 OBD-equipped vehicles upon arrival at the I/M test lanes in Wisconsin30.  Parallel EVI240
and OBD-I/M testing was conducted in the Wisconsin test lanes by Envirotest Systems
       30 Under Purchase Order No. 7CS124NTSA, "Status of OBD Systems Upon Arrival at I/M Lanes." Report No.
SR98-10-02, "Summary of Test Results from Wisconsin EPA OBD Project," October 16, 1998.

                                           27

-------
Corporation (under subcontract to Sierra Research, Inc.). The original purpose of the study was
to conduct testing of OBD systems on MY 1996 and newer vehicles, with the intent to:

    •  Gain practical experience in conducting OBD-I/M tests;

    •  Use this experience to develop guidance on how to perform OBD-I/M tests properly;

    •  Estimate the average time required to perform an OBD-I/M test;

    •  Determine the frequency of readiness for each OBD system monitor;

    •  Determine the reliability of the MIL as an identifier of vehicles likely to fail other I/M
       tests;

    •  Determine the frequency and nature of DTCs stored in OBD computers; and

    •  Identify any problems among OBD-equipped vehicles that could interfere with proper
       testing, including those specific to particular vehicle models.


4.2.2  Results

       Between May 20,  1998, through July 25, 1998,  2,583  paired OBD/IM240 test records
were collected. Only initial tests were used because Wisconsin does not currently fail vehicles
on the basis of the OBD-I/M check  .  Information was gathered concerning test time, OBD
readiness, DTC and/or MIL frequency, etc. However, due to  problems with the OBD software
and hardware  used by Wisconsin's testing contractor at the time of the Sierra study, no useful
information was gathered by Sierra concerning OBD readiness or DTC and/or MIL-on rates.
Nevertheless,  useful information was gathered concerning test time and the general ability of
inspectors to locate difficult-too-find DLCs. Table 14 below  provides a summary of the data
gathered, divided by make, model, and vehicle type.

       As  can be seen from Table 14, the average OBD-I/M test time was roughly 31 seconds,
including the time to locate the OBD connector, connect to the system, interrogate it,  and
                                  on
download the  resulting information  . Care should be taken in interpreting the test time
estimates, however, due to possible variance in how inspectors conducted the test. Envirotest's
inspectors were instructed to locate the OBD connector as soon as they were prompted to do so
       31
         While Wisconsin does not fail vehicles on the basis of OBD (yet), the State does fail vehicle based upon their IM240
results. The Sierra study focused on initial tests only to avoid double counting vehicles which failed their initial IM240 and then
returned for a retest.

       32
         It should be noted that these test time calculations do not include the time needed to record vehicle information, such
as VIN, license plate number, etc. Such information was gathered as part of the overall testing process, which also included
performance of an IM240, as previously discussed. No vehicle in the Wisconsin program received just the OBD test. The test
times discussed here, therefore, reflect only the time spent conducting the OBD-I/M portion of the overall test process.

                                             28

-------
estimates, however, due to possible variance in how inspectors conducted the test. Envirotest's
inspectors were instructed to locate the OBD connector as soon as they were prompted to do so
by the test system. The system tracked the time from when the inspector prompt appeared to
when the connection to the OBD system was established. However, it is obvious from the short
connect times recorded for some vehicles (e.g., 1-2 seconds) that certain inspectors were locating
the connector in advance of the prompt.  Actual connect times (and thus overall test time as well)
may therefore be slightly longer than the data in Table 14 suggest. However, given how the test
is structured, it is believed that this would add only about five seconds at most to some of the
recorded  OBD-I/M test times.

       The time it takes to locate the DLC is a relevant variable in assessing the time it takes to
perform an OBD-I/M inspection because DLC location varies from manufacturer to
manufacturer, and from model to model. Attempts to standardize DLC location are reflected in
the Society of Automotive Engineers (SAE) Recommended Practices  J1962, which specifies the
following with regard to DLC location:

       3.1  Consistency of Location - The vehicle connector shall be  located in the passenger
       compartment in the area bounded by the driver's end of the instrument panel to 300 mm
       beyond the vehicle center line, attached to  the instrument panel, and accessible from the
       driver's seat.  The preferred location is between  the steering column and the vehicle
       centerline. The vehicle connector shall be mounted to facilitate mating and unmating.

       3.2  Ease of Access - Access to the vehicle connector shall not require a tool for the
       removal of an instrument panel cover,  connector cover, or any barriers. The vehicle
       connector shall be fastened and located so as to permit a one-handed/blind insertion of
       the mating test equipment connector.

       3.3. Visibility - The vehicle connector  should be out of the occupant's (front and rear
       seat) normal line of sight but easily visible to a "crouched" technician.

Even with this guidance, however, vehicle manufacturers have been anything but consistent  with
regard to  where they place the DLC.
                                          29

-------
Table 14: Summary of Wisconsin Data. 5/20/98 to 7/25/98
Make
Acura
Audi
BMW
Buick
Cadillac
Chevrolet



Chrysler

Datsun

Dodge



Eagle
Ford



Geo

GMC


Honda

Hyundai
Infmiti
Isuzu

Jaguar
Jeep
Lexus
Lincoln
Landrover
Mazda

Mercury

Mercedes
Mitsubishi

Oldsmobile

Other
Plymouth

Pontiac

Porsche
Saab
Sterling
Subaru
Suzuki
Toyota


Volkswagen
Volvo
TOTAL
Type
LDGV
LDGV
LDGV
LDGV
LDGV
LDGV
LDGT1
LDGT2
HDGT
LDGV
LDGT1
LDGV
LDGT1
LDGV
LDGT1
LDGT2
HDGT
LDGV
LDGV
LDGT1
LDGT2
HDGT
LDGV
LDGT1
LDGT1
LDGT2
HDGT
LDGV
LDGT1
LDGV
LDGV
LDGT1
LDGT2
LDGV
LDGT1
LDGV
LDGV
LDGT2
LDGV
LDGT1
LDGV
LDGT1
All
LDGV
LDGT1
LDGV
LDGT1
LDGT1
LDGV
LDGT1
LDGV
LDGT1
LDGV
LDGV
LDGV
LDGV
LDGT1
LDGV
LDGT1
LDGT2
LDGV
LDGV

Number Tested
19
3
23
117
41
182
101
75
6
42
25
105
17
88
115
30
6
3
144
117
44
10
23
2
21
17
5
127
20
4
2
15
2
4
80
14
26
5
33
2
98
33
15
14
3
56
11
1
32
48
168
13
1
7
61
9
2
199
45
4
45
5
2,583
Connect Time (sec)
28.9
6.0
25.3
12.8
23.5
14.8
12.4
12.6
24.0
14.9
11.6
15.9
29.3
16.5
10.7
14.4
19.5
18.3
13.5
18.6
16.6
36.7
16.0
5.5
11.8
10.7
19.4
12.8
33.5
2.7
1.5
9.8
24.5
21.3
17.1
13.2
17.3
19.5
17.1
23.0
15.3
11.7
22.2
19.7
32.3
18.1
16.0
40.0
14.2
9.6
15.8
8.9
80.0
20.6
13.5
53.7
31.5
12.2
22.0
32.0
18.2
14.2
15.6
Comm Time (sec)
13.3
14.0
14.7
12.8
17.3
15.0
15.8
13.9
5.8
18.0
18.2
16.7
18.6
15.6
16.0
16.1
12.7
16.0
13.7
15.7
13.0
13.2
14.3
17.0
14.4
15.2
13.8
16.9
15.9
15.7
16.5
13.8
15.0
14.3
21.3
18.1
13.2
14.0
14.1
27.0
13.4
17.1
15.1
15.9
14.3
14.7
14.0
12.0
16.1
15.4
14.2
14.8
14.0
14.6
17.8
15.4
11.5
15.7
14.8
18.0
18.4
15.0
15.6
Total OBD Time (sec)
42.2
20.0
40.0
25.6
40.8
29.8
28.2
26.5
29.8
32.9
29.8
32.6
47.9
32.0
26.7
30.5
32.2
34.3
27.2
34.3
29.7
48.9
30.3
22.5
26.2
25.9
33.2
29.7
49.3
18.3
18.0
23.6
39.5
35.5
38.4
31.3
30.5
33.5
31.2
50.0
28.6
28.8
37.3
35.6
46.7
32.8
30.0
52.0
30.3
25.0
29.9
23.8
94.0
35.1
31.3
69.1
43.0
27.9
36.8
50.0
36.6
29.2
31.2
                         30

-------
               Table 15: DLC Location Problems in the Sierra Wisconsin Data
Make
Acura
Audi
BMW
Buick
Cadillac
Chevrolet
Chrysler
Datsun
Dodge
Ford
Geo
GMC
Honda
Hyundai
Isuzu
Lexus
Mazda
Mercury
Mercedes
Mitsubishi
Oldsmobile
Pontiac
Subaru
Suzuki
Toyota
Volkswagen
Volvo
TOTAL
Total in sample
19
3
23
117
41
364
67
122
239
315
25
43
147
4
17
14
35
131
15
17
67
179
9
2
248
45
5
2,583
# DLC location problem
14
3
18
2
3
2
1
1
3
7
1
1
69
1
2
1
1
3
12
7
3
1
1
2
20
25
4
208
% DLC location problem in
sample of make
11.8%
100%
78.3%
1.7%
7.3%
0.6%
1.5%
0.8%
1.3%
2.2%
4%
2.3
46.9%
25%
11.8%
7.1%
2.9%
2.3%
80%
41.2%
4.5%
0.6%
11.1%
100%
8.1%
55.6%
80%
8.1%
       Ultimately, OBD-I/M test times were found to be highly dependent on the ease with
which the inspector located the DLC, which was itself found to vary greatly among the various
makes and models included in the Sierra Research study. During the first phase of the
Wisconsin analysis, Sierra Research found that it took considerably longer to locate the DLC on
some makes and models than it did for others.  Apparently, despite attempts to standardize DLC
location, some manufacturers have interpreted SAE J1962 more broadly than originally
anticipated. In fact, many vehicles were identified by inspectors as "untestable" because they
could not locate the DLC in a timely manner. Out of the 2,583 vehicles involved in the test
                                          31

-------
program, 208 or 8.1% were identified as "untestable," largely because the inspector was unable
to locate the connector within a time period that was commensurate with the throughput demands
of the test network.  In fact, DLC location is not so much an issue of "testability" as it is an issue
of throughput. Table 15 above provides a breakdown of these so-called "untestable" vehicles by
manufacturer.

       Designating a given vehicle as "untestable" proved to be highly subjective.  While nearly
all of the test records represented in Table 15 contain a connector location of "99" (signifying a
location other than somewhere under the front dashboard), many of the records also contain
inspector comments indicating they were either unable to find the connector or it was found in an
abnormal location. Some inspectors were apparently able to test them, while others could not
locate the connectors, and still others could locate the connectors but indicated it was too much
effort to do so.  For example, on the BMW 318i a cover panel must be removed with a
screwdriver before the OBD connector can be accessed.

4.2.3   Conclusions

       In general, the OBD connector was located by one or more inspectors on a very high
fraction of test vehicles. However, without further efforts, the difficulty in accessing  some of the
connectors could have a significant impact on the future  success of large-scale OBD-I/M testing.
Accessibility time is particularly an issue in a high-volume test-only I/M environment.
Inspectors are under continuous pressure from both motorists and management to be as fast and
efficient as possible in completing required inspection procedures. Any connectors that take
over 15-30 seconds to locate and access are a problem in this environment.  While it is expected
that inspectors will quickly learn the abnormal DLC locations for higher volume makes and
models (e.g., behind the ash tray on Honda passenger cars), EPA believes that the potential for
start-up
problems is nevertheless significant. To address this issue, EPA has therefore developed a
database of atypical DLC locations based upon the Wisconsin data, as well as manufacturer-
supplied information. This database is available electronically at:
www.epa.gov/otaq/regs/im/obd/obd-im.htm

4.3     Vehicle Readiness

4.3.1   Background

       The OBD system monitors the status of up to 11 emission control related subsystems by
performing either continuous or periodic functional tests of specific components and vehicle
conditions. The first three testing categories - misfire, fuel trim, and comprehensive components
- are continuous, while the remaining eight only run after a certain set of conditions has been
met. The algorithms for running these eight, periodic monitors are confidential to each
manufacturer and involve such things as ambient temperature as well as driving times and
conditions. Most vehicles  will have at least five of the eight remaining monitors (catalyst,
evaporative system, oxygen sensor, heated oxygen sensor, and exhaust gas recirculation or EGR
system) while the remaining three (air conditioning, secondary air, and heated catalyst) are not

                                           32

-------
necessarily applicable to all vehicles.  When a vehicle is scanned at an OBD-I/M test site, these
monitors can appear as either "ready" (meaning the test in question has been run), "not ready"
(meaning the test has not been run yet), or "not applicable" (meaning the vehicle is not equipped
with the components in question).

       Current Federal regulations for OBD-I/M testing require that I/M programs reject from
further testing  any MY  1996 or newer OBD-equipped vehicle that is found to have one or more
unset readiness codes. It is important to note that "rejection" is distinct from "failure." In the
context of the OBD-I/M check, rejection is triggered by a vehicle's readiness status while failure
is related to the presence of DTCs that command the MIL to be lit. If DTCs are present and the
MIL is commanded on, the vehicle is  failed, the initial test process is considered complete,  and
an official test report is generated. If, on the other hand, unset readiness codes are present,  the
vehicle is rejected and the test process is aborted until such time as all readiness codes are set.

       The reason vehicles with unset readiness codes are rejected but not failed is because an
unset readiness code is not necessarily an indication of an emission problem.  Rather, it is an
indication that certain monitor(s) that  are intended to  determine whether or not there may be an
emission problem have not been run to evaluate the system. In the case of rejection, the issue of
whether or not the vehicle requires repairs is deferred until the readiness code(s) have been set
and the monitor(s) run.

       There are many reasons why a readiness code may not be set when an OBD-equipped
vehicle arrives at the I/M test site - some of them wholly legitimate and beyond the control of
the motorist. For example, if the battery is disconnected during servicing or the monitors are
turned  off with a scan tool, it takes a varying amount  of time for the monitors to reset, and some
may still not be ready when the vehicle shows up for its I/M inspection.  It is also possible that
the battery was disconnected on purpose in an attempt to fraudulently extinguish the MIL and
clear DTCs prior to OBD-I/M testing. While it is true that disconnecting the battery will
temporarily clear any DTCs that are present, many of these will be quickly reset (in particular,
the continuous monitors discussed above). In fact, readiness codes were developed specifically
to prevent vehicle owners from evading the OBD-I/M test by disconnecting their batteries just
prior to testing. In many cases, exercising the monitors to set a readiness code may be as simple
as operating the vehicle on a dynamometer or on the highway for a certain amount of time, while
in other cases,  readiness is more difficult to establish because of design issues with certain makes
and models of vehicles.

       To determine the extent to which vehicles may be appearing for their OBD-I/M check
with unset readiness codes in the real world, EPA looked at OBD readiness data from
Wisconsin's I/M program for the last eight months of 1999. The program  data EPA analyzed
included EVI240, gas cap, and OBD MIL illumination and readiness data for over 116,000 MY
1996 and newer vehicles. The data was analyzed to determine the size of the readiness problem,
the number of model years affected, and the approximate percentage of vehicles that would be
rejected under a variety of possible readiness criteria. EPA also looked at the frequency of MIL
illumination across model years and vehicle types, and compared the relative failure rates of the
OBD-I/M check versus lane-based EVI240s.
4.3.2   Results

                                           33

-------
4.3.2.1 Readiness

       Since August 1998, Envirotest Systems Corporation (the I/M contractor in Wisconsin)
has been sending EPA staff OBD-I/M check and IM240 test data collected on MY 1996 and
newer vehicles coming through the Wisconsin test lanes. The data provided by Envirotest
included vehicle identification numbers (VEST), and EVI240, OBD-I/M, and gas cap test results
(for a full list of the 40 data fields included in the Wisconsin data, see Appendix 8). Because of
the OBD software and hardware problems discussed earlier, EPA limited its analysis to data
gathered beginning with May 1999, by which time the software and hardware issues had been
resolved.

       Table 16 below provides the average mileage accumulation for MY 1996 and 1998
vehicles from the Wisconsin data set. Because Wisconsin did not include odometer data until
recently and only tests vehicles every other year, the data EPA has available for MY 1997
vehicles does not include odometer readings. However, because the data we have for MY 1998
and 1997 represents vehicles that are being tested at the same age (i.e., when they are one year
old) we can assume that the average mileage accumulation for MY 1997 vehicles at the time of
their first test is similar to that of MY 1998 vehicles at the time of their first test (i.e., between
20,000 to 22,000 miles, depending upon vehicle class).

       Because of the different emission standards for LDVs versus LDTs, these vehicle classes
were analyzed separately. Looking at the three model  years and two vehicle classes represented
in the Wisconsin data therefore forms six vehicle categories: 1996, 1997, and 1998 LDVs and
LDTs. There is one caveat concerning these groupings, however. Because Wisconsin used the
same IM240 cutpoints for some light trucks as it did passenger cars (mostly four cylinder SIOs,
Rangers, etc.) these LDTs were listed as LDVs for these analyses (see Appendix 8 for a
discussion of EPA's analysis methodology).
         Table 16: Average Mileage Accumulation, by Model Year and Vehicle Type
Vehicle class
LDVs
LDTs
1996
45,385
51,018
1998
20,745
22,962
       Table 17 below presents the "not ready" status for MY 1996-98 LDVs and LDTs in the
Wisconsin data set. Note that the majority of the "not ready" vehicles are MY 1996 LDVs
(6.9%) and that the majority (77%) of all "not ready" MY 1996 LDVs were not ready for the
catalyst monitor, while MY 1998 LDVs were more frequently "not ready" for the evaporative
system monitor. Note further that by MY 1998, the "not ready" rate for LDVs dropped over
five-fold - from 6.9% to 1.3% - while the overall "not ready" rate for MY 1996 vehicles (5.8%)
dropped more than four-fold by MY 1998 - to 1.4 %.

       We can speculate that this difference in "not ready" rates among the three model years
reflects a maturation curve for OBD technology, with the systems improving as manufacturers
gained experience with what did and did not work in the real world. By the same token,
                                          34

-------
however, it is also possible that the lower occurrence of readiness problems among the newer
model years could be the result of differences in relative age and/or mileage accumulation.  A
test of this latter hypothesis would be to look at vehicles from different model years when they
are at the same age and have accumulated comparable mileage. As discussed above, the
Wisconsin data includes test results from MY 1997 and MY 1998 vehicles that were receiving
their first tests on their one-year anniversary of purchase.  Unfortunately, as we also discussed
above, EPA does not have mileage accumulation data for MY  1997. Nevertheless, in the
absence of any compelling reason to assume that MY 1997 vehicles were driven more or less
than their MY 1998 counterparts in their first year of operation, we believe it is reasonable to
assume that MY 1997 and MY 1998 vehicles exhibited comparable, accumulated mileage.  The
fact that the data in Table 17 still  shows a significant decline in "not ready" rate from MY 1997
to MY 1998 - from an overall average of 2.3% to 1.4% - suggests that manufacturer learning
curve is at least a likely explanation for the significant trend toward improvement in observed
"not ready" rates.

       As discussed in the background section above, some instances of vehicle unreadiness are
due to vehicle design issues which EPA and CARB are still working with vehicle manufacturers
to resolve.  In the interim, it does  not seem right to penalize motorists for something that is
beyond their control. One logical solution is to allow states the flexibility (and the discretion) to
not reject certain vehicles if the only problem is that they have unset readiness codes.  The
natural question then is, how do you allow these exemptions from the readiness criteria without
opening the door to motorist fraud? In discussing this issue with the states and other interested
OBD and I/M stakeholders, EPA concluded that the key is to limit the use of readiness
exemptions - first,  by model year, and secondarily, by the number (and possibly category) of
unset readiness  codes allowed.

                    Table 17: "Not Ready" (NR) Status for MY 1996-98

96LDV
%
96LDT
%
96 Total
%
97LDV
%
97LDT
%
97 Total
%
98LDV
%
98LDT
%
98 Total
%
TOTAL
%
Total
Tested
27,313

16,423

43,736

14,946

7,656

22,602

27,615

22,716

50,331

116,669

Not Ready
(NR)
1,873
6.9%
651
4.0%
2,524
5.8%
360
2.4%
171
2.2%
531
2.3%
361
1.3%
350
1.5%
711
1.4%
3,766
3.2%
OneNR
1,155
4.2%
169
1.0%
1,324
3.0%
58
0.4%
34
0.4%
92
0.4%
101
0.4%
69
0.3%
170
0.3%
1,586
1.4%
TwoNR
884
3.2%
64
0.4%
948
2.2%
30
0.2%
14
0.2%
44
0.2%
61
0.2%
32
0.1%
93
0.2%
1,085
0.9%
Catalyst
1,435
5.3%
471
2.9%
1,906
4.4%
87
0.6%
88
1.1%
175
0.8%
105
0.4%
221
1.0%
326
0.6%
2,407
2.1%
Evap
475
1.7%
184
1.1%
659
1.5%
209
1.4%
77
1.0%
286
1.3%
287
1.0%
182
0.8%
469
0.9%
1,414
1.2%
O2
826
3.0%
74
0.5%
900
2.1%
38
0.3%
11
0.1%
49
0.2%
59
0.2%
32
0.1%
91
0.2%
1,040
0.9%
Heated O2
880
3.2%
186
1.1%
1,064
2.4%
102
0.7%
31
0.4%
133
0.6%
61
0.2%
55
0.2%
116
0.2%
1,313
1.1%
EGR valve
1,041
3.8%
72
0.4%
1,113
2.5%
33
0.2%
18
0.2%
51
0.2%
55
0.2%
17
0.1%
72
0.1%
1,236
1.1%
                                           35

-------
       Determining the optimum combination of limiting factors, however, required real-world
data.  Therefore, in performing its analysis of the Wisconsin data, EPA also looked at the impact
of adjusting "not ready" rates based upon a variety of possible readiness waiver scenarios. For
example, does it make more sense to exempt vehicles based upon a certain number of "not
ready" codes? Or would it be better to limit the exemptions to vehicles presenting as  "not ready"
for specific OBD monitors? In Table 17 above, the column headed "One NR" reflects the "not
ready" rate by model year and vehicle type adjusted to reflect a waiver of the "not ready"
rejection requirement if only one monitor is listed as "not ready." The column headed "Two
NR" reflects a similar adjustment of the "not ready" rate, but this time assuming a waiver of the
rejection requirement if up to two monitors are listed as "not ready." Table 17 also breaks out
the readiness  status of the vehicles in the Wisconsin data by monitor.

       Table 17 shows that if any one monitor is allowed to be "not ready" the overall rejection
rate among MY 1996-98 vehicles goes from 3.2% to 1.4%.  If exemptions are allowed for
vehicles with up to two unset readiness codes, the overall rejection rate goes down even further -
to 0.9%.  Because Wisconsin did not fail vehicles on the basis of the OBD-I/M check - which
was being conducted on a purely advisory basis at the time this data was collected - vehicles
were also not being rejected for unset readiness codes. As a result, no attempt was made at the
test lanes to exercise these monitors prior to continuing the test.  EPA therefore believes that the
relative "not ready" rates reflected in Table 17 represent the worst-case scenario for these model
years, and that the frequency of unresolved "not ready" codes in a fully implemented  OBD-I/M
program will  be even lower.
4.3.2.2 MIL-on and EVI240 Failure Rates
       Table 18 below compares the relative failure rates for the OBD-I/M check versus the
EVI240 test observed in Wisconsin, and the degree to which the test results overlap. As can be
seen from the data, the OBD-I/M check almost always fails slightly more vehicles than does the
EVI240 (MY 1998 LDVs are the only exception).  There are several obvious reasons for the
marginal difference in failure rate between these two tests:

       1)     The cutpoints for OBD are more stringent than the EVI240 (i.e., 1.5 vs. 2 times the
              certification standard);

       2)     The EVI240 only monitors vehicle performance for approximately 4 minutes over
              a limited number of operating modes, while OBD performs ongoing monitoring
              of vehicle performance over the full range of operating conditions; and

       3)     OBD monitors individual systems and components for any sign of degradation
              thus allowing it to identify necessary maintenance prior to the vehicle's producing
              high emissions, while the EVI240 can only identify vehicles which have already
              become high emitters (see discussion on "OBD and Preventative Maintenance" in
              section 2).
                                          36

-------
       Table 18 also makes clear that the agreement between IM240 and the OBD-I/M check is
exceedingly low for all model years and vehicle types. What is not clear from this data is which
of the two tests is more beneficial to the environment.  Traditionally, relative failure rate has
been used as a crude indicator of test effectiveness in I/M programs, with the assumption being
that the more vehicles that are failed, the more emission reductions are being achieved. Using
simple, relative failure rates, Table 18 suggests that the OBD-I/M check is the more effective
test, environmentally, because it has the greater overall failure rate.

       When we use gross failure rate as our indicator of environmental effectiveness, however,
we are ignoring one very important factor: false failures. After all, any test can be made to have
a high failure rate - up to and including 100% - if one just makes the cutpoints tight enough and
does not choose to worry about false failures and their impact on overall program acceptance.
And, as indicated above, the OBD-I/M check does have more rigorous cutpoints than does the
EVI240. However, as suggested by the data presented in section 2 of this TSD, substantial
evidence suggests that lane-based EVI240s  can produce false failure rates at least as high as that
resulting from OBD-I/M testing on OBD-equipped vehicles, due to improper preconditioning,
infrequent and/or inadequate quality assurance, etc. Conversely, section 2 also suggests that the
vast majority of OBD-identified failures did trigger needed repairs and/or maintenance.  Finally,
manufacturers have an incentive to minimize MIL illumination when no detectable problem
exists. Therefore, we expect false MIL illumination to be a decreasing problem. Taken together,
these findings suggest that comparing the two tests - the OBD-I/M check versus EVI240 - the
OBD-I/M check will have no higher and perhaps less of a false failure rate than the EVI240. The
OBD-I/M check may therefore have a marginally higher absolute failure rate and ability to
identify problem vehicles when compared  to the IM240.

                              Table 18:  OBD vs. IM240 Fails

MY/Class
1996LDV
1996LDT
1996 Total
1997LDV
1997LDT
1997 Total
1998LDV
1998LDT
1998 Total

Total Tested
27,313
16,422
43,735
14,944
7,656
22,600
27,616
22,716
50,332
OBD Fail
(number)
645
436
1,081
91
66
157
118
123
241
OBD Fail
(percent)
2.4%
2.7%
2.5%
0.6%
0.9%
0.7%
0.4%
0.5%
0.5%
IM240 Fail
(number)
569
383
952
71
51
122
223
47
270
IM240 Fail
(percent)
2.1%
2.3%
2.2%
0.5%
0.7%
0.5%
0.8%
0.2%
0.5%
Failed Both
(number)
59
100
159
7
0
7
7
0
7
Failed Both
(percent)
0.2%
0.6%
0.4%
0.2%
0.0%
0.0%
0.0%
0.0%
0.0%
4.3.3   Conclusions
                                           37

-------
       Based upon its analysis of the Wisconsin lane data, EPA has concluded that although
readiness seems to be a concern among the earliest OBD model years, flexibility in the form of
readiness-based exemptions can go a long way toward minimizing the problem in the short term,
while improvements in OBD implementation by the manufacturers will likely eliminate or
greatly minimize the problem in the long run. To provide the needed flexibility to states to
ensure smooth implementation of their OBD-based I/M programs, EPA is taking action to allow
states to complete the testing process on MY 1996-2000 vehicles with two or fewer unset
readiness codes; for MY 2001 and newer vehicles, the testing process can still be considered
complete provided there is no more than one unset readiness code. This does not mean that these
vehicles are exempt from the OBD-I/M check.  The complete MIL check and scan must be run in
all cases, and the vehicle still must be failed if the MIL is commanded on. The vehicle should
continue to be rejected if it is MY 1996-2000 and has three or more unset readiness codes or is
MY 2001 or newer and has two or more unset readiness codes.  This allowance is consistent with
a FACA OBD workgroup recommendation. It is intended  to reduce the potential for customer
inconvenience during this start-up phase of the transition to OBD-I/M testing. We believe that
the environmental impact of this exemption will be negligible, given the small number of
vehicles involved, the  likelihood that at least some of these readiness codes will have been set in
time for subsequent OBD-I/M checks, and the fact that an unset readiness code is not itself an
indication of an emission problem.
4.4    Gas Cap Testing vs. OBD-I/M
4.4.1   Background

       Unlike the OBD exhaust test versus the EVI240, where the failure criteria for OBD are
tighter than the failure criteria for the EVI240, the OBD failure criteria for leak detection are
known to be more lenient than the gas cap pressure test currently in use in several states.
Although in theory this difference in test stringency should result in a greater number of failures
for the gas cap test than for the OBD-based evaporative system test, it is not obvious that
vehicles actually develop such "in between" leaks in the real world (and, if so, whether the
frequency of such leaks is significant enough to warrant recommending the continuation of the
gas cap test in conjunction with OBD-based testing).

       To shed light on this issue, EPA decided to look at the Wisconsin data, focusing on the
relative failure rates for the OBD-based evaporative system test versus the gas cap pressure test.
Unlike the previous discussion concerning the use of gross failure rates as an indicator of test
effectiveness when it comes to analyzing tailpipe tests, comparative failure rate  is a fairly
reliable index of a leak detection test's relative effectiveness — even when the failure criteria of
the two tests being compared are different.  Compared to tailpipe tests which are dynamic and
relatively complicated, leak detection tests aimed at identifying leaks in the gas  cap are
straightforward. Whereas vehicle operation (i.e., whether steady-state, transient, loaded or
unloaded), vehicle preconditioning, fuel composition, tire inflation, and multiple-instrument test
equipment calibration can all have a considerable impact on the pass-fail decisions made by a
traditional tailpipe test, none of these factors will have much effect on the traditional gas cap
pressure test.

                                           38

-------
4.4.2   Results
       Table 19 below compares the relative number and percentage of OBD evaporative system
failure rates found in the Wisconsin vehicle sample versus the gas cap pressure test failure rate.
Note that the gas cap failure rate is several orders of magnitude higher than the OBD failure rate
for the entire evaporative emission system.  Clearly, some of this is due to the fact that enhanced
OBD  evaporative system monitoring was phased in over the model years being looked at in this
sample33.  Furthermore, as described in our earlier discussion on OBD readiness, the overall
OBD  readiness on MY 1998 LDVs was dominated by vehicles which showed up at the test lane
without their evaporative system monitors having run. Even with these caveats taken into
consideration, however, EPA believes that the difference in failure rates is pronounced enough to
warrant consideration of retaining the gas cap pressure test as a complement to OBD-I/M testing
in those areas needing VOC reductions to attain and/or maintain their clean air goals.

                Table 19: Gas Cap vs. OBD Evaporative System Failure Rates
MY/Class

1996LDV
1996LDT
1996 Total
1997LDV
1997LDT
1997 Total
1998LDV
1998LDT
1998 Total
Total Tested

27,313
16,422
43,735
14,944
7,656
22,600
27,616
22,716
50,332
Gas Cap Fail
(number)
291
245
536
83
48
131
170
155
325
Gas Cap Fail
(percent)
1.1%
1.5%
1.2%
0.6%
0.6%
0.6%
0.6%
0.7%
0.6%
OBD Evap Fail
(number)
7
3
10
2
1
3
6
1
7
OBD Evap Fail
0.03%
0.02%
0.02%
0.01%
0.01%
0.01%
0.02%
0.004%
0.01%
4.4.3   Conclusions

       For the reasons discussed above, EPA believes that separate pressure testing of the gas
cap test using the test procedures currently employed in many I/M programs should be continued
in conjunction with OBD-I/M testing on MY 1996 and newer OBD-equipped vehicles.
Retention of the gas cap pressure test is the only exception to EPA's standing recommendation
regarding the dual testing of MY 1996 and newer, OBD-equipped vehicles.
       33
         The phase-in requirements for MY 1996, 1997,1998, and 1999+ are 20%, 40%, 90% and 100%, respectively.

                                           39

-------
Appendix 1: Test Sequence Used at Laboratories
                    1)     Procurement and acceptance into the program
                    2)     LA-4 cycle (preconditioning for IM240 test)
                    3)     IM240 test
                    4)     Drain in-use fuel
                    5)     Fill with indolene (40% fill)
                    6)     LA-4 cycle (preconditioning for FTP test)
                    7)     12 hour soak (no diurnal heat build)
                    8)     FTP test (no evaporative test)
                    9)     EVI240 test
                    10)    Repair if necessary
                    11)    OBD Readiness codes cleared thru operation of vehicle
                    12)    Repeat starting at step 4

-------
Appendix 2: Breakdown of the broken parts found for FTP-passing, MIL-on vehicles
Systems/Components
O2 Sensor
EGR.Valve
Ignition System (spark plugs, ignition wires, other)
Transmission components
PCM, Reprogram or Replace
Miscellaneous Wires
Engine, Mechanical (cylinder head, harmonic balancer, valve springs)
Vacuum Leaks
Thermostat, Cooling System
Fuel Pump
Transmission Unit
Cam Sensor
Secondary Air Combo Valve
Throttle Position Valve
Exhaust Leak
Catalyst
LDV
3
4
9
3
4
3
2
1
1
0
0
1
1
0
0
0
LPT
24
6
0
5
3
0
1
1
0
1
1
0
0
1
1
1

-------
Appendix 3: Lane IM240 vs. Lab IM240

       There are a number of differences between the way an IM240 test is conducted in an
inspection lane and the way that the test is conducted in an emissions laboratory. Some of them
are:

       1) quality of the test equipment

       2) frequency of calibration of test equipment

       3) skill of technician

       4) control of ambient conditions

       5) control of tire pressure

       6) operating temperature of the vehicle
       The first five items are of critical importance for a certification test in the laboratory but it
is our opinion that they not crucial for the I/M function. By far the greatest importance is item
six. There is a large variation in emissions between a partly warmed vehicle and a fully warmed
vehicle. In the laboratory an LA4 ( 1372 seconds )test cycle is run before the LAB240 test to
assure that the engine is fully warmed up and the catalyst hot. Vehicles arriving at I/M inspection
lanes are assumed to be at operating temperature due to the driving prior to arrival at the lane
(this may or may not be true).  Attempts have been made in I/M systems to address this
preconditioning problem through various methods.

-------
Appendix 4: Description of Vehicles/Trucks Assumed to Fail FTP
 CDH4, 1996 S-10
 Pickup MIL off
 (computer
 commanding MIL
 "On")
Truck could not accelerate and would
stall in 3rd gear on FTP
Lab IM240 results:
(THC/CO/NOx)
11.8/147/0.02
Black plume of smoke from
tailpipe
 ATL78, 1999
 Malibu
 MIL illuminated
 74,000 miles
IM240 test of the vehicle caused
closure of test cell due to hydrocarbon
contamination of instruments.
Decision made to not run FTP.
Lab IM240 results:
32.1/45.6/0.14
Raw fuel out of the tailpipe

-------
Appendix 5




                           Raw Data - OBD/Tailpipe Pilot

-------
Appendix 6




             Detailed Data From The 30-Vehicle OBD Evaporative Pilot Study

-------
Appendix 7

This section will primarily discuss issues with the Mazdas (150, 182), Hondas (153, 188), and
Fords (155, 194).  Reports and data are still pending from Honda and Ford, respectively.

       Use of external mounted thermocouples instead of installing internal thermocouples is a
common EPA practice for in-use evaporative emission testing. Without this simplification,
instrumenting the vehicle in strict accordance with the EPA certification requirements for
locating thermocouples and fuel drains can require cutting access panels through the floor of the
vehicle's trunk compartment. ATL's practical experience in using surface mounted
thermocouples is that this thermocouple location does not compromise testing accuracy because
accurate measurement of the internal liquid temperature at the mid-point of the 40% fuel level is
achieved. Vehicles with plastic fuel tanks used thermocouple probes installed through the
bottom of the fuel tank.  Any fuel tank modification that comprised the integrity of the OEM
tank was resolved by installing a new fuel tank or fuel sending unit before the vehicle was
returned to the owner.

       The FTP evaporative emission running loss test requires that  the measured fuel tank
temperature track the target temperature within 3 degrees F  over the dynamometer driving
portion (Urban Dynamometer Driving Schedule (UDDS), New York City Cycle (NYCC),
NYCC, UDDS) of the running loss test.  In general, the measured fuel tank temperatures denoted
as "Actual F" (failed) or "Actual R" (repaired) in Table A-5 in the Appendix indicate close
agreement with the vehicle manufacturer supplied fuel tank temperature profile. Manufacturer
supplied  fuel tank target temperatures and ATL measured temperatures for starting and ending
segments of the running loss test are summarized in Table A-5 in the Appendix.

       Exceptions to meeting the 3 degree tolerance were observed for vehicles 150, 154, 184,
and 189.  The 3 degree tolerance is not straightforward to meet for in-use evaporative emission
testing.  The deviations for vehicles 150, 154,  184, and 189  range from slightly over 3 degrees F
to about 7 degrees F. Not withstanding test work with vehicle  188, a 1999 Honda Accord
(discussed below), these deviations from the target temperature profile and the short time of the
excursion are not thought to be important because their effect on running loss results is judged to
be minor.

-------
Appendix 8: Wisconsin I/M and OBD Data Fields
Date/time
Mod yr.
Make
Model
Vin
Test
HC Stan
Co Stan
NOx
HC
Co
NOx
Em fsec
Em res
Pr cap Stan
Pr cap ini
Prcap
Pr cap res
Onboard
Obd res
Trno
Codel
Code2 - 6
Ready misfire
Fuel
Comp
Cat
Heat
Date and time the vehicle was tested
Model year
Make
Model
Vin number
Test number, 1 for the first time vehicle has been tested in this test cycle, 2 for the first retest. A very few vehicles
have been retested four times.
Final cutpoints, 0.6 grams per mile for cars
Final cutpoints, 1 .5 grams per mile for cars
Final cutpoints 0.7 grams per mile for cars, Wisconsin does not fail for NOx
Actual emissions total grams divided by the total miles, at the time the test was terminated
Actual emissions total grams divided by the total miles, at the time the test was terminated
Actual emissions total grams divided by the total miles, at the time the test was terminated
Number of seconds that the test ran. ( "0" for the full 240 second test)
P or F, pass or fail the 240 tailpipe test
Pressure cap standard, inches of water
Initial pressure, inches of water
Final pressure, inches of water
P or F, pass or fail pressure cap test
Whether the technician could find the OBD connection. No cases where he could not after October 98
Pass or fail, if MIL was illuminated. Should correspond to column AM
Number of codes present (sum of V through AA) but is sometimes wrong
The next six columns list the OBD trouble codes, if any
(Blank)
The next 1 1 columns list the readiness codes. 0 means that the monitor is not fitted. 1 means that the monitor is fitted
but not ready. 2 means that the monitor is ready
Fuel trim
Various circuits necessary for the other monitors to work
Catalyst
Heated catalyst

-------
Evap
Sair
Acsys
Oxy
Hoxy
Egr
Obd Mil
Odo
Evaporative system
Secondary air
Air conditioning
Oxygen sensor
Heated oxygen sensor
Exhaust gas recirculation sensor
Mil light, 1 if lighted, 0 if not. Should be same as column T
Odometer reading to nearest 1,000 miles (truncated)

-------