United States                                       May 11.
Environmental Protection                                   1993
Agency	EPA-AA-AQAB-93-01
Air

v5r  EPA   Evaluation of a Four-Mode
            Steady-State Test With
            Acceleration Simulation Modes
            As An Alternative Inspection and
            Maintenance Test for Enhanced
            I/M Programs
                  WmiamMPidgeoo
                   Daniel J. Sampson
                  PaulH.BuifoageIV
                  LanV C. T ^nAman
                  William B. Clemmens
                     Erik Heizog
                  David J. Bizezinski
                   David Sosnowski

-------
 1.    Executive Summary	1
      1.1   Purpose	1
      1.2   Findings	2
            1.2.1  Ability to Correctly Identify Vehicles Needing Repair ..2
            1.2.2  Ability to Distinguish Sufficiently Repaired Vehicles
                   From  Insufficiently Repaired Vehicles	4
            1.2.3  Ability to Distinguish Between Functioning and
                   Malfunctioning Evaporative Canister Purge Systems	4
            1.2.4  Test  Costs	S
            1.2.5  Adequacy of the ASM for Enhanced I/M Programs	5
2.    Background	6
3.    Test Procedures	12
4.    Data Description	13
      4.1   Data Listings	13
      4.2   Database Statistics	14
      4.3   Quality Control (QC)  Protocol	17
5.    Analyses/Discussion	19
      5.1   Introduction	19
      5.2   Analyses Techniques	19
            5.2.1  Reducing Four Steady-State Modes to a Single Score per
                   Pollutant For Comparison to One Cutpoint per
                   Pollutant	19
                  5.2.1.1  Reporting Overall ASM Results Versus Reporting
                           Individual Mode Results	20
                  5.2.1.2  Determination of Individual Mode Scores	21
            5.2.2  Multiple; Linear Regressions to Find ASM Coefficients ...21
            5.2.3  Applying ASM Coefficients	22
            5.2.4  ASM Concentration Measurements	23
            5.2*. 3  Explanation of the Criteria Used To Compare I/M Tests  ..23
                  5.2.S.I  Excess Emission Identification Rate  (IDR)	23
                  5.2.5.2  Failure Rate	24
                  5.2.5.3  Error-of-Commission (Ec) Rate	24
                  5.2.5.4  Two-Waya-To-Pass Criteria	25
                  5.2.5.5  Discrepant Failures (DFs)	26
                  5.2.5.6  Unproductive Failure  (UF) Rate	27

-------
             5.2.5.7  Vehicles with Malfunctions That Were Not
                     Counted as Ecs and DFs	27
             5.2.5.8  Weighting Factors to Correct Biased
                     Recruiting	29
 5.3   Comparison of IM240 Versus ASM Using Cutpoint Tables	31
 5.4   Comparison Using Scatter Plots and Regression Tables	40
      5.4.1   Using the Coefficient  of Determination  (R2  ) and
              Standard Error of the  Estimate  for  Objective
              Comparisons	40
      5.4.2 Advantage of Using Weighted Modes	42
      5.4.3 Observations of Scatterplots	43
      5.4.4 Poor ASM HC Correlation	43
 5.5   Derivation of ASM Coefficients and Mode Contribution
      Variations Prom Sample to Sample	45
      5.5.1   ASM Versus IM240 As  The Dependent Variable  For
              Determining ASM Coefficients	45
      5.5.2   Variability of ASM Coefficients	51
      5.5.3   Significance  of Mode Contributions	53
      5.5.4   Conclusions on ASM Mode Contributions	57
5.6   Repair Analyses	57
      5.6.1   Contractor Repairs	57
      5.6.2   Commercial repairs	71
            5.6.2.1  Introduction	71
            5.6.2.2  Database/Analysis	71
            5.6.2.3  Results/Conclusions	73
      5.6.3   In-Us« Emission Reductions froa Real World Repairs	77
      5.6.4   One-Mode Repairs on  ASM	90
5.7   Purge Analyses	99
      5.7.1  introduction	99
      5.7.2   The Database	100
      5.7.3  The Results	100
5.8   IM240 Improvements and the Four-Mode IM240	104
      5.8.1 Reduce Test-to-Test Variability	105
      5.8.2   Statistical Techniques to  Improve the IM240's
              Correlation With the FTP	107
                                ii

-------
6.    Test Programs by Other Organizations	110
      6.1   Colorado Test Prograa	110
      6.2   California Test Program	110
7.    Test Costs Comparison	112
8.    Evaluation of the Adequacy of the ASM for Enhanced I/M Programs	115
      8.1   Introduction	115
      8.2   MOBILESa Analysis	116
9.    Appendices Table of Contents	120
                                      iii

-------
 1.    executive

       1.1   Purpose

    On  November  5, 1992, the U.S. Environmental Protection Agency (EPA)
 promulgated a  regulation1 for state-operated enhanced Inspection and
 Maintenance (Z/M) programs.  This regulation established the IM2402 as the
 benchmark  I/M  test, against which any alternative test must be found
 equivalent,  or nearly  so but with compensating improvements in other program
 aspects.

    EPA performed testa on over 1500 vehicles in Mesa,  Arizona to evaluate a
 four-mode/  steady-state procedure utilizing two Acceleration Simulation
 Modes3.  (This four-nod* teat procedure will herein be referred to as the
 "ASM" test,  although only the first two modes are strictly ASM modes.)  This
 evaluation  was designed for determining whether the ASM is a suitable
 alternative  to the ZM240 for enhanced I/M testing.

    The ASM test utilizes equipment costing about half of the anticipated cost
 of the  equipment required for IM240 testing.  This equipment is less expensive
 because the  ASM  does not involve transient driving and the equipment only
 approximates mass emissions via pollutant concentration measurements.  In
 contrast, the IM240 is a transient test requiring more expensive equipment
 measuring true mass emissions during typical driving.

    The purpose of this document is to provide:

      - EPA's evaluation regarding the effectiveness of the ASM test;

      - a description of the analysis techniques EPA used;

      -  the data used in the evaluation; and
1  Inspection/Maintenance Program Requirements; Final Rule 40 CFR Part 51,
Federal Register, November 5,1992

2  William M« Pldgeon, and Natalie Dobie, "The XM240 Transient I/M Dynamometer
Driving Schedule) and The Composite I/M Test Procedure," EPA-AA-TSS-91-1,
January 1991

3  Thomas C. Austin and Larry Sherwood, "Development of Improved Loaded-Mode
Test Procedures for Inspection and Maintenance Programs,* Sierra Research,
Inc. and California Bureau of Automotive Repair, SAB Technical Paper No.
891120, May 1989.

-------
       -  a deacription of the teat program.

    This is the only ASM study conducted in an official Z/M station.  The
 vehicles were randomly selected and tested under the widely varying ambient
 conditions and preconditioning that normally  attend official I/M tests.  Many
 more cars were tested than in any other ASM study.  Also, this is the only
 study  to use one sample to develop the ASM mode  weighting factors and an
 independent sample to evaluate their effectiveness. EPA strongly believes that
 this study should be given far more weight than  all previous ASM studies.
      1.2   Finding*

    EPA's findings are based on performance  comparisons between the ASM and
the IM240 regarding five important considerations:

      -  their relative ability to fail malfunctioning vehicles  (needing
         exhaust emission control syatea repairs)  and to avoid failing
         properly functioning vehicles;

      -  their relative) ability to distinguish repaired vehicles (exhaust-
         repairs) that are sufficiently repaired from those that are
         insufficiently repaired;

      -  their relative ability to distinguish between functioning and
         malfunctioning evaporative canister purge systems;

      -  their relative) coat a; and

      -  the adequacy of the) ASM foe Enhanced Z/M Programs using MOBZLESa.
            1.2.1   Ability to  Correctly  Identify  Vehicles  Heeding
                     Repair

    EPA commonly/ use* the) rate) of exceaa emissions identified during an Z/M
teat to objectively and quantitatively compare Z/M teat procedurea.  Excess
emissions are thoae FTP-meaaured emissions that exceed the certification
emission standarda for the) vehicle under consideration.  For example, a
vehicle certified to the) 0.41 g/mi HC standard whose FTP result was 2.00 g/mi,
would have exceaa emisaiona equalling 1.59 g/mi HC  (i.e., 2.00 - 0.41 • 1.59)

    The exceaa emissions identification rate 
-------
 excess emissions.   The more excess emissions an I/M test identifies,  the
 better the test.

    EPA uses IDR instead of merely comparing the number of  vehicles that
 correctly fail  and correctly pass.  The IDR better contrasts the relative
 merits of competing- I/M test procedures because failing vehicles with high
 emissions is more  important than failing those that are only slightly above
 their certification standards.  For example, take two I/M procedures  that
 correctly failed 100 of the 500 vehicles that had FTP emissions greater than
 their certification standards/ but only 50 cars failed both tests.  If the
 fifty cars that failed Test A were high FTP emitters, and the other 50 cars
 that  failed Test B had FTP emissions only slightly above their standards,
 obviously Test  A would be preferred, and its IDR would reflect its better
 performance.  Test A's better performance is not evident in comparing the
 number of vehicles that correctly fail.

    The ASM does not find high emitting vehicles as well as the IM240.  Some
 high  emitters which could be caught with the IM240 give low ASM scores.  Table
 1.2.1 shows the percent decrease in the excess emissions identification rate
 that  would accompany substituting the ASM for the IM240.  For example, an
 IM240-based I/M program's HC and HOx IDRs will suffer nearly a 20% decrease by
 substituting the ASM test at the same failure rate  (18%) that is produced by
 EPA's  recommended  outpoints for biennial I/M programs.
    Table)  1.2.1    Loea  in  Identification Effectiveness  With  ASM  Test
               Scenario
                                     • C
             CO
                                                            MOi
Failure Rate Held at 18%
(0.8/15/2.0 •«• 0.50/12.0 XM240
Outpoints)
19.0%
                                                 9.5%
                                                                18.5%
14.0%
 Best ID** with ECS Held
 Below 5%
These value* ar* % differences.  For example:
Source:  Table 5.3.1, Section 5.3
                                                    14.3%
                                                           17.5%

                                                          * 100  •  19.0%
    An aggressive I/M program, tolerating both higher failure rates and higher
false-failure rates would relinquish about 15% of its inspection effectiveness
by substituting the ASM test.
    Additional related findings are listed below:

-------
 •     The ASM fails cars that are actually clean more often than the IM240.
      About 1 in 10 cars failed by the ASM did not appear to need repair,
      compared to about 1 in 30 for the IM240.  EPA knows from other testing
      that more preconditioning can eliminate IM240 errors; we are not sure
      whether it can for ASM failures.

 •     Making ASM outpoints more stringent in an attempt to get the same
      effectiveness as the IM240 increases the failure rate and/or the error
      rate beyond what EPA believes any I/M program would want or is willing
      to commit to in binding regulation form.

    The comparative ability to identify vehicles needing repair is fully
discussed in Section 5.3. why IDRs and associated criteria are important,  how
the criteria are derived, and the tradeoffs associated with increasing
outpoint stringency to increase IDRs are discussed in Section 5.2.
            1.2.2   Ability  to  Distinguish  Sufficiently  Repaired
                     Vehicle*  From  Insufficiently  Repaired. Vehicle*

    Vehicles that do fail the ASM teat and get repaired, can pass ASM
outpoints with repairs that are not  as effective as  the repairs needed to pass
IM240 outpoints, even when repaired  in good faith.   Also, the ASM modes are
prone to "adjust to pass/readjust  after* strategies  like the idle and
25007idle tests.

    Several of the 17 cars which failed  the Arizona  test and the ASM were
repaired in local shops, after which they passed the Arizona and ASM test but
still had high IM240 emissions. This is  the same pattern seen in 2500/idle  I/M
programs.  Repair analyses are discussed in Section  5.6.
            1.2.3   Ability  to  Distinguish  Between  Functioning  and
                     Malfunctioning  Evaporative  Caniatec  Purge
                     Systeas

    In purge testing,  the ASM  and the  IM240  should do equally well  in
identifying malfunctioning purge systems, so their comparative ability to fail
vehicles with malfunctioning purge systems has not been an issue.  Therefore,
the research issue has been whether, and how many properly functioning
vehicles would fail.  That is, EPA is more concerned with errors-of-commission
than with errors-of-omission.   About 4-6% of the vehicles failed the ASM
evaporative canister purge system test but were actually properly functioning.
This is about 38% to 52% of all cars that failed the ASM purge.

-------
 About 1% of the vehicles failed the ZM240 purge system test,  but were actually
 properly functioning.  This is about 12% to 18% of all cars that failed the
 IM240 purge.

    Unlike transient IM240 testing,  which requires vehicles to operate through
 a wide range of speeds and loads, the four steady-state modes of the ASM do
 not provide a purge opportunity for a significant portion of the fleet.   The
 purge system test results are discussed in Section 5.7.
            1.2.4   Zest  Cost*

    The 180 seconds required for this four-mode ASM test  is  the  same  as would
be needed for the XM240 if  special algorithms are used to pass obviously clean
cars and fail obviously dirty cars early in the cycle. So, the ASM does  not
save test time or reduce the number of lanes required. A shorter test based on
fewer than four modes would have even less benefit.

    The only cost advantage for this ASM test is that up  to  about half the
equipment cost can be avoided by not having variable inertia capability in the
dynamometer and low-concentration measurement capability in the gas analysis
instruments. This savings works out to about 75 cents per test in a
centralized program.  Test  costs are discussed in Section 7.

            1.2.5   Adequacy of  th«  ASM  for  enhanced  I/M Programs

    The MOBILES* analysis results show that even in a maximum annual  program,
covering all weight classes, with ASM, purge, and pressure testing of all
model years and comprehensive anti-tampering inspections, the ASM test yields
insufficient benefits to meet the performance standard for HC, CO, or NOx.

-------
 2.    Background

    EPA began development of a transient I/M test  procedure, named the IM240,
 during 1989.  EPA published a Notice of Proposed Rulemaking on July  13,  1992
 which proposed  a  performance standard for enhanced I/M programs that assumed
 the use of  the  ZM240 test procedure.

    On May 8, 1992,  ARCO Products Company released a report4 recommending that
 an  alternative  to the ZM240 be allowed for enhanced I/M programs.  The
 operating modes for ARCO's alternative procedure were not conclusively
 determined, but the modes were based on "Acceleration Simulation Mode*
 procedures developed by the California Bureau of Automotive Repair and Sierra
 Research, Inc.

    In contrast to ARCO's report which was somewhat ambiguous  on which modes
 should be included in the alternative test, the earlier BAR/Sierra report5  had
 recommended an  ASM I/M test that included the traditional 2500/idle  test (2
 modes) with two loaded dynamometer modes, at IS mph and at 25 mph.  The
 authors concluded that these two dynamometer modes were needed for NOx
 correlation with  the FTP*, and that the 2500/idle modes were necessary for
 good HC/CO correlation.

    ARCO's report,  which reached different conclusions,  was baaed on test
 results from five newer vehicles that were tested with and without implanted
malfunctions, resulting in 30 tests.  ARCO's conclusions are directly quoted
below:
4  Kenneth L. Boekhaus, Brian K. Sullivan, and Charles B. Gang, "Evaluation of
Enhanced Inspection Techniques on State-of-the-Art Automobiles," ARCO Products
Company, May 8, 1992.

5  Austin and Sherwood.

* The Federal Test Procedure (FTf) is a mass emissions test created to
determine whether prototype vehicles comply with EPA standards, thus allowing
production vehicles to b« certified for sale in the United States.  The  FTP
has become the "gold standard* for determining vehicle emission levels,  so it
is also used to determine the emission levels of "in-use"  vehicles.  The FTP
is too costly to use for Z/M because vehicles must be maintained  in a closely
controlled environment for over  13 hours.  The FTP driving cycle  includes 31
minutes of actual driving which  takes 41 minutes to complete  due  to a 10
minute engine shut-off between the second  and third modes  of  this 3 mode test.

-------
       1.   An «"NfT«d IM program utilizing steady-state exhaust emission testing is
           as effective in identifying can needing repair as is the EPA's proposed
           IM240test Because the cost of the IM240 equipment is four times that of
           an enhanced VM test, the enhanced VM test is far more cost effective.

       2.   Canister purging can be tested as effectively in an enhanced VM test as in
           anIM240test

       3.   The current BAR90 exhaust emissions test conditions of idle and 2500
           rpm/no load are not effective in identifying most malfunctions in state-of-
      4.    The ASM5015 steady-state test condition is effective in identifying
           malfunctions for HC, CO and NOx and should be included in any
           enhanced VM test developed.

      5.    Further development work is needed to develop one or more other steady-
           state test conditions to complement the ASM5015 test.

      &    The IM240 test correlates better with the FTP test in predicting absolute
           emissions levels than does the ASM5015  test  With one  or more
           additional steady-state test conditions, steady-state testing would likely
           correlate as well as the IM240 lest

       7.   The BAR 90 Test Analyzer System, with NOx analyzer, can be used for
           enhanced VM testing incorporating a steady-state dynamometer.^

    On November 5,  1992, EPA promulgated the final I/M rule  establishing the
IM240 as the benchmark I/M test, against which  any alternative test  must be
found equivalent.   The ZM240 is a  transient test which measures true mass
emissions during typical driving.   In contrast,  the Acceleration Simulation
Mode  procedures  only approximate mass emissions via pollutant concentration
measurements during several steady-state modes.   A BAR90 HC and CO analyzer
with  an NO analyzer ia sufficient.  Emissions measurements are not made during
the accelerations  and decelerations between the steady-state driving modes
because such measurement a  require  more expensive equipment including a
constant volume  sampler to dilute  the exhaust and measure flow, and analyzers
capable of accurately measuring the  resulting  low concentration samples.

    The  purpose* of  IP A'* alternative  I/M test procedure  study is to  evaluate
whether the DC240's performance as an I/M test  can be  attained, or nearly so,
by a  multi-mode, steady-state procedure (including two ASM modes)  that
utilizes equipment coating about half of the anticipated cost of the equipment
required for IM240 testing.
6  Boekhaua, et  al.

-------
    Due to widespread interest and the need for states to move forward with
specific testing plans, EPA prepared to initiate a test program to evaluate a.
steady-state loaded I/M test as a potential alternative to the IM240 for
enhanced I/M.  EPA's alternative test study focused on the Acceleration
Simulation Mode procedures.

    EPA needed to select a practicable number of steady-state operating modes,
but there was disagreement among the proponents of steady-state testing for
enhanced I/M.  ARCO concluded that the 2500 rpm 6 idle modes are not effective
for identifying most malfunctions.  In contrast, BAR/Sierra recommended an I/M
test consisting of the following modes: 5015, 2525, 2500 rpm, and idle.

    EPA's  desire to evaluate a single steady-state procedure agreeable to all
interested parties led to a conference call with the interested parties on
July 27, 1992.  The participants included:  ARCO, Sierra Research, California
Bureau of Automotive Repair,  Allen Test Products/SAVER, EPA's Testing
Contractor (Automotive Testing Labs), and EPA.

-------
    The parties reached a consensus that the steady-state test to be evaluated
 should include four modes  (a fifth mode was only to be performed on the first
 50 cars with automatic transmissions):
         15 mph (ASM S01S)*
         25 mph (ASM 2525)"
         50 mph at road load horsepower***
         idle (automatic transmissions using Drive rather than Neutral)****
         idle (first 50 vehicles with  automatic transmissions using Neutral
        rather than Drive)****

    This test procedure will herein be referred to as the "ASM*  procedure,
 although only the first two modes are strictly ASM modes.
* This is a steady-state 15 mph mode  (501S).  The dynamometer load is set to
simulate 50%  (5015) of the power  required to accelerate the particular
vehicle being tested at 3.3 mph/second at 15 mph.  The ASM does not include a
true speed changing acceleration during emissions measurement, instead the
speed is held constant while the dynamometer load is set to simulate the power
required to accelerate the car.  The 3.3 mph/second acceleration rate is the
maximum acceleration rate during the Federal Test Procedure (FTP).  The FTP is
the transient (accelerations and decelerations) procedure used to certify that
vehicles comply with Federal emissions standards, which is required before the
manufacturer can offer them for sale.  The  IM240, for the most part, is taken
directly from the FTP.  The 5015 mode usually requires a higher load setting
than the 2525 or the 50 mph road load modes.

** This  is  a steady-state 25 mph node (2525).  It is analogous to the ASM5015
mode in that the dynamometer load set to simulate 25%  (2525)  of  the power
required to accelerate the particular vehicle being tested at 3.3 mph/second
at 25 mph.

*** This  is  a  SO  mph mode with the dynamometer set to the power required for a
vehicle to maintain SO mph on level road taking into account  air resistance,
tire losses, bearing friction in the drivetrain, etc.  Air drag is  the major
resistance) at 50 mph.

**** Because) the)  vehicles) were  to be operated on the dynamometer,  it was
judged that th* vehicles could be safely tested at  idle in Drive.   Because
automatic-transmission-equipped vehicles idle  in drive during the FTP,  and
some ECM algorithms for the emission control system change with transmission
selector position, idle in Drive  is expected to yield better  correlation with
the FTP than idle in Neutral.  Since  all known idle emissions tests had been
run in Neutral prior to EPA's ASM evaluation,  the first 50 vehicles,  or more,
were also run in Neutral to allow comparison with other databases.

-------
     Having reached a consensus on the procedure to  be evaluated, EPA issued a
 work  assignment7 on July 30, 1992, directing EPA's  testing contractor to
 implement the  new procedure.   Shakedown testing began in August  and the  first
 official ASM test was  performed on September 10,  1992.   The last as-received
 ASM test was performed at the  I/M lane  on March 19, 1993.  This  analysis
 includes tests that were performed through February 17,  1993.

     Another issue EPA must consider is the impact of approving alternative I/M
 procedures on  the automobile manufacturers.  The  Motor Vehicle Manufacturers
 Association, in written comments on the Z/M NPRM,  said:

            6.0.0 ALTERNATIVE TEST METHODS

            MVMA agrees that EPA should not allow enhanced I/M areas to
       implement alternative tests until they have submitted SBfrfflui"** *^a supporting
       the quality of the test, v>4 showing that die test produces *m'TtKff reductions
       equivalent to those of the IM240 Test

            One such report by ARCO describes an acceleration simulation mode
       (ASM) that was compared to the IM240 test A substantial cost advantage win
       this alternative test is that it does not require the use of a constant volume
       sample (CVS) sampling system. The report references an earlier study by Sierra
       Research tt*n* cahj***flt*t T*tTf ffT"Tf*A"> by multiplying a "constant* times
       "emissions concentration'' times "inertia weight". Yet during the comparison of
       the two test methods, the mass emissions for the ASM were measured utilizing a
       CVS.  For • more accurate comparison, the ASM data should have been
       calculated in the method prescribed for use in the field, Le., with BAR-90
       readings and without the use of CVS equipment

            Probably of greaser concern, however, is the cutpoints selected for each
       test process.  Since cntpoints are aa important criteria in comparing and
       evaluating test processes, realistic outpoints have to be determined before an
       accurate comparison can be made. The IM240 Test cutpoints selected for this
       [ARCO's] comparison are extremely low and thus 'create' false Culms. IB
       contrast, die (ARCO] selection process for die ASM cutpoints is not well
       explained and remains ambiguous, making IM240 Test versus ASM Test
            ARCO used only five vehicles in the study. Their objective was to
       'evsJnsieilBviafcffiryofanalienMto                It appears much
       work is irxBiiiTxt before such an alternative could be iMuperly defined and
       evaluated. In the NPRM preamble, EPA stated that if this ASM test can be
       shown to be  aa effective at die IM240 Test, it could be permitted aa a
7  Statement of Work Change) 1,  July 30,  1992;  Work Assignment 0-2,  Contract
No.  68-CI-0055, "Teat Procedure) to Evaluate the Acceleration Simulation Mode
and  the Emissions  Measurement Capabilities of  a BAR90 Certified Analyzer  with
An Integrated Fuel Cell  Type NO Analyzer."
                                            10

-------
       "substitute*. MVMA is oooceroed that "substitute" tests could lead to several
       alternative tests with varying degrees of effectiveness. MVMA requests that
       EPA continue to critically assess any alternative tests proposed by enhanced VM
       areas. This review process will help assure that any alternative tests are able to
       properly identify failing vehicles.

     EPA is  also puzzled by  ARCO's conclusions which seen contradictory.
 ARCO's first conclusion states:

            An enhanced IM program utilizing steady-state exhaust emission testing
       is as effective in identifying cars needing repair as is the EPA's proposed IM240
    Their fourth conclusion  states that:

            The ASM5015 steady-state test condition is effective in identifying
       malfunctions for HC. CO and NOx and should be included in any enhanced VM
       test developed,

    ARCO's fifth conclusion  states that:

            Further development work is needed to develop one or more other steady*
       stale test conditions to complement the ASM5015 test

    These statements suggest that ARCO's testing indicated that the  ASMS015
was not as effective as the  XM240,  that the  other  nodes they evaluated were
not helpful,  and additional  work was required to identify better  alternatives.
This  report  will document  additional work performed by EPA  to evaluate the
ASM5015 and  three additional steady-state nodes.
                                            11

-------
 3.    Test  Procedure*

     The bast way to compare I/M tests is to utilize actual results from an I/M
 station in conjunction with FTP3  run on a  subset o£ the vehicles also tested
 at the I/M station.   A highly inferior  method is to compare the procedures
 based only on  test data collected in a  laboratory which is not subject to the
 range of  vehicle operating conditions which normally precede actual I/M tests,
 nor the range  of ambient conditions encountered during actual I/M tests.

     It is widely acknowledged that  a given vehicle's emissions can vary widely
 with changes in vehicle operating conditions  that precede emissions tests, and
 a given vehicle's  emissions can vary widely with ambient conditions
 encountered during an emissions test.   So, in contrast to laboratory test
 results,  the results  from pilot tests run  in  an official I/M station provide
 significantly  more confidence that the  pilot  test  results will accurately
 represent future results  when the procedure is mandated for official I/M
 testing.

    For these reasons, EPA carried  out  IM240  and ASM testing (through a
 contractor) in an  I/M station in  Mesa,  Arizona, with FTP testing  in a
 contractor-owned laboratory also  in Mesa.  In this  respect, EPA's  results  have
 much greater applicability to the real  world  than  results from recent  ''ASM*
 testing by Environment Canada8, ARCO, California Air Resources Board, and  the
 Colorado Department of Health.

    The test procedures are discussed in detail in Appendix A.
*  Vera F. Ballantyne, Pratt. St-party State Testing Raporfc an
-------
 4.    Data  Description

    From September 10, 1992 through March 19,  1993, EPA'3 contractor,
 Automotive Testing Laboratories  (ATL), conducted a vehicle  testing program in
 Mesa,  Arizona, a suburb of Phoenix, on mostly 1983 and newer vehicles.

    This program included several tasks designed to produce data for an
 analysis comparing the ASM test to the ZM240 test as  predictors of actual FTP
 emissions.  These tasks included the operation of an  Arizona I/M  inspection
 lane.  Vehicles at this lane received IM240s with a functional test  of  the
 evaporative canister purge system  (referred to as the purge test  in  the
 remainder  of this report), ASMa with the purge test,  Arizona I/M  tests,  and
 fuel system pressure tests under real-world I/M testing conditions.   In
 addition,  vehicles were recruited from the I/M lane for additional tasks,
 which  included:

       •  FTP laboratory testing
       •  IM240 laboratory testing
       •  Contractor IM240-targeted repairs
       •  Commercial repairs obtained by vehicle owners to pass the official
         Arizona I/M teat.

    Choosing vehicles for laboratory testing was driven by  the importance of
 testing and assessing emissions from—and the impact  of repair on—dirty in-
 use vehicles.  A random sample of vehicles visiting the I/M station would
 result in the contractor recruiting mostly clean vehicles (see Section
 5.2.5.8). But most excess emissions come from a relatively small percentage of
 vehicles known as high to super emitters.  To avoid the problem and cost of
evaluating a majority of vehicles that will ultimately be assessed as clean,  a
stratified recruitment plan was employed to deliberately over-recruit dirty
cars,  based on lane-IM240 results at the Mesa lane.  A nominally  50/50 mix of
 IM240-clean and DQ40-dirty vehicles were to be recruited for FTP exhaust
testing.  In actual practice, nor* dirty cars than clean have been recruited
which is shown in Table 4.2.2.

    Specifica_aonceraing the recruitment criteria and the teat procedures for
these task* are> discussed in Appendix A.

      4.1    Data  Listing*

    Appendix B provide* a listing of the data used for the outpoint
effectiveness analysis* the contractor  repair analysis, and the commercial
 repair analysis, which are all discussed in Section  5.
                                       13

-------
    Data for the over 1400 vehicles that only received one  set of  lane tests
 (no laboratory tests and no after-repair lane tests)  are only available  on
 disk.   These data  include the purge analysis data, and the lane  data used to
 calculate ASM  coefficients.  The available disk(a) will include  all IM240 and
 ASM lane data  including lane data for the laboratory tested vehicles. These
 can be  requested by contacting:

    William M.  Pidgeon
    U.S. EPA
    National Vehicle and Fuel Emissions Laboratory
    2565 Plymouth Road
    Ann Arbor,  Michigan   48105-2425
    Tel. No.  313-668-4416
    Fax. No.  313-668-4497

    Fax requests for data disks are preferred and a form is provided at  the
end of Appendix B; questions can be addressed by phone.
      4 .2   Database  Statistics

    The first official ASM/ZM240 test series was run on September 9,  1992.
Data collected up to March 17, 1993 were considered for these analyses.
During that period, 1574 vehicles received 1758 ASM/IM240 test series at the
Arizona I/M lane.  Priority for testing was given to 1983 and newer model year
fuel injected vehicles.  The following table illustrates the model year and
fuel metering distribution of the tested fleet:
                                      14

-------
                                Table  4.2.1
               Lana  Data  By  Modal  Year and  Fual  Mataring
MYR 1
81
82
83
84
85
86
87
88
89
90
91
92
To**1"
m 1
.
.
12
31
38
94
103
100
119
133
ISO
104
88S
TBI
.
2
6
30
42
S3
SO
61
77
48
35
11
415
ICARB
3
3
26
46
49
45
46
36
17
2
-
.
273
Totals
3
5
44
107
129
192
201
197
213
183
18S
US
1574
    Of tha 1574 vahiclaa tastad 27 were  recruitad for tha commercial rapair
program and 127 vahiclaa wara racruitad to tha laboratory for additional  testa
and for contractor repairs whan tha rapair criteria tiara mat (Section  S.6).
    Tha following liat summarizes  tha  criteria used for recruiting laboratory
vehiclea and for data completeness:

      •  Tha IM240 and tha ASM wara designed to distinguish between
         malfunctioning and properly functioning newer technology cars,  so
         only 1983 and nawar fuel-injected (no carbureted)  cars wara used.

      •  One-half of the) laboratory vehicles were to exceed 0.80/15.0/2.0
         (HC/CO/NOx) oa their laae-XM240.

      •  Ona-half of the) racruitad vehicles were to have tha lana-XM240
         pexforaad price to tha ASM.

      •  Only vehiclea having aa as-received FTP, aa aa-received lane-IM240,
         and aa as-recaived lana ASM teat wara used.  Vahiclaa missing any one
         of theaa three) taata wara not included in tha analysis.

    Tha resulting database consisted of 106 fuel-injected.   Table 4.2.2 lists
actual distribution statistics for these laboratory vahiclaa.
                                      15

-------
                                Table  4.2.2
             Distribution  of  Laboratory  Recruited  Vehicles
Lana-IM240
Passed: SO. 80 / 15.0 / 2.0
Failed: >0.80 / 15.0 / 2.0
Totals
Fuel
Metering
PFI
TBZ
PFI
TBZ

ASM Prior
to IM240
18
5
18
14
55
IM240 Prior
to ASM
14
3
23
11
51
Totals
32
8
41
25
106
    Table 4.2.3 shows the model year and fuel metering distribution for the
106 laboratory recruited vehicles.
                                Table  4.2.3
                Lab Data  by  Modal Tear aad  Fuel Metering
MYR 1
83
84
85
86
87
88
89
90
91
92
Totals
PFI
6
12
7
13
9
5
7
6
7
1
73
1 TBI
2
8
7
6
1
4
2
2
1
.
33
Totals
8
20
14
19
10
9
9
8
8
1
106
   Table  4.2.4 provide* FT* HC/CO emitter group statistics  for these
recruited vehicles.   FTP emitter groups are defined based on FTP  emissions as
follows:
                                     16

-------
I    Normals!    HC<0.82    and
                                                  CCX10.2
Highs :
0.82£HC<1.64
HC<1.64
and
or
and
CCX13 . 6
10 . 2SCO<13 . 6

Very Highs:
1.64SKX10.2
HCX10.2
and
or
and
C0<150
13.6SOX150

Supers :
HC210.2
or
CO2150
                                  Table  4.2.4
           Lab  Vehicle  FTP  BC/CO Emitter Category  Distribution
Ndramb
67
Highs
13
Very Hi«h
25
Supen
1
    For more detailed information on the data used Cor these analyses refer to
Section 5.  For details on the data excluded from these analyses refer to
Section 4.3 and Appendix C.

      4.3   Quality  Control  (QC)   Protocol

    This Section provides a general description of the QC process.  For more
detailed descriptions of the QC criteria and data excluded  from these analyses
see Appendix C, which lists the QC criteria in detail and the vehicles removed
due to the QC protocol.

    Data were received from ATT, in  two forms.  Calculated cycle-composite
values for all test* (lab and lane),  except the  ASM tests,  and second-by-
second data for lane-IM240s and ASHs  were provided.  The calculated data and
the raw seeond-by-second data followed separate  but similar QC processes.   The
calculated data* were processed using a program which performed checks on FTP
data and 11424ft (lab and lane) data.  These checks included bag result
comparisons/ fuel economy checks,  test distance  checks, dynamometer setting
checks, and test-to-test comparisons.  For details  on  these checks see
Appendix C.

    The second-by-second data were processed by  a separate  program which
performed similar checks for the raw data.  The  QC checks for the second-by-
second data included checks for acceptable speed, correct test/mode duration,
                                      17

-------
 sampling continuity,  reasonable ambient background concentrations, acceptable
 purge  flow, and reasonable  fuel economy.  Again details concerning these QC
 criteria are  included in Appendix C.  The second-by-second QC program also
 calculated composite  values for the IM240 and ASM tests.

    In addition to the QC program comparisons,  the calculated results reported
 by ATL were compared  to those results calculated from the second-by-second
 data.  Significant differences were investigated.  All lab vehicles violating
 the QC criteria were  hand checked by EPA staff and the data were corrected or
 removed, as warranted.  Due to the volume of lane data, lane vehicles that
 violated QC tolerances were removed from all pertinent analyses, without
 further  attempts to "save"  the data unless solutions were obvious.  These
 unutilized data will  be checked, as time permits, for future use.  In
 contrast, because the vehicles that received FTPS were relatively precious,
 significant effort was expended to correct data that were identified by the QC
process.

   Vehicles  removed from the sample are discussed in Appendix C on page C-4
through C-8.
                                      18

-------
 5.    Analyses/Discussion

      5.1   Introduction

    The purpose of Section 5  is to present the analysis EPA used to assess
 whether the ASM test ia sufficiently effective in identifying high emitting
 cars needing repair when compared to the  ZM240 test, and the findings of that
 analysis.  Additionally, it provides a comparison of the repair issues for
 those vehicles that were identified aa needing repairs.

    Section 5.3 compares the  ability of the ASM and the IM240 to identify
 vehicles needing repair, and presents EPA'a major findings regarding the
 effectiveness of the ASM aa an alternative to the IM240 for enhanced I/M
 programs.  It discusses comparisons of IM240 versus ASM using information from
 cutpoint tables.  Section 5.2 provides information needed to understand how
 the cutpoint tables were derived.

    Section 5.4 compares the  correlation of the IM240 and ASM with the FTP
 using traditional regression analysis. Section 5.5 discusses the somewhat
 specialized issue of how four ASM scores  are combined  in one score and the
 uncertainties and sensitivities in this process.

    Section 5.6 discusses the repairs performed by the contractor and repairs
performed by commercial repair shops.

    Section 5.7 discusses canister purge system test results and Section 5.8
discusses methods that will be explored to improve the power of the  IM240.

      5.2   Analyse*   Techniques

    This  section discusses the methodology and criteria EPA used to compare
the ability of the ASM and the IM240 to identify  vehicles  needing repair.
This section explains why the criteria are  important,  how  the  criteria  are
derived,  and indicates the tradeoffs associated with these interrelated
criteria.  Than* Section 5.3 contrasts the ASM and the IM240 using the
criteria explained ia Section 5.2.
            5.2.1   Reducing Four Steady-State  Modes  to  e  Single
                     Score  per  Pollutant   roc Comparison, to  One
                     Cutpoiat per  Pollutant

    This section explains how the final ASM  score is computed.   Two questions
will be answered ia this section:
                                      19

-------
       1.   Should the four mod* scores for each pollutant be combined to
           calculate a single result or score  for each pollutant, or should a
           separate score be reported for each of the four modes, and apply
           those separate scores to separate outpoints for each of the four
           modes?

       2.   How is the score computed for each  ASM mode?
             5.2.1.1   Reporting  Overall  ASM  Results Versus  Reporting
                       Individual  Mode  Results

    There are three alternatives for reporting overall ASM test results.  The
 first alternative does not combine the scores from the separate modes,  so this
 alternative  is analogous to the way 2500/idle test results are reported.  The
 HC  and CO scores are reported for the 2500 mode and separate HC and CO scores
 are reported for the idle mode for a total of four scores and up  to four
 outpoints.   For the four mode ASM test, this is too complicated.  With NOx
 added, three outpoints are needed for the 3 dynamometer modes and two
 outpoints (HC and CO only) for the idle mode, necessitating 11 separate
 outpoints.   (Because NOx emissions are insignificant during an idle test, NOx
 is  only considered for the 3 dynamometer modes.)   This first alternative  is
 too unwieldy for a four mode test.

    The second and third alternatives are two different  ways to report  a
 single score for each pollutant by combining one pollutant's scores from all
 the modes.

    For the  second alternative,  the single score  would be the sum of the
 scores from each mode, using a weighting of 25% for each mode.  For example,
 to  calculate the single ASM score for HC, the equation would be as follows:

ASM HC »(0.25 * 3015 HQ + (0.23 * 2525 HQ + (0.25 » 50RL HQ + (0.23 * idle HC)

    In the third alternative,  which EPA used, a single score is determined
 from the sum of the individual mode scores, but the weighting or coefficient
 for each was determined by regression techniques.  A multiple regression was
performed wherein all four of the mode scores are independent variables that
were regressed against FT* scores.  The regression produced coefficients for
each mode, plus a constant.  These coefficients weight each mode more
appropriately than the second alternative's method of just assigning each mode
a weighting of 25%.   BAR/Sierra used this regression method, and  likewise
EPA's analyses for this report also used this regression method,  with one
difference that is discussed in Section 5.5.  This yields an equation to
                                      20

-------
 calculate a single ASM score for each pollutant.   For example, the equation
 for calculating a single ASM HC score is as follows:

 ASM HC - (x * 5015 HQ + (y * 2525 HC) + (z * 50RL HQ + (t * idle HQ + Constant

             the x, y, z, t, and constant terms are listed in Table 5.2.2.

    While ASM advocates have used this  concept, none have proposed specific
 coefficients  for EPA to evaluate.  Thus,  EPA had to develop coefficients.

    The remaining question is:  "How were  each of the individual mode scores
 determined?"
             5.2.1.2   Determination  of  Individual  Mode  Scores

    Emission concentrations were measured on each of the four ASM modes (see
Section 5.2.3 for more details).  These concentration measurements were then
converted to simulated grams/mile emissions, because concentration
measurements do not provide a reliable indication of the magnitude of
pollutants emitted per mile traveled.  At the  same  exhaust  concentration
level, a heavy vehicle will emit more per mile than a light vehicle.

    To calculate simulated g/mi results,  EPA followed BAR/Sierra's method,
which was also followed by ARCO, wherein the measured concentration values are
multiplied by the Inertia Weight (engine displacement for the  idle mode)  of
the vehicle.  ttie Idle Mode was not considered tor  NO*  since it  is a no load
test.   EPA also divided these simulated g/mi  results by the scaling factors
listed in Table 5.2.1.  Using these factors yield overall ASM  scores that have
magnitudes similar to FTP and IM240 magnitudes.

              Table)  3.2.1:   feeling r act ore Used  to  Keep
              Regression  Coefficients  of  Bqual  Magnitude)
Pollutant
HC
CO
NOx
Modes 1-3
rcoNci * » / x
10*
10'
10<
Mode 4
[COMC1 * DiSP(L) / X
103
10°
NA
            5.2.2   Multiple Linear  Regressions  to  rind
                     Coeffioienta
    As previously discussed*  multiple linear regressions were performed using
the four modes (three for NOx) of the ASM test as the independent variables
                                      21

-------
 (XI,...,X4)   The one difference mentioned in Section 5.2.1.1 above  is  that the
 IM240 (rather than the FTP) was used as the dependent (Y)  variable*.   This was
 done for  testa on which the ASM was run first only,  because the corresponding
 IM240s are pre-conditioned, and thus more closely resemble an FTP.

    Vehicles that were recruited to the lab or received commercial  repairs
 were not  included in the database used to develop coefficients,  because  these
 are  the cars  to  which the coefficients were applied.  EPA determined that
 including these  would cause the developed coefficients to mask the  test
 variability of the ASM.   (This is also discussed in Section 5.5.)

    The multiple linear regressions were run on a database of 608 lane ASM
 tests versus pre-conditioned lane-IM240s, giving the following coefficients
 for  each  mode.
  Coefficient*
                Table  5.2.2
Developed  from  Multiple  Regression  ASM
    (see Table 5.2.1 for scaling factors)
Versus  IM240
Mode
Constant
5015
2525
50MPH
Idle
Adjusted Rz
HC
0.083
0.025
O.OS9
0.136
0.124
29.0%
CO
2.936
0.040
0.043
0.356
1.350
50.1%
NOx
0.258
0.061
0.219
0.352
NA
59.1%
            5.2.3   Applying  ASM  Coefficients

    The coefficients were then used to calculate composite ASM scores for all
lab vehicles and coonercially repaired vehicles.  These are the ASM scores
that arei reported in the. ensuing outpoint tables, scatterplots, and
regression*.
*  Why the XM240 waa used aa the dependent variable, rather than the FTP, is
explained in Section 5.5.  Thia ia not discussed here because the purpose of
this section ia to explain how, rather than why.  Also, this issue requires a
lengthy discussion and reliea on information presented in Section 5.5, so
repetition ia also avoided.
                                      22

-------
             5.2.4   ASM  Concentration  Measurements

    The ASM concentrations were measured over a 40 second period.  Because the
 exhaust sample delay to the most downstream analyzer cell is almost 10
 seconds,  the first 10 seconds of data were ignored.  The  concentrations that
 are used  in  the composite ASM score calculations  are actually reported
 averages  over the last 30 second period.  For various reasons, the time
 allotted  for measured concentrations was occasionally less than  30 seconds.
 In  these  few cases, EPA calculated the average  concentrations over this
 shortened period and reported these values. No ASM  tests were accepted with
 concentrations averaged over a period of less than 20 seconds.

             5.2.5   Explanation  of the Criteria Used  To  Compare  I/M
                     Tests

    In assessing the overall effectiveness of the ASM relative to the ZM240,
 it  is important to determine their effectiveness  in  measuring and determining
 a variety of  factors, including the excess emissions identified, the  failure
 rate, the error-of-commission rate, the two-ways-to-pass criteria, the
 discrepant failures, and the, unproductive failure rate.   Each of these  is
 discussed below.  These criteria are used in Section 5.3 to  compare the
 effectiveness of the two procedures.

            5.2.5.1   Excess  Emission  Identification  Rate (ID*)

    EPA commonly uses the rate of excess emissions identified during  an I/M
 test to objectively and quantitatively compare  I/M test procedures.   Excess
 emissions are those FTP-measured emissions that exceed the certification
emission standards for the vehicle under consideration.   For example, a
vehicle certified to the 0.41 g/mi HC standard whose FTP result was  2.00  g/mi,
would have excess emissions equalling 1.59 g/mi HC  (i.e., 2.00  - 0.41 - 1.59).

    The excess emissions identification rate (IDR) equals the sum of  the
excess emissions for thsj vehicles failing the I/M test divided by the total
excess emissions) (because* of imperfect correlation between I/M tests and the
FTP, some I/Mfpsssing vehicles also have excess emissions which are used for
calculating thei total excess emissions).  Thus, assuming an I/M area that
tests 1000 veaieles, 100 of which arc emitting 1.59 g/mi excess emissions
each,  while the) I/M test, fails  (identifies) 80 of the excess emitting
vehicles,  the excess emission identification rate can be calculated as
follows:

    8Q failing vehicles * 1.59 a/m\ excess per  vehicle     A   an.
       100 vehicles * 1.59 g/mi excess  per vehicle
                                      23

-------
    EPA uses  IDR instead of merely comparing the number of  vehicles that
 correctly fail and correctly pass.  The IDR better contrasts the relative
 merits of competing I/M test procedures because failing vehicles with high
 emissions is  more  important than failing those that are only slightly above
 their certification standards.  For example, take two I/M procedures  that
 correctly failed 100 of the 500 vehicles that had FTP emissions greater than
 their certification standards, but only SO cars failed both tests.   If the
 fifty cars that failed  Test A were high FTP emitters, and the other 50 cars
 that  failed Test B had  FTP emissions only slightly above their standards,
 obviously Test A would  be preferred, and its IDR would reflect its better
 performance.   Test A's  better performance is not evident in comparing the
 number of vehicles that correctly fail.

            5.2.5.2   Failure  Rate)

        As the IDR increases with different test procedures or different
 outpoints,  the opportunity to identify vehicles for emission repairs also
 increases.  However, this measure is not sufficient for determining which is
 the more  efficient and  cost-effective I/M test.  Other criteria must also be
 addressed before such an assessment can be made.  One such criterion is the
 failure rate,  which ia  calculated by dividing the number of failing vehicles
 by the number  of vehicles tested.  For example:

    50 vehicles failed I/M ., ,A_   «•.,/«* ji
     1000  vehicle. tested  * l°° " 5% 1/M f*ilur* rat-

        The ideal  I/M teat ia on* that fails all off the dirtiest vehicles
 while passing  those below the FTP standard or close: to it but still above it.
 The potential  emission  reduction benefit decreases as emission levels from a
 vehicle approach the standard, because the prospect for effective repair
diminishes.  Thus,  achieving a high IDR in conjunction with a low failure rate
 (as a result of identifying fewer vehicles passing or close to the standard)
efficiently utilise* resources.

            5.2.9.3   Brxos-o£-Commission   (Bo)   Rat*

    Properly functioning vehicles which pass FTP standard* sometimes fail the
 I/M test;  theae are referred to aa false failurea or errors-of-commission
 (Bcs).  when error-of-commission vehiclea are sent to  repair shops, no
emission control system malfunctions exist.  Often, the repair shop finds that
the vehicle now paasea  the teat without any change*.  Theae false failures
wast* resources, annoy  vehicl* owner*, and may  lead to emission* increases as
a result of unnecessary and possibly detrimental "repairs."  Automobile
manufacturers  see  this  aa a significant problem, sine* it  can contribute to
                                      24

-------
 customer dissatisfaction and increased warranty coats.   An I/M program seeking
 larger emission  reductions through more stringent emission test  standards may
 actually increase the number of false failures.  The error-of-commission rate
 is,  therefore, an important measure for evaluating the  accuracy  of  I/M tests.

    To see how an error-of-eommission rate is  calculated,  assume an I/M area
 which  tests  1000 vehicles, of which 100 fail the I/M test, although only SO of
 those  100 failing vehicles also exceed the FTP standards.   The error-of-
 commission rate  equals the number of vehicles that fail the I/M  test  while
 passing  the  FTP  divided by the total number of vehicles which  were  I/M tested:

         50 vehicles failed I/M but passed FTP „ ,AA    c.  _
         	100Q v.hicl.3       	 * 100 -  5% EC rat.
    As the error-of-commission rate decreases,  vehicle  owner  satisfaction and
acceptance of the I/M program increases.  Thus, while it is relatively easy to
improve the IDR by making the I/M test standards more stringent,  this
*improvement* comes at the cost of potential increases in the error-of-
commission rate.

            5.2.5.4   Two-Ways-To-Pass  Criteria

        The theory behind the two-ways-to-pass criteria for the IM240 is as
follows.  Assuming that the IM240 test was correctly performed in the first
place, the most likely reason that a properly functioning vehicle would fail
an IM240 is that the evaporative canister was highly loaded with HC molecules
and that they were being purged into the engine during the test.  This has
been a significant cause of false failures in existing I/M programs and it has
been shown that highly loaded canisters can cause both high HC and CO
emissions, even though the feedback fuel metering system is functioning
properly.

    Since the canister is being purged during the IM240, the fuel vapor
concentration fro* the canister continually decreases during IM240 operation.
The decreasing; fuel vapor concentration results in decreasing HC and CO
emissions.  So£ emissions during Mode-2 (the last 136 seconds of the 239
second cyclep should be) lower than the composite results.  On the other hand,
if the vehicle is actually malfunctioning, Mode-2 emissions should remain
high.

    Catalyst temperature can also affect test outcome.  Emissions are
generally highest after a cold start, before the catalyst  has had a chance to
warm up.  If a vehicle is standing in line for a prolonged period of  time,  or
was not sufficiently warmed up before arriving at the  test lane, this can
                                       25

-------
 cause the vehicle to fail,  whan,  in fact, it should be passed.  Under the two-
 ways-to-pass criteria, Mode-1  acts as a preconditioning mode, thus providing
 insurance against this particular variety of false failure.

    NOx outpoint  criteria are not included in EPA'3 two-ways-to-pass
 algorithm.   So a  vehicle which meets the ZM240 NOx outpoint  (i.e., composite
 NOx £ 2.0)  only fails if both  its composite emissions exceed the HC or CO
 composite outpoints, and its Mode-2 emissions exceed the HC or CO Mode-2
 outpoints.   In other words, a  vehicle can pass by having low HC/CO emissions
 in Mode-2 even if its Mode-1 HC/CO emissions were high.  EPA is mandating this
 approach  to IM240 outpoints.

    The ZM240 outpoint tables,  in Appendix E and Table 5.3.1 in the next
 section,   were calculated using the two-ways-to-pass-criteria.

    The two-ways-to-pass criteria were optimized only at the outpoints  EPA
 recommends  for biennial  enhanced  Z/M programs, which are referred to as
 "standard*  or "recommended* ZM240 outpoints.  For composite emissions, the
 standard  outpoints are 0.80 g/mi  HC, 15.0 g/mi CO and 2.0 g/mi NOx.  The Mode-
 2  criteria  for the standard outpoints are 0.50 g/mi HC and 12.0 g/mi CO.  The
 Mode-2 outpoints  were carefully selected from EPA'a ZM240 data collected in
 Indiana,  to pass  properly functioning vehicles while continuing to  fail
malfunctioning vehicles. (The Mode-2 criteria were not redetermined for this
new Arizona sample.)  The) Mode-2  criteria, listed in the outpoint tables in
Appendix  r and Table 5.3.1, simply increase proportionally with increasing
composite outpoints  (i.e., become less stringent) and decrease proportionally
with decreasing composite outpoints (i.e., become more stringent).  The point
 is that these Mode-2 criteria have not been optimised at every stringency
 level to provide  the) beat tradeoff among IDR, failure rate,  and Ecs, so it is
probable  that the effectiveness of the ZM240 Mode-2 outpoints can be improved.
            5.2.S.S   Discrepant  Failures   (DFs)

    Discrepant failure* are vehicles that fail an Z/M teat for HC and/or CO
and pass the> ffs? foe HC and CO, but fail the FTP for NOx, or vice versa.   The
table below illustrates one) possible discrepant failure  scenario:
Teat
Short Teat
FTP
HC or CO
Pass
Fail
NOX
Fail
Pass
    Zn thia example, a false failure for NOx happens to occur on a vehicle
which is a false pass for HC/CO.
                                      26

-------
    Repair diagnostic routines are frequently selected on the basis of which
 pollutant caused the I/M test failure.  Given that HC/CO and NOx move in
 opposite directions with changes to the A/F ratio, there is  not much reason to
 expect  that  fixing a NOx problem will reduce HC/CO emissions.  Therefore,
 these scenarios  represent an error of sorts for the short test.  If a vehicle
 was to  fail  the  short test for NOx, whereas the only high FTP pollutant was
 CO, chances  are  the mechanic will be looking for a problem that causes high
 NOx.  In this case, the problem that is causing high CO  emissions  is not
 likely  to be found.

             5.2.5.6   Unproductive  Failure  (UP) Rate

    The unproductive failure rate represents the percentage  of vehicles that
 will be identified as needing repair, but either repair  is not needed  (Ecs),
 or it is not likely the reason for repair will be found  (DFs).  The
 unproductive failure rate is calculated by adding errors-of-conmission to
 discrepant failures, and dividing the quantity by the total  number of vehicles
 which were I/M tested.  Keeping with the same example as above, take an  I/M
 area which tests 1000 vehicles.  100 fail the I/M test,  50 of those  100
 failing vehicles are Errors-of•Commission, and S are Discrepant Failures:

           50 Bea + 5 DFs    » ,««   . ... .,_.*
        1000 vehicle, tested * 10° ' 5 5% OT  rat«

    Unproductive Failure
            5.2.9.7   Vehicles  with  Malfunctions that Were Mot
                       Counted a*  EC*  and  DFs

    Errors-of-commission in I/M programs have been most  often  caused by test-
to-test variability or incompatibility between the I/M test procedure and
vehicle emission control system*  (e.g., air pump switching), so attempting to
repair Ee vehicles) were) fruitless.  With the IM240, however, EPA has found
that some vehicle* that had failed the ZM240 and passed the FTP actually did
have malfunction*, so they were correctly identified and air quality would
suffer by ignoring the*).  By the strict definition of Ecs, the IM240 is
penalized despite it* successfully identifying malfunctioning vehicles.

     A likely reason for vehicle* passing the FTP despite a malfunction is
that malfunction* are sometime* intermittent,  vehicle 3172 provides a good
example.  This vehicle had a number of IM240s performed, some with high NOx
and others with low NOx.  The mechanic indicated that the  vehicle had a sticky
EGR valve.  The mechanic's diagnosis was not influenced by the FTP result
because the contractor had been instructed to report only  IM240  scores to the
                                      27

-------
 mechanics,  not the FTP score.   This has become standard practice to allow the
 contractor-repairs to simulate  commercial repairs/ where mechanics will not
 have access to FTP results.

     For this analysis, EPA did not count vehicles as Ecs when they had a
 malfunction that would logically explain the IM240 test failure.  To
 facilitate  a fair comparison between the ASM and the IM240, the ASM failing
 vehicles  that  passed the  FTP, but had malfunctions, also were not counted as
 Ecs when  their malfunction would logically explain the ASM failure.

     EPA was very conservative in that a vehicle was counted as an EC unless
 the malfunction clearly explained the ASM or IM240 test failure.  For example,
 vehicle 3239 failed the IM240 with 2.4 g/mi NOx yet passed the FTP.  The
 vehicle was diagnosed as  having a slow responding 02 sensor and it was
 replaced.  Because (1) a  report of a slow-responding 02 sensor does not
 indicate  that  objective criteria were used,  (2) NOx failures are not strongly
 associated with defective 02 sensors, and  (3) all of its other IM240 tests
 had passing NOx, the car  xaa. counted as an EC despite the mechanic's judgement
 the the O2 sensor  should  be replaced, which he did.

    Using the similar logic, some vehicles with discrepant failures were also
 not counted aa  DFs when their malfunctions could logically explain the  I/M
 test failure and a proper repair could be expected to reduce FTP emissions of
 the affected pollutant even though FTP emissions of that pollutant were
 initially below FTP standards.  For example, the vacuum leaks on vehicle 3154
 could cause a lean air/fuel ratio which can lower the catalyst's NOx
 conversion efficiency and cause higher combustion temperatures, both of which
 can causa high  NOx on the ZM240 and ASM.  FTP NOx emissions should also be
 affected but perhaps not  enough to cause an FTP failure.  Because it  is
 logical for a mechanic to check for vacuum leaks on a car that fails NOx,  and
this vehicle did have vacuum leaks, the I/M tests shouldn't be penalized for
correctly identifying the malfunction,  on the other hand, if this vehicle had
failed CO on aa I/M test  and NOx on the FTP, the mechanic would  look  for
problems causing a rich air/fuel ratio, which would probably preclude looking
 for vacuum leak*.

     Table 5.2  list* the five vehicles the met the strict definitions for  Ecs
or DFs, but were not counted for the reasons discussed.  Note chat  while these
vehicles were not  counted a* Scs or DFs in the outpoint  tables,  they still do
count toward the failure  Rate.
                                       28

-------
    Table  5.2.3.1:   Cars  not  Counted  ••  «cs  or OWm
   Vehicle
      Original  Status
Malfunctions  Explaining
    I/M  Test  Failure
 3154
3172
3200
3216
3244
Discrepant failure;  failed
IM240. and ASM NOx, but
failed FTP CO only.
Error-of-Coaudaslon;  IM240
NOx.

Discrepant Failure;  failed
IM240 and ASM NOx, but
failed FTP HC only.
Injector seals leak  at
intake manifold;
distributor advance  vacuum
hose broken.

EGR valve sticks,  EGR valve
vacuum line plugged.

EGR position sensor  out of
range.
Brror-of-Commission/ ASM HC  ECM malfunction
£rror-o/-Commission/ XM240
and ASM NOx.
Injector malfunctions
intermittently.
            5.2.5.•   Weighting  Factor*  to  Correot  Biased Recruiting

    The criteria  used to recruit  vehicles  for laboratory testing heavily
biased this laboratory sample toward ZM240 failing vehicles.   Sixty-two
percent of the 106 laboratory vehicles had failed the lane-!M240 criteria
(>0.80/15.0/2.0), whereas only 19% of 2,070 cars tested at  the Ian* failed the
IM240.  This resulted in a laboratory sample that was highly  biased toward
failing vehicle*.  (Two-ways-to-pass criteria was not considered for
laboratory recruiting.)

    Using this biased database results in  unrealiatically high excess emission
identification rates, and unrealiatically low error-of-commission rates.   So
the laboratory database- most be corrected to represent the pass/fail vehicle
ratio in the*ia-ose fleet to correctly determine iDRs, failure rates,  and Ecs.
The database- was) corrected using the weighting factors presented in Table
5.2.5.2.

    Weighting factors are used as follows:  If  the 66 failing vehicles that
received FTP tests had excess HC emissions which totaled 100 g/mi, the
database would be corrected in this case by multiplying 100 by the 5.97
weighting factor, resulting in corrected total excess emissions of 597 g/mi
for the dirty vehicles.  In comparison, the total excess emissions of the
                                      29

-------
 IM240-clean vehicles have to be multiplied by 41.9 to make their excess
 emissions representative.  The total simulated excess emissions  are the  sum of
 the simulated excess emissions from the clean and dirty vehicles in the  I/M
 lane sample.  The number of vehicles tested was similarly adjusted with  the
 factors for the purpose of calculating failure rates.  The sample of 40  clean
 vehicles provides confidence in conclusions about a teat's relative tendency
 to avoid failing clean cars.

                               Table  5.2.5.2
      Weighting  raetors  Used  To  Adjust the  Laboratory  Database
IM240 at Lane
Paaa:
rail:
SO. 80/15. 0/2.0
>0. 80/15. 0/2.0
t at Lane
1676
394
f at Lab
40
66
Weighting Factor
41.90
5.97
    The resulting weighted database was  used to produce the  realistic
estimates of IDRa, failure rates, and Ecs that are listed as outpoint  tables
in Appendices D 4 E.  These outpoint tables are sorted by failure rates.   For
the outpoints that produce the same failure rate,  the results are sorted first
by HC IDRs (in descending order) and then by NOx IDRs.
                                      30

-------
      5.3   Comparison  of  XM240  Versus  ASM  Using  Cutpoint  Tables

    In assessing the overall effectiveness of I/M test procedures,  as
discussed in Section 5.2, it is important to determine the test's
effectiveness in terns of IDR,  the failure rate, discrepant failures and
unproductive failure rate.

    Appendices D and E list the same criteria for many different outpoints.
Table 5.3.1 provides a summary of these criteria to  compare the ASM with the
IM240 for the following three important scenarios:

      -  ASM outpoints selected to achieve the  same  18%  failure rate  (using
         the outpoint tables that are  reweighted to  correct the lab sample
         bias)  that result from EPA's  recommended  IM240  two-ways-to-pass
         outpoints of .80/15.0/2.0 + 0.50/12.0. Among the ASM outpoint
         combinations with this failure rate (see  Appendix B), a combination
         was selected that produced the maximum IDRs for all  the pollutants
         simultaneously,  so there was  no need to set priorities among
         pollutants.

      -  ASM outpoints selected to achieve IDRs similar  to the IDRs  that
         result  from EPA's recommended IM240 two-ways-to-pass outpoints of
         .80 /  15.0 / 2.0 + 0.50 / 12.0.  Because  ASM CO and NOx  IDRs could
         more favorably be presented by excluding HC, two ASM outpoint sets
         are presented* on* to provide matching ASM and  IM240 HC  IDRs
         (resulting in better IDRs for CO and NOx),  and the  second to provide
         matching ASM and IM240 CO « NOx IDRs.

      -  ASM and IM24O outpoints selected to achieve th* highest  IDRs possible
         whil* keeping th* unproductive failure rat* below 5%.  This case was
         addressed oa th* possibility that an aggressive I/M program might be
         willing to op*rat* with such a high Be rat*.
                                      31

-------
                                Table  5.3.1
   Comparison  of  th* Ability of  IN240  and ASM  to  Identify  Vehicles
   Whoa*  Emission*  Exceed  Certification  standards  Baaad  on  106  Lab
           Vehicle*  Weighted  to  Represent  1676  Lan*  Vahiclaa
Eicin Emissions
Failure Identified
Test Scenario Rate HC CO NOz
% % % %
IM240 Staadnd IS 92.2 67.5 83.4
Ctpcs.
ASM SaaMFafl IS 74.7 61.1 68.0
Rat*
ASM SimilatHC 42 92.4 78.1 95.0
IDR
ASM Similar CO* 24 80.4 66.2 89.4
NOx IDR.
ASM BeetlDRa 2S 82.5 67.0 80.1
w/UF9<3%
IM240 BestlDb 33 95.9 78.2 97.1
w/UFe<5%
Weighted f
ofVehielaf 1676
Unproductive,
Discrepant Failure
EC* Failures Rate** Cutpotnts
# # %
0 12 0.6 .80/15.0/2.0
+
0.50 / 12.0
42 6 2.3 1.00/8.0/2.0
174 180 17.1 1.00/8.0/1.0
84 48 6.4 1.00/11.0/1.4
48 48 4.6 .40/8.0/1.5
60 12 3.5 JO / 9.0/1.7 +
.19 / 7.0

*  Exclude* Bo vahlclaa that had malfunctions that cauaad an I/M tast failure,
but bacauao thoy wr« intaraittant malfunctions, did not fail th« FTP.  FTPs
war* always performed on a diffarant day.  Sine* thay w*r* corractly
idantifiad by th* I/M t**tf they ar* not vahiclaa that will "ping-pong".

** The Unproductive Failure Rate  includes th* traditional EC vehicles and the
discrepant failures,  without including th* traditional Be vehicles that were
found to have intermittent malfunctions that were not identified by th* FTP
test.
                                      32

-------
    For the first scenario with an 18% failure rate for both  tests,  the ASM
 statistics in Table 5.3.1 show that the IM240 identifies 18 percentage points
 more of the excess HC emissions and 15 percentage points more of the excess
 NOx •missions than the ASM identifies, with a significantly lower unproductive
 failure rate.  Expressed differently, an lM240-based program would relinquish
 about 19% of its HC effectiveness and 18.5% of its NOx effectiveness by
 substituting the ASM test at the same failure rate.  Some relatively dirty
 vehicles are missed by the ASM and replaced by relatively clean vehicles.
 This scenario is illustrated in Figure 5.3.1.

    The second scenario shows that in order to achieve HC IDRs similar to  the
 IM240'3 at an 18% failure rate, the ASM'a failure rate must be increased to
 42%, resulting in an unacceptable EC rate of 17%.  To achieve similar CO and
 NOx IDRs with the ASM, an ASM failure rate of 24% is necessary, and that also
 results in an unacceptable unproductive failure rate of 6.4%.  This scenario
 is illustrated in Figures 5.3.2 and 5.3.3.

    The last scenario compares the tests at the maximum IDRs achievable while
 limiting the unproductive failure rate to less than five percent.  Again, the
 IM240 IDRs are significantly higher than the ASM's, with a lower unproductive
 failure rate.  The IM240 HC IDR is 14% higher, the CO  IDR is  14.3% higher, and
the NOx IDR is 17.5% higher, with an Ce rate that is 1% lower.  Expressed
differently, an aggressive IM240-based program with a  3.5% unproductive
 failure rate would relinquish about 14% of its HC and  17.5% of its NOx
effectiveness by substituting the ASM test at at an even higher unproductive
 failure rate.  This scenario is illustrated in Figure  5.3.4.

    These statistics indicate that the ASM test is significantly less
effective than the IM240 as an I/M test.

    The second scenario, wherein the ASM's HC IDR is raised to match the
IM240's HC IOR of 92%, is anticipated to raise the following question:

          Why dkta't EPA neks the ASM*s HC onpoim mom stringent to increase
      the ASM's IDR wfatant toatntaf the arafency of the ASKTi NOx aupoini.
    The answe»-i« that eight vehicles  (see Table 5.3.2) have a major impact on
the ASM HC XDR, but their ASM HC scores are  less than 0.3 g/«i.   Although
their ASM HC score* are very  low,  they account for roughly 10.5% of the total
excess FTP HC emissions.  These eight  vehicles also have ASM CO  score* below
8.0 g/sd.  While developing the ASM outpoint tables, EPA found that ASM
outpoints below 0.3/1.0 caused  failure rates and Ccs to increase excessively,
so the final cutpoint tables  did not include tighter outpoints.  So to achieve
                                       33

-------
the IM240'3 HC IDR, the only "practical" way to  identify these cars is through
the NOx outpoint.

    Five of the eight  cars with  high FTP HC that pass the ASM HC outpoints are
failed by a NOx outpoint of 1.5  or less.   These  five cars account for 7.6% of
the total excess HC emissions.  So the 1.0 g/mi  ASM NOx outpoint achieves an
HC IDR comparable to the 92.2% achieved by the IM240 at EPA'3 standard
outpoints.

    Table 5.3.3 summarizes the ASM outpoint table in Appendix E to show that
the only way for the ASM to achieve the IM240's  HC IDR of 92.2% at the
recommended outpoints for biennial programs  is to lower the ASM NOx outpoint
to 1.0.
                   Table  5.3.2:   Vehicle*  that Paaa a
            0.30  g/ai  ASM  HC Outpoint While Failing FTP HC

Vehicle   HC  FCT   CO FTP   NOx  FTP   HC  ASM   CO  ASM   NOx  ASM
3180
3192
3195
3199
3201
3254
3257
3259
Table
ASM
0.96
0.49
0.51
0.53
0.94
1.87
1.26
1.94
5.3.

Cutpolats
0.30 / 8.0 t
0.30 / 8.0 >
0.30 / 8.0 /
0.30 / 8.0 1
0.30 / 8.0 /
0.30 / 8.0 /
0.30 / 8.0 /
0.40 / 8.0 /
1.00 / 8.0 /
1.00 / 9.0 /
1.00 / 8.0
' 2.0
1«*
' 1.9
1.4
1.3
1.2
1.0
1.0
1.2
1.0
/ 1.0
9.75
6.31
5.80
10.90
19.73
35.87
8.57
14.95
3: Alterni
Idoatifkatloa
HC 1 CO
88.0% 69.2%
88.2% 69.3%
89.0% 71.0%
90.3% 73.1%
90.3% 73.2%
91.4% 76.4%
96.6% 82.1%
92.4% 78.8%
84.1% 71.9%
91.3% 74.9%
92.4% 78.1%
1.22
0.53
0.66
1.53
1.72
1.16
0.90
0.53
0.15
0.20
0.18
0.29
0.17
0.29
0.25
0.23
itiva ASM Outpoints
••tea
1 NOx
74.9%
78.2%
82.7%
89.3%
89.3%
89.9%
93.0%
93.0%
89.8%
93.0%
93.0%
Fallart
Rat*
29%
30%
33%
38%
40%
42%
48%
43%
32%
40%
42%
Be
Rat**
4.3%
4.3%
4.3%
6.4%
6.4%
6.4%
8.7%
8.7%
4.0%
8.4%
8.4%
3.64
6.42
3.26
3.89
4.73
7.37
4.65
4.61
For High
I Dlscrtpaat
1 FallarM
0
0
42
42
84
12C
132
138
132
221
180
0.68
0.92
1.27
1.53
1.22
1.13
1.03
1.22
HC XDR4
1 UF
1 Rat*
4.3%
4.3%
6.4%
8.4%
10.4%
12.4%
13.0%
13.3%
10.4%
19.1%
17.1%
                                      34

-------
    To achieve an HC ZDR rat* greater than 89% a NOx outpoint of less than 1.5
is necessary.  To achieve an HC ZDR rate greater than 91.4% a NOx outpoint of
less than 1.2 is necessary, and to achieve an HC IDR rate greater than 92% a
NOx outpoint of 1.0 is necessary.  Once a tight NOx outpoint is used to fail
these cars with excess HC, the HC outpoint no longer determines the result, at
least in this sample.  So, ASM outpoints of 1.00/8.0/1.0 are the least
stringent ASM outpoints that can achieve a 92% HC ZDR.

    Another consideration is that the ASM outpoints have been optimized for
this database.  Zn contrast, the ZM240 recommended outpoints were optimized
for the Indiana ZM240 database.  Because of sample to sample differences, the
optimum outpoints are expected to vary slightly from one database to another.
So the optimum ASM outpoints are being compared to standard ZM240 outpoints,
which while optimum for the Indiana data, are not the optimum outpoints for
this database.   Applying ASM outpoints optimized for this data base, to  a
different database, is expected to further lower the ASM's performance.
                                       35

-------
                               Figure 5 J.I
                        Comparison of ASM to IM240
                      Using IM240 Standard Outpoints
             With Maximum ASM IDRs at Equivalent Failure Rates
100%  -i-
 90%
 80%
70%
60%
50%
40%
30%
20%
10%
 0%
                                83%
                                   N
-------
                   Figure 5.3.3
            Comparison of ASM to IM240
    With ASM CO & NOx IDRs Matching IM240 IDRs
                         89%
           COIDR     NOxIDR
IIM24O«RecommeodcdQflpointi  O ASM it Mttchfcf CO * NOx IDRs
                     38

-------
                Figure 5.3.4
         Comparison of ASM & IM240
          Maximum IDRs <§> EC <5%
                 97%
       COIDR     NOxIDR    FtfloreRuo   UiyuJuuiw
• lM240Bc«IDIU9<3%Ec  D ASM B«« HMU « <5% Be
                  39

-------
       5.4   Comparison  using  Scatter  Plots  and  Regression  Tables

    The objective of this analysis was to check the correlation of both the
 IM240  and  the ASM test with the FTP.  The correlations  are  illustrated in
 Figures F-l through F-9,  in Appendix F.  Appendix F  includes  regression tables
 along  with these scatterplots.  The regressions  show similar  R2 over the
 entire data range, but the ZM240 correlates much better to  the FTP for
 vehicles emitting closer  to the FTP standards.

            5.4.1   Using the  Coefficient of Determination  (R2  )  and
                     Standard Brror  of the  Estimate for Objective
                     Comparisons

    R2   represents  the percentage of variability  in the dependent variable
 (FTP result) that is explained by the independent variable  (I/M  test  result)
 and is often used to compare one I/M teat's effectiveness with another's,  but
 R2  can often be misleading.  Since R2  is often  used in correlation studies,
 it does provide an indication of comparative test accuracy  that  would be of
 interest to readers accustomed to seeing such comparisons.  More important,
 however, is how well these tests discriminate between malfunctioning  and
 properly functioning vehicles at an I/M station,  which is best measured using
 the techniques discussed  in Section 5.2.

    For a vehicle to fail an XM240, it must fail  the two-ways-to-pass-criteria
 developed by EPA (see Section 5.2.2).  The R2 values presented in this  section
 are for composite IM240 scores only and do not account for  this.  Two-ways-to-
 pass affects the quantitative correlation between IM240 and FTP  significantly
 because the Mode-2 HC and CO values are often more representative of  vehicles'
 actual FTP emissions.  However, EPA believes it is not appropriate to mix and
match composite and Mode-2 scores into one quantitative correlation analysis.

    Additionally,  the R2 comparisons presented here  do  not  account for  the
 sample's bias toward high emitters  (discussed in Section 5.2.5.8).  The 106
 vehicles that were recruited to the lab for FTP testing were purposely biased
to include a high number,  of dirty vehicles,  when regressing the I/M test
 scores verso* the) CTP to  determine R2 values, these high emission values
disproportionately influence- some regression statistics, given the typical
distribution of in-use emissions data.  Thus the emission values close to the
 FTP standards) (where comparing I/M tests  is most important), have less
 influence on the R2  statistic than desirable for determining the actual
merits of these tests.  Cutpoint tables account  for this sample bias by
 weighting each vehicles'  emissions according to  the population of vehicles
 tested at the I/M lane.
                                      40

-------
     To account for these limitations the sample was divided into the  following
 three groups:

       •   All Vehicle*.  This database is not  very useful  for comparing
          correlation because the cleanest and dirtiest vehicles dominate the
          R2  statistic.  Both tests correctly differentiate these.  More
          pertinent are the vehicles with emissions closer to the FTP  standard,
          where the capability of I/M tests is not masked by the very  clean and
          very dirty vehicles.  Also, vehicle 3211 is a CO outlier for both
          tests.   It has a major effect on the regression equation and the R2 ,
          thus masking  the typical capability for both procedures.
       •  Vehicle 3211 Removed  for HC,  CO, NOx.   This database better
         characterises the correlation of both short tests with the FTP,  but
         for the reasons discussed, it is not the most relevant for comparing
         the effectiveness of the tests.

       •  Marginal  Emitter*:  Only  vehicles that are not very  clean  or  not
         very dirty on FTP using following criteria:

         HC  20.30 g/mi and <1.5  g/mi on the FTP
         CO  22.5  g/mi and <25.0 g/mi on the FTP
         NOx 20.5  g/mi and <2.25 g/mi on the FTP
         Also, Vehicle 3211 was excluded as an outlier.

         All vehicle* with FTP emissions less than 0.30 HC, 2.5 CO, and 0.5
         NOx passed the ASM and IM240 tests, for all the cutpoint sets
         evaluated in Section 5.3.

    The standard error is an objective measurement of test variability
expressed in the unit* (g/mi. in thi* case) of the variable* used in the
regression.  Because R2  are expressed as percents, standard errors have an
advantage of being leaa abstract.

    Table 5.4.1 provide* a summary of R2  and standard error* for Figures F-l
through r-9 in- jppenrtim F, divided  into the 3 group* just discussed.  The
"Marginal Emitters* group indicate* that the R2  for the IM240 are
considerably higher for HC and NOx, and somewhat higher for CO.  Likewise,  all
the standard error* are lower for the ZM240, most notably for HC.
                                      41

-------
            Tabla  5.4.1   Summary  of R2    and Standard Errors
Data Sat :
n :
Proeadura:
R* .
BC
Std. Error •
*' -
CO
Std. Error •
R* -
NOx
Std. Error •
All Vehicles
106
IM240
82%
0.62
54%
13.4
70%
0.65
ASM
73%
0.76
68%
11.2
71%
0.64
Vahicla 3211
Removed
105
IM240
83%
0.61
75%
10.0
70%
0.66
ASM
74%
0.75
80%
8.9
71%
0.64
Vehicles Naar
Standards
43
IM240
63%
0.19
25%
4.3
46%
0.34
ASM
17%
0.28
13%
4.6
26%
0.39
    Tha standard arrors liatad in Tabla  5.4.1 can ba usad to astinata tha
lowast FTP value that would confidantly pradlct a dirty vahicla.  For example,
tha HC standard arror La 0.28 g/mi for tha "Marginal Emitters" group.  Since
95% of th« tin*, a vehicle's result will be within  ±2  standard errors, this
suggests that tha lowest ASM HC score that confidently predicts an HC-dlrty
vehicle (i.e., FTP HC > 0.41) ia the ASM HC score that yields  (using the
regression equation) an FTP HC of 0.97 g/mi [0.41 + (2 * 0.28)1.  In contrast,
using the IM240 error of 0.19 g/mi means tha loweat IM240 score that
confidently predicts a HC-dirty vehicle is 0.79 g/mi,  over 18% lass  than  the
score neaded to confidently predict an ASM HC dirty vehicle.

      5.4.2  Advantage  of  Using Weighted  Mode*
    The ASM t«a% ia given a big advantage in the way the regressions are
performed baeaoa* each mod* is weighted separately according to the XM240.
the other hand, tn« XM240 score is a non-weighted scon.  EPA developed the
ZM240 to contain similar driving conditions as the FTP.  However, the
frequency of each condition ia not proportional to the FTP.  By weighting
different modes of the ZM240 to the FTP similar to the way EPA has weighted
the 4 modes of the ASM test, EPA has found the R2  for the ZM240 to improve
immensely.  The current score reported for the IM240 is something like
On
                                      42

-------
 weighting each mode of the ASM test 25%.  This would hurt the correlation of
 the ASM with the FTP,  because, as is shown in Section 5.5, the 50 mph mode
 accounts for roughly half of the composite ASM scores for each pollutant.

     5.4.3  Observations of   Scatterplots

     The  scatterplots for the first two sets of data (Figures  F-l  through  F-6)
 do not  appear much  different for either test, mainly because the high emittera
 cause the emissions close to the standards to appear as a tight pack of data.
 The plots for vehicles near the standard only  (Figures F-7 through F-9),
 however,  suggest the following:

    HC - The  XM240  identifies the dirtier cars much better.  Notice on the ASM
          HC plot how many high emitters (FTP HO0.82 g/mi) still score
          relatively low on the composite ASM score.  Six vehicles pass the
          very tight ASM HC cutpoint of 0.3 predicted g/mi, yet have FTP
          emissions  greater than twice the FTP standard (0.82 g/mi).

    CO - Neither test  appears to correlate very well over this emission  range
          for CO.  Two  issues come into play that explain why this is.  First,
          cars with  loaded canisters will have high IM240 Mode-1 CO emissions
          at the  lane,  causing the short test to have a high score while the
         FTP  at  the lab is relatively low.  The second scenario is cold start
         problems.   Two vehicle* in the database (Vehicles 3175 and 3227)
         appear  to  have cold start problems, with high Mode-1 FTP CO
         emissions,  and low Bag-3 FTP CO emissions.  Since the lane test is a
         hot  start  test, these vehicles will show up clean at the lane, and
         the  cold start FTPS will be significantly dirtier.

  NOz -  The IM240 has  a slightly  tighter  fit to  the regression line, and more
         of the FTP dirty care fall to the upper right of the scatterplot
          (i.e., fail the teat properly).
    5.4.4  Poosr- ASM  WO  Correlation
    AS discussed in Section 5.3, ASM HC scores do not correlate very well with
FTP HC scores).  Thia section briefly discusses theoretically why some of these
vehicles had very low ASM HC emissions, yet failed the XM240 and FTP for HC
emissions.  Because the contractor'a mechanics were not aware of the ASM
scores, vehicles were not diagnosed with the objective of determining the
cause of the performance differences on these I/M testa.

    The first four vehicles in Table 5.4.2 were found to have ignition
problems.  Thia ia logical considering that misfire, which causes high  HC, is

                                      43

-------
 sometimes related to load.  A3 load increases the voltage required to  jump the
 spark plug gap alao increases.  Some portions of the IM240 load the vehicle
 more heavily than any of the ASM modes, so a marginal ignition  system  that
 only causes misfire during the higher IM240 loads will not be identified by
 ASM HC.

    vehicles 3180 and 3264 were found to have bad 02  sensors and other
 malfunctions.  Slow responding O2 sensors are more likely to be identified
 during a transient test, because changing throttle position tends  to cause
 air/fuel ratio excursions that will cause high emissions unless the fuel
 induction system rapidly compensates to maintain the optimum air/fuel  ratio.
 So slow responding 02 sensor might explain the high HC on the IM240 and low HC
 on the ASM.

    The disconnected vacuum lines on vehicle 3201 could have caused lean-
misfire during accelerations on the IM240 that would not be apparent  on the
steady-state modes of the ASM.
    These explanations can not be proven with the  existing data,  but they
should indicate that a steady-state test suffers from known disadvantages
identifying vehicles with these types of malfunctions.

          Table  3.4.2    Vehicles  with  Poor  ASM  HC Correlation
                                                                  in
VIH   ASM HC
              HC   XM240  HC   Problem  round
3259
3257

3155
3210
0.23
0.25

0.34
0.35
1.94
1.26
 .25
 .40
3180
3264
3254
3201
3165
0.15
0.49
0.2§
0.17^
0.3f '-
0.9*
1.3C
1.87
0.94
1.96
1.50     Ignition Module
1.92     Plug Mires,  Plugs  Transducer,
            Ignition Coil Transistor
2.77     Incorrect  Plugs* Torn wire boot
1.04     O2 Sensor,  Spark Plugs,  fuel Hose,
            Catalyst
1.33     02 Sensor  and Injectors
2.1C     02 Sensor,  Vacuum Switching Valve
2.2S     ECU Intermittent,  Catalyst
1.15     Vacuum Lines Disconnected
1.59     Dirty Injectors
                                       44

-------
       5.5   Derivation  of  ASM  Coefficients  and  Nod*  Contribution
             Variation*  From  Sample  to  Sample

    This section discuss** why th« ASM coefficients  that EPA basad its
 analyses on war* developed using the IM240 as the dependent variable  rather
 than  the FTP.   BAR/Sierra and ARCO used the FTP to develop ASM Coefficients.

    Also discussed are the rather large variations in the  ASM  coefficients
 with  different  samples, and the variation in the contribution  of each ASM mode
 to  the final ASM score  (expressed as percent contribution).

             5.5.1   ASM  Versus  ZM240  A*  The  Dependent  Variable For
                     Determining  ASM  Coefficient*

    EPA faced a dilemma in determining the best method  for developing the ASM
 equation coefficient*.  No ASM advocate has recommended specific coefficients,
 on  which, EPA should accept or reject the ASM approach.  So two options were
 to: 1)  perform the multiple  regressions on all the lab recruited vehicles  for
 ASM versus FTP.  Or, 2)  perform the multiple regressions for  ASM versus IM240
 on  a  subset  of  the lane sampler excluding all lab recruited cars.

    Obviously,  the ideal method is to regress the ASMS  versus  FTPS  (i.e.,
 option 1),  but  this raises * problem.  To evaluate how w*ll the ASM identifies
 FTP failing  vehicle*, the coefficient* must be applied to the lab recruited
 vehicle* to  calculate simulated grams/mile scores.  However,  good statistical
 practice mandate* applying the coefficients to a different sample than those
 from which they were developed.

    This interlinking method,  wherein the coefficients  are applied to the sane
 vehicle* from which they were developed, would minimise the effect* of the
 test's variability.  This improper interlinking is illustrated using results
 from Vehicle 3211.  Thi* vehicle'* lane-XM240 CO score wa* 93 g/mi and it* ASM
CO score wa* 65 g/mi u*ing the coefficient* from the 60S vehicle sample listed
 in Table 5.5.1  (The- relevance of the other sample* in thi* table will be
discussed latex!.  It* FW CO score was only 10.8 g/mi.
                                      45

-------
                                 Table  5.5.1
           CO Coefficients  Developed from Different
lea
Mode
Constant
5015
2525
50MPH
Idle
Adjusted R*
CO Sample 1
2.814
0.035
0.072
0.425
0.891
34.2%
CO Sample 2
2.836
0.116
-0.058
0.391
2.014
58.4%
CO All 608
2.936
0.040
0.043
0.356
1.350
50.1%
COvsFTP
5.533
-0.047
0.565
0.050
1.968
80.6%
    The scatter plots below show that  using the IM240-developed coefficients
 cause this vehicle to be easily identified as an  outlier.   In marked contrast,
 using the FTP-developed coefficients make it  look like this vehicle's ASM
 score highly correlates to its FTP score.  The ASM mode scores are weighted
 differently, so the high scoring mode(s) are  de-emphasized. But  these same-
 sample FTP-based ASM coefficients are obviously peculiar to this  sample, and
 highly dependent on it containing this one particular car.   (See  Tables  S.S.I
 and 5.5.6.)

 Still not answered is which ASM coefficients  better  indicate whether the
 vehicle is malfunctioning or not.  Some could argue  that this vehicle should
 not be an outlier.  Instead, the XM240-developed coefficients inappropriately
 make it appear as an outlier.  Attempting to  resolve this,  the  raw ASM
 concentration measurements were checked.  This vehicle's SO mph mode CO
 concentration was 4.96%, which is higher than 97% of vehicles recruited  to the
 lab (103 of the 106 vehicles).

    Using the same-sample,  FTP-based ASM coefficients  prevents this vehicle
 from being an outlier because they adjust themselves to minimize the  effect of
 the SO mph mode score from all cars. Additionally, the very high concentration
measurement (4.96%) proves that the vehicle had a malfunction causing very
 high emissions that had been inappropriately minimized.  (This  was also
 verified during the mechanic's inspection which found a defective O2  sensor
 and that an KM PROM update was required.)  This evidence strongly supports
 EPA's properly using the preconditioned ZM240s as the dependent variable for
 developing AST coefficients to compare the ASM and XM240 correlations with the
 FTP.
    This evidence also casts doubt on conclusions developed from test programs
that used interlinked coefficients,  interlinking makes the correlation
between the ASM, or any other test, and the FTP significantly better than
could be expected in an official I/M program.  Since I/M programs will apply
ASM coefficients to vehicles that were never FTP tested, the opportunity for
                                       46

-------
interlinking will not exist, so ASM performance should not be evaluated using
interlinked coefficients.

    Another reason for EPA's not using FTP-based coefficients is because some
are negative, which means that as ASM501S emissions increase, FTP emissions
decrease.  This is counter-intuitive.
                                      47

-------
                        ASM Coefficients Developed
                                  from
                 608 Lane ASM vs Pre-Conditioned IM240s
                        ASM Coefficients Developed
                                  from
                        106 Lane ASM vs Lab FTP
8
                                 48

-------
    EPA decided that the best method, using the Arizona data, was to regress
 the four  steady-state  nodes  (three  for NOx) against lane-only, preconditioned
 !M240s.   There were three major  factors leading to this decision.  First, this
 allowed applying  the coefficients to a different subset of data  (the lab
 recruited, vehicles).   Second, the sample size was considerably larger  (608 vs.
 106 tests).   EPA's  FTP sample was too snail to divide and use half to
 determine coefficients and the other half to evaluate ASM effectiveness, which
 is  supported by the negative coefficient yielded for the ASMS015 listed in
 Table 5.5.1.   Third, only preconditioned. IM240s were used because they
 correlate better  with  the FTP than  non-preconditioned. IM240s.  The only
 significant  compromises in using XM240s instead of FTPS is that the composite
 ASM score does not  include a cold start excess  (which would  be independent of
 warmed-up ASM node  concentrations anyway) and that the mix of speeds and loads
 in  ZM240  is  not exactly like that in the full FTP driving cycle  (a hardship
 borne by  the IM240  in  its own correlation to the FTP).

    Figure 5.5.1 illustrates that preconditioned !M240s strongly correlate
 with the  FTP.  These data are from  the 106 lab recruited vehicles, but  are
 restricted to  IM240s that were performed following the ASM at the lane, making
them preconditioned IM240s.

    NOx has the worst correlation because of a few outliers  at the high end,
but this  ia not a concern for the ASM since the NOx coefficients are
relatively stable,  which is discussed in the next section.
                                       49

-------
M. JLJL1	Hicrli Correlation of Preconditioned Lan«-TM240a with FTPs
                       IM240vsFTP
                   Preconditioned IM240s

14
1.5 10 15 3.0
IM240HC
3.5 4.0
4.5
                       XM240vsFTP
                   Preconditioned IM240s
      10
20
30 40 50 60
IM240CO
IM240vsFTP
Preconditioned IM240i
70 80
        UO
24
3.0
IM240NOz
4jO
54
64
                             50

-------
            3.3.2   Variability  of  ASM Coefficients

    The objective of  the  following analysis was to investigate the stability
of the coefficients used to calculate composite ASM simulated grants/mile
scores.  The database was divided into four samples for comparison.

    Sample 1 was  developed by using a random number generator to  select  304
vehicles from the lane-only fleet of 608.  The remaining 304 vehicles became
Sample 2.  Sample 3 waa all 608  vehicles,  and the fourth sample was  the 106
laboratory vehicles.   The ASM coefficients for the first three samples were
developed using the IM240 as the dependent variable and the FTP sample used
the FTP for the dependent variable. The resulting coefficients are listed in
the following tables.  (Table 3.5.3 is a duplicate of Table 5.5.1.).
                                Table  3.5.2
           HC  Coefficients Developed  froa  Different  Saaples
Mode
Constant
5015
2525
50MPH
Idle
Adjusted R*
HC Sample 1
0.080
0.045
0.047
0.147
0.084
21.7%
HC Sample 2
0.073
0.008
0.059
0.123
0.585
38.9%
HC All 608
0.083
0.025
0.059
0.136
0.124
29.0%
HCvsFTP
0.291
-0.261
0.507
0.238
0.154
79.4%
          CO
                 Table  3.3.3
Coefficients  Developed froa
Different  Samples
Mode
Constant
5015
2525
50MPH
Idle
Adjusted R»
CO Sample 1
1814
0.035
0.072
*• 0.425
0.891
34.2%
CO Sample 2
1836
0.116
•0.058
0391
2,014
58.4%
CO All 608
2.936
0.040
0.043
0.356
1.350
50.1%
COvsFTP
5.533
•0.047
0.565
0.050
1.968
80.6%
                                      51

-------
                                Table  5.5.4
          NO*  Coefficients  Developed  £ro» Different  Sample*
Mode
Constant
5015
2525
50MPH
Adjusted Rz
NOX Sample 1
0.230
0.088
0.206
0.386
602%
NOX Sample 2
0.279
0.045
0.212
0.333
57.9%
NOX All 608
0.258
0.061
0.219
0.352
59.1%
NOX vs FTP
0.190
0.148
0.093
0.291
71.1%
    The negative coefficients are highlighted in bold.  One could infer  from
the negative coefficients that  increasing the emissions during that mode of
the ASM would lower the composite  score.

    These coefficients  were used with the ASM data, froa each of the 106 lab-
recruited vehicles, to calculate the emissions for each mode and the percent
of the total emissions that  each mode contributed.  These mode contributions
give a better indication of  each modes  importance in  the final ASM score, than
the coefficients, which are  more difficult to interpret.  The results are
listed in the following tables:
                                fable  5.5.5
          Average)  Contribution of  Total  HC
•sioos  by  Mod*
Mode
Constant
5015
2525
50MPH
Idle
HC Sample 1
17%
20%
17%
43%
2%
HC Sample 2
17%
4%
22%
39%
18%
HC All 608
19%
12%
22%
43%
4%
HCviFTP
11%
-75%
116%
45%
3%
                                      52

-------
                                Table  5.5.6
          Average  Contribution of Total    CO emissions by  Mod*
Mode
Constant
5015
2525
50MPH
Idle
CO Sample 1
26%
4%
7%
57%
6%
CO Sample 2
27%
12%
-6%
53%
15%
CO All 608
29%
4%
4%
51%
11%
COvsFTP
28%
-5%
55%
7%
15%
                                Table 5.5.7
         Average Contribution  of  Total HOx emissions by Mod*
Mode
Constant
5015
2525
50MPH
NOX Sample 1
15%
11%
24%
50%
NOX Sample 2
20%
6%
27%
47%
NOX All 608
17%
8%
27%
48%
NOX vs FTP
20%
22%
13%
45%
    The HC coefficients in particular are very volatile, and that the negative
FTP-developed coefficients are counter-intuitive.  When applying the
coefficients from Sample 1,  the idle mode,  on average, only contributes 2% to
the total score.  This contribution jumps to 18% when the coefficients from
Sample 2 are applied.  Similarly the ASM5015 contribution drops from 20% to
4%.  These examples indicate that the  largest sample  (€08 vehicles) with
preconditioned ZM240s was the best sample available for developing ASM
coefficients.

            5.5.3   Significance  of  Mode  Contributions

    The ASM mod* contributions also vary as the composite ASM score moves from
low values  (fos) which the constant term will be the primary contributor to the
composite sce*e> to relatively high vslues  (for which the constant term will
be a relatively small contributor to the composite emission).  This is
illustrated in Figure* 5.5.2 to 5.5.4.   For CO, th* ASM501S and ASM2S2S ar*
combined, because of th* negative contributions of 2525 and the small
contribution of the ASM5015 in relation to  the 50  mph mode.
                                      53

-------
                  rigur*  5.5.2
           Hod*  Contribution*  for  BC
U
                staf Sample I Coeffiden*
    I Constant Hldk     QSOMPH  12525
                              5015
FTPHC4
                            FTP HC >
               Ifofaf SMplt 2 Cotffidtnto
               ldb     DSOMPH  B2523
                               5015
                                           FTP HC > \M
                         54

-------
                    rigor*  5.S.3
            Mod*  Contribution*  for  CO
                Using Sample 1 Coefficients
       I Constant
         DSOMPH

I 5015+2525
FTP CO S fcM
tM < CO  S  13.M
134 < CO i 15*
                Usfaf Sanple 2 Coefficients
                   lldte
         DSOMPH
I 5015+2325
FTP CO *
                          55

-------
                             rigux*  5.5.4
                     Mod*  Contribution*  for  HOx
100%


 80%


 60%


 40%


 20%


  0%
                         Using Sample 1 Coefficfentt
! Constant DSOMPH  B2525
                                                  5015
           -I-
LJ
        .lS  1.15
-------
    The fact that the 50 nph mod* contributes so much to the composit* score
 for each  pollutant  is also reason for concern.  This opens the opportunity for
 mechanics to adjust vehicles to lower emissions for just one mode (namely the
 50 mph),  which will be  further discussed in Sections 5.6.3 and 5.6.4.
             5.5.4   Conclusions  on  ASM  Mode  Contributions

    While not a mode that was recommended by the ASM developers, the 50 mph
 mode  at  road load horsepower appears to be more important for identifying
 dirty vehicles than the lower speed, acceleration simulation modes (ASM5015
 and ASM2525).  Surprising was the small contribution of the ASM5015 (mode 1)
 for identifying dirty vehicles considering that BAR/Sierra and ARCO both found
 this  mode to be the moat effective.  This suggests that the first mode in a
 four  mode sequence serves mainly to precondition vehicles for the following
 modes.   Randomizing the order of the modes may be useful in determining the
 best  sequence.

    For the outpoint analysis in Section 5.3 and the regression analysis in
 Section  5.4, the ASM scores used were those calculated from the coefficients
 developed from the 608 ASM versus preconditioned lane-ZM240s.  However, the
 variability  of the HC coefficients between the two random subsets of 304 tests
 suggest that a different sample of 608 tests might produce substantially
 different equation coefficients.  The resulting change in HC (and in some
 cases CO and NQx) ASM scores would produce different failure rates, ZDRs, and
 Be  rates in  the outpoint tables, and different R2   values in the regressions
 of ASM versus FTP.  So the volatile coefficients may vary from sample to
 sample, or worse yet region to region, resulting in disparate X/M programs
which would be hard to evaluate on a consistent basis.
      9.6   Repair  Analyses

            S.C.I   Contractor  Repairs

    The objective) of this analysis was to investigate the performance of both
the XM240 antf the) ASM testa as predictors of changes  (i.e., decreases or
increases) in FTP emissions following contractor-performed, IM240-target*d
repairs.

    Of the IOC  vehicles used in the outpoint analysis (Table 4.2.2), 56
exceeded the lane-IM240 0.80/15.0/2.0 + 0.50/12.0 outpoint and were repaired
by the contractor.  Of these, 52 received each of the three following tests
both prior to repairs  (i.e., as-received) and following repairs:

                                      57

-------
       -  a  Lane-IM240,
       -  a  Lane-ASM, and
       -  an FTP.

 These  52  were used in this analysis and are included with the data  listed in
 Appendix  B.  The resulting database of those 52 fuel-injected vehicles  has the
 following distribution:

Fuel
Metering
PFI
TBI
Order of
ASM Prior
to XM240
14
11
[•ana Testing
ZH240 Prior
to ASM
16
11
    The contractor was instructed to perform the minimum repairs  necessary  in
order that each vehicle's IM240 emissions after repair (as tested at the
contractor's laboratory) meet the following criteria:

      -  composite ZM240 HC  £  0.80 g/mi,
      -  composite XM240 CO  £ 15.00 g/mi, and
      -  composite ZM240 NOx S  2.0 g/mi.

The contractor was allowed multiple repair attempts if the first set of
repairs did not reduce the XK240 emission levels enough.  The repairs were
limited to 91,000 per car.  And, the contractor was instructed that "the
mechanic should only be aware of the XM240 scores for the XH240-targeted
repairs.*  Because ASM outpoints, that could distinguish malfunctioning
vehicles from properly functioning vehicles, were not yet developed, only
IM240-targeted repairs were performed.

    These IM240 emission repair criteria were met at the contractor's
laboratory foe all cars prior to the second and final FTP, with  the highest
after-repair laboratory ZM240 composite  HC emission score of 0.56, CO of
10.82, and WOm of 1.93  (g/mi).  The effects of  those IM240-targeted repairs on
FTP emissions are illustrated in the following  table:
                                       58

-------
                                        «t 1 . 1
                 FTP  Emissions  Prior  to  and  Following
                         IM240-Targeted   Repairs

FTP Emissions
HC As -Received
After Repair
CO As-Received
After Repair
NOx As-Received
After Repair
Mean
1.458
0.326
19.707
3.331
1.649
0.739
Rang*
Xmisi
Minimum
0.16
0.10
0.28
0.63
0.20
0.05
• of
lions
Maximum
13.07
0.75
113.40
8.82
7.56
1.81
The resulting FTP emissions after the IM240-targeted repairs were essentially
independent of the as-received FTP emissions.  (That is, the R-squares
associated with before and after HC, CO, and NOx were only 0.1%,  1.2%,  and
1.0%, respectively.)

    The data from these repaired vehicles can give insight into the question
of whether the IM240 test and outpoints cause repairs to be made which also
reduce FTP emissions.  In other words, does the IM240 and the FTP correlate
well on a single vehicle?  This correlation is to be expected based on the
realistic nature of the ZM240 driving cycle, and the good correlation found in
samples of vehicles not repaired.

    For each of those 52 vehicles (all 1983 and newer fuel-injected cars),  the
change in each pollutant (HC, CO, and NOx)• following contractor repairs, was
calculated for each of those) three test cycles.  Regressing the reductions in
the lane emissions against the reductions in FTP emissions produced Tables
5.6.1.2 and &.6.I.3.  The «ix graphs  (Figures 5.6.1.1 through 5.6.1.3) that
follow those) regression tables illustrate the results of this analysis.
                                      59

-------
                                  S fi 1 g
     Ragreaaion o*  Changaa  in  Lana-XM240  Kaiaaiona  relieving
Contractor  Rapaira  Varan*  Corraaponding Changaa  in  FTP  Bmiaaiona
Dependent variable
R2 .81.9%
is: A(FTP HC)


s- 0.8320 with 52-2.50 degrees of freedom
Source
Regression
Residual
Variable
Constant
AUM240 HO
Sum off Squares df
156.693 1
34.611 50
Coefficient «.e. of Coeff
-0.365173 0.1524
1.4106 0.0938
Mean Square
157
0.69222
t-ratio
-2.4
15
F-ratio
226

Dependent variable
R2 .47.5%
is: A(FTP CO)


s. 18.50 with 52-2-50 degrees of freedom
Source
Regression
Residual
Variable
Constant
AHM240 CO)
Sum of Squares df
15469.4 1
17110.1 50
Coefficient e.e. of Coeff
4.64057 3.103
0.846373 0.1259
Mean Square
15469
342.203
t-ratlo
1.5
6.72
F-ratlo
45.2

 Dependent variable is:              A(FTP  NOx)
 R2 .64.5%
 s .  0.8846 wtth  52-2-50  degrees of freedom
 Source  -
 Regression
 Residual

 Variable
 Constant
 AHM240  NOx)
Sum of Squares     df
    71.1008          1
    39.1265         50

  Coefficient   s.e. of Coeff
   -0.275563       0.1747
   0.738523        0.0775
Mean  Square   P-ratio
    71.1         90.9
  0.78253

   t-ratlo
    -1.58
    9.53	
                                  60

-------
      R«gx*a«ioa  ot Caang««  ia  Lan«-ASM Kaissioaa  rollowing
Contrmctor  Repairs V«r«u«  Corresponding  Chang**  ia FW  Xalsaion*
Dependant variable
R2 -71.7%
is: A(FTP HC)


s- 1.040 with 52-2-50 degrees of freedom
Source
Regression
Residual
Variable
Constant
AfASM HO
Sum of Squares df
137.187 1
54.1169 50
Coefficient e.e. of Coeff
0.270944 0.1633
2.18967 0.1945
Mean Square
137
1.08234
t-ratlo
1.66
11.3
F-ratlo
127

Dependent variable
R2 -79.5%
is: A(FTP CO)


s- 11.55 with 52-2-50 degrees of freedom
Source
Regression
Residual
Variable
Constant
AfASM CO)
Sum of Square* df
25906.3 1
6673.29 50
Coefficient t.e. of Coeff
5.71775 1.775
1.0868S 0.078
Mean Square
25906
133.466
t-ratlo
3.22
13.9
F-ratlo
194

Dependent variable
R2 .70.8%
is: A(FTP NOx)


s - 0.8016). wtth 52-2-50 degrees of freedom
Source
Regression
Residual
Variable
Constant
A(ASM NO*)
Sum of Squares df
78.0956 1
32.1317 50
Coefficient e.e. of Coeff
-0.013624 0.1392
0.829714 0.0753
Mean Square
78.1
0.642635
t-ratlo
-0.098
11
P-ratlo
122

                                  61

-------
                   rigor*   5.6.1.1

      Decreases in HC Emissions Following Repairs
                     AFTPVSAIM240HC
                    Refresnoo Lin*
                   R-Squar«d«81J%
-1
135

  AIM240 HC Emisrioos (a/ml)
                     AFTPTflAASMHC
                 AASMHCEBtatottfefel)
                           62

-------
                                 3.6.1.2
           Decreases in CO Emissions Following Repairs
                          AFTPvf AIM240CO
   120

   100
3
  RcfradoB Line
R-Squarcd> 47.5%
           .  N
                         20       40        60

                      AIM240 CO Emtafou (g/mO
                                  80
100
                           AFTPnAASMCO
                         20       40        6t

                       AASMCOBmhriow(i/»i)
                                           100
                                63

-------
             Figure  5.6.1.3
Oecreaaea  in NOx Emissions  Following Repairs
              AFTPvsAIM240NOx
        I            1
       Rtfnttioo Line
      R-Squ*rtd = 64.5%
       135
          AIM240 NOx Embriou (gfaQ
               AFTFvsAASMNOs
       1            3           S
           A ASM NQx Eatarion (|M)
                     64

-------
    Comparing the two graphs that examine the changes in HC emissions  (Figure
 5.6.1.1),  it is apparent that one vehicle  (vehicle number 3190)  exhibited a
 reduction  in FTP HC emissions substantially greater than any of the other 51
 vehicles  (12.88 g/mi HC reduction compared to only 3.86 for the next larger
 FTP HC  reduction).  Since it is possible that one such vehicle could
 substantially affect the regression analysis, a second set of regressions were
 performed  on the remaining 51 cars (i.e.,  with vehicle number 3190 deleted) to
 determine  the effect.  The effects on the  slopes of the regression lines are
 given in Tables 5.6.1.4 and 5.6.1.5.

                               Table  5.6.1.4
                     Kffeot  oa  ZM240  Regression  Line
                              For HC Emissions
                         Of  Deleting  Vehicle  3190

Constant
Coefficient
R2
Bassd on Al
52 Vehicle*
-0.365173
1.4106
81.9%
Based on 51
Vehicle*
0.018227
0.93431
74.7%
                               Table)  5.6.1.5
                      Bffect  oa  ASM  Regression  Lin*
                             For HC  Emissions
                         Of  Deleting  Vehicle  3190
                                 52
                    on  Al
                 Vehicles
Baaed  on 51
  Vehlclea
                                   0.270944
I  Constant

Ifft         I   71.7%
   0.452113
                                   2.18967
   1.35391
                                                67.8%
From TaJble» 9.(.1.4 and 5.6.1.5, we see that deleting that potential HC
outlier (vehicle) number 3190) baa a similar effect on each regression  line.
The slop* of the) ZM240 regression line decreases 11.6 degrees,  and the slope
of the ASM regression lino decreases 11.9 degrees.  Since delating the change
in HC emissions of vehicle 3190 from the sample has the same effect on both
regression lines, it would be) advisable to use the equations based on  51 cara
to estimate changee in FW HC emissions based on IM240 and/or ASM HC changes,
for IM240 and/or ASM HC changes between -1.0 and +4.0 g/mi.
                                      65

-------
    Comparing the two graphs that examine the changes in CO emissions  (Figure
 5.6.1.2),  it appears, at  first glance, that the composite IM240 tends to over
 predict the repair benefit to CO emissions for some vehicles with relatively
 small FTP  CO repair benefits.  However, the actual situation is that several
 relatively cleaner vehicles  (though still exceeding FTP standards) had
 unusually  high  XM240 results on their first test.  IM240 CO was a lot lower
 after repair, but the FTP emissions had comparatively little room to improve.
 The five vehicles in Figure 5.6.1.2 that exhibit this problem  (vehicles
 numbered:   3157, 3175,  3211, 3213, and 3214) all have as-received composite
 FTP CO less than IS g/mi.  For two of those five, most of the high composite
 IM240 emissions resulted  from the first mode  (i.e., the first 93 seconds) of
 the IM240.  For this reason, EPA has recommended that vehicles which fail the
 composite  HC or CO outpoint be given a second chance to pass by examining the
 Mode-2 emissions  (see *Two-ways-to-paas" in Section 5.3).  A similar situation
 cannot happen for the ASMs as analyzed in this report because the weighting
 factors, in effect, cause the CO scores on the first mode  (5015) to be
 ignored.   One vehicle that deserves special note is vehicle number 3211.  That
 vehicle exhibited the largest ZM240 CO reduction (91.70 g/mi), but an FTP CO
 reduction  of only 8.00  g/mi.  This high lane-IM240 CO reduction resulted from
 a high initial  (i.e., as-received) lane test score of 93.07 g/mi, but an
 initial FTP CO score of 10.79.  (However, the lane score was confirmed by an
 indolene-fueled lab-IM240 following the FTP which had a CO result of 52.48
 g/mi.)   The ASM tests on  this vehicle did not exhibit a large CO  reduction
 following  repairs because bath the initial ASM and the ASM following repairs
 exhibited very high CO  emissions  (more than 5%) during the 50 mph cruise mode.
 (Thus,  the  ASM did not  over estimate the CO repair benefit on vehicle 3211
because  the ASM over estimated both the initial FTP CO emissions, as well as,
the FTP  CO  emissions following repair.)  In apite of the few over predictions
of emission benefits from repairs, it should be noted  (as  illustrated in Table
 5.6.1.1) that following the IM240-targeted repairs, no vehicle  was  left  with
high unrepaired FTP emissions.

   Most of the} vehicle*, which exhibited very little if any HC or CO
 improvement following the) IM240-targeted repairs, had been recruited for
repairs becmae> they exhibited, on the lane-lM240 test,  low HC and CO,  but
high NOx-  Therefore, no  significant improvement in either FTP HC or CO was to
be expected.

   Comparing the two graphs that examine the changes in HOx emissions  (Figure
 5.6.1.3),  it is apparent  that on* vehicle  (vehicle  number  3202) exhibited a
 reduction  in FTP NOx emission* greater than any of the other  51 vehicles (6.31
g/mi NOx reduction compared to 4.98 for the next larger  FTP NOx reduction).
 Since  it is possible that one such vehicle  could substantially affect the
 regression  analysis, a  second set of  regressions were performed on the
 remaining  51 cars (i.e.,  with vehicle number  3202  deleted)  to determine the


                                       66

-------
 effect.   The effects on the slopes of the regression lines are given in Tables
 5.6.1.6  and 5.6.1.7.  From Tables 5.6.1.6 and 5.6.1.7, we see that deleting
 that potential NOx outlier (vehicle number 3202) has virtually no effect on
 either regression line.  The slope of the XM240  regression line decreases only
 3.3  degrees, and the slope of the ASM regression line decreases less than half
 a degree.

                               Table  5.6.1.6
                     Effect  oa  XM240  Regression  Lin*
                             For  HOx  Emissions
                         Of  Deleting  Vehicle  3202

Constant
Coefficient
R2
Based on Al
52 Vehicles
-0.275563
0.738523
64.5%
Based on 51
Vehicles
•0.183932
0.652265
57.4%
                               Table  5.6.1.7
                      Bffeot  oa  ASM  Ragressioa  Liae
                             rox  MO*  Emissions
                         Of  Deleting  Vehicle  3202

Constant
Coefflelen
R*
Based on Al
52 Vehicles
•0.013624
0.829714
70.8%
Based on 51
Vehicles
•0.003336
0.816328
60.1%
    Six vehicles (vehicle numbers:  3172,  3200,  3212,  3239, 3240, and 3244)
exhibited large decrease* ia lane NOx emissions,  but little  if aay change in
FTP NOx •ttiaeioas).  These six had a number of factors ia coonon:
         All six had low as-received FT? HC (for five of the six HC £ 0.37,
         and BC • 0.59 for the sixth), CO (CO £ 3.47), and NOx (NOx * 2.34) .
      -  All six had low as-received lane-IM240 HC (HC 1 0.29) aad CO
         (CO £ 3.33),  but high laae-XM240 NOx (NOx *  1.14).
                                      67

-------
              six had  low as-received ASM composite HC (HC £ 0.24)  and CO
          (CO    0.80  and  Mode-2 ZM240 HC  >  0.50.

      2) the 16 vehicle* that were recruited  (and repaired) because their
         initial lane-XM240 exceeded the CO cutpoint of:
           Composite XM240 CO > 15.00  and  Mode-2 XM240 CO > 12.00.

      3) the 30 vehicle* that were recruited  (and repaired) because their
         initial lane-DC40 exceeded the NOx cutpoint of:
           Composite IM240 NO* > 2.00.

As previously discussed, two vehicle*  (vehicles  numbered 3211  and 3190)  could
be deleted from the -HC-Repaired" and from the "Co-Repaired" data bases due  co
questionable test result*.  Additionally, vehicle number 3202  could be  deleted
from the "NOx-Repaired" data, base for similar reason*.   Thus,  in addition to
performing regression analyse* on the entire  52  car  data base,  we can also
perform regression* on the 32/16/30  (HC/CO/NOx)  subset*,  a* well a*,  (after
deleting the questionable vehicle*) on the  30/14/29  car subset*.  Within these
various data seta, we performed 1C linear regressions,  the  results of which
are summarized in Table* 5.6.1.8 through 5.6.1.10.
                                       68

-------
                         ^  g
R«gr«««ioa  Lin««  of  ABC for  Short
                                            V«r*ua  FT*


Conatant
J^A^^fl^BlA^A
bovniCMni
R-Squaiad
....... ••••^»
Baaad on
MB
VaMclaa
-.365173
1.41060
81.9%
BaMd on
32
ExcMdlng
InftlalHC
-.932339
1.63036
82.7%
BaMd on
32lft«»
TWO
0.014741
0.958254
633*

BaMd on
AIS2
VaMClM
.270944
2.18967
71.7%
Baaad on
32
ExcMdlng
inMalHC
0.371352
2.14853
672%
Baaad on
32 Minus
Two
0.7929
1.11862
903%
                            a. a. i. a
R*gx«»ioa  LiM«  of  AGO  for Shore  T«at«  y«rao«


Conatant
CoafUdanf
R-Squavad
•••••••••
Aisa
VaMdM
4.64057
0. 841373
47.5%
— IM240-
«
Eicaartlnq
InttMCO
17.S097
0.611959
18.7%

BaMd on
lOlflnua
Two
-0.36712
1.28527
55.1%
~.~_. •«•*
Baaad on
Alfl2
VaMdM
5.71775
1.08685
79.5%
Baaad on
tt
Excaadlng
imunco
13.9744
0.95755
73.0%
Baaad on
l6Mnua
Two
13.4061
0.962594
74.1%
                           69

-------
                                       5  £ I  1ft
          Regression  Lines  of  AMOx  for Short Tests  Versus  FTP


Constant
Coefficient
R-Squarad

Baaadon
AI82
Vahtolaa
-.275563
0.738523
64.5%
___ IU94A _
Baaadon
a
Exceeding
Initial NOx
-.778908
0.886525
54.2%

BaMdon
On«
-0.45849
0.733666
39.9%

Baaadon
Aiaa
Vehicles
-0.013624
0.829714
70.8%
.... AQU ...
Baaadon
30
Exceeding
Inttlal NOx
0.209571
0.763686
62.3%

Baaad on
30 Minus
Ona
0.273927
0.711124
45.7%
    Examining the slopes and y-intercepts  (i.e., the "coefficient* and
"constants" in Tables 5.6.1.8 through 5.6.1.10)  of the 18 regression  lines, we
make the following observations:

      -  Limiting the analysis to only those vehicles  whose initial lane-!M240
         test exceeded the outpoint for the pollutant  being examined:

            — had virtually no effect on the regression line predicting FTP
               HC changes based on ASM HC changes, and only a relatively small
               effect on the line predicting FTP CO changes based on  ASM CO
               changes;

            — had moderate effects on the two regression lines predicting FTP
               HC and CO changes based on IM240 HC and CO changes; and

            — had only relatively small effects on the regression lines
               predicting FTF NOx changes based on ASM or XM240 NOx changes.
               Again, the) effect was larger for the IM240 case.

      -  Deleting the one or two questionable vehicles prior to performing the
         regression analysis:

            — produced only small effects in the two NOx cases (IM240 and
               ASM) and in the ASM CO case and

            — produced substantial effects in the two HC cases and in the
               IM240 CO case.

    In  summary, this analysis indicates that the change in ASM scores before
and after repairs correlates with changes in FTP emissions* about as well as
                                      70

-------
 for the  IM240.   However, because ASM outpoints were  not  recommended by ASM
 proponents and EPA did not have outpoints to use as repair targets/ the
 contractor repairs were performed to attain XM240 scores  that complied with
 the standard ZM240 outpoints.  So the contractor repairs  offer little insight
 into the primary question of whether vehicles repaired to pass an ASM test
 will be  as effective ae vehicles repaired to pass the IM240 test.  The next
 two sections will further discuss repair issues.
            5.C.2   Commercial  repair*

            5.6.2.1   Introduction

    The purpose of this analysis was  to compare the effects of commercial
repairs, for vehicles that failed the Arizona I/M test,  on IM240  and ASM
after-repair test results.  Experience has shown that commercial  repairs
geared to steady-state I/M tests have not met expectations for  in-use emission
reductions.  Because vehicles are operated only at steady-state,  repairs have
been geared to reducing emissions at those operating conditions.   As a  result,
emissions over the full range of operating conditions are often not
effectively reduced, even when vehicles are repaired to pas* a  steady-state
I/M test.  Thia is one reason EPA has established a transient test for
enhanced I/M.  Since the XM240 requires vehicles to perform over  a wide
variety of real-world operating conditions, IM240-successful repairs must  be
effective in reducing emissions over a wide range of operating  conditions.

    By comparing the effects of commercial repairs on ASM and IM240 test
results at selected outpoints, an evaluation of the comparative repair
effectiveness can be made.  As discussed above,  EPA analyzed the results  of
repairs performed to pass the Arizona I/M test to determine whether such
repairs would significantly reduce ASM emissions without significantly
reducing FTP emissions.  Sine* the ASM test and the Arizona I/M test are
somewhat similar in that they are steady-state tests, repairs for the Arizona
I/M test may provide) information on whether ASM-successful repair are as
effective as> XM240-effactive repairs.   The data show that successful repairs
for the Arizona) I/M teat are more likely to be successful for the ASM test
than for tha- IH240.

            S.C.2.2   Database/Analysis

    EPA1a commercial repair program in Mesa consisted of offering incentives
to owners of 1983 and newer vehicles that  failed the Arizona I/M taat, but
were not needed or declined to participate in laboratory testing, to return to
EPA'3 IM240 lane for after-repair ASM and  IM240 tests.  To receive their
incentive, they were told to return with a receipt for commercial repairs.  No

                                      71

-------
 instructions were given to owners regarding where or how to get their cars
 repaired,  and owners were not compensated for the actual repair itself.

    As of April 1, 1993, before- and after-repair data were available for  23
 of these vehicles.  One vehicle, 113239  (CRf 24) was removed from the database
 due to unacceptable speed deviations on  its initial ASM test, leaving 22
 vehicles available for  analysis.  For this analysis, five other vehicles were
 excluded because  they continued to fail  the Arizona test after repairs.   The
 resulting  database consisted of 17 successfully repaired, 1983 and newer
 vehicles.

    Cutpoints were applied to the IM240 and ASM data to determine pass/fail
 status.  The pass/fail  determinations were then compared to evaluate the
 effects of commercial repairs.  Three different outpoint sets were used to
 make the comparisons.   Since the Arizona test measures HC/CO only, the first
 comparisons  involved only HC and CO criteria.  Two additional comparisons were
 made which included NOx cutpoints.  All  three are listed below  (Section 5.3
 discusses the relevance of these cutpoints.):

      •  IM240 recommended cutpoints for HC/CO with ASM cutpoints that produce
         the highest IDRs at the same failure rate as the IM240 recommended
         cutpoints:

         IM240 -  0.80 /  15.0  -I- 0.50 / 12.0
         ASM - 1.00 / 8.0

      •  EPA recommended XM240 cutpoints including NOz with ASM cutpoints  that
         produce the same 18% failure rate.  These cutpoints are listed below:

         ZM240 -  0.80 /  15.0 / 2.0 + 0.50 / 12.0
         ASM - 1.00 / 8.0 / 2.0

      •  ASM and  ZM24O outpoints selected to achieve the highest  ZDRs possible
         whil*> keeping th* probable EC rate below St.  These cutpoints  are
         listed below:

         IM2«r-  0.30 /  9.0/ 1.7 + 0.19  / 7.0
         ASM - 0.40 / 8.0 / 1.5

    For each set of cutpoints,  a comparison of th* initial and  final test
results were mad*.  To *valuat* th* effects of  repairs on  a  specific Z/M  test
a vehicle must b* Identified by th* Z/M  test for repair*.  Thus,  while  the
initial test  result comparison allowed th* identification  ability of these two
X/M tests to b* compared, th* final test result allow* an  evaluation of the
relative repair effectiveness of th* Z/M tests.  Th* data  were  restricted to


                                      72

-------
 vehicles  which were identified by all teats for the comparison of  final test
 results.   Using these common vehicles allows the comparison of repair effects
 to be clearly illustrated.

    The results, which are discussed in the next  section, indicate that the
 IM240 is  superior at identifying vehicles requiring repair  and that for
 vehicles  which initially fail both the ZM240 and ASM,  steady-state repairs ace
 more  likely to result in ASM passing scores than in XM240 passing  scores.

            5.6.2.3   Results/Conclusions

    Initially, all 17 vehicles used in this analysis failed their  initial
 Arizona I/M test.  However, the XM240 and ASM identified slightly  different
 sets  of vehicles as needing repairs.  Vehicles of interest  are those that pass
 the initial IM240 and fail the initial ASM and those that fail the initial
 IM240 and pass the initial ASM.

    As shown in Table 5.6.2-1,  for the initial HC/CO only comparison, one car
 passed the  IM240 and failed the ASM and four cars passed the ASM and  failed
 the IM240  (see Appendix B for data listings).  These errors-of-onission
 support the assertion mad* in Section 5.3 that the ASM is weaker than the
 IM240 at  identifying malfunctioning vehicles with HC and/or CO emission
 problems.

                               Table  5.6.2-1
                   Initial  Pass/Fail  Status  Comparison
HC/CO only
Cutpointe
IM240 ASM
PASS PASS 3
IM240 ASM
FAIL PASS 4
IM240 ASM
PAS* PAIL 1
ZM246& ASM
nas FAIL 9
Comma Failure
Kate Cut points
ZM240 ASM
PASS PASS 2
XM240 ASM
FAIL PASS 2
ZM240 ASM
PASS FAIL 0
ZM240 ASM
FAIL FAIL 13
Optim*! XDR/MU
Be Cutpoiate
ZM240 ASM
PASS PASS 1
ZM240 ASM
FAIL PASS 2
IM240 ASM
PASS FAIL 0
ZM240 ASM
FAIL FAIL 14
    Vehicle 13504 (CR* 25) failed the ASM and Arizona test duo to a CO problem
which appear* to occur only at idle operation.  Because the ZM240 driving
cycle includes little idle operation, this vehicle was not identified by the
IM240 HC/CO only outpoints.  An air/fuel mixture adjustment reduced emissions
sufficiently to pass the HC/CO outpoints for the ASM and Arizona teats.
However, this vehicle did exhibit excessive NOx emissions that were identified
by the addition of a NOx outpoint.  Incidentally, the fuel mixture  adjustment
4
                                      73

-------
 did little to address  or reduce this vehicle's NOx emissions on either the
 IM240 or the ASM.

     In contrast to the IM240, which failed to identify only one vehicle,  four
 vehicles passed the ASM  HC/CO only outpoints and failed the IM240 and Arizona
 outpoints.   Three of these vehicles are examples of ASM errors-of-omission and
 illustrate  the superior  identification ability of the IM240.  The fourth
 vehicle  failed NOx and will be discussed after the three that passed.

    Vehicle 13471 (CRt 27) failed the Arizona and IM240 tests because of high
 CO  emissions,  but passed the ASM test.  Vehicle 13125  (CRt 12) failed HC on
 both the IM240 and Arizona tests and was not identified by the ASM outpoints.
 Vehicle  13202  (CRf 15) failed the HC and CO idle modes of the Arizona test.
 On  the IM240,  vehicle  13202 failed HC and NOx but passed CO due to the two-
 ways-to-pass algorithm.   The ASM identified this vehicle for NOx emissions
 only.

    The fourth vehicle that initially passed only the ASM test was vehicle
 12771  (CRt  8).  This vehicle exemplifies the weakness of steady-state X/M
 tests and is discussed in detail in Section 5.6.3.   Vehicle 12771 failed CO
 on the loaded  mode of  the Arizona test but passed the CO outpoint on both the
 IM240 and the  ASM.  However, the car failed NOx and HC on the  IM240 and  failed
 only NOx  on the ASM.  After repair, this car passed both the ASM and Arizona
 tests even when ASM outpoints were tightened.  These repairs did not
 sufficiently reduce emissions over the full operating range of the vehicle,
demonstrated by the vehicle continuing to fail both HC  (1.01 g/mi) and NOx
 (3.01 g/mi) on the ZM240.  This supports the assertion made in the
 introduction that repairs to pase a steady-state test may not  be effective in
reducing emissions over normal driving conditions and, therefore, do not
effectively reduce in-use emissions.

    To illustrate the effects of commercial repairs on ASM and IM240 after-
repair test results, data were restricted to vehicles that  failed both  the
initial ASM and ZM240  (see Table 5.6.2-1).  The results of  these comparisons
are graphically depicted in rigures 5.6.2-1 thru 5.6.2-3.
                                      74

-------
                Figure  5.6.2-1
 ircial  Rapaira  Paaaing ASM  and  IM240  Cutpointa
             HC/CO  only Coapariaon
                                              Pass ASM
                                              But Fail
                                               IM240
               Figura  3.6.2-2
•roial  Rftpaica Vaaaiag ASM  and  ZM240 Catpointa
                        Rat*  Coapariaon
                                              P«S« ASM
                                              But Fail
                                                IM240
                      75

-------
                               Figure  5.6.2-3
                rcial  Repairs  Paasing  ASM  mnd  ZM240
                              High  Ko  Comparison
                  14
Cutpointa
                                                   Pass
                                                   Final
                                                   IM240
        Pass ASM
        But Fail
          IM240
    These graph* show that vehicles can and will be repaired to pasa the ASM
test but will continue to fail the IM240.

    For the first comparison using only HC and CO outpoints, three vehicles
passad the ASM but continued to fail the IM240.   The  second comparison added
the NQx outpoint which in combination with the HC/CO  outpoints produced the
same failure ratea for the XM240 and ASM.  Again,  three vehicles  passed the
ASM but continued to fail the XM240.  For the comparison using the most
stringent outpoints for the IM240 and ASM, five vehicles passed the  ASM but
continued to fail the IM240.  For all of these comparisons, there were no
vehicles that failed the ASM and passed the IM240 after commercial repairs.
This indicates that repairs which are sufficient to pass the  ASM  test are  not
necessarily sufficient to pasa the IM240, indicating  that the repair
effectiveness of the IM240 ia superior to that of the ASM.

    Based on these results,  repairs to pass the  steady-state Arizona I/M test
are significantly more effective at reducing ASM emission scores  than XM240
emission score*.  Although the sample of successful commercial repairs is
small, these reaulta indicate that the ASM test, if implemented,  will result
in significantly lower identification rates and emission reduction benefits
than those of the ZM240.  A more detailed investigation of ASM emission
reduction benefits ia discussed in Section 5.6.3.
                                      76

-------
             3.6.3    Xa-Use  emission  Reductions  from Steal  World
                      Repairs

    On* of the concerns with any t«st is the ability of  an observed reduction
 on the test  to  reflect  real and permanent in-use reductions.   Two  particular
 concerns  are (1)  can  unscrupulous mechanics find repair strategies that  would
 allow a vehicle to  temporarily pass the I/M test without resulting in
 permanent in-use  reductions  (i.e., temporary repairs would be undone  after
 passing the  test),  and  (2) is the test sufficiently imprecise such that
 honest, but  insufficient, repairs would not be detected by the I/M retest.

           Teat   Defeating  Strategies

    It is common  knowledge that the current idle test  can be,  and  is  being,
 defeated  by  a variety of methods.  Most can be used only in the privacy  of a
 test-and-repair station.  Some of the common ones that can be used in a  test-
 only  station include  creating a vacuum leak to lean out the air-fuel  ratio for
 CO failure*, and  raising the idle speed to create a similar effect.   A logical
 question  is, what ia  the likelihood that test defeating strategies can be
 developed by unscrupulous mechanics for the ASM or for the XM240 I/M tests.

    On the surface, the ASM teat appears easier to beat than the ZM240 because
 of ita steady-state nature and number of limited operating modes.  In theory
 at least, the mechanic could employ a similar method to the idle test for ASM
 CO failurea.  The proceaa would include creating a vacuum leak and disabling
 the feedback control  system.  Since it is assumed that moat shops would have a
 dynamometer  in an ASM I/M scenario, the mechanic would simply need to operate
.the vehicle  on the  dynamometer and adjust the leak until the car was under the
 outpoints.  Moat  likely the driveability of the car would be quite poor;
 however,  it would only need sufficient driveability to drive to the test
 center and return,  where the teat beating repairs could be undone.

    If the vehicle could drive to the test center, then it could certainly
 drive the steady-state teat, since driveability ia not required oa the ASM,
 and emisaioae>•*• not recorded during the transitions between ASM teat modes.
 Conversely*  tme> emiaaione are measured during driving transitions on the
 IM240, and the* lack of driveability would require more throttle movement  with
 a  likely  subataatial  increeae ia CO emissions.  If miafire occurred  during
 driving tranaitiona because of the leaa condition, the HC, and  possibly NOx,
 would increase oa the XM240, but would not  on the ASM (because  emissions  are
 not measured during driving transitions).

    Another potential teat defeating atrategy that could occur  oa the ASM for
 NOx failurea deela  with ignition timing.  Retarding ignition timing  has long
 been  aa approach  to reducing  HO*.  Retarding the  ignition timing  excessively,


                                      77

-------
 however,  reduces driveability.  Once again, however, driveability is not
 required on the ASM.   A severe  loss in driveability on the IH240 would be
 expected to increase  CO significantly, but would be expected to have little
 effect on the ASM CO  levels.

    Some may point out that many new cars do not have adjustable distributors,
 and others do not have distributors at all.  Therefore, it would not be
 possible  to retard the timing,  so such a test defeating strategy would not
 exist  on  these cars.   What many may not know is that all cars with non-
 adjustable distributors  and those without distributors have a base timing mode
 that can  be activated.   Activation of base timing will severely retard the
 timing in most cases,  and could be used to lower NOx emissions.

    Since these are only a few of the less creative methods that might be
 attempted to defeat an ASM or XM240 test, it would be useful to verify if the
 theoretical potential really could occur.  Currently there is no data on
 purposefully test defeating repairs.  However, data from vehicles tested in
 Arizona and sent  for  commercial repair may shed some light on the potential
 for test defeating or improper  repairs to be identified by either the ASM or
 the XM240.

    In our commercial repair data base,  twenty-two vehicles failed the Arizona
 X/M test which includes  a steady-state loaded mode and an idle mode.  All of
 these vehicles received  a 4-mode ASM test and an IM240 test.  The vehicle
 owners took the vehicles for commercial repair, and volunteered for repeat ASM
 and XM240 tests when  they returned for their Arizona retest.

    Five) of the 22 vehicles were excluded from this analysis because their
 four-mode ASM  emissions  did not exceed the outpoints of 1.0/8.0/2.0
 (HC/CO/NOx).   The resulting 17  vehicles represent the portion of the  22 car
sample that would have failed a four-mode ASM test if that had been the
official teat.  Not*  that this  group of 17 vehicles  represents  a different
portion of the> sample) of 22 coonercially repaired vehicles than the 17
vehicles used  for analysis in Section 5.6.2.  The analysis in Section 5.6.2
excluded five* vehicle* that ultimately did not pass  the Arizona X/M test  after
repairs.  TBe>  analysis in this  section excluded five vehicles that passed the
 initial ASMVtaat, but included  those vehicles that did not ultimately pass the
Arizona X/M test  after repairs.

    The repairs conducted on th« 17 vehicles are listsd in Table 5.6.3-1.
From these  repairs and the) resulting ASM and IM240 scores* the  possibility of
test defeating atrategies can be evaluated.  Mote that the multiple repairs
 represent retest  failures on the Arizona X/M teat.   Also, four  of the 17
vehicles that  initially  failed  the ASM test did not  ultimately  pass the
Arizona X/M teat.

                                      78

-------
     The before and after repair emission results are graphically represented
 in Figures 5.6.3-1  though  5.6.3-3.  From the repair data reported by the
 commercial garages  (in Table 5.6.3-1),  it is clear that many vehicles had the
 air-fuel  ratio  adjusted or received repairs that would likely affect air-fuel
 ratio.  Many of  the vehicles were feedback carbureted; however, this should
 make no difference  for the purposes of  evaluating the effect of air-fuel ratio
 on the test type.   Only vehicles CR-07  and CR-16 had reported commercial
 repairs that would  not likely affect air-fuel ratio (it was assumed that the
 "tune-up" repairs in Table.5.6.3-1, in  some cases, could have involved
 adjustment of air-fuel ratio).

    On these  other vehicles, the degree of the effect oa air-fuel ratio is
 unknown.   But, from the CO emission results in Figure 5.6.3-2, it is clear
 that in general, a  repair  that resulted in reduced CO on the XM240 also
 reduced CO on the ASM. However, there  are some exceptions.  These are
 vehicles  CR-10,  and CR-25.   Vehicle CR-10 failed the before and after IM240,
 failed the before-ASM, but passed the after-ASN.  Whereas vehicle CR-25 passed
 the before and after IM240,  failed the  before-ASM, and passed the after-ASM.

    Since CO  is primarily a  function of air-fuel ratio*  the observation from
 these two vehicles  ia that the air-fuel ratio during the steady-state test  can
 be different than the average over the  transient test.  To some extent, this
 observation also appears to be evident  in the CO results for vehicles CR-03,
 CR-06, and CR-22 (see Figure 5.6.3-2).  In the case of vehicle CR-10, the air-
 fuel  ratio during steady-state operation is sufficiently lean after  repairs to
 allow the vehicle to pass  the ASM, but  rich enough overall during transient
.driving to cause an IM240  failure.  The opposite is apparently true  for
 vehicle CR-25, where the before repair  air-fuel ratio during steady-state is
 apparently sufficiently rich to cause an ASM failure, but lean enough during
 average driving  to  allow the vehicle to pass the XM240.

    Certainly, the> CO leTol  can also be affected by the catalyst.  But the
 catalyst  was) the> same ia all of these tests, so the catalyst effect  should
 wash out.   Alaev catalyst  efficiency can be somewhat gauged by HC levels as
 seen in Figure* 5.€.3-1. The after  repair HC levels oa  vehicle CR-10 clearly
 pass  the  ASftt eotpoiat. Thsj after repair  IM240  HC  status parallels the CO
 status.   la other words), based oa the  ZM240 this vehicle was  still broken, but
 was passed oa the ASM. The HC levels oa vehicle CR-25  were low for all IM240
 and ASM tests.   Based oa ZM240 results, this vehicle should not have been
 failed for HC or CO.  However, vehicle  CR-25 did have serious problems as
 evidenced by the MOac emissions ia Figure  5.6.3-3.

    The emissioa results oa these two vehicles, reinforce the  following point.
 Air-fuel  ratio caa  affect  the CO levels on both tests.   In particular, the


                                       79

-------
                                  s.
air-fuel ratio during a steady-state mod* can be different than the overall
ratio during transient operation.  Therefore, it is likely that with willful
intent, a mechanic could purposefully create a vacuum leak, and adjust it so
that a car could pass the ASM, but not the XM240.  Whereas the amount of
leanness in vehicle CR-10 was not sufficient to pass CO on the ZM240, it was
sufficient to pass the ASM.  Furthermore, the amount of leanness was not
sufficient to cause vehicle CR-10 to fail either the IM240 or the ASM NOx
outpoints.  Therefore, the results on vehicle CR-10 support the theoretical
possibility that unscrupulous mechanics could, with proper adjustment of
vacuum leaks, be able to adjust vehicles to temporarily pass the ASM CO
without increasing the NOx emissions sufficiently to cause an ASM NOx failure.
As indicated previously, the likelihood of such improper and temporary repairs
would be exacerbated in a program where the ASM was the official test, because
unscrupulous repair centers could conveniently maladjust a vehicle on a
dynamometer to pasa the steady-state modes of the ASM test.
                                      80

-------
                                Tablt 54.3.1
                            Vehicle  Repairs
VEttNQ             1ttPq»ir
CR-01     Adjusted aovTuel mixture on
CR-02

CR-03

CR-04

CR-05

CR-06

CR-07

CR-08

CR-09

CR-10

CR-13
CR-15
CR-16

CR-21
                  I ata/fuel mixture on
           Adjusted aa/tod mixture on
                  r. Replaced heat valve.
                  vacuum leak and adjust
           ignition timmf.
           Bfptactxt O2 Sensor and performed
           TuneUp.
           Rpl O2
                      cap and rotor,
           Rpl te bd^aa, fud flt AdJ
           QfltflUL ^^IttDflBQ OlL
           TUM« op, replaced fuel fltar,
                   ^» 4na^^*
                   » nnBT.
          AdJuswJ attnet nrixtnre and kite
          AdjusttdaoTTud mixture, idte speed.

                               •fearer*
           Set
                                           Rpred electrical short in
                                           harness from ECU to mix
                                           control.
Adjusted Ukxaa/fud
mixture, cleaned fuel
injectors.
A one year waiver was
J^^M^»dl &.« *kl« -•-•-• —
giauiBu tor uiii vrnrMii
                                           Adjusted ata/tal mixtnre
                                           and kflA!

                                           Performed Tdne-ap.
                    iper operation of choka    Scoped <8
-------
                               -''«,
                                    lUJ.t


                           Commercial Repair Effects

                                  to MC briMtam « CM

                                   MMHC.CO.WNOl
      3   I   I   \  1   I   S3   \   I   I   I   II   I   I
8
                           Commercial Repair Effects
       I   I   I   I   I  I   II   \   II   I   I   3   I   I   ?
                                   VtMeto Hunter
                                     82

-------
                            Commercial Rtpair Effects
         33333333333353333
            Insufficient  Repaira

    Another concern with an I/M teat ia the ability for the teat to cause
'proper and sufficient repaira to be performed if the vehicle faila the I/M
teat.  For this analyaia, proper and aufficient repaira are conaidered to be
repaira sufficient to pass the IM240 outpointa.  The commercial repair data
used in the preceding section on test defeating strategiea can alao provide
some inaight into thia iaaue.

    Of internet* ia the) coapariaoa of teat modea between the 4-mode ASM and the
Arizona I/It tee*.  Both have aa idle mode, and both have a steady-state loaded
mode.  The. Adaona loaded mode is similar to the ASM 2S2S mode.
    Using the general aisdlarity of the teat (i.e., idle and loaded modes),
the general sufficiency of ASM repaira can be approximated by observing the
results  fro» vehiclea used ia  the previous section that failed the initial ASM
teat and the initial Arixona I/M teat.  A caae history oa vehicle number CR-
08, which  initially failed the ZM240 HC and NOx outpoint* (aa well aa the
Arizona  CO outpoint and the ASM MQx outpoint), illustrates the concern about
the ability of  the  ASM  teat to cauae proper and sufficient repaira to occur
in-use.
                                       83

-------
     After vehicle number CR-08 had  failed the Arizona Z/M test for CO on
 January 4, and had received initial tests  on the ZM240  and ASM,  it was
 enlisted in the commercial repair program.   Two weeks after the  commercial
 repairs (January 14),  the vehicle returned for  its after-repair  IM240,  ASM,
 and Arizona Z/M retest (and for the owner  to obtain the recruitment  incentive
 payment).  At that time, it was  discovered that two days after the initial  I/M
 test (which was conducted on January 4), and following  repairs (listed  in
 Table 5.6.3-1), the repair center had taken the vehicle to another Arizona
 test lane for an Z/M retest. At this other z/M lane, on January 6 the  vehicle
 easily passed the Arizona outpoints of 1.2% CO  and 220  ppm HC.   However,  when
 retested on January 14 at the IM240 test lane,  this vehicle failed the  Arizona
 HC outpoint by a wide margin (see Table 3.6.3-2).  The  owner was demonstrably
 upset (even though a valid Arizona  passing certificate  had been  issued),  and
 left the teat center abruptly.   However, the owner returned again in another
 two weeks (January 26).  At this time, the vehicle passed  all of the Arizona
 outpoints.  The owner did not divulge any  information on corrections or
 repairs that may have occurred between January  14 and January 26.

                                  Tim»l&3-2

                        Test Data • Vehicle No. CR-08
                                  (Test — I
                                      k*e     I— M240 — I    I— ASM —I
                            HC  CO  HC  CO   HC CO  NO*  HC  CO NO*
                            MB  * met  ft   IM IM  «M   *M »M aM
       MMflt   '•r'Mre 2771   M L£  S7 O3I  JJL 124  2J1  0,4*  S.7
       1AW9J   SfettTe*    —    40 044 H« 0.7»   _   _  _   _  _  _
       1/I4/M   LM 0*240 2977   MOCIlUaOTLil 4.1  2&OL333J  1.63
       1/26W   LM 04240 3161   7S 03t  41 O.M  Ltt 4.4  101   IM240 Ms  actually increased slightly from the first test to the
last.

    The) moat peaaimlatlo scanario on thia vehicle ia that one* the vehicle
failed the Arizona teat for CO* the) mechanic maladjusted the vehicle, took it
to an X/M Ian** where it passed,  and  then undid the maladjustments.  These
undid maladjustments  were then observed on the January 14 Arizona retest.  A
more benign conclusion  ia that the mechanic performed incomplete repairs, but
                                       84

-------
 the repairs  were ultimately sufficient to pass the Arizona test.   Also,  th«
 repairs  were sufficient to pass the ASM NOx outpoints (HC and CO  were always
 below the  ASM outpoints), while they were not sufficient to pass  the ZM240 NOx
 outpoint,  nor were they sufficient to pass the ZM240 HC cutpoint.  In fact,
 the ASM  would not even have identified this vehicle as a high HC  emitter.

    Clearly,  on this particular vehicle,  conraercial repairs were  not
 sufficient to pass the IM240, but were sufficient to pass the ASM test.

    Reviewing the data for all 17 vehicles that failed the initial ASM test,
 all of these vehicles, except CR-07, CR-21, and CR-22 eventually passed the
 Arizona  HC I/M retest.  Also, all vehicles, except these three vehicles  and
 CR-06 eventually passed the Arizona CO I/M retest.

    However,  vehicles CR-03,  CR-04,  CR-06, CR-08,  CR-09,  CR-10, CR-15, and CR-
 16  which initially failed the IM240, continued to fail HC on the IM240 after
 all conmercial repairs (see figure 5.6.3-1).  Further, after all connercial
 repairs, these same vehicles passed the ASM HC cutpoint  (note CR-8, CR-9,  and
CR-15 passed the initial ASM test, see Figure 5.6.3-1).  Given the
similarities of the ASM and the Arizona test, these data suggest that the
 level of HC  repair on the ASM would be similar to the current Arizona I/M
test.  This  assumption oa test similarity and stringency of repair
effectiveness is further supported by the fact that the three vehicles that
failed to pass the Arizona test after repairs  (CR-7, CR-21, and CR-22) were
also the only vehicles that failed the after-repair ASM test  (see Table Figure
5.6.3-1).

    Another method of looking at the ability of the ASM to enforce proper and
sufficient repairs is to look at the test status of the ASM results before and
after repairs relative) to the before and after IM240 status.  The test status
for the 17 vehicles) initially failing the ASM for at least one pollutant  is
listed in a  truth-table) forsmt in Table 5.6.3-3 by pollutant  (i.e., HC, CO,
and NOx).  The) roughly sqoara boxes in a diagonal row  represent  tost  results
where tha  ZM240 and MM status before and after repair were identical.
Deviations) frozfrth* diagonal row, obviously  represent  results where the status
differs betveeetths) XM240 and ASM.  The fuzzy horizontal rectangular box  in
Table 5.6.3-*highlights those vehicles which passed the ASM test after
repair, but  vare* still failing the ZM240  for HC, CO, or  NOx.
    Frost the Table, a total of 11 vehicles continued to fail thsj IM240 HC
cutpoint aftar repair.  Of these 11  vehicles,  8  vehicles  (or 73%)  passed the
ASM aftar repair  (threa of the eight also passed the  initial ASM tast).   All
eight vehicles also passed the Arizona HC cutpoint after  repair.   As
previously mentioned, in the) 11 vehicle sample that continued to fail the
IM240 HC cutpoint, 100% of the vehicles that failed the Arizona HC cutpoint

                                       85

-------
  (CR-07, CR-21, and CR-22), also failed the ASM HC cutpoint.  Thus, every
 vehicle that continued to fail the Arizona HC cutpoint  also continued to fail
 the ASM HC cutpoint.  In other words, in this sample of commercial repairs,
 the ASM did not fail anymore retest vehicles than the Arizona  Z/M test.

     CO test status represents somewhat  of  a mixed bag.   Seven vehicles
 continued to fail the 1M240 for CO after repair.   Only  one vehicle in this
 group of seven (or 14%) passed the ASM for CO after repair.  This vehicle also
 passed the Arizona retest.  In this seven vehicle sample that  continued to
 fail the ZM240 CO cutpoint, six vehicles continued to fail the ASM,  and four
 vehicles (CR-06,  CR-07, CR-21, and CR-22)  failed the Arizona retest.  In this
 case, the ASM found 2 more vehicles than the Arizona I/M retest after
 commercial repairs.

     However,  it should be noted,  that four other  vehicles  had anomalous CO
 results.  Two vehicles, CR-15 and CR-16 failed the initial IM240 for CO,
 passed the initial ASM, and subsequently passed both the IM240 and ASM for  CO.
 Two other vehicles (CR-09 and CR-25), passed the initial IM240, failed the  ASM
 for CO, and also subsequently passsd both the IM240 and ASM reteats.  If the
 ASM was as good as the XM240 in identifying vehicles that should fail a retest
 (at the same overall failure rate), one might expect a  random  scatter on each
 side of the diagonal boxes, particularly for vehicles just marginally failing
 or passing (which all of these were, except CR-25).  Even so,  all of these
 vehicles also passsd the Arizona CO retest.   So they do not represent any
 additional retest failures following commercial repairs that the ASM would
 have found over the standard Arizona test.  Also note,  that these four
 vehicles initially failed and continued to fail the IM240 for  HC or NOx,  and
-that the commercial repairs reduced the IM240 CO levels in all cases.

    A total of  5 vehicles ia this sample of  17 continued to fail IM240 NOx
 after commercial repairs.  Two of the five (or 40%) passed the ASM after
 repairs.   The) Arizona Z/M tost doss not teat for NOx, therefore, it is  more
 difficult to judge) thm effectiveness of the ASM  (using the Arizona test as a
 surrogate)  to fores) proper and sufficient commercial repairs.

    This analymis) began with a concern  about the ability of the ASM test to
 foster  propss> and sufficient commercial repairs following aa Z/M failure.
 Because of thm general similarity of tho Arizona Z/M tost to thm ASM,  it was
 expected that repairs) targeted by the) commercial repair industry towards the
 Arizona test would bo similar to those that would be targeted towards the ASM,
 at least for HC and CO.  Thus, if the ASM were more effective ia forcing
 better  repairs than ths> Arizona tost, tho ASM retest should fail more cars for
 a  given pollutant than thm Arizona retest.  Further, if tho ASM were very
 effective in forcing proper repairs, it would fail as many ears, for a given
 pollutant,  as aa ZM240 rstsst.

                                       36

-------
    The analysis shows that for this small sample,  the ASM fails no more cars
 for HC after commercial repairs than the Arizona retest,  and fails only
 twenty-seven percent of those that failed HC on the IM240 after repair. These
 results imply that the ASM test would not force the repair industry to make
 any more.repairs for high HC emissions than the current  Arizona test,  and
 obviously, not as many HC repairs as the ZM240.  Therefore, the data from  this
 sample suggest that the repair effectiveness credits for HC in the MOBILE
 model  for commercial repairs on the ASM should be no greater than that
 currently given for existing basic I/M programs.

    The analysis for CO retest failures,  indicates  that the ASM found  two  more
 vehicles  than the Arizona test.  The Arizona I/M retest found about 57 percent
 of  the IM240  retest failures, and the ASM found about 86 percent of the retest:
 failures.  Thus in this small sample, it appears that an ASM retest would have
 forced the repair industry to make additional CO repairs over and above those
 that would have been required to pass the Arizona outpoints, but again, not as
 many as the  IM240 outpoints would require.  These results suggest that the
 repair effectiveness credits for CO in the MOBIL* model for commercial repairs
 on the ASM should probably be given additional credit over that currently
 given  for existing I/M program*.  The additional credit would be approximately
 equal to  €0 percent of the) difference between that currently given for
 existing  I/M programs and that given to I/M programs employing the IM240.
 However, given the potential ease) that unscrupulous mechanics could defeat the
 CO portion of the ASM retest, assigning additional CO repair effectiveness
 credits in the model for ASM over those currently given for existing  I/M
 programs would be difficult to rationalize at this time.

    The analysis fox MOx retest failures is somewhat hampered by the fact  that
 the Arizona test only fail* vehicles for HC and CO.  Iven though the  repair
 industry was not repairing vehicles; to an MOx standard, the ZM240 and ASM
 retests after commercial repairs can be used to determine whether either
 retest would hav« forced the repair industry to make  additional repairs.
 Clearly, beta testa would have required some vehicles to  get additional
 repairs fo» high NOx emisaiona.  However, the results from this sample
 indicate thafc as> ASM retest would only require  (0  percent of the  vehicles that
 failed the O04t retest to get additional NOx repairs.  Therefore,  this result
 would suggest that the repair effectiveness credits for MOx in the MOBIL*
model for commercial repairs on the ASM should be  about only sixty percent  of
 that given for the IM240.
                                       87

-------
IM24O
HC

CO

NOx
  HC
  CO
  NOx
  HC
  CO
  NOX
HC
CO
N0»
                              Table 5^3-3
           Effect of Commercial Repairs on Test Status
                                   ASM Status
                    (Statu*  B«£ox« Repair - Statu* Aft«r Rapalr)
              CR-07.CR-21.
                 OI-22
            CR-04.CR-OZ
            Ot-21.0t-22
            CR-9.OM3,
               Ot-25
CR-03.CR-04,
CR-06.CR-10,
   CR-16
   CR-10
                                            aux.ai-1*
                                          Ot-05,OU13
                                          CR-01.CR-05,
                                          CR-13.CR-26
                                             Ot-02
                                            au».at-2S
                                                        OI-08.CR-09,
                                                           CR-15
                                                        CR-01.CR-26
                                                        CR-15. CR-16
                                                           ai-
-------
            S.C.4   One-Mod* Repairs  on  ASM

    The objective of thia analysis was to investigate the theoretical effects
 of targeting the ASM repairs to a single mode.  That is,  if it  were possible
 for a mechanic to reduce the emissions sufficiently on  a  single mode while
 leaving the remaining three modes unaffected:

      •  Could an ASM failing vehicle, with such a repair,  be made to pass  the
         ASM composite outpoint?

      •  What are the emission characteristics of such  passing  vehicles?

    Examining the 106 laboratory test vehicles,  we can  determine whether the
 as-received NO* emissions met or exceeded an FTP NOx standard of  2.0 g/mi.
 Also, we can determine FTP HC/CO emission range.  That  is:

      •  Pasa
           FTP HC $   0.41    and  CO *   3.40 (g/mi)

      •  Marginal (Failing) Emitters
           FTP HC >   0.41    or   CO >   3.40 (g/mi)
           and
           FTP HC i   0.82    and  CO i  10.20 (g/mi)

      •  High Emitters
           FTP HC >   0.82    or   CO >  10.20 (g/mi)
           and
           FTP HC S   1.C4    and  CO S  13.60 (g/mi)

      •  Very High Emitters
           FTP HC >   1.64    or   CO >  13.60 (g/mi)
           and
           FT* EC S  10.09    and  CO £ 150.00 (g/mi)

                Bitter*
                : >  10.00    or   CO > 150.00 (g/mi)

   Claaaifying the)  laboratory vehiolea in thia way producea tea strata/
however, two of thoM strata arc «qpty,  and  one stratum has only  • single  test
vehicle.  Uaing the weighting  factors (Table S.2.5.2), we)  eaa  model the lane
vehicles and characterise the)  emissions  of that simulated  lane sample  of 2,071
 1983 and new fuel-injected passenger ears.   (Actually, the lane sample was
 2,070 ears* the, additional vehicle  resulted  from rounding  off  the) estimated
                                       89

-------
 number of vehicles in those eight strata.)
 following table.
          The distribution is given in the
     NOXS2J
     NOx>2J>
Laboratory Sample
FTPHC/COEmlsetone
Pees
27
0
Meroinel
36
4
HUM
10
3
V.Htah
18
7
Super
1
0
Simulated Lane Fleet

Piea
808
0
FTP HC/CO Emleelene
Uiralnel
934
24
Won
98
18
V.Htah
143
42
Super
6
0
    An ASM outpoint of 1.00/8.0/2.0 (i.e., composite ASM HC £ 1.00,  composite
 ASM CO S 8.Or  and composite ASM NO* S 2.0) will fail 372 vehicles in that
 simulated lane fleet.  The distribution of those 372 vehicles is given in the
 following table.
     NOXilO
     NO»»2J
                                                    QO/1
                     FTP HOCO Emleetone
12
                                        Super
54
                   FTPHCCOfmlMlofw
V
                                •
144
                   18
30
              12
72
              36
                         Super
6
    If mechanics were able to repair those 372 vehicle* so that the emissions
on the 2525 mode, the 50 mph mode, and the idle mode remained unchanged, but
.(he emissions  (HC, CO, and HOx) oa the 5015 mode were reduced by 80 or 90
percent  (the model yields the same result for each), then only 42 of those 372
would be able  to pass the> 1.00/8.0/2.0 outpoint.  Thus, a repair strategy that
targeted only  the) 501S mod* would result in "successfully* repairing only
about 11 percent of the) originally failing vehicles.   The distribution of
those 42 passing vehicles is given in the following table.
                                       90

-------
              Passing  after  Reducing  5015  Mod*  by  10  or  90%
    NOXS24

0
0
Laboratory 81
FTPHG
R/mfBinsI
1
0
vomit
HWw
0
0
unpls
Mien*
V« Hlflll
0
0
Super
0
0
PtS*
0
0
Simula!
FTPHC
MWOjnsI
42
0
•dLHM
COEml
Htah
0
0
> Fl«*t
Mton*
V.Htoh
0
0
Super
0
0
    Rather than attempting to reduce the emissions on the 5015 mode by a flat
percentage* the mechanic could target  the  typical emissions on the 5015 mode
of vehicles whose ASM composite emissions  pass the  1.00/8.0/2.0 outpoint.
Such a repair strategy would not change a  single failing vehicle into a
passing vehicle in our model.

    If mechanics were able to repair those 372 vehicles so that the emissions
(HC, CO, and HOx) on th* 2S2S mod* were reduced by  80 percent  (while th*
emissions on the other thre* modes remained unchanged),  then only 30 of thos*
372 would be able to pas* th* 1.00/8.0/2.0 outpoint.  Th* distribution of
those 30 "successfully* repaired vehicles  is given  in th* following table.

                 •ta*  i%4«4»^4tmfe4.«
                 Passing after Reducing 252S Mod* by  80%
    NOKȣ0
                                 V.
Super
                                               12
Super
Reducing th* 2525 mod* emi**ion* by 90 percent would add 41 vehicles (42
marginal HC/CO emitter* with NQx 1 2.0 and 6 very high HC/CO emitters with
NOx > 2.0) to th* 30 who** estimated ASM composite score would pass th*
outpoint.  7b**» • repair strategy that targeted only th* 2525 mod* would be
successful, omUaly 21 p*xc*at of th* originally failing vehicle*.
             at*
    Rather them attempting to reduce the emission* oa the) 2S2S mod* by a flat
percentage, th* mechanic could target th* typical emission* oa th* 2525 mode
of vehicle* who** ASM composit* emissions pass th* 1.00/8.0/2.0 outpoint.
Such a repair strategy would not Chang* a single failing vehicl* into a
passing vehicl* in our mod*!.

    If mechanics wer* able to repair those 372 vehicle* so that th* emissions
(HC, CO, sad HO*) oa th* 50 mph cruise mod* w«r* reduced by 80 percent  (while
                                       91

-------
 the emissions on the other three modes remained unchanged),  then 264 of those
 372 would be able to pass the 1.00/8.0/2.0 outpoint.  The distribution of
 those 264 passing vehicles is given in the following table.

                  Fla^fe  niafcfibutlon  Tailing  1  QQ/i  Q/2 Q
             Passing after Reducing  50  mph  Cruise Mode  by  80%
     NOx>24
Laboratory Sample
FTP HC/CO Emissions
Pas*
3
0
MeVQIfWl
5
2
IMialisi
mynv
3
2
V.Htah
3
2
Super
0
0
Simulated Lane Fleet
FTP HOCO Emissions
Pass
54
0
• f^-,,1 |
•"•WgsTMef
138
12
High
18
12
V.HIgh
18
12
Super
0
0
Reducing the 50 mph cruise emissions by 90 percent would add 6 vehicles (all
with very high HC/CO emitters and NOx > 2.0) to the 264 whose estimated ASM
composite score would pass the cutpoint.  Thus, a repair strategy that
targeted only the 50 mph cruise mode would result in "successfully* repairing
about 73 percent of the) originally failing vehicles.

    Rather than attempting to reduce the emissions on the 50 mph cruise mode
by a flat percentage, the) mechanic could target the) typical emissions on the
50 mph cruise mode of vehicles whose ASM composite emissions pass the
1.00/8.0/2.0 cutpoint.  Such a repair strategy would result in "successfully"
repairing 162 (44%) of the) originally failing vehicles.  The distribution of
those 162 passing vehicles is given in the following table.

                 yia^fe  Pt«»flhn»ia«   railing   1  QQ/t 0/2  Q
      Passing after ftedaoing  50 mph Cruise)   Mod* to  Meaiaal Seer*
    NOx i 10
    NOx»laV
                                 If IMnh
                                 T. nign
Super
       42
rTF HC/CO Inhesions
            V.
84
24
12
            Super
    It menlmfntr v*re> able) to repair those 372 vehicle* so that the) emissions
(only HC and 00} OB the> idle) mode were reduced by 80 or 90 percent  (the model
yields the) ssaa> result foe each) while the emissions oa the other three modes
remained unchanged/ them only 9C of those 372 would be able to pass the
1.00/8.0/2.0 cutpoint.  Thu*> a repair strategy that targeted only the idle
mode) would result in "successfully* repairing only  about one-fourth of the
originally failing vehicles.  The distribution of those) 9C passing vehicles  ia
given ia the) following table).
                                       92

-------
                                                 1-QQ/-Q/2.O
              Passing  after  Reducing  Idle Mod*  by  80 or  90%
     NOx>24
Pass
0
0
Laboratory Si
F1PHG
levokwl
2
0
ICOBflt
UbrfM
niflnv
1
0
rnpls
MM!)*
V. Htah
1
0
Super
0
0
Pis*
0
0
Slmulat
FTPHC
mAW^MfemAft
MVQV1M
84
0
•dLam
CO fin
UbeJt
nion
6
0
»PlMt
MlOTM
V.HIflh
6
0
Super
0
0
    Rather than attempting to reduce  the emissions on the idle mode by a flat
percentage, the mechanic could target the typical emissions on the idle mode
of vehicles whose ASM composite emissions pass the  1.00/8.0/2.0 outpoint.
Such a repair strategy would result in "successfully*  repairing only 60  (16%)
of the originally failing vehicles.  The distribution  of  those 60 passing
vehicles is given in the following table.

                                    a  Tatlia   1 QQ/t.Q/2  fl
            Passing  aftes  Kedveing  Idle  Mode)  to Woainal Scor*
    NO»24
                     FTPHOCOI
                                 V. HlflH
Super
             FIFHC/COImtaelorw
                                                     42
                           12
Super
    Tightening the ASM outpoint from 1.00/8.0/2.0  to a more stringent outpoint
of 0.40/8.0/1.S produce* similar results in our model.

    An ASM outpoint of 0.40/1.0/1.5 will fail 587  vehicle* in that  simulated
lane fleet.  Tha distribution of those 587 vehicle* is given in the following
table.
    A* with th* 1.00/8.0/2.0 outpoint, single-end* repair* that reduced the
5015 mod* emission* by 80 or 90 percent would succeed in "successfully"
                                      93

-------
repairing only 108 (18 percent of the 587  failing vehicles)
of those 108 vehicles is given in the following table.
                                                             The distribution
                         ni •«•«•**»««; ion  railing   0.40/8 ft/1 S
              Passing after Reducing 3013  Mode  by  80  or  90%
     NOxiiO
     NO»2J

1
0
Labori
FTPHC
•ML*eiMa»feAl
•NsWglTW
5
0
rtory Si
CO Em
***-*--
rwyiM
0
0
unple
V. Woo
0
0
Super
0
0
Pass
6
0
Simula!
FTPHC
Marginal
102
0
ad Lane Rest
"CO Emi
Htah
0
0
•won*
V. High
0
0
Super
0
0
    Rather than attempting to reduce the emissions on the 3013 node by a flat
percentage, the mechanic could target the typical emissions on the 5013 mode
of vehicles whose ASM composite emissions pass the 0.40/8.0/1.5 outpoint.
Such a repair strategy would result in "successfully* repairing only 54  (16%)
of the originally failing vehicles.  The distribution of those 54 passing
vehicles is given in the following table.

                                                  Q  1Q/S
            Passing  aftec  Reducing 3015  Mod*  to  Vomlnal  Seer*
    NOXS24
    NOX»24
                     FTP HC/CO Emissions
                                 V. Hlflli
                                       Super
FTP HCVCO Emissions
            V
           . T«
                                                     48
    If mechanics were able to repair those 387 vehicles so that the emissions
on the 2525 mod* were reduced by 80 percent  (while) the emissions  on the other
three mode* remained unchanged), then only 138 of  those 587  would be  able to
pass the 0.40/8.0/1.3 outpoint.  The distribution  of those 138 vehicles is
given in the) following table).
                                       94

-------
                 y^pjft*  ptaferttmfctan  >allln«y  Q  ^Q/g ft/T jt
                  Passing  aftax  Reducing  2525 Mode by  10%
     NOx»24
                     FTP HC/CO Emissions
                                 V.Mgh
                      9
1
     Super
                  Slmulstsd Lane Rest
                  FTP HC/CO Emissions
Mugjral
 126
V.HIgh
  6
Sui
                                       0
Reducing the 2525 mods; emissions by 90 percent would add 12 vehicles (6
passing HC/CO with NOa 1 2.0 and € marginal HC/CO emitters with NOx > 2.0) to
the 138 whose estimated ASM composite score would pass the outpoint.  Thus, a
repair strategy that targeted only the 2525 mode would be successful on only
about 26 percent of the originally failing vehicles.

    Rather than attempting to reduce the emissions on the 2525 mode by a flat
percentage, the mechanic could target the typical emissions on the 2525 mode
of vehicles whose ASM composite emissions pass the 0.40/8.0/1.5 outpoint.
Such a repair strategy would result in "successfully* repairing only 102  (17%)
of the originally failing vehicles.  The distribution of those 102 passing
vehicles ia given in the following table.

                                                  0 .4Q/t .0/l.S
            Passing  aftex  ftedueing 2525  Mods)  to  Vominal  Score,
     NOx s 2.0
                     FTP HC/CO I
                                 V.HiQh
     Super
                   96
              V.Htah
        Super
    If mechanic* wer« eblsj to repair those 587 vehicles so that the emissions
on the 50 mpb exvia* mod* were reduced by 80 percent  (while  the emissions on
the other thne> sadee remained unchanged), then  3€5 of those 587  would be able
to pass the>».40/§.0/1.5 outpoint.  The  distribution  of those 365 passing
vehicle* ia> gins* ia tte following  table.
                                       95

-------
                 fit
Failing   0.10/8 0/1 S
            Passing aft** Reducing  SO  aph Cruise Nod*  by  80%
     NOXS2J
Laboratory Sample

PMO
6
0
FTP HOCO Eml
Marginal
12
1
HUM
4
0
aalone
V* Hlflll
2
0

Super
0
0
Simulated Lano Float

Paao
72
0
FTP HOCO Emteelons
Marginal
251
6
IHnli
man
24
0
V. Htah
12
0
Super
0
0
Reducing the 50 nph cruise emissions by 90 percent would add 12 vehicles  (all
with NOx > 2.0; 6 of which with marginal HC/CO and 6 with high HC/CO)  to  the
365 whoso estimated ASM composite score would pass the outpoint.  Thus, a
repair strategy that targeted only tho 50 nph cruise mode would result in
"successfully* repairing about €4 percent of the originally failing vehicles.

    Rather than attempting to reduce the emissions on the 50 mph cruise mode
by a flat percentage, th* mechanic could target tho typical emissions on  tho
50 mph cruise mode of vehicles whoso ASM composite emissions peso  tho
0.40/8.0/1.5 cutpoint.  Such a repair strategy would result in "successfully'
repairing only 275 (47%) of tho originally failing vehicles.  Tho  distribution
of thoso 275 passing vehicles ia given in tho following table.

                                                  0.4Q/a fl/l.S
      Passing aftox Reducing  50 mph Cruise)  Mod*  to  nominal  Score
    NOX »10
                                 V.HIoh
Super
                                               48
Simulated Lane Pleat
FTP HOCO Emission*
            V Utah
            w» I Iflill
              191
       24
12
    If mechanic* w*r* abl* to repair thoso 587 vehicle* so that tho omissions
on th* idl* mod*) w*r* roduced by 80 or 90 percent  (tho model yiolda tho same
result for each} vail*) th* omission* on tho other thro* modes remained
unchanged, thoatoaly 42 of thoso 587 would bo able to paaa th* 0.40/8.0/1.5
cutpoint.  Ttaav • ropaix strategy that targeted only th* idl* mod* would
result ia •succ*aafully* repairing only about seven percent of th* originally
failing vohicl**.  Th*) distribution of those 42 passing vehicle* is given in
tho following tabl*.
                                      9<

-------
                                       Fallin   Q.AO/i.Q/1 Jt
              tassiag  after  Reducing  Idle Mod* by  SO  ox 90%
Pie*
0
0
Laboratory Si
FTP HO
Itaratna!
1
0
rcocmi
Htahs
0
0
mpto
W Utah
»• nvn
0
0
Super
0
0
PaSO
0
0
Sbmilat
FTPHC
" 	 • 	 •
NeWuWtv
42
0
•dLam
COImJ
Htah
0
0
I Fleet
•stone
V.HIgh
0
0
Super
0
0
    Rather than attempting to r«duc« th« •missiona  (only HC and CO) on the
idle mod* by a flat percentage, the mechanic could target the typical
emissions on the idle mode of vehicles whose ASM composite emissions  pass the
0.40/8.0/1.5 outpoint.  Such a repair strategy would have produced exactly the
same result (i.e., 42 passing vehicles) as would reducing the idle emissions
by a flat 80 or 90 percent.

    From the preceding two examples (i.e.,  using outpoints of 1.00/8.0/2.0  and
0.40/8.0/1.5), the only potentially effective "single-mod* ASM repairs" are
those repairs targeted at the 50 mpb cruise mode (reducing emissions  by 90%).
However, the model predicts that those repairs would not be successful on 27
to 36 percent of the originally failing vehicles.  The distributions, of those
vehicles that ar* still failing th* respective ASM outpoint after repairs
targeted on th* 50 mph cruis* mod*  (reducing emissions by 90%),  are given in
th* following table.
0.40/1  o/i.a
Cruise Mod* by  90%
                 After  Reducing  50
    NOX»2*
                                                 SIU Ming Outpoint of 0^40/10/1 J
                                        Super
                                          6
                               Super
                                 6
                           „ w* can see that th* vehicle* that th* model
predict* will continue) to exceed th* respective ASM outpoints, even  after
single-mod* repairs targeted at th* 50 mph  cruis* mod*,  ar*  among th* highest
emitters in th* simulated Ian* fleet.
                                       97

-------
 5 . 7   Purge  Analyses

             S. 7.1    Introduction

    In purge  testing, the concern is not that too many malfunctioning vehicles
 will  pass a teat.   Instead/ the major concern ia too many properly functioning
 vehicles will fail  a  test that attempts to replace a test with real-world
 driving  behavior with a  few steady-state modes. These steady-state modes may
 not provide some vehicles with the opportunity to purge.  Thus, the purpose of
 this  section was to compare the canister purge system false failure rates
 (errors-of-commission or Bcs) for the ASM and IM240.  The data indicate that
 the ZM240 is  significantly less likely to falsely fail vehicles for purge than
 the ASM.

    Vehicle evaporative emissions contribute significantly to the VOC
 inventory.  Because vehicle fuel tanks and carburetors must vent to atmosphere
 for proper vehicle  operation, cartoon canisters are added to collect
 hydrocarbon molecules which would otherwise escape.  Because the carbon
 canister  has  a  finite capacity, which if exceeded, allows hydrocarbons to
 escape, the canister  must be kept purged of stored hydrocarbon molecules.  The
 evaporative control system includes a purge system which draws stored
 hydrocarbons  into the engine where they are burned.

    Most properly functioning canister purge system* do not purge constantly;
 instead, most only  purge; when their CCM computer algorithms call for purge.
 Driveability  or emission problem* accompany purge that initiates during
 unfavorable conditions,  so purge algorithm* are designed to take advantage of
 opportune condition*.  The purge algorithm* are known to vary widely from
model to model.  So,  the main problem for an I/M teat is to provide vehicle
operation that  will coincide with the conditions necessary to induce the
system to activate)  canister purge.  EPA ha* found that some vehicle* only
purge during  acceleration* or deceleration*, which is problematic for ateady
state testa such •• the) Aflat and could result in falsely failing vehicles with
purge system* that  are) properly functioning.
    Also,  aeaafevehielea have) timer* that don't allow purge for several minutes
after the) eofl*** i* ataxtad or * specified operating temperature  ia  reached,
so all el*e> being equal* tha longer the teat duration, the lower  the
probability of purge) falaa failure*.  This also make* tha teat order
important, sinca tha taat that vaa performed second i* more  likely to achieve
purge than tha initial taat.  Thia ia one reason tha teat procedure  in our
study ia Maaa required tha taat order to be reversed each time another car  was
tested and why tha angina waa raatartad just before each taat.
                                       98

-------
    The ZM240 purge flow was summed over the  full  ZM240.  The ASM purge flow
 was measured and sunned over the full four nodes of the ASM including
 transient  segments of the ASM cycle, in contrast to the ASM exhaust  emissions,
 which were only measured during the four ASM steady-state modes.  Because the
 flow measuring equipment is the same for both transient and steady-state
 tests, it  was practical to measure purge flow on the ASM during the  three
 accelerations and the single deceleration needed to complete the  four ASM
 modes.

             5.7.2   The  Database

    The database for the lane purge analysis  was restricted to  vehicles that
 met all of the following criteria:

      -  as-received purge data were available for both the XM240 and the ASM,
      -  the test order was known,
      -  and the data passed the purge QC criteria (see Appendix C).

    The resulting database consisted of 1170  vehicle*.   Of  these, 577 received
 the ZM240  first and 593 received the ASM test firat.  The comparisons made for
 this analysis included failure rate comparisons, and comparison* of vehicles
 for which the ASM and XM240 purge) status did not agree, or "false failures".
 in addition,  these comparisons were made for data stratified by test order.
The standard used for these comparisons was 1.0 liter/test.

            5.7.3    Th*  Reaulte

    The results of the failure rate comparisons are as follows:

      •  Overall purge) failure rate*:
         - The XM240 failed 7.43%  (87 vehicles).
         - The ASM failed 11.45%  (134 vehicles).

      •  Initial test failure ratea:
         • Tbei IM240.1st failed €.93%  (40 vehicles).
               AM. 1st failed 11.13%  (6C vehicles).
      •  Second teat failure rateas
         • The) DC40.2nd failed 7.92%  (47 vehicles).
         - The ASM.2nd failed 11.79%  (68 vehicles).

    These higher failure ratea for the ASM raise the question of whether the
ASM correctly identified non-purging vehicles that the ZM240 missed,  or
whether the ASM incorrectly failed vehicles.  Passing the DC40 purge test
requires either that purge actually occurs or that the) measurement system


                                       99

-------
 falsely indicate that purge  is occurring.  Sine* the ASM and IM240 were run
 with the same measurement  system, and results were reported electronically
 without human intervention,  it is not conceivable that a measurement error
 made some cars pass the  ZM240 and fail the ASM.  Consequently, the ASM-
 fail/!M240-pass  cars must  be considered improper fails by the ASM, and vice
 versa,  with a possibility  that test order was a contributing factor in
 specific cases despite the engine restart for both tests.  However, since the
 sample  has essentially an  equal number of each test order, test order should
 not  be  a relevant factor overall.

    The next set of statistics implies that both the ASM and the IM240 falsely
 fail vehicles, but the ASM falsely fails more vehicles.

      •   Overall false failure rates  (fails one test but not the other):
          - 1.1%  or 13 vehicles failed the ZM240 but passed the ASM.
          - 5.13% or 60 vehicles failed the ASM but passed the IM240.

      •   False failures  on initial test:
          - The IM240.1st falsely failed 1.21%  (7 vehicles).
          - The ASM.1st falsely failed 4.22%  (23 vehicles).

      •  False failure*  on second test:
         - The ZM240.2nd falsely failed 1.01%  (6 vehicles).
         - The ASM.2nd falsely failed  €.07%  (35 vehicles).

    Figure 5.7.1 graphically illustrates the comparison of false failure
rates.
                                      100

-------
   7.0% • •

   6.0% ••

   5.0% • >

   4.0% •>

   3.0% •-

   2.0% • -

   1.0% ••

   0.0%
  Overall
  N-1170

5.1%
                                  Figure 5.7.1
                    Comparison of  False  Failure  Rates  for
                            ZM240  purge  vs  ASM purge
                                                        ASM.2nd
                                                        N-577
         1.1%
ASM. 1st
 N-593
                      4.2%
ZM240.1st
  N-577
                               1.2%
            Overall - False
              Failure  Rate
                      Initial - False
                       Failure Rat*
                         6.1%
IM240.2nd
  N-593
                                                      1.0%
                        9 recondit ioned
                         False Failure
                              Rate
     DASM Purge - False Failure Rat*    fl ZM240 Purge - False Failure Rate
    The moat relevant comparison is the initial tests  (ASM. 1st • ZM240.1st)
because they are more representative of the conditions and vehicle
preconditioning expected in official I/M programs than the preconditioned
tests.  The purge results for the initial tests were similar to  the overall
results; the ASM. 1st'a false failure rate was 3 percentage points  higher than
for the IM240.1st.

    With a 3 to 4% falsa) failure rate*  the ASM purge test  could  cause severe
problems to I/M programs) la the form of frustrated consumers and skeptical
mechanics.

    As disease** in the introduction,  test order was expected to be important
because th* tee* that we* performed second would be more likely to achieve
purge than the teat that was performed first.  Contrary to expectations,
however, the ASM.2nd exhibited a 0.63% increase in failure rate and a 1.85%
increase in false failures when compared to the ASM. 1st.  The the IM240.2nd
also produced a higher failure rate (+1.0%) than the DC40.1st, but the
lM240.2nd's false failures decreased 0.2% compared to the XM240.1st.

    In addition, the false failure rate for the XM240.2nd is markedly better
than for the ASM.2nd when viewed as a percentage of failures.  Figure 5.7.2
                                      101

-------
 shows that the false failure rat* dropped from 17.5% of the failing IM240.1st
 vehicles to 12.8%  of the  failing ZM240.2nd vehicles.  In contrast,  the  false
 failure rate increased to 51.5% of the ASM.2nd failing vehicles  from 37.9% of
 the ASM.1st failing vehicles.  So although the false failure rate for both
 tests is expected  to decrease further if the engine restart is avoided  before
 performing the second-chance test, these data suggest that retesting is more
 effective in reducing  ZM240 false failures than ASM false failures.
                                   Figure  5.7.2
               raise  Failure  Rates  as  a Percentage  of Failures
                          for  XM240  purge vs  ASM  purge
             Overall - False
              Failure Rate
Initial - False
 Failure Rate
Preconditioned
- False Failure
     Rate
      OASM Purge-Fals* Failure Rat*    fl IM240 Purge-False Failure Rate
    Overall,  74 vehicle* failed both th* ASM and IM240.   Th* ASM falsely
failed 60 additional vehicles while the IM240 falsely failed only 13
additional vehicle*.  A* shown in Figure 5.7.2, 44.8% of th* 134 vehicles
failing they AaMpurg* wer* false failure* compared to 14.9% of th* 87 vehicles
failing th*rXltttt purg*.  In addition, 37.9% (29 of 
-------
 failures.   However,  it  is speculative whether this will significantly reduce
 the ASM false failure rate.  Also, second-chance testing adds cost.   Since the
 false failure rate increases from the ASM.1st to the ASM.2nd and decreases for
 the IM24.2nd compared to the ZM240.1st (See Figure 5.7.2),  the data  indicates
 that second-chance testing may not be as effective for the ASM as for the
 IM240.

     Second-chance testing costs will be lower for the ZM240  because  it  fails
 fewer cars  initially, thus, requiring fewer retests and some vehicles just do
 not purge during steady-state operation.  For these vehicles, an alternate
 cycle such  as  the IM240 would be required.  Dynamometer costs would then
 increase because inertia simulation is needed, but the more expensive IM240
 exhaust measurement  systems would not be needed.

     In summary, the comparison of ASM purge and XM240 purge shows that  the
 IM240 is superior in correctly identifying vehicles with malfunctioning purge
 systems,  with false failure rates of 4 to 6% for the ASM (3 to 6 times higher
 than IM240  false failure rates), an additional second-chance test will be
 required.  And since some vehicles simply do not purge on the ASM steady-state
modes, even with purge) measured during the accelerations between modes, an
alternate cycle such as the ZM240 may be required for retests.  In conclusion/
the  ASM purge  test is substantially less effective than the ZM240 purge test.
      5. •   ZM240  Improvements  and  the  Four-Mode  ZM240

    The purpose) of this section is to convey that refinements are possible
which would make the XM240's performance a "moving-target," and to further
reiterate why one sample should bet used to develop the ASM-mode weighting
factors and an independent sample used to evaluate the ASM's effectiveness.

    EPA's recommended ZM240 outpoints of 0.80/15.0/2.0 + 0.50/12.0 represent a
compromise between failing high emitting vehicles and not failing clean
vehicles.  fts outpoints ax* tightened* the iDRa generally increase at the
expense of inenasing the) possibility of errors-of-commission.   Increasing  the
power of the) tawe (i.«., the ability to distinguish between malfunctioning  and
properly fTimtqiiirg vehicles) serves the public good, in that the high
emitters not identified by the) test are not  repaired*  so the  cost of testing
such vehicles is not rewarded by air quality improvements that accrue from
identifying and repairing such vehicles.  More tangible  is that  vehicle  owner
satisfaction and acceptance) of I/M programs  increase with lower  errors-of-
coonission.

    EPA is not content foe all tias with the absolute performance of the ZM240
as now defined.  Although ia a relative sense, its performance is superior  to

                                      103

-------
                                        . Jfc.
 any of the alternative I/M tests, it can be improved. Consider for example the
 IM240's 15 g/mi CO outpoint.  This is more than four tinea the FTP CO
 standard,  and unlike the  FTP  which includes a cold start9,  the IM240 is  to be
 performed on fully warmed up  vehicles.  So the errors of omission (vehicles
 which pass,  but should not) are higher than if a more stringent CO standard
 were used.  EPA testing has shown that tighter outpoints will identify more
 high emitters,  but also fail  some properly functioning vehicles.  Although the
 IM240 is considerably better  than any alternative I/M test, in this regard,
 there is no  question that its performance can be improved.  The IM240's
 performance  can be improved in two areas:

 -   Reduce  the teat-to-test variability so that outpoints can be tightened
    without falsely failing clean vehicles.

 -   Us* statistical techniques to improve the IM240rs correlation with the FTP.

    Such improvements will serve the public interest by increasing the air
 quality yield per  teat-dollar, so alternative testa should be evaluated
 againat the  state-of-the-art  of IM240 testing rather than the IM240
 performance,  aa it existed, when the I/M Rule was published.  Proponents of
 alternative  I/M teata nay point out that if the XM240'a performance, aa it
 stood in November  1992, waa good enough to meet the performance standard, then
 this performance standard should be the standard for alternative teata.   While
 such a policy may  indeed  "level-the-playing-field* for alternative testa and
 is in  fact what  la allowed by EPA'a I/M Rule, it ia difficult to argue that
 this approach promotes  the general welfare and should guide state and local
 decision-makers  concerned aa  much about clean air aa about meeting minimum
 requirementa.

            S.t.l   ftedvoe) Teet-to-Test  Variability

    Teat-to-teat variability ia the primary reason why the DC240'a outpoints
are so much lea* stringent than the FTPs.  The FTP controla a number of
variables that are) widely known to affect a given vehicle'a emiaaiona.  Some
variables the*, are) tightly controlled for FTPs were either more loosely
controlled oaxnot  controlled  in EPA'a ZM240 lane teata.  Theae  include, among
othera:
9  CO (and HC) eaiaaione are considerably higher during waraup than during
fully warned-up operation.
                                      104

-------
       -  ambient temperature in the test cell
       -  humidity  in the test cell
       -  engine temperature  (FTP indirectly controls with preconditioning and
          ambient conditions)
       -  catalyst  temperature (FTP indirectly controls with preconditioning
          and ambient conditions)
       -  vehicle operation prior to the emissions test (can affect emission
          control system timers for purge, air switching,  etc.,  and other
          variables affecting emissions)
       -  evaporative canister loading  (FTP indirectly controls with ambient
          conditions and vehicle operation during the 12 hours preceding the
          FTP  emissions test)
       -  tire pressure
       •  speed excursions from the nominal speed (±2 mph on FTP vs. driver
          discretion for EPA's pilot XM240 testing to date)
       -  exhaust system backpressure (NOx can be adversely affected by a
          constant  volume sampler if quality control is not adequate)
       -  fuel composition

    EPA has already made improvements that I/M programs will be required to
implement, but were not implemented during EPA's testing.  For example, FTPs
are voided if speed excursions from the nominal speed exceed ±2 mph.  In
contrast, much of  EPA's data are from vehicles with speed excursions that
exceed ±2 mph.  In a committee that included I/M contractors, state X/M
program, officials, XM240 equipment manufacturers, and automobile
manufacturers, a consensus was reached on requiring this tighter speed
tolerance along with additional tighter controls that will reduce test-to-test
variability10.

    There are also variables that can not be controlled,  such as ambient
temperature and canister loading, but can be compensated  for to better
distinguish between malfunctioning and properly functioning vehicles. Given
enough data,  computer algorithms can be developed that consider the more
important variables) and apply adjustment factors to the official  IM240  test
results.

    SiapllstiOY approaches] such as setting tire pressure or providing second-
chance testa fox vehicle* that are within l.S  times the outpoints seem costly
to implement  because) additional X/M lanes and  personnel are  needed, but are
judged to be  cost  effective) since vehicles that should not  fail but do,  must
be retested after  •repairs' anyway.
10  Draft High-Tech Test Procedures, Quality Control Requirements, and
Equipment Specifications, April S,  1993.
                                      105

-------
     More sophisticated algorithms utilizing sensors to measure variables  like
 ambient temperature and catalyst  temperature allow relationships to be
 developed and used to compute scores.  These are more efficient because they
 can increase the power of the test  without requiring additional I/M lanes and
 personnel.  Developing these techniques will require substantial data.  When
 the IM240 is implemented, much data will become available to allow development
 of such algorithms.
             5.9.2    Statistical  Techniques  to  Improve  the)  XM240's
                      Correlation  With  the  FT»

    Presently/ the ZM240 score constitutes the sun of the mass emissions
 divided by the distance accumulated.  Because almost every second of operation
 is taken from various segments of  the FTP, the two tests correlate better than
 any existing alternative I/M test.  But their correlation can be improved
 using multiple regression.   For example, the uncontrolled variables that
 attend XM240 testa probably make it appropriate to de-emphasise the initial
 operation of the IM240 in computing the score and emphasizing the later
 operation.  The later operation is  somewhat preconditioned by the initial
 operation.   Also, the IM240 has a  higher average speed than the FTP, so de-
 emphasising the  high speed  portion* should produce a better correlation with
 the FTP.

    The data itself can be used to determine the more appropriate weighting
 through the use  of regression techniques.  For example, EPA divided the IM240
.into  four modes  as followss

    node 1:     0-60 second*
    Mod* 2:   61-119 second*
    Mod* 3:  120-174 second*
    Mode 4:  175-239
    A* for thevASM* coefficients are developed by performing a multiple
regression Mtisjuin the) result*  from four mode* are  the  independent  variables
and the) m> reevlte 1* the) dependent variable, which allow* the  data to
determine the> appropriate; mod*  weighting.
    EPA tried thi* uciAg the only substantial database with the information
needed  (FTP* with ZM240 4 ••ode results  or second-by-second result*),  which
happen* to be the vehicles on  which this  report  focuses.   So the coefficients
had to be developed on the asm set of  data to which they were applied.   EPA
condemns this practice, as discussed  in Section  3.5, but  having  no
                                      106

-------
 alternative, the results are presented only to provide an indication of how
 the  IM240's performance is enhanced through this approach.

    Unfortunately,  the legitimate performance  increase gained by using
 multiple regression to determine appropriate mode weighting can not  be
 isolated from the inappropriate application of these coefficients  to the  same
 vehicles from which they were developed. So the performance presented in  Table
 5.8.2 gives an overly optimistic view, but also reiterates  that ASM-advocates
 who do not accept EPA'a judgement that it is inappropriate  to  apply
 coefficients to the vehicles from which they were developed, should  then
 compare the ASM performance with inter-linked coefficients  to  the  IM240 also
 utilizing inter-linked coefficients.

    The multiple regression was performed on the first  91 vehicles in the
 database only (this analysis was not repeated when additional  data became
 available).  The results are presented in Tables 5.8.1 and 5.8.2.  The
 negative coefficients in Table 5.8.1 indicate that insufficient  data is
 available for developing logical coefficients, which will compensate, to  some
degree for the inter-linked performance listed in Table 5.8.2.
                                 Table)  5.8.1
                       Coefficients  Developed  from
              Multiple  Regression  of  4-Mode  ZM240  vs.  FT»
Mode
Constant
1
2
3
4
R*
me
0.03
-0.30
1.16
0.2C
0.09
90.3%
CO
-0.28
-0.18
1.70
-0.01
-0.08
86.0%
M0«
-0.02
-0.07
0.37
0.02
0.5C
69.6%
    Tabl« S.tvX illustrate* how the 4-Mode test improve* the tradeoff between
iDRa and Be ntoa at «quivalent failure rates.
                                      107

-------
                                •  •».
                                 Table  5.8.2
              4-Modsj  ZM240  Performance  Veraua  Normal  ZM240

rail
Rate

12%
13%
14%
15%
19%
20%
23%


I
4 -Mod*
88%
90%
91%
92%
91%
94%
95%
C
Regular
88%
90%
91%
91%
93%
93%
93%
IDRa

CO
4-Mode
65%
66%
66%
66%
72%
72%
73%
Regular
63%
65%
66%
66%
68%
69%
69%


NO*
4 -Mod*
72%
75%
78%
75%
83%
86%
88%
Regular
75%
76%
78%
78%
82%
82%
83%
Ic»


4 -Mode
0.0%
0.0%
0.4%
0.4%
0.4%
0.4%
0.7%
Regular
0.4%
0.4%
0.7%
0.7%
1.4%
1.8%
3.9%
    Notice the performance increase in that the iDRs increase and errors-of-
comaission decrease.

    In conclusion,  developer* of alternative  I/M tests should not consider the
performance of the ZK240 to be fixed,  while better than any existing I/M
tests, XM240 improvements are possible and desirable.  CPA's mission to
improve air quality and enhance the public welfare necessitates evaluating
alternative tests,  not against the performance of  the  XM240 as  it was in
November 1992 when the I/M Rule was published, but instead, against the  state-
of-the-art.
                                      108

-------
 6.    Test  Programs  by  Other  Organizations

       6.1    Colorado  Tost  Program

    Tho Colorado Department of Health  (COH)  completed an evaluation11
 comparing the FTP and the  IM240  to the following eight  X/M test modes, in the
 order the modes were  performed:

       -   35 mph road  load
       -   50 mph road  load
       -   ASM 2545
       -   ASM 2525
       -   ASM 5015
       -   Idle tost
       -   2500 rpm

       Their conclusions included the following:

           "The loaded mode [IM240] tests (both [93] second and 240 second]
       identify significantly man of the excessively emitting vehicles and man of the
       excess eaimkxttthaa do any of ttw steady-state tests. They also have fewer
       etnas of commission sod less sensitivity to differences between FTP sod short
       test emission levels. Win w ote consideration, eitter the 95 second or the
       240 second version of dw QM240] would be the clear choice for the most
       accarate sod effective kkntifkation of excessively emitting vehicles.1*


       €.2   California  Test  Program

    EPA has received a preliminary analysis12  from the California Air
Resources Board (CAM) comparing the ASM5015  and the ASM2525 to  the IM240.
The GARB  analyeie looks) favorably on the ASM  modee and concludes that the ASM
tests  are as effective ae  the  ZM240.  However,  there are significant concerns
with CARS's data.  The** concerns) include the following:

       -   They CUB database la  not representative of the newer  fleet.
11
    Ragazzi, et al.
12  Draft Memorandum from Jeff  Long, Manager, Analysis Section to Mark
Carlock, Chief, Motor Vehicle Analysis Section,  "Comparison of Excess
Emissions  Identified by IM240,  ASMS015 and ASM252S Tests,* California Air
Resources  Board, not dated,  received April 15,  1993.
                                        109

-------
      -  GARB'a tasting is not representative of actual I/M testing.
      -  All of CARS'a testa were preconditioned.
      -  GARB'a ASM equations, when applied to EPA's data, demonstrate poor
         performance.

    EPA is preparing a separate document that will consider CARS's ASM teat
program in more detail.
                                      110

-------
 7.    Teat  Coata Comparison

     Supporters  of  the ASM have frequently suggested that it would be a  more
 cost-effective test than the IM240, given that the equipment cost is
 significantly lower.   As Table 5.12.1  shows, the equipment package for the ASM
 aeries, with purge and pressure  testing, does have a lower total cost than the
 IM240,  purge,  and pressure  equipment package.

                                  Table  7.1
              Equipment Costs  for  the  ASM  Series  and  IM240

                IM240                                 ASM
                            £flflt            V.epi < pmont-
      Pressure Rig          $600           Pressure Rig         9600
       Purge Meter          $500           Purge Meter          $500
           VDA             $1,000              VDA             $1,000
       Dynamometer         $25,000          Dynamometer         $20,000
     CVS £ Analyzers      $79,000"      BAR90 fi NOx Bench      $19,000
          Total           $106,100                             $41,100

    These figures reflect the most recent cost information that EPA has
 received from industry.   EPA has published previous estimates of the per
 vehicle costs of  ASM and ZM240  testing  in  "I/M Costs,  Benefits, and Impacts,"
 in November,  1992.   EPA  found,  and independent analyses confirmed, that
 equipment costs,  when spread over the useful life  of the equipment, constitute
 a relatively small  portion of the per vehicle cost of  an Z/M test; labor and
-overhead costs are  considerably higher.  In analysing  the) current average per
 vehicle inspection  cost  in s centralised program of $8.50, EPA estimated that
 equipment accounted for  219, labor for  96C, 82C went to defray construction
 costs,  the  state  oversight fee  averaged $1.25, and the remaining $5.26 went to
 cover various overhead cost* (for a  full discussion of EPA's cost estimation
 assumptions and methodology  the reader  is  referred to  Sections 5.2 and 5.3 of
 "I/M Costs,  Benefits,  and Impacts,*  contained in Appendix M).  Current testing
 stations have) aa  average) peak capacity  of  25 vehicles  per hour and enough
 stations are> constructed to  avoid long  lines on peak '*••""'< days.  Given the
 typical pattern of  owners' choices about when to cone  for inspections, this
 results in  aa average* actual throughput of 12.5 vehicles per hour  which
13  Letter from. Kenneth W. Thomas, Marketing Manager, Z/M Systems, Horiba
Instruments  Incorporated,  to Bill Pidgeon,  U.S.  Environmental Protection
Agency, April 7,  1973  and  Quotation  from Scott P.  Corrunker,  Salee Engineer,
Combined fluid Products Company to Dan Sampson,  U.S.  Environmental Protection
Agency, January 27,  1993.  These are attached as Appendices J and X.

                                      Ill

-------
 translates into 39,000 vehicles per year per  lane, and costs are spread over a
 multi-year period, five years in most  cases.

     Throughput is  the most  critical  variable in estimating costs since it
 determines the size of the. inspection  station network needed for a given area
 and the number of vehicles  over which  costs for each lane are spread.
 inspection lanes usually have more than one position, with different parts of
 the inspection performed at each one.   Hence, throughput is governed not by
 the time required to perform the total test sequence, but by the time required
 at the longest position. Whether the  test sequence consists of the IM240 with
 purge and pressure testing  or the ASM  with purge and pressure testing, the
 longest part of the sequence is the tailpipe  emissions test.

     The combined IM240 and  purge test  takes approximately three minutes (using
 fast-pass and fast-fail) to perform on the average.  Allowing an additional
 minute to maneuver the vehicle onto the dynamometer and otherwise prepare the
 vehicle for testing the total time at  the longest position is estimated to be
 four minutes.  This translate* into a  peak lane capacity of IS  vehicles per
 hour and an average actual  throughput  of 7.5  vehicles per hour.  The ASM
 consists of four modes lasting 40 seconds each with a few seconds in between
 to change speed.  This works out to approximately three minutes per test.
 Allowing, again, aa additional minute  to maneuver the vehicle onto the
 dynamometer and otherwise prepare it for testing, the total test time is about
 four minutes, hence,  the throughput rates for the ASM is the same as for the
 IM240.   Average throughput  for both tests is  7.5 vehicles per hour.  Assuming
 that stations operate €0 hours per week, 52 weeks per year, and costs are
 spread over a five year period, then equipment costs are spread over  a total
•of 117,000 vehicles.

    The optimum lane)  configuration for both tests is a three position lane
 staffed by thre« inspectors.  Consequently, as shown in "I/M Costs,  Benefits,
 and Impacts,* staff,  infrastructure and overhead costs  are essentially the
 same for both tests.   The) only difference is  in  the cost of equipment.   Table
 5.12.2  show* th« estimated per vehicle costs  for performing the ASM and the
 IM240.   The> eoats are) derived using the same  methodology and assumptions as in
 Appendix B.  Overhead costs for ZM240  and ASM testa are estimated by factoring
 the overhead fo* current centralized programs by the change in throughput.
 Equipment, and construction costs are  obtained by dividing thos« costs over
 the total vehicle traffic la a five year period.   Staff coata are obtained by
 dividing inspectors1  hourly wagea (36.00) by the average number of vehicles
 inspected in a hour.   State oversight costs are estimated at 91.75 per vehicle
 but could vary depending upon the intensity of the state oversight program;
 they would not vary between the two test types.
                                       112

-------
    Despite the difference between the coats of the equipment packages
required for the two tests,  the total cost per vehicle, factoring in all
necessary costa involved in a testing program, differs very little between the
two tests.  In a high volume test program the per vehicle cost difference is
estimated at 74*; the per vehicle cost for the ASM is about 5 percent less
than for the IM240.

                                      Table 7.2
     Coefe  Components  and  Cost per Vehicle  for  the  ASM  and  XM240
    TM24Q

    $2.40
    $1.75
    $1.39
    $1.71
    $9.12
    $16.37
         Inspection Staff
         State Oversight
          Test Equipment
Building Modification/Construction
          Other Overhead
        Total Cost Per Test
$2.40
$1.75
$0.65
$1.71
$9.12
515.63
                                     113

-------
 8 .    evaluation  of  the  Adequacy  of the  ASM  for  Enhanced  I/M
       Prograa*

       8 .1    Introduction

     The preceding chapters  show that the four-mod* ASM teat  is  not equivalent
 to the IM240 on a par-car basis.  Evan  if ASM outpoints are selected so that
 the same number of  cars are failed, they will represent a smaller portion of
 the fleet's excess  emissions,  and the cars will not be repaired as effectively
 as if the IM240 were used for  reinspection after repair.  However,  to some
 extent this loss of emission reduction  can be compensated for by improving
 other I/M program features  to  make them more stringent than would otherwise be
 required to meet the emission  reduction performance standard in EPA's rule for
 enhanced I/M programs.   Among  these other features are the inspection of
 heavy-duty gasoline-fueled  vehicles, the use of the ASM test for all 1981 and
 newer vehicles rather than  just the 1986 and newer vehicles which are assumed
 to be tested with the IM240 in the model I/M program, a higher failure rate
 for pre-1981 vehicles,  purge testing for more model years than in the model
 program, and more comprehensive tampering inspections.

        Whether these improvements are  enough to offset the loss of benefit
 from the ASM is the decisive question that determines whether areaa subject to
 the enhanced I/M prograa requirement can rely on ASM testing instead of IM240
 testing.  Also of interest  ia  whether it is possible to use the ASM and still
 operate a biennial  prograa. To answer  these questions, EPA examined annual
 and biennial scenario*  in which the ASM outpoints were made a* stringent as
 EPA believe* is consistent  with good engineering practice and the possible
.offsetting  prograa  improvement* were made as large as EPA considers reasonably
 possible.   If this  hypothetical best-possible ASM prograa cannot satisfy the
 enhanced I/M performance standard* then no ASM prograa can.

        Regarding beat-possible ASM outpoints, EPA ha* assumed that the
 failure rate associated with the  most stringent IM240  outpoint* for which EPA
 ha* provided eaiaaioa reduction credit* is the limit of good engineering
 practice in a* I/M  prograa. These IM240 HC/CO/NOx outpoints are 0.6/10.0/1.5,
 compared to- the> 0.1/20.0/2.0 used in the model enhanced I/M prograa.   Th* ASM
 outpoint* tha* aatched  thi* failure rat* in the full Mesa lane saaple  were
 0.40/8.0/1.•„  The**) ASM outpoint* can  be expected to  produce  a higher error
 of commission rat*  than the 0.6/20.0/2.0 IM240 outpoints, but  ia the interest
 of exploring the limit* of  ASM testing, EPA assumed that thi*  did not  make
 thea unacceptable.   KPA calculated MOBILES* I/M credit* for the*e ASM
 outpoint*,  uaing the saae basic approach as originally used foe the  IM240
 credit*.  We thea used MOIILSSa with  these credit* and appropriate aaaumptions
 for the offsetting  prograa improvement* to determine  the overall benefit of a
                                      114

-------
 beat-possible hypothetical ASM program.  Further description of this process
 follows.

       8 .2   MOBIUSa  Analysis

    The Z/M credits for the ASM test procedure were determined using the
 identification rate from  the Arizona test sample.  The laboratory sample was
 weighted as described in  Section 5.2.5.8 to reverse the effect of the
 recruitment bias.  The fraction of total emissions identified by the ASM test
 with best-possible outpoints and the ZM240 test with its standard outpoints
 were determined for that  sample*.  Using the ZM240 results for the Arizona
 sample, the ASM identification rates were converted to a fraction of the IM240
 results.  These fractions were then used in the Z/M credit model to adjust the
 ZM240  identification  rates used in MOBZLK 5 to represent the effect of the ASM
 test.

    For repair effects, based on current information,  EPA can only give the
 ASM teat the same repair  effect as the 2500/Zdle test procedure for HC and CO.
 For NOx, the ASM test was temporarily assumed to have the same repair effect
 as the ZM240 test procedure using a 2.0 NOx outpoint, the nearest available to
 the 1.8 ASM outpoint.  At this time, we made this temporary assumption for NOx
 so that the ASM program can be analyzed for all three pollutants even though
 the repeir effectiveness  problems found for HC and CO appear to be similar for
 NOx.  Unlike HC and CO, there  is no set of alternative repair effectiveness
 numbers available that could be used since steady-state tests have never been
 used for NOx control  in the past.

    Using the ASM credit set described above,  we proceeded to perform MOBZLESa
 runs for four separate Z/M program scenarios: a no-I/M run, an enhanced Z/M
performance standard  run,  and  two ASM runs, one assuming an annual testing
program, and the) other, a biennial program.  All four scenarios assume
national default inputs for the) local area parameter record - including
vehicle registration mix,  ambient temperature, average VMT, fuel KVP, average
 speed, etc. - and cover evaluation years ranging from 2000 to 2011.  Depending
on the osone> classification, statee must show in the 1993 SZP that  the  Z/M
program selected meat* the) performance standard in these evaluation years.

    Both ASM runs) were) identical, with the exception of the. above-noted
difference) in teat frequency.  The other program parameters assumed for the
* For convenience in calculations, MOBXUKSa Z/M credits for a particular test
and outpoints are developed starting with the total emissions identification
rate, rather than the) excess emission identification  rate used  in earlier
sections, to more readily display the) relative effectiveness of tests.  The
difference does not affect  the) final result.

                                      115

-------
ASM run* include* • program start year of 1983, a test-only network, and ASM
tasting of model year 1981 and latar light-duty vehicles and light-duty
trucks.  The ASM runs also assunad evaporative system purge and pressure
testing, and visual inspection of the catalyst, inlet restrictor, gas cap, air
pump, EGR, tailpipe lead test, and PCV system on all 1971 and later model year
vehicles.  Full purge benefits were given for ASM testing, since ASM purge
testing will fail virtually all cars that would fail the IM240 test.  A pre-
1981 stringency of 40% was assumed, along with a 3% waiver rate and a 96%
program compliance rate.

   Once these MOBILES* runs were complete,  we compared the results for the
enhanced X/M performance standard run and the ASM runs with the no I/M case to
determine what percent reduction was required to meet the performance standard
and what reductions could be expected from the annual and biennial ASM
programs modeled.  The results are shown in Table 8.1.
                                      11C

-------
                                 Table  1.1
      MOBXLBS* emission Factors  and  Reduction*  from ASM Testing

2000 No I/M
Enhanced Performance Standard
Maximum Annual ASM
Maximum Biennial ASM
2001 No I/M
Enhanced Performance Standard
Maximum Annual ASM
Maximum Biennial ASM
2003 He Z/M
Enhanced Performance Standard
Maximum Annual ASM
Maximum Biennial ASM
2004 He I/M
Enhanced Performance Standard
Maximum Annual ASM
Maximum Biennial ASM
.2001 No I/M
Enhanced Performance Standard
Maximum Annual ASM
Maximum Biennial ASM
2011 So I/M
Enhanced Perfoxaeme* Standard
Maximum Afuraafc.iam*
. _ «e-
voc
g/m Redux OK?
2.88
1.96 32.0%
2.00 30.5% NO
2.07 27.9% NO

2.66
!.<• 36.6%
1.11 31. i% NO
1.17 29.4% NO
2.52
1.53 39.2%
1.71 32.3% NO
1.7< 30.1% NO
2.47
1.47 40.3%
I. ft 32. <% NO
1.72 30.4% NO
2.3»
1.3* 41.1%
1.60 33.1% NO
1.68 30.9% NO
CO
g/m Redux OK?

22.23
13.98 37.1%
15.08 32.1% NO
15.79 29.0% NO




NOx*
g/m Redux OK?
2.27
1.97 13.5%
1.93 15.0% YES
i.9« 13.9% res

2.10
1.77 15.8%
1.76 1C. 3% YES
1.78 15.0% NO
2.02
1.C7 17.2%
1.C7 17.0% NO
1.70 15.7% NO
1.97
1.62 17.8%
1.63 17.4% NO
1.66 16.1% NO
1.94
1.56 18.8%
1.60 17.6% NO
1.62 16.3% NO
*  with t
ry assumption for NOx repair benefits, am described la text.
    By comparing the ASM results to the performance standard/ we conclude that
neither the •"H*"**^ annual nor the ITU if"™ biennial ASM program would meet the
performance standard for BC or CO  for any of the milestone years.  For MO*,
the biennial ASM program with the  temporary assumption for NQs repair benefits
meets the performance standard in  2000, but misses it for each successive
                                     117

-------
milestone, while the annual ASM program meets the performance standard through
the 2003 milestone.

    These NOx results include a caveat,  however.   The degree to which the ASM
NOx benefit in the table exceeds the performance standard is quite small,  if
the percent NOx repair benefit for ASM testing is anything less than 90%
(i.e., 13.51/15.0%) as good as for IM240 testing, the Maximum Annual ASM
program will not meet the NOx performance standard in 2000.  The corresponding
"actual values* for the Maximum Biennial ASM program in 2000 and the Maximum
Annual ASM program in 2003 are 98% (13.51/13.8%) and 97%  (15.81/16.2%),
respectively.  While EPA for the present reserves judgment on exactly how much
NOx repair benefit is lost with ASM testing,  (while we consider a test program
to further explore this question) it is clear from Section 5.3 that the  loss
is certainly at least 10%.  Thus, ASM testing cannot meet the performance
standard for any pollutant for any milestone date, and therefore is not  an
acceptable test in any enhanced X/M program.
                                       118

-------
                       9.  Appendices  Table  of  Contents

 Appendix A  Teat Procedures	A-l

 Appendix B  Emissions Data Listing for Vehicles Receiving Both Lane and
             Laboratory  Tests	B-l

 Appendix C  QC Criteria for ASM/IM240 Database	C-l

 Appendix 0  IM240 Cutpoint Tables	D-l

 Appendix E  ASM Cutpoint Tables	E-l

 Appendix F  Scatter Plots and Regression Tables	F-l

 Appendix Q  ARCO, Sierra, Environment Canada Data Analysis	G-l

 Appendix a  Estimated Coat of High-Tech I/M Testing	H-l

Appendix X  ASM and IM240 Credits for State Implementation Plans With
             MOBILES Runs	1-1

Appendix J  Emission* Analyser Price  Information from Horiba	J-l

Appendix K  Centrifugal Blower Price Quotation from. Combined Fluid
             Product* Company	K-l

Appendix l>  fae*-Paa» and Fast-Fail	L-l
                                      119

-------
Appendix A:    Test  Procedures

-------
     Test  Procedure  to  Evaluate  the  Acceleration  Simulation  Mode  and
     the  Emissions  Measurement  Capabilities  of  a  BAR90  Certified
        Analyzer  With  An  Integrated  Fuel  Cell  Type NO  Analyzer

    1.0    Objectives

 The  objective of this ASM test project was to collect data to compare the
 effectiveness of a four-mode steady-state test procedure as an alternative I/M
 test to the ZM240.  Emissions and canister purge  flow data were collected
 using the  following vehicle operating modes:

      -  Two Acceleration Simulation Modes (5015  and 2525)

      -  A 50 mph steady state mode at road load

      -  An idle mode in Drive

      -  An idle mode in Neutral

    These modes  will  subsequently be referred to as  the ASM test.  The same
data were collected for the IM240 test.

    The  lower cost  of the emissions measurement equipment is the salient
feature that makes the ASM attractive to its proponents.  Therefore, EPA made
ASM emissions measurements using a certified BAR90 analyzer  (for HC, CO, and
C02)  with an integrated NO analyzer of the fuel cell type for NO measurements.
For the IM240 a CVS-based emissions measurement system was used.

   Canister  purge  flow measurements were made with  the same 0-50 liter per
minute for both the ASM and ZM240.

   The  testing  was carried out  in two locations,  a  single I/M lane in an
official Arizona I/M station and at a laboratory  owned by Automotive Testing
Laboratories (ATI.).  Both were located in Mesa, Arizona.
    2.0   Phoenix Lan« Procedure

    The following is  a  description of  the  I/M  lane procedures.

      -  This procedure was restricted to 1983 and newer light duty vehicles
         with fuel injection, when available.   Carbureted 1983 £ newer
         vehicles were tested when fuel injected vehicles were unavailable.
                                     A-2

-------
   Pre-1983 light duty vehicles were tested only when 1983 and newer
   vehicles were unavailable.

-  Each light duty vehicle received:

      1. The ASM test that included the following modes in the sequence
         listed:

      -  ASM5015 with purge,
      -  ASM2525 with purge,
      -  50 mph at road load, with purge,
      -  idle test (automatic transmissions in drive),
      -  idle test (automatic transmissions in neutral) for the  first SO
         cars.  Car 51 and subsequent cars will not  get the 5th  mode.

These four or five modes will be referred to as the  ASM series.

       2.    An IM240 with purge.

       3.    A pressure test.

       4.    An Arizona State I/M test.


2.1    Procedure  Sequence

-  In general all odd numbered vehicles got the IM240 as  the  initial
   test and all even numbered vehicles got the ASM as the initial test.

-  Data collected included a number 1 or 2 in a field named "Test.Order"
   to designate whether the ASM series procedure was run first or
   second.  Discrepancies between the Test.Order entry and even/odd
   vehicle numbers are resolved by relying on the Test.Order entry, as
   this was ATI/s primary means to identify test order.


2.2    Measurement  Equipment

-  For the ASM series, a certified BAR90 HC/CO/COa exhaust emission
   analyzer was used to measure HC, CO, and CO2, with an integrated NO
   analyzer using a fuel cell sensor.  ATL only acquired a NO
   analyzer/BAR90 analyzer combination that provided second-by-second
   data for HC, CO, C02, and NO.  The data output for the ASM test went
                                A-3

-------
   to  3-1/2 inch floppy discs that included run number,  time (sec), mode
   number, vehicle speed, purge flow, NO (ppm),  HC (ppra),  C02  (%), CO
    (%), actual torque, required torque, actual horsepower  and  required
   horsepower.

 -  A 50 liter/min Sierra flow meter was used to measure  total  canister
   purge flow.  The flow meter system output was the cumulative second-
   by-second data for total flow recorded on the 3-1/2"  floppy discs
   discussed above.

 -  For the ZM240, normal measurements with the CVS system continued at
   the lane.  The data collected included time (sec), bag number,
   ambient measurements, NOx (grama/second), HC (g/sec), C02 (g/sec), CO
   (g/sec), and purge in standard liters.

 2.3    Procedure  Details

 -  An  electric Clayton dynamometer was used for both the IM240 and the
   ASM series.  The dynamometer horsepower settings for the ASMS were as
   follows:
       •  5015           HP - (ETW / 250)
       •  2525           HP - (ETH / 300)
       •  50 Mph         HP - Road Load

   The horsepower and inertia weight settings for the ZM240 were as
   normally performed.  The minimum inertia weight setting  (2,000 Ibs.)
   was used for the ASMS.

-  Manual transmission vehicles were tested in second gear  for both the
   ASM5015 and th* ASM2525.  The 50 mph road load mode used the top non-
   overdrive gear, typically 4th gear on a 5-speed,  4th gear on a 4-
   speed,  and 3rd gear on a 3-speed.  Drivers used a lower  gears for
   vehicles that were lugging.

-  The engine was shut off prior to the IM240 and the ASM5015  (as will
   normally b« don* by I/M programs to connect the purge meter),
   regardless of which procedure was performed first, and restarted just
   prior to initiating these procedures.  The engine was not shut off
   between ASM modes, and the vehicle was accelerated from  the current
   mods up to the next mod* speed, without first returning  to  zero.

-  The ASM emission sampling period and the canister purge  flow
   measurement period were as follows:
                                A-4

-------
         1.  Each ASM mode was initiated after the vehicle  speed had achieved
         the nominal speed (15, 25, or 50 mph, and 0 mph idle) ±2 mph.  Once
         up to speed, emissions sampling of one second average concentrations
         continued for 40 seconds.  Emission scores for HC,  CO, C02 and NO
         were reported for each second.  Emissions scores for the first 10
         seconds of each mode were ignored to allow the dynamometer to
         stabilize and to allow for transport time to the analyzers.

         2.  The purge flow reported was the second by second cumulative  flow
         over the entire ASM cycle, including transient accelerations.  The
         nominal acceleration rate was 3.3 mph/sec., with a minimum
         acceleration rate of 1.8 mph/sec and a maximum of  4.3 mph/sec.   The
         table below lists the minimum, nominal, and maximum acceleration
         times used to accelerate from one mode to another. For  example, the
         table shows that the time to accelerate from 25 mph to  50 mph should
         be 7.6 seconds., but can take as long as 13.9 seconds.,  and as little
         as 5.8 seconds.  The zero to 60 mph time is provided  to  indicate how
         the specified acceleration times relate to a commonly  known reference
         of vehicle performance.  ATL used a video driver's aid with the
         nominal acceleration rat*.




Minimum
Nominal
Maximum

Acceleration
Rat*
(mph/sec)
4.3
3.3
1.8
Time
0-15
mph
(sees)
3.5
4.5
8.3
to Accelerate from- to:
15-25
mph
(sees)
2.3
3.0
5.6
25-50
mph
(sees)
5.8
7.6
13.9
0-60
mph
(sees)
14.0
18.2
33.3
         During the accelerations between modes, the dynamometer load setting
         did not exceed road load.  This was specified to enhance the
         opportunity for canister purge during the ASM accelerations.  The
         combination of the ASM load and the base 2,000 Ib. inertia may load
         some, vehicles to heavily to allow purge to initiate.
    3.0   Lab Reeruitmant

    Light duty vehicles that received all of the lane tests (IM240, ASM
series, and Arizona I/M test), were recruited  for testing at ATL's laboratory.
Cars were categorized as passing or failing using the IM240 cutpoints  in  the
table below:
                                      A-5

-------
               Phoenix Lane IM240 Outpoints for Lab Procurement
Model
Years
1983+
HC
0.80
CO
q/mile
>15.0
NOx
q/mile
>2.0
    The following table provides the laboratory  recruitment goals for the
pass/fail categories listed as a percentage of the total number of  cars
recruited to the lab for this task.  The initial recruitment  target was 100
vehicles.

           Phoenix Lab  Recruitment  Goals Using Lane ZM240 Categories
Model
Years
1986+
1983-85
HC/CO
Pass
15%
10%
HC/CO
Fail
15%
10%
NOx
Pass
15%
10%
NOx
Fail
15%
10%
    4.0    Commercial  Repair  Recruitment

    Owners of vehicles  that failed the Arizona  I/M teat,  and received and
IM240/ASM series, were offered 950 to return to the lane for after-repair
tests.  These vehicle owners were only offered this incentive if they  refused
to participate in the laboratory testing program or if their vehicles  were not
needed for laboratory recruitment.  Recruiting vehicles for laboratory tests
was a higher priority than for commercial repair participation.

    The owners were informed that they must  return with repair receipts
indicating repairs by a commercial establishment with itemized labor and parts
costs to qualify for the $50 incentive.  ATL  included either the original
receipts or copies in the vehicle test packets that ATL provided to EPA.  In
addition, ATL provided summarized comments and data for these vehicles on
electronic disk.

    Vehicles  returning  after commercial repairs followed the  same procedures.
    5.0   Lab  Procedure

    The lab procedure is summarized in Attachment 1,  so this section will only
add explanations to the procedure listed in Attachment 1.
                                      A-6

-------
    5.1   Two  Group*

    The vehicles recruited to the lab  were separated into two groups:

 1.    Those whose initial lane test was the IM240  and were  repaired to IM240
      targets.  For the vehicles in this group,  the ZM240 always precedes the
      ASM series (see Attachment 1).

 2.    Those whose initial lane test was the ASM series  and  were repaired to
      ASM targets.  There were not enough data to  set ASM repair targets, so
      IM240 targets were used for both groups. For the  vehicles in this group,
      however, the ASM series always preceded the  IM240 (see Attachment 1).


    5.2   Repair Target*

    The repair targets were to achieve 0.80/15.0/2.0  on the  IM240  for both the
ASM-targeted group and the IM240-targeted group.  Initially, repair  targets
were to be provided to ATL for the ASM targeted group to replace the IM240
targets.  However,  due to time and data constraints this proved impossible.

    For the  initial  repair attempt, the mechanic was  only aware of the lane
IM240 score for both vehicle groups  (initial lane  test: ASM or IM240).  For
subsequent repair attempts, the mechanics were only be aware of lane and  lab
IM240 scores.  FTP scores were not provided to the mechanics for either group.

    Repairs  were limited to 91,000.
    5.3   Laboratory  Te«t  Equipment

    Due to time and financial constraints,  EPA was unable to develop lab ASM
capability.   Th« IM240 and FTP were measured with a CVS system.  Modal or
second-by-second CVS capability waa not available at the laboratory.
                                      A-7

-------
                                   Appendix  A:   Attachment  1

                                        ASM/IM240 Lab Procedure

                                    Revision Date: 10/21/92
                                    Number tested *
                                    Recruitment:  1983+fuel injected only.
                                    Repairs: Get IM240 Indolene to
                                    .8/15/2.0. The mechanic should only be
                                    aware of IM240 scores for the IM240
                                    targetted repairs. $1,000 repair limit/car
                                    - catalyst if necessary, aftmrkt preferred.
                                    Develop explanations for any IM240
                                    failures that pass FTP, while veh is still at
                                    lab.
                                                Tank Fuel
                                             On-Road Warmup
                                             Tank Fuel IM240
                                       9.0 RVP Indokoe As-Received
                                          LA-4 Prep cycle @80°F
                                                No Diurnal
                                               FTP Exhaust
                                               No Hot Soak
                                    IM240 Indolene (with purge if available)
                                   Repair to get IM240 Indolene to
                                   .8/15/2.0, The mechanic should only be
                                   aware of IM240 scores - not FTP scores,
                                   •UI0 QQiy DGTZQRB IDilltllZUIXl FCDftlTS
                                   necessary to achieve targets.
                                   Report After-First-Repair Indolene
                                   IM240 regardless of outcome. Mechanic
                                   will only be aware of lane IM240 score
                                   for first repair, not lab tank fuel score.
                                   Continue repairs if necessary. Dont
                                   perform FTP until .8/15/2.0 is achieved.

                                   9A RVP Indolent After-Repair to
                                   IM240 01/15/10
                                   3 LA-4 Prep cycles® 80°F for all
                                   vehicles
                                       Top off to 40% fill-don't drain.
                                               FTP Exhaust
                                   IM2U tatatene RM1 (w/purge if
                                   available)
                                       Stop repairs even if failing FTP.
Indolent Lane Tests For Vehkkf
Whojt Initial LSUM Test Wss IM240
Indolenc Ltnt Testt Procedure for
VeUdes Whow Initial Lant Test Was
ASM Serial
          On-Road Warmup
        Lane Indolene IM240
            ASM Series
          On-Road Warmup
            ASM Series
        Lane Indolene IM240

-------
 Appendix  B
Data  Listings

-------

  w
  0)

  o
 -,-t
 x:
 o
 0)
 an
 CO

 co

•H
 (0

s

 0)

3
 (TJ
EH

.U
 c
•H
 O
 a


o
      53
        8

        %
                 OOOCMOi-IOt-IC4Oi-«i-IC4<-l<-IO>-li-IOOCMOOMOOCMi-«i-ICMOOO
            ,-ir-i««-icMoartint~f-ii-ir>

            OOiHOOOOO«-IOOOOOOr-«OOOOOOOCS|OOi--liHOr4i-IOOO
             8
                                                                                                                        .   . «o r-
                                                                                                                                                     (N

                                                                                                                                                      I

                                                                                                                                                     03
3
           c<it-ii*»eiic«i«*c
           oddodoooooooooooooooooooooooooooo
                o>r»

                Oi-«OHr4r4i-lf-l^r4f-ltHr4<^i^f-«
-------
 CO
 0)
 r*
 o
 •H
 £


 3

 TJ
 CU
 j-i
        3
                  mONCM<*»vor»cor»o»oocot-if-4f*r»r-^<r><»>>nvofM0»o*o*a
f-tcNNvvOi-tveoom

O«*OOO>HOf-1
<-\
 <0

5

 0)
l-l
X3
 m
H
                                                                                         •»  «  u,  CM  W.


                                                                                         «  rH   '  CO  °*
          CM
        1

        S


        a
                                                                                                                                                                 en
                                                                                                                                                                  i
                                                                                                                                                                 CO
2
                    cMO»tHO»^t»a»p«co«»»1_vo»cort(o
                                                                                inCOCOM'*'
                                                                                                                                       CO  CM  O»  CM
                  Aj*j'Oij'OT3tJ>OT}'O^4JW4J4J-a-OAJ4jjo-ouT3'O'Ofc»*jjO4j'a'aiOw


                  ddodoooooaoooooeoooooodoeeeoeoooo
      §
            ((MCMCMCMf^(MCMCMCMNCMCOCOm(0«*>
-------
 w
 0)
iH
 o
•H
ff,
 0)


T3
 oic>if*»nf><»>o>io>ir-i<-i»«-i<-iinin»-i.-4evj

            dcsioodooodoooooooooooooooooooooooo
   01
   8
 (0
 >
rH
 (0

5

 (I)
rH
£1
 (0
£-•
       \

       C
       I

       M

       94
              OOOOOOOOOOO°f-«OOOOOOOO
                                                                      i/) l/> P»  U">  U")    CNJ ri>r'r«coa>ooaoeo>4r4r4C4(N
                    rHfHi-li-«<-4*4<-liHr4<-l«-lr-lr4i-li-l^rHr4f>lf-t(^r4i-«fHr«4f-lf-«MCi|O

                    fHf-lr4i-tr4r4i-lf-lr4r4<-t>H<-lf-l<-l<-4r>oo«o«o«A9>o\o<-«(M«0«mvmioininin«vaaot0iOf<4(N
        !
            Mo«oiPfo«n«
                                                                                        in 10 in  m  in i/)

-------




3
I










Bi
i
a
o

'






i
1
II






8
4
•T
|
1
I
I
I
!






i
i
i
i
0

j
1
1
1
(
o

2
o
(
1
CM

I



9
t
;
|
£


























g
s
5
•
8
c
c
|l
y
:
•
I
*
*
C


?
s

£
2

g
W

a
9

jj
*
8

y
V

j*
1
3
t
|j
2
4*
5
f*
4j
S
3
*
41



CM in envo vo r- o vo * M VOM cocn MOO OM co * MM <-ico
vo cn MVO * eo m vo OM in vo coco cocn in o vo o m * co<-i
OO COO MM OO MM MO OO MM i-IM OM OO rHiH

* * voen co» o en r* m *"! * «*** * eo voo '"") r~ COM o M
J^co COM r- m vo* coen 2 *"* co "^ covo men r?"1 coco cn co


in co voen * en * M mm men eoen coen enen r*vo r-ien ooo
COM coo MM cncM MM enM * M MM coo m M MO CMO
MO OO OO OO OO OO OO OO OO OO OO OO
^^ ^^ ^* f^ QO ^^ i*4 ^* ^* ^0 ^0 ^^ ^ft *^~ ^^ GO ^* ^J*
•M * * ^^ * * f^ * * * f\ * rt\ * * * * * «M * * * irt *
^^ ^^ ^^ ^^ ^* f^ ^^ ^^

cn r-i o vo co co vo r- m M vom mo com oen en co cnm MOO
r-M MM MM oen *M MO COM MM mo r»M oo voo
OO OO OO MO OO MO OO OO MO OO OO OO
mo Mr* * en o m r— vo moo oo* vor* covo r*>en o* or*
en* in oo CM m CM r* OM enM *M CM co *en cn* cn* men
OM *O CM M MO MM MO OO CMM MO OM COO MM

« oo o * * r- 1 « M « «-. cn « vo eo * cn * ^ m M oo « vo
P{M Mr* ** <5* vo* J^1^ ** MM r»o *o MO ***
*en o* *M r-o m«o ** *M MVO oo oo mr» en* vo*
*M MM co en r- cn * M coo C*>M MM mo COM MO r*<-t
MO OO OO CMO OO MO OO OO MO OO OO OO
en m MOO *m vo r» r» M * en MCO voen ooo VOM co M oovo
men co eo OM M en OM M m COM M* men en 00 r- m coo
OO mo MM OO MM MO OO MM MO OM OO MM

co m * n « cc
MM MCM MM MM M)W MM MM MM M«M MM MM MCM
OO OO OO OO OO OO OO OO OO OO OO OO
22 22 22 22 22 22 22 22 22 22 22 22
MM MM MM MM MM MM MM MM MM MM MM Mfi
com com om 0*0 en m 000 o co om mm r»m vo en eo*
MM MM m M mo MM en M enM oo M OM MM oo CMM
OM OM OM O M OM OM OM OM MM OM MM OM
MM MM MM MM MM MM MM MM MM MM MM MM
n cn cn en cn cn cn en cn. cn on 01 on cn cn en on on OK on o^ o^ en c
5 {/j *"J *jj •jj t/\ *2 co S co T! Co c/i co *^ co co ^« c/
14 m. CO CQ w] £^ Vt A? CO ZZ CO 22 ^O K^ ^O c« ^O ^O CO £^ CO K^ vi «
H^ HH H*^ H"^ H*^ H"^ H^ H^ HH H^ H*^ H^
M» vom cnm mM vo vo r*« MM enm OM r*r» min oo*
C*»M on* mM mo MM »M mm *M M* P«M onm oor-
vo cn VOM r* cn r- tf» r» cn r* o r* o r- en oo M vo on com vo ^
MM MM MM MM MM MM MM MM MM MM MM MCM
OO MM »» MOT vow c»t» encn oo mio cnon MM *•*
mm mm mm mm mm mm mm vovo vo vo vo so r» r* r* r-
MM MM MM MM MM MM MM MM MM MM MM M <->
mm mm mm mm mm mm men com mm mm mm mm
























i
ca

























-------
   CM

    t
i
m
I
            co  CM  vo  vo  vo  m
            o  CM  *  c*  in  m  oo

            fH  rH  rH  O  •-*  t-l
                vo  o
                                 vo
                        ^  co
                                      in
                vo  ro  OD   vo  v
                CM  CM  »   <-i  m  oo
            O  0  0  O  0  O     •»
                        O  O  r-t
           in  oo  m
           m  oo  oo
                         en  vo m
                         vo  CM vo
           r-l  O  O  O    oo  o
               O  O  O  t-4  rt  CM
                        -»
           «  a
           S  S
                                «  -
                                 s
                             S  S
                       o  o
               •O  fl
           nee
           0  0  0
                         M  e   «

                         o  o  o
               cn  vo
               O  fH
                                    CM
            CM  CM  CM   CM  CM  CM  CM
            O  O  O   O  O  O  O
                CO  CO   CO  CO  CO  CO
            en  en  en   en  en  en  en
           H  K  H  fr«  H  (t  frt
           n  o>  M  co  co  o)  co
           *   K  H  H  H  H  H

               o  oo  in  o  en  o
               in  co  r*  eo  t-«  co
           CM  CM  »  *  »  m  m
               CO  CO  CO  CO  CO  CO
           r«  en  o   ,-»  CM  »  m
           in  m  vo   vo  vo  vo  vo
           M  CM  CM   CM  CM  CM  CM
           O  CO  CO   CO  CO  CO  CO
                                                                                                                                                         in
                                                                                                                                                          i
                                                                                                                                                         en

-------






gj
I
XM240



4
w
4



|




• Scores
3
•
i
<
u

«
M
(
1
M
I
a

•
4.
•r
a











i
1
•
8
2
«
3
2
M
g
s
ft
r;
a
j

a
in
|
X
8
a

u
1
0
j
Q
tt
I
1
»r * oro eo » or» vo  " * ™ <* oo M v vo en en t-* CM M
™ * 2 m com £ m *<*» M ^ 2 ^ ^ <*> *- * N ™ "* * rr)'r
p- co oom in vo MO M vo o en <-i vo Oi-i vo M mm p> r- co M
rH O «T |H r-im r-( rH VO M t-  coo r-i«o r»co CM i-t o m
o» c»o om OM cMt-i p-rH ini-i coo mo c\im inm CMCM
I-IO rH O rH O rH O CO O IT) O (NO OO rH O OO OO OO
mo or* v (N vnm o ve on vo ov vo r- i- m r» o CM  i-i i-i«o t- vo rn m voo oo vco coin covo
OO rH rH CM rH O rH OO OO irt O MO MO MO rH rH r- rH
•vo °!^ incB °. r^ * r* • r» ^rHmmmco^m coco ^M
5°* In0* co» IN" S01 <5N rHrt «oo coo i-»o com mr-
om * p- m P- enr* Mm voo co vo eno I-H m cn« inert mm
O>*> COO mm M rH O rH O rH r- rH O P- O MO OO rH O OO rH O OO
VO rH VOOO M » O M coo vo » r- rH in CM
OO M rH rH O Ov4 OO OO lOO MO rH O OO rH rH P-rH
•^ en ^,4 co oo "Jen ®r4 °. w "iMinr-^enrnvo f.,H ^rn
J3 p« p» o»« Jg«n ^in 5® ^r^m^o^o^ S* ^^
OM OM voo> mm on m n0* P>V en« p- co eft M «r> OM
«« Oft M  VO OT» » in » o P>** r- co in eo ao o p- p» r-trn
OO rH M rH O rH O MO MM CM rH MO MO M rH MO O rH
rH M rH M rH M MM MM MM MM MM MM MM MM MM
MM MM MM MM MM MM MM MM MM MM MM MM
en en en en en en en en en en on en cnen encn en en en en o\ en cnen
MIT M£! MM M vj MM MM MM MM MM MM MM MM
Hn ** 55 H H H 5 ** ** H ** H H ^ ^ ^ fr* *•*• t4*4 H *^
w M w ZJ cn o» co 2 coca coca coco coco coco coca coco coco
P. *i vo vo mm * « OM M vo OM COM COM mr> p- co mr»
o vo voo o m MOB vo ao vo m COM mm en m «m moo p>co
en«*> enM o* OM M M»
MM MM MM MM MM MM CMM MM MM MM MM MM
mm cnen oo mm coco oo vovo r- r- co co oo MM MM
r»r» r»r» co co eo eo coca en en cnct cnen en en oo oo oo
MM MM MM MM MM MM MM MM MM MM MM MM
mm m m m m m m mm mm mm mm mm mm mm m<*









^o
i
CQ










-------

















(0
m
a
£
flj
a
0)
os

o



M
, t
4J
c
o
CJ
*
M




















,
•3









8
I
M
O
i










«
u
1
0i
B






I
^

H
3
3
i




i
c
t
w
*
1
c

u

c
c
a
N

C


H
|

0
a
I
0

























{
*
1


8
u
4
3

i
§



a
a
u
1
2






U



g
«

8


y
M

M
41
0
t
^
o

41
'
m
I
*n
1
cMen * fi COCM p- co mm * m mvo mft Ofi o oo eoen men
m * oco p- vo i-tm vo* oen mvo om co p. p- * CMCM ocr\
m ft r-ift oo mo oo *»• o oo CMO r-tfi ft o mo mo


•«• en * * ^ °*» vom ^ * P- co oen vOft  ft OCM * fl
OO r-l O r-IO OO OO OO OO OO OO OO t-tO OO
^ 22 "2 ^0 IMO 22^-'op^°*o £° *"


r-i CM enm CON oo *N mm en CM men *en mm mm voco
i—co ftco coo en ft eno i-ifi mft coo meo eno en o oo
oo mo ft o oo fto oo oo fto oo oo oo f-to
P-N p-m fift P-N coco mm p-m *en CM* co* p- vo five
mm en m *vo en ft mco ft o> 1-1 o enm CM m p- * TO o vo
*fl Oft OO NO OO mo flfl CMO CMfl ft O *O *O
^cnmenf

• t ^^ • • ^^ t • • ^^ • ^^ • « ^^ ^^ • • ^^ •
*4 P* P* fi en ^^ ft CM ft fi ^^ fi ft

*p> oen VON CON N* m* *co o ft men mco m* *co
p*eo NN ooo Oft *o fifi m ft o ft mp* oo ft o oo
OO COO ft O ftO ftO OO OO NO OO ft O fl O flO

p»o CMfi ft vo ft* p*o ft m mco ON CM vo vop* cove *m
ftm en ft men CM m mm ft o enp* mm oco mm CM m com
*fl O ft OO ftO OO flfl OO CMO CM ft fl O COO r-IO
^
~ p* ^ r^ r^ , %M t ^9 w» p» rn ** % \o p* ^ t 'W t ^F *o \o
* * f* * CO • ^j • f^ • • • • • _^ • • • _.. • ^ • • •
^^
vom m* GO N o* coco p-m mo mm or* N* oo mm
en p* co * co 01 * co * co co co co co fi fi co vo N c^i co CM P* ft
OO COO .NO ftO OO OO OO ft O OO ftO ft O OO
aa aa a c an c e ce c c aa c a ee aa aa
flfl flfl NN f»t* NN NN CMN flfl NN NN flfl flft
OO OO OO OO OO OO OO OO OO OO OO OO
SIS! SISI 3 $ «I9 9>£! 51 51 9*1 «l«l 53 9 99 99 99
op> cno m ft CON coo *N m CM m* m * eo r» en* fifi
m fi o ft o ft ON o ft o ft Oft Oft o ft Oft ON ft CM
ft CM NN NN NN NN NN NN NN NN NN NN NN
NN NN NN NN NN NN NN NN NN NN NN NN

fl £1 fl N fl N fl £1 fl N fl N ft CM ft N ft N fl £1 <-l|* *••**
H R HH HH H R HH HH HH HH HH N SJ H£ H£
w rj COM co to car* con coca coco coco con co r* wzf COL
H^ HH HH H ^ HH HH HH HH HH H*^ H^ H*^
r-t en en en * eo p-m o>-t N* com en ao vo m *o fto meo
vo vo fi m CM vo co* co^o mot voen vo ft p- ft NP» * eo coo
Ntn co* co* co«o co* co* m* mm coto *io *ve *vo
NN NN NN NN NN NN CMN CMN NN NN NN NCM
cn
-------






n
i
XM240



•
u
I
a
£
•



!
I
4


i
i
1
i
(
0
:•
41
•f
n
<
c.

«
u
(
a
N
t
a

•
4-
•4
•
!











i
\
»
s
i
»
%
n
«
s
B
|
z
a

s
z
s

M
|
2
2
*1
2
«
1
r- T mo a>o co "T en CM en CM vo vo m o p» vo en vo en CM n IH
o co m co en » m* co m en *• vo co o m en en oo vo m o co
r-tO » v«n
oa> ^o r» moo H<-I CM CM oo 1-1 <*>m
OO OO OO OO OO OO OOOO OO r-t O OO r-l O
o\o CM oo •» rH oo CM i- co (M<-i oo m ro o *-tcn m p- vor> mi-i
«"O\ m»-« f>m cn»  CM oo
oo o f-t Oi-i oo CM o>-« «o vo o CM o r-t t-t m CM en i-i en«>
OO OO r-IO OO OO OO OOOO OO -1 O OO  r^oo oop» P»CM com cnmi-ico men vo o *p» ovo
p»eo ft CM en co •-• CM * » vocn cnromco CM vo o CM co i-« en CM
OO r-l ft <-IO CMO OO r^ O r-IOi-IO CMO f-lfl <-IO OO
mo p- vo oo« m oo ^-i eo p- vo * oo co m co vo rH m "!<-• vocn
i-ti-i CM*M m*^ «orM «n« »o ^O«O»H co^ mt-i JJ^ 00*0
TO p»«o en m r»r4 CM m vo<-i T CM en p» oca moo co CM voo
CM co CM CM  CM  W *J T3 -O MM T3 T3 -O T3 4J aJ 4J 4J 4J *» W AJ
ec c c a« n« ee «n c c c c « « n n n n n n
oo oo oo o o oo oo oooo oo oe oo oo
22 22 22 22 22 22 22 22 22 22 22 22
MM MM MM MM MM MM MM MM MM MM MM MM
M *«n m» VO«VOM p»«» r«r» «or» com
r-l CM i-tfM MM M(M <-• CM M(M MCMOCM OM OM M(M CMCM
CMCM CMCM CMCM CMCM CMCM CMCM CM CM CO CO CO <*> CO «n CO CO COCO
en en en en en en en en en en en en cnencncn en en en en en en encn
^2 ^£1 "^£1 «^£! ^2J ••«£? M£*MIM MCM ^CM «-»fM <-«M
"B "S -s -£ ^s "^ »g «^ «« «s -^ ^^
H*^ K1* H" H** H •* H** K^HK HH HH HH fr»H
o*« CM«O encn p» o *OM en en co -i o m mmmwrncMcor*
en» enp> en vo or* MC* cMr> »oocoo o «<0 m «o m *o m vo mvoooM coo aai-« en •-» CM«O
CMCM CMCM CMCM CMCM CMCM CMCM CMCMCMCO CMCO CMO> CMCO COCO
encn oo CMCM com * » mm r« p- co eo en en oo CMCM P» P-
CMCM CMCM CMCM CMCM CMCM CMCM CMCMCMCM CMCM CMCM CMCM CMCM
cof> row «•>«•> «*>«•> coco coco cocoroco coco coco coco co
-------
2!
2

    8
            (M  10

            (M  en
                    vo  m
                    p.  «o
                         m  o
                         m  M
                        O  O      i-lfl
            «*>  M


            o  o
                        oo
                         o»  co


                         C3  O
4*  VO

00  M
                                                CM  O
                                                in
M  O


  •    •

r-(  O
            m  o



            r-l  O
                        OO
                                    r-IO
                                                MO
            oo  m
            co  en
            oo
                   oo en
                   vo p.
                     •   »
                   o o
                        VO «O



                        rl O
«-l  P»

CO  O
  •    •

co  o
           Ot-i
                   r-l  10


                   o*  o
                        VO CO
                        r-l CO


                        M O
en  rH
*•  o

M  o
           in p-
             •   •

           OO
           2"
           i-lO
                        p» en
                        co p»
                               vo  m
                               o  o
                       OO      f-lt-l
                       M o

                         •   •
                       o o
                               8
                               vo  en

                                 •    •

                               ~4  O
                                           m
                                             •

                                           M
                                    O O

                                    ^^ ^H
                                      •   •

                                    M O
           •o -o
            G  e
           M M

           O O
                   «   II



                   O  O
                         a  «
                        r4 fl

                        d d
??
M  (M

O  O
           O  rH
           M  M
           O  O

           cn  en
               m
               S
           o oo
           m<-i
           CM m
en  cn
m  in
M  M
co  co
                   M  N
                   •-•  N
                   M  W
                   O  O
                        f-l M
                        M M
                        O O

                        en 01
r»  o
t-t  f-i
M  CO
O  O

en  01
                                    <-4  M      r4
                       co
                       vo vo
                       MM
                       CO CO
                                    co  co
                               MM
                               CO CO
                                                    H
                                                COM
                                                HH
                                                co  co
m  m

M  M
co  co
                                                                                                                                                        I

                                                                                                                                                        03

-------




a
3
*







a
J
c
ti
e
M







4i
a
i*
3£
H

J
*
•r!
j
4





1
j
i
s

'
I
i
»








i
c
ia
4
•;
1
!
c

i
«
0
M
Oi
1
a


j
]
c



J
1
*
^
d
M

Jl
jf
j
1
S


















J
1
1
1
t

(
c,
_
i
t

S
j
»
„
c


s
9

g
2
8

3


8



ij





o
9

M
•
5
0
,
3
j
B
^






»

VOOP- P- «-i mo on o rH co en p. ooenen om CMP- CMP* om
m vo o rHp- oo rHvom vo*m rH m co vo m voon oen CM c~
OOO CM rH OO CM rH rH OOO rHrHO OO OO rH O CMCM

•CM* rHcn °. o p- m * °*. °. T * ^ «n "! T ^ CM m. m. * rH
* ' " * ^ la * * * _* .^ «• ift t* * *^ O * ** • *
^4 ^^

VO CM rH *rH VOm *OrH *CM* VOrHCM CM in VO rH *rH P- VO
OOO OO rHO OOO rHrHO rHrHO rHO OO (MCM OO
oo0 00 go ooo ooo ooo oo °,0 oo o0
. T, os rHoi • *"J moo ... ."1*^ ^^ • oo ^T "I1"1
|g}£p-' CM'M* *§{ on** SeoS VOCMO mcM ^^ OCM 2®
rH rH
00 rH rH O rH Ul VO CM VO rH 00 CJV Cft *MCM CM rH OO* O Vrt rHCM
MP>rH OM P»O OrHOn VO rH VO mOP> OOn m rH rH CO OrH
• •• •• •• ••• ••• ••• •• • * •• • «
rH O O OO CM rH rHrHO CM CM rH fO CM rH CM O Irt O m* rHrH
oo ro CM moo rHO vo en rH o en r- rHooo oo vo CM vo en CM I-HOD
* VO * *P- rHM OOinO mO* *OrH CMO P«P» rHM P»OO
OOO CM rH OO MMO OOO rH CM rH OO OrH rHrH OO

oonon vo p- on rH I-H ao rH o on m *vo* co o vo m oen vo o
o m o OM VOM moo * o m vooi-t o en CON o vo *CM
*•• •• •• ••• ••• ••• .• •• •• «•
rH O O OO CM rH rH rH rH CM N rH OMN NO l/> O VO m rH rH
floon rHN rH oo p» vo o rH * onvoon rH ono o*
rH O o rHO ° Q OOO OM* NON *o ° O (-00 OrH
OO OO O OOO C* f~ «-t NOO P> rH O OO

lA ^5 IA lA ^5 IA INI r\i ^ n- O IM «^
**^ m rH **^* OOS* 5*2 rHrHrH 3^« *2 ^«J N*
*^
mr* MM oo m o ao rH m o*m om M cove vo p.
rH On ._ OO * f3 m VO O rH O rH m m rH OOn P- _, m rH *O
• • ^^ • • • ^^ • • • • ft * . . • • • • ^~ • • •
NN OO VO rH O O O O MOO *O 00 rHN rHO
oncnorn o r-r-co

T3^^ ^ ^ T3 "0 «U^ on o r* on o p- on o r~ on r* on p» on p> en
rHrHN ft ft ft ft rH rH N rH rH N rHrHN rHrH ft ft ft ft rHrH
HHH H H H H H H H H H H H H H H H H H H H H H
CO CO CO COCO V» (0 CO CO CO CO W 10 CO CO CO COM COM (OM COCO
00 VO v 000000 OO OO 00 OO rHrH
on en on oo * * p» r» r» on on on en en en mm vo vo vo vo eo eo
ao ao oo vo vo vo vo P. r» r» r»r»f P-P»P» ao ao oo oo onon onen
rHrHrH NM MM MMM MMM NNN NM MM MM NCM
rH rH rH rHrH rH rH rH rH rH ft ft ft rH rH rH rHrH rHrH rHrH rHrH
























0
1
CO




























-------




















(TJ
(TJ
Q
•H
m
a.

r-t
•H
O
M
e
g
o
(j

^
M























2
9







a
8
o
3
0
*
1







u
•
"

Jg
^
H
M
H
it




|

1
3

[
I








•
c
(1
a
4
•r
a
1
C

2
c
£
CM
e
j


0
4J
3
j
3



•
a
*
«
3
M


1
12
*n
3
















i
£
•
§
il
p
o
1
fj

1
4
N
s


rj
*

£
SB

8

0
a

8



a



8

,i
*
M

3
0

u
2

•

1

^
I
«•


mm i« m I-ICM * vo CMVO <-ivnoo VOCM en oo r» CM mm vm
moo of cMoo voen mi-i 01-10 IH r> menr>m * in en *r
OO rH CM r-IO r-l O mm r-l iH l-l CO r-l OOOO OO OO


om COCM mm ^ CM m* °°.  o CM T oo
rH en CM CM v CM o en vo in in o r* m vo en rH ^ oo CM CM rH rH
OO OO OO i-l O OO r-l i-l r-l r-l O •-! r-l O iH OO OO
00 go °,S m S °, g oSS 0° ".*.«« 00 §0
•• CM* "O VOfH ** O^OCO VO* **f-'{J^ '* p«!*
^ iH __J CM CjV _A -— _. CO VO ,. , 	 . -.. 	 . PO ^* CM w» CT\ ^*4 O !» ^"
^ r^ ^ ^* CM r^ CN " . . CM ^ ***
cneft ^^> ro«-» p*^ 01 GO CM o in cnv CM m co v r*cM CMP-
r-* o ^*f*4 eneo co vo CM o ^* CM co COCM CMCM^*^* oo CM •-*
• * •• •* • • •• ••• •• •••• •• ••
r-io oo oo CM o r-i r-i in m m CM r-i m * r-i r> oo oo

oo CM v p» vo rH en om oomen mo o m m CM co + rnm
CMVO in v p» CM meo mm vo vo vo CM i1 CM CM r-i •-! r~ oo CM m
OO O r-l r-l r-l rH O mm r-l r-l rH * CM O O r-l O OO O r-t

r^N 3 c4 2« S2 S^ ciicQ 2^ S^S^ ^2 Jjjer!
VOCM r-i en vo oo mo mi-t CM •* r> CM » oo m •« oo enen «* c~
rH rH * 1-1 OO CMCO en rn; o r> oo » en eoeoeov oo CM rH
• • •• •• •• •• ••• •• •••• •• ••
•HO OO rH r^ m O rH rH VOmm m rH m « rH OO OO OO
o^ 00 °.o ^o ovo »r»o com CM CM m eo ooo oo
o * o o ocMOOrHommomcM

r-i ° r~ CM ^ vo * vo CMOO en «H ^ °° m mm^m CM



CM CM lH •» rH i-l VO (- VO O\ - *f in VO OO P- Cn rH OO rH rH
o o vo m v en tH enen cn^cM * CM vo en eo m ^ o voo
O rH O OO i-l O OO O°tH OO OO VO iH en O iHO

men ^ ° * '"o °*N  ^ en r^*t-»'^ *° OCM

ce ee «« oe cm m m m c c m m « « me mm
CMC* CMOi r-lrt CM CM CM«H H»Hi-l CMCM rH^fH^ t-»CM i-lr-l
* • •• • • •• • • • * • • • • • • • * * ••
OO OO OO OO OO OOO OO OOOO OO OO
22 gg gg 2g gg ggg 33 gggg 33 33
r^ F"T p^ ^^ i^ ^^ ^^ p1^ f^ ^^ r^ r^ r^ p*f p^ ^^ ^^ P^ ^^ p^ ^^ ^^ ^^
OP^ CM r* CMCM) *0**  en en mm rH ri
oo oo CMCM CMCM * v oo vo vo vo oo «««« r* r» r» c-
oo <-i rH rH i-i tH fi CMCM CM CM CM mm mmmm enen vv
m en en en en en en en en en en en m en 
r-tfH) r-t »-l rH tH! lH5 «H) i-l r-» ,H ,H r-l rH |H «H «H; r^ r-t lH fH. iHiH
r-iiH v» CMCM enen mio i-ii-
-------
















ITJ
4-1
03
a
M
•H
(0
a
$

rH
(0
•H
U
M
0)
0
u ;
•
HI
M
M «





(
M
,

1
<
1
i
<
f
•1
!



(
3j (
W «
'

C,


1
I
1
G
iN
• &
2«
8
CO
0 *
* I

i t

C



\
jj
*
a '
" S
*J
• Jj
1 Azrizoi
[Loaded





g
4
i*
1
j
3






<
i
M
(
1
<
c
1
1
(

I

§

t


J
3


8


f*
3

8



•
8

8


4
!
4

|
1
2
^
1
|

w
J
•
j
«
« «
v co

^ «
w •


CO i-l
o o
0 0

-------
ASM/IM240   Phoenix  Lane   Data   Request   Form
  Please  Fax  or Mail  Your Data  Requests to:

                      Attn:  William M. Pidgeon
                      U. S. Environmental Protection Agency
                      2565  Plymouth  Road
                      Ann  Arbor, MI  48105-2425

                      Phone #: (313)  668-4416
                      Fax #:  (313) 668-4497
            Requestor:

          Organization:

                Phone #:

                  Fax #:


      Mailing Address:
                                                            4
 Specify Disk  Format
 (3.5  inch only)
   EH High Density
  Specify File Format:
                                        pi ,BM   r-i M
                                        U IBM   U Mtc
                                             Low Density
Text ASCII File
(tab  separated)
                CSV ASCII File
                (comma  separated)
                                                        SYLK File
Lotus File (.WKS)*
                                                Excel  File (v2.1)*
                                                                            4
  *Note:  Oldest available formats were chosen for maximum compatibility.
   These formats should  be compatible  with newer versions.

-------
            Appendix  C:
QC  Steps  for ASM Analysis Database

-------
    The phoenix lane data used in these analyses were reported to EPA by the
 testing contractor as total values (concentrations,  mass,  or  flow) for the
 entire test mode as well as in second-by-second form.   The following automated
 quality control  (QC) checks were performed by EPA on the data.  Tests that
 were flagged by one (or more) of these QC checks were then manually verified.


 Second by Second ASM  Tolerance  Check*:

    •  Speed Tolerance - ± 15% of nominal speed for Modes 1,2,3.  Allowed
      tolerance to be exceeded for less than 3 seconds  in  duration. Also
      checked Idle for Modes 4,5

    •  Mode Length - Checked to ensure that each mode contained at  least  20  and
      not more than 30 "stable" seconds.

    •  Hp/Torque Tolerance - Compared required and actual horsepower  (Hp)  and
      Torque and flagged differences > ±10% for at least 5 seconds.

      All vehicles with test weights above 4000 pounds  exceeded this  tolerance
      because of the capacity of the dynamometer.  These cars were not  removed
      from these analyses.  Smaller vehicles exceeding  this tolerance were
      removed.

    •  Calculated average concentrations and cumulative  purge for all  ASM
      modes. Average concentrations were calculated as  the average
      concentration from second 10 to second 39 of each mode.  The first 10
      seconds of each mode were ignored to allow for the dynamometer
      stabilization and exhaust transport time,  vehicles  with less  than 30
      seconds per mode were noted and vehicles with less than 20 seconds per
      mode were excluded.  Purge values were calculated as the total  purge  in
      liters over the entire ASM including transient accelerations.

Second  by  Second ZM240  Tolerance  Checks:

    •  Speed Tolerance - ± 4 mph at ± 1 sec of nominal speed.  Allowed
      tolerance to be exceeded for less than 3 seconds  in  duration.  Also
      speeds exceeding 70 mph, and less than 0 mph were flagged.

    •  Background Concentration Tolerances - Flagged background readings
      outside the following ranges:

                          1.8 < HC  < 10.0 (ppm)
                        -10.0 < CO  < 30.0 (ppm)
                        - 0.5 < NOjc < 1.25 (ppm)
                          0.0 < C02 < 0.15 (percent)

    •  Test Length - Checked to ensure that the full 240 seconds were present.

    •  Distance Tolerance - Flagged distance >  ± 5% of nominal distance

                        Bag 1:  0.532 < dist 1 <  0.588
                        Bag 2:  1.393 < dist 2 <  1.469
                                      C-2

-------
    •   Fuel  Economy Tolerance - Flagged fuel economies < 10 mpg and >50 mpg

    •   Sample  Continuity and Integrity - Ensured that the sampling was
       continuous  (i.e., sec(I) - I for I - 1 to 240) and that gram and
       concentration values were non-zero (HC, CO and C02 cannot all be zero
       for fuel economy calculations or dilution factors).


       Non-zero concentrations were not mandatory for the Phoenix data because
       second  by second concentrations received were calculated, not measured.
       The calculated concentrations were based on the reported grains per
       second  results.  These vehicles were still flagged for low
       concentrations but were not removed from the analyses for this reason.


    •   Comparison of composite and bag results calculated from the second by
       second  data with composite and bags results received from ATL.
       Differences of > 10% were flagged.


Purge  Flow Data  QC

    •   Comparison of second-by-second purge flow to the reported cumulative
       purge flow and pass/fail status reported by ATL.  All significant
       differences were flagged.

    •  Vehicles exhibiting a non-zero constant purge rate for more than 20
       seconds and at various speeds were flagged.  Purge data was rounded to
       nearest 0.01 liter/second prior to processing.


Bag FTP  Tolerance  Check*:

    •  The ratios of corresponding emissions  (HC, CO, and NOX) and fuel economy
       for each of the three bags that were not within expected ranges were
      flagged.

    •  The temperatures, barometric pressures, and distances that were not
      within expected ranges were flagged.


Bag IM240  Tolerance  Checks:

    •  Bag-1 emissions (HC, CO, and NO*) and  fuel economy were compared to the
      corresponding Bag-2 results (based on  regression  analyses previously
      performed on the Indiana data).  All significant  differences were
      flagged.

    •  The Bag-1 and Bag-2 fuel economies were also compared to the test
      weight.   All fuel economy values that  were not within an expected  range
       (based on test weight)  were flagged.
                                      C-3

-------
    •  The Bag-1 and Bag-2 distances not with the  following ranges were
      flagged:

                        Bag 1:  0.545 £ dist 1  £  0.586
                        Bag 2:  1.365 S dist 2  Z  1.435

Bag  IM240/FTP  Tolerance  Checks:

    •  For the laboratory recruited vehicles, the  composite IM240 emissions
      (HC, CO, and NOX) and fuel economy were compared  to the corresponding
      FTP results (based on regression analyses previously performed on the
      Indiana data).  All significant differences were  flagged.

Dynamometer  Loading  Tolerance   Checks:

    •  The test weights and horsepower settings  had to be within 10% for all
      tests performed on each vehicle.
Excluded  Data  Summary

    This section of Appendix C details the  vehicles excluded from the various
databases.

    Purge: Analysis  -  1725  of the 1758 lane tests contained the necessary
data to be included into this analysis.  Of these 153 were removed because of
a malfunctioning purge meter, 184 tests were repeat tests for  vehicles
previously tested and were removed, 5 cars had purge flow status  fields which
indicated missing data, 95 additional vehicles had no indication  of test order
and were removed, and 118 of the remaining vehicles exhibited non-zero
constant purge rates over varied vehicle speeds and were removed.   The  result
was a database of 1170 lane tested vehicles.

    Outpoint Table Analysis  -  This analysis required  laboratory FTP  data.
Therefore, only lab recruited vehicles were considered for this analysis.  Of
the 127 recruited vehicles, 17 did not receive initial ASM tests and one  (veht
3258) did not receive an as-received FTP.  Of the remaining 109 vehicles one
vehicle  (vehf 2177) was removed because the ambient FTP temperature exceeded
allowable tolerances, one vehicle  (veht 3253)  was removed due to extremely
low HC emissions at the lane caused by a flame-out in the FID HC analyzer, and
vehf 3164 was removed due to unacceptable speed deviations on its initial ASM
test.  The resulting database contained 106 lab recruited vehicles.
                                      C-4

-------
    Commercial  Repair Analysis - Of the  27  vehicles recruited for  this
 program only 23 had completed after repair tests at the time of this analysis.
 One vehicle, #13239  (CRt 24) was  removed from the database due to unacceptable
 speed deviations on its initial ASM test, leaving 22 vehicles available for
 analysis.  For the analysis of Section 5.6.2,  5 vehicles failed to pass the
 Arizona state test on the subsequent retest and were removed.  The resulting
 database used for this analysis consisted of 17 vehicles which received
 "successful" commercial repairs.  For the commercial repair analysis of
 Section 5.6.3, only vehicles initially failing ASM outpoints were included.
 The result was 17 vehicles.  Commercial repairs did not have to be successful
 for this analysis and the two data sets contained slightly different cars.

    Regression  Coefficient  Analysis -  For this  analysis all lab  recruited
 vehicles and commercial repair vehicles were removed from the analysis to
 prevent the application of coefficients to data used to develop those
 coefficients.  Therefore, 1422 of the 1758 vehicles were considered for
 inclusion into this analysis.  Ten vehicles were removed because the composite
 IM240 data was not available.  The following vehicles were removed because
 there was insufficient second by  second data to calculate composite IM240
 results:
Run t
1027
1855
2231
3066
3077
3079
3081
Reason for Removal
Test has only 93 seconds
Missing second by second
Test has only 93 seconds
Sampling Discontinuity
Sampling Discontinuity
Sampling Discontinuity
Sampling Discontinuity
    Of the 1405 remaining vehicles,  1192 passed all QC tolerances.  Purge
tolerances were not considered foe this analysis.  The following table lists
the QC tolerances checks for which vehicles were removed  from this analysis.
                                      C-5

-------
Tolerance
riao/aed
ASM Speed
Short ASM Mode
ASM Horsepower
ZM240 Speed
IM240 Fuel Economy
IM240 Background
IM240 Sample
Nuaber of
Vehicles
8
2
10
14
4
163
18
    Note:   1405 minus the above vehicles does  not  equal  1192 because  some
vehicles exceeded more than one tolerance

    Six hundred and eight (608)  of the 1192  tests  remaining received  the IM240
second and were chosen for this analysis.
                                      C-6

-------
      Appendix D
IM240  Cutpoint  Tablas

-------
5  s
    ggooooooooooooooovovovooovomooooomomomotoio

     voooooooooooooooooooooooodoooo'oodooo
            oooooooooooooooooooooooooooooooooo
                                                                                                                                                                 a
                                                                                                                                                                 X
                                                                                                                                                                 o
    i
                                                                                                                                                                 o
                                                                                                                                                                 0)
                                                                                                                                                                 in
        ooooooooooo
ooooooooooooooooooooooooooooooooooo
            oooooooooo
                                                                                                                                                         fl       U
    •3  ("•  in  m  in  in  en  en  en  en  r*   r*  r^  r*  fi  fi
    ainr>vor»r*voenenenooaBooeoaBaB
    CbCMCMCMCMCMCMCMCMCMCMCMCMCMCMCM

a
9
H
2

£
•H
*

O
5
Q
M
1
!








5
S z
0
5 0
{! W
s
- X
_ Q
i z
'i 8
X U
I
9 U
^^•^•i^M

CM
i;
H
®



0
V

m


CO

CM
00

»
en
0
VO
x
in
0
in
CM
O
o
CM


m
to

ft
en
CM
CO
00
CM
CM

in
m
m
o
VO
g
O
CM
O
O
X
§


en
•w

CM
CM

00

CM

m
m
o
VO
X
CM
O
m
CM
0
o
o
o


m
in

ft
m
CM
CO
co
CM
CM
in
en
"»
•w
in
o
m
CM
O
oo
x.
o


m
m

ft
3
CM
CO
00
CM
CM
•n
en
e
CM
m
o
m
CM
X
O
m
o
CM


m
m

*•
CM
0
oo
aa
CM
VO
m
en
0
vo
in
o
CM
0
O
X
o
CM


00


m
m
en
s
en
CM
CM
in
n
0
VO
X
s
o
tn
x
0
o
o


m


VO
»
5
en
CO
CM
CM

in
in
m
vo
en
x
O
•f
CM
X
O
CM
X
s


in


VO
v
VO
an
oo
CM
CM

m
in
«
V0
A
•x
m
o
CM
X
O
CM
X
O
CM


in


VO
*

en
co
CM
CM

m
m
01
«*
CM
X
CM
VO
O
tf)
CM1
X
0
CM
X
s


in


vo
V

CM
CM
CM

in
m
en
H»
an
x
m
0
M
X
O
CM
X
O
M


m


00
CO
r»
CO
CM
CM
f

M
n
«
*
x
3
e
«
CM
X
O
00
X
s


in


00
3
r»
00
CM
CM
f

•n
n
0
CM
^»
O
«
CM
X
0
in
x
S


m


00
!3

oo
CM
CM
f*

lit
Ct
V
•w
x
O
K»
CM
•x
O
00
x
S


in


oo
3

OB
CM
CM


l»xxxxxx
OOOOOOOOOOOOOOOOO
CMCMCM>OOBaBCO»««voaBV*aBvOCOO
-------
        v
           OOOOOOOOOOOOOOOOOOOOOOOC40IC4MOOOOOOOO    ~
                                                                                                              a
                                                                                                              X
                                                                                                              o
           ooooooooooooooooooooooooooooooooooo
        •  v0t0l0VOV0V0V0V0ak<00tV00«^9l>0O<'')£t0OOOO*0v0*0t0t0*a*0>0

        yddodoooooooooooooddodddcNiMcNCNdddoodoo
        c2
 9
O


I
o
X
        o^1^^1^1^^^
                        
                                o a o o  e
                                                   ooooooooooooooooo
o  o o o e
                      ooooooooooooooooooooooooooo
          eoinoo««>oooo«B«


          ^^'X^>»>.1^>,>»>i^>,

          OOOOOOOOOOQO
                                                                                       (M d  CM  r4
                                                          OOOOOOOOOOOOOOOOOO
       i

-------
             i  *
             &  ^
                 OOOOOOOOOOOOOOOOOOOf1r-r-<9coaoaoooo<.oao
i
^
           fl  (O  t-l
           5  in  in
                OOTO*OOOOOOF
                                                                                                                                                                co  co  co   ^CNin(nmm«i»Jwo«»»wi~c->or—   or—  o«orr'itorfi»jr'«r«»»>  in  «i   »i»i»>



         OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO


         -»--»--f>  +   -»'-f'-»'-f'-r'-»>*   +  -f'*-f'-»'-»'-*'>-»--t--«--f'-»--t>*  +   -*'*'f>'-»'**>
                ooooooooooooooooooooooooooooooooooo

                inoBincainoooo«>K>MO**iniAMr>«ooM-lOOOOOOOOOOOOOO
                                                                        O
                                                                        «0
                                                                                                                                                                 o  o   o
         «e  va   so  
-------
                                                    •Mk
O




I

£ ° °
' 1 0 0
£
I » *
3 co co
a ° °
| V0 V0
£ vo vo
£ co co
M^H»
£ C- f
Zoo
CO OB
8SS
\£ |M
y oo oo
^H j^. .fc-
^^ *^l V'f
01 01
«s * *
Z CM CM
8 CO CO
CM CM
Ok Ok
in m
y o o
••BUB
O *

M m m
" CM CM
|«0
* tn m
V CM CM
I® °-
o oo
CM 0
o o
o o
I..
is=

0 0
0 0
CO CO
0 0
VO «0
co co
f r-
o o
CO 00
o o

00 CD
3 ot

« <0
CM CM
CM CM
Ok Ok
in in
0 0
O M>

8m
ot

m m
ot ot
o o
m ot
o o
o o
0 0
00 OB

O O O
0 0 O
.
o o o
OB CM CM

^O C^ C*l
^1 t^t ^1
00 CO 00
0 0 0
3 3 3
CM CM CM
CM CM CM

oo oo co
oo oo oo
CM CM CM
OB OB OB
m in in


£22




•N *• ^
So o
CD CD
O O O
000

0 0
0 0
V0 VO
o o
CM CM

CO CO
CM CM
CO CD
O O

CM CM
got

CD 00

s :
m in

*-4
22





SS
0 0
00 S

0
o
Ok
o
OB

co

oo
in

CM
CM
*"


Ok
M

-»
A





3
o
•*

o
o
Ok
o
CO

CO

00
m

CM
3



01
r*
•A

-
s





o
«t
o
"*

o
o
Ok
o
OB

CO

OB
m

CM
M
^


r»


K





8
o
*••

o
o

o
OB

CO

00
•n

CM
CM
***


Ok
M


5





s
0
*

o
0

o
CM

CO

00
in

CM
CM
Ok


Ok
r»


8





o
o
*

o
o

o
OB

CO

OB
•?

CM
CM
^


Ok
in


CO





o
o
~
••••
o
0

o
CM



OB
m

CM
CM
*^


Ok
in


s





o
0
o

OOOOOOOOCMCMCMCMCMOOOOOO
ooooooooooooooooooo

OOOOOOOOCMCMCMCMCMOOOOOO


««COCOCOCOrOCOOOBOBCMCMOkOkOkOICMCM
OBaoaocDOBCBaocoaBc»r»cococoa»oocooBa»


COM>M)OOOOCMCM
CMCMCMCMCMCMCMCMOOOOkOtmCOrOCOCMCM


CMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCM



*^* ^9 f4 eft l^ ^* ^* ^^ ^9 »^* o^* P§ ift •> lO lA Jrt *^ ^w
Vi •» (^ ^^ 1^ CO ^l ^^ In ^^ ^^ '•W »^* ^§ ^
-------
ii
I*
OOOOOCMCMCMCMCMCMeMCMOOOOOOOOOOQOOOC4CMCMCMr4CMeM NOi
3 28f
s Emissions
C CC
7 585
M S m
O
o
t-l
^^ ^4
J 0
1 ;•
1 :
s °.
3
o
m
o
1 i 2
0

O
00

00

CM
00

00

CM
CM


00
CM
00
m
0
o
oo
o
co
•v
o
CM
t-l
S
o
CM
0

o
00

00

CM
00

oo

CM
CM


00
CM
u>
00
in
r»

t-4
n
o
CM
•N.
O
CO
S
o
CM
O

o
00

00

CM
00

00

CM
CM


00
CM
Ul
00
m
0
2
In
o
CM
•x
0
m
o
\o
o
01
0

o
00

00

CM
00

00

CM
CM


00
CM
U)
00
m
p»
vo
<*»
CM
r-
O
en
•x
O
CM
S
0
A
i-l
O

CM
in
en

o
CM
00
o
r»

vo
t-i


00
CM
CM
P-
m
m
vo
.
o
o
t-l
o
VO
o
A
0

CM
m


*
CM
OO
CM
VO

CM
§


00
CM
OO
m
in
09
CM
VO
O
A
•-I
•x
O
o
o
o
t-l
A
t-l
0

CM
in
§

^
CM
00
CM
VO

»
0
O
o
CM
**
A
O

CM
m
A

*4
CM
O
VO

CM
O
*"

00
CM
m
vo
in
in

3

o
en
*v
o
o
t-l
o
o
ft
A
fl
O

CM
in
A

**
CO
o
vo

CM
§


co
CM
in
in

m

o
CM
•X
O
O
o
CM
t-l
A
t-l
oooooooooooooooooooooooo

CMCMOOOOOOOOOOOOOOCMCMCMcoooooaoaDoocoooa>aooor>r>a»aBooaoooao
CMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCM





•*f-t*4O.1>V-N.-X>«.^>.'X'X^>.>X*^*X'>«.>»'*«'^
oooooooooooooooooooooooo
oooooooooooooooooooooooo
v4*4OOOOOOOOOOOOOOOOOOOOOO
AAOOOOOOOOOOOOOOOOOOOOOO
r4v4
-------
          acKcnenAen«oenenenenenenenen0«eneKcneKcnenenenenovotenenenencneftAen

     ^CMCMNCMCMCMCMCMC9OOOOOCMOOOOCMCMCMCMCMCMCMCMCM(MCMCMCMCMoic4
                ooooooooooooooooooooooooooooooooooo
                                                                                                                                                                       •0
                                                                                                                                                                       c
                                                                                                                                                                       •0

                                                                                                                                                                       a.
                                                                                                                                                                       X
                                                                                                                                                                       a
            •  <«enenenehenir>ir4C\ii-4.-<<-li-lr4.-4VOr4l-lr-l,"4VOtaVO
                                                                                                                                                                 X
                                                                                                                                                                 II
                                                                                                                                                                 u

                                                                                                                                                                 01
                                                                                                                                                                 a
                    CD  CO  CO  CO   CO

                    ft  fl  ft  ft   ft
                                      VYOOOOOOCMOOOOOOOOOOOOOOOOCMCMCMCM
                                      o«nfOcocofOco»«t>fo«nrof><*><»)<'i  u>  «e  <-«

	_.  ckeMcnenencnoooocDaocoaDaBoaoooBa»i-ir>r>r>vovo<«vovovoiA>evor>r~««r«r>r-r-
               CM  CM   CM  CM  CM  CM  CM

               CMCMCMCMCMCMCMf^«n«n«»>C»»«nn«»»CMCMCMeMCMCMCMCMCMCMCMCMC>li4<-ioinr>r>r>i»r>i»c>)CMCMCMCMinininmininininincMr«c>><«m<«vo
        CD  CD  OD  CD  oo  oo  oo  ^  en  en  en  en   en  en  en  o  o  o  o  OB  CD  oo  do  oo  OB   QD  oo  oo  f>  ft  en  ^4  ^4  ^  *+
        inmmmmininiAinmK>ininininwv>u>Q)inininininininininv>«oin«e«a<««
           rip^r^r^r^i^r^r^infl^^**1*
           Si0vov0<0v0tawi0r»r»r-
           i**'  c^  co  co  co  co  co  co  c*o  co  co  wO
                                                       >-«fir4f-««Dcocoa»r>i^(~r~r-r>r>i—r-r«CMi-,XiX>>.X.X,>,X,X,XJX,X,X1X,X»>»>»X.X^X.>.


                                                           OOOOOOOOOOOOOOOOOOOOOOO

                                                           incMOCMCMCMCM^»-toO'Hfioooenenoencnenen
               ooooooooooooooooooooooooooooooooooo
          I
               CD  A  MY  in

               ddoooooo
                                                          ooooooooooooooooooooooooo
        OOOOOOOOflftftftftfl«4ftftftf)^^^^4t^flflftflCMCMCMCOCOrOin
        CMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMOtCICMeMCMCMCMCMCMCMCMCMCMCMCMOl

-------
*c CM Ol O O CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM (O (O CM CM CM CM CM CM CM CM CM CN
&N


s
£ooooooooooooooooooooooo25o°oooooooo
a
£

•n
OB 	 	
uCMCMOOCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMOOCMCMCMCMCMCMCMCMCMCM
c3
•o
c
•o
Q.
X
1)
u
O
*~*
in
c
0
l_(
o
0)
01
                OOflOflOOOOOOOOOOOOOOOOOOOOaDOBOOOOOOOOOO
                                                                           O

                                                                           V0
I


Q
M
                *»cocof'>c'»«ncofn(n<'>c'>cnr»r-f-r«r»r»t»r-
           2  o  o  r»  P»
                at  at  ao  ao
                              ^»«n«n«nco«»»cn«nco«ncn«n(nfo


    BfococofofOfotntncnrtMfnfofnf'tcMNCMCMCMCMCMCMCMCM'i'w^'w***^'*''!'
         atatatatatatatatatatatatatatatatataiatatatatatatatatatatatatatatatateK
   ••^IB



     !*ooo>atooooooooaor«i<>>r>r>r»r*«a»aiatatatatetaiatataioooooooo
    Q*4,4Ototoo«aocoooooaocooeoo«oototototototO!O!Oiotoo*4r4'4>4<-i>4<-ir4
    ZOOrOCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCNfMfMCMCMCMCMCntnfnfOfnOlinPltntO
••^•v«ri^inmminin«M«MCMC<««M(Makataiatataiaiati-ir4>«\Aatatatatatatatci>
                     inr>r*r>r>ri>inininininincMCMCMCMCMCMCM-«F4t-4atatatatakakatatatatcROQoooQoooowve«0>«vo««va«a«ovo«0<0«o«0<0«o<0vo

1


&
               UCMCM^4«4r4^4>4i^«4^4»4v4p4^4v<00
           *
                                                                                              00  00  00  0»  00  «0  ««


                                                                                              C^  f^  ^^  CO  CO  CO  CO
                                                 o  o  o  o   o

                                                 oo  oi  oo  ot   CD
                                                                    o o  o  o  o  o  o

                                                                    OD CM  oi  oo  oo  at  at
                                                                                                      o   o  o

                                                                                                      oo   o  o
o  o  o  o  o

*0  r*  *tf  *o  *£
                                                                                                                                                                       o
                                                                                                                                                                        I
                                                                                                                                                                       a
               OOO
                                                 0OOO
                                                                                    OOOO
                                                                                                             C9OOOeOO
                                                                                                                                                    «9Oe90
                        ^  ^4  ^  ^  ^  ^  ^  Cf  W  €*•  Ci  d  Ol
               O  O  O  O  O   O
                                                 oooooeoo
                                                                                     ooooooooooooooooo

                                                                                     d^^docMCMcoaocftcDooaotao'oDco
           i
               ooooooooooooooooooooooooooooooooooo


                 'ddooodddoodoooooddddddddddododoodo
                CMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCM(M(M(MCMCMCMCMCMCMC<4CMCMCMCM
-------

         oooooooooo
                                                                      o  o   o
                                                                                                  o  o   o   o  o
                                                                                                                                                  o   o   o  o  o   o
         oooooooooo

         •^•XV.'S.'X.^'X.'X.N.'V.
                                                                  oooooooooo
                                                                                                                                    Ul0tUI0kUI***
                                                                                                                                    oooooooo
         IS>  O  •-   10   O
                                                                                    OOft^MOfOH*
         ooooooooooooooooooooooooooooooooooo

         >^>»'S.>v.1S,'N'S,«V'>V>v'»»'>.*»«^^V'V'S..'XXi'X>.^^.1^^.^.'>.»-I—   l-»l-«^l-l-'l-'»-'^t-'l->>-«(-»Mh*l—   >-»l-»MI-'t->»-»l->ls>MM»O|iJ|>J
         ooooooooooooooooooooooooooooooooooo
                                                                                                                                                                     *
         ooooooooooooooooooooooooooooooooooo
                                                                                                                                                                                   B
                                                                                                                                                            0k   0k
                                                                                                                                                                                   M

                                                                                                                                                                                   0
                                                                                                                                                  o   o   o   o  o   o
                                                                                                                                                                               2
                                                                                                   CD  »  W   CO
                                                                                                                                                                                           M

                                                                                                                                                                                           O
re
(5

n
ID
X
                                                                                                                                                                          n
oooooooo
                                                             ooooooo«eoooooooooooooo
         o  o   o   o
                            10  a>
                                               oooooooo
0
X
T3

B>
3
a
                                                                                                                                                                         1

-------
       o   o  o  e  e
                                eeeoooooooeooeooooooooooooooo
       UlbJU>WOIU>UIUIU>U)U)U)U>WWU>b>U>UICt>UIUIUI
       ooooooooooooooooooooooo



                                                                              09  10  flB  «  OB  <0
                                                                                                           UI4b*>fa>fa»WU>b>U>U>W
                                                                                                           ooooooooooo

                                                                                                           xx,xx,xxxxxxx


                                                                                                           O»O>tOO»»IOO»-'t<>OM
       ooooooooooooooooooooooeeeoooooeoeoo

       -^XXXXXXXXXX-^XXXXXXXX>X.XXXXXXXXXXXX.XX

       ^^,^l_>,_>»«»->t*»<»*l_>l_>l->h*»*l->»->l->**»jMlotOI->l-*l-'l->l->l-«H>l->l->»->IOIO

       u>uiuiuiuioio*«J^0tvioi«J«>i«JCDaBtowoo»*»>>uiuiuiuioBOBOBtowJO»woo»too0k»J0kw0t«j0k«j

       ooooooooooooooooooooooo
                                                                                                               o  o  o
                                                                                                                               O  OB  JO  O  OB  SO

                                                                                                                               o  o  o  o  o  o


                                     MIO«d«J«J
                                                 24ft  4ft  4ft  ^^  1^  (ft  4ft  4ft  4ft  4ft  4ft  4ft  4ft 4ft  4ft  4ft  4ft  4ft  4ft 4ft  4ft  4ft  4ft  4ft  4ft
                                                 »»***UimUIUtUIUlUIUI«»»»*MMIOIOIOIO»»>IO
            U>»»»«MI-*l»H<»»OOOOOO**»*»»*»W«Ju)u>u>uiu>wi<)«j->j^»>tOIOIS»IOIOMMMIOIOIOIOOOOOOOOO

                                                 OOOOOtOMMMMMIOIO^'k'MI-'OOOOOOOO
        tototototototototototoiotototototo«oioio>ioioioioioio

        »**J«J>JMIOIOM0teO«IOIO»>>IOMIOMM«>**»UIUIUIUIUIUIUIUI
        tototo«ototetotetotote«otetotototoiototoioiowtetototooaaBaBo»soBaBO9       =
                     -                         ~    ~   ~   ~   ~   ~       •k*«»*>»*UtUIU>W4>J->o"io>oi*->->-'»-'e'>oNO>ioie>>o>ooo
                                                                                          OOOOBOBO»OBUIUIUIUIUIUIUIUI
E

        *J  ^J  (ft  4ft  4ft   4ft  4ft  4ft  4ft  4ft  Ift  4ft   4ft  4ft  4ft  4ft  4ft  4ft  4ft  4ft  4ft   4ft  4ft

        uju>H'»~H'»~H<»<**io»tototo
                                                                                                           StOtOtOtOtOMtOtOtOVDtoB.
                                                                                                           aBo»o»aBaBO>eDOBO»OBO>p
o
V.
10
s>
n
o
3
x
•o
        a\f>0>O\O\0\O\0>O*O\0i9iO\Oi9\0iO>9*9>9\O>99>0>&O\9i0>&>0*0>O*&'O*
        ooooooooooooooooooooooooooooooooooo
        fOMIOMIOIOIOIOIOIOIOIOtOIOIOMMIOIOIOIOIOIOIOIOMMIOMIOMIOIOIOlO

        tetotototototototototo«o»«w«to«toto>etototototoIOfO**Jb*IS>|OlOlS>fOK9lOK>


                                                                                           V>^OtO^Oti9tO^OtO^O>OtO^O^O^O^O
                                                                                                                                                             M

-------
            o  o  o  e  o

            o  o  o  o  o
                                     oppppppppppooooopopooooooo

                                     oeoooooeoooooooooeoooooooo
        ooooooooooeoooooooooooooooooooooooo

        V.X.X.X.X.X.X.X.X.X.X.'X.X.V.V.'N.'VV.X.'VX.V.X.X.X.XV.V'X.X.X'X.X.'V'V.
                                                                                                              MMMMIOMMMIOM
                                                                                                                                   •W  WW  W^   w  r^  ^M _^A


                                                                                                                                   M  ^  U   ^  ^  f f
uiuimwiuto\o»et0\0i->J-J-J«J«j«aBOD»aDwwww>woooo

•*-  +  <4-  +  +  H-'f   +   'f*  +  +  't-*  +  *   +   -«.   +  ******'«-*  +  *  +  **   +   *   +   3J§

OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOS
            ooooooooooooooooo

o
 I
                g  s  s  s
                                                         $  s  s  s   s  s  g  s  s  s  s  s  s   s  $  s  s  s

        oooooooooooooooooo
                                                                                                                                    ...


                                                                                                                                    000"
        Mi^MMMooooooooooooooooooooooo
                                                                                                                                       b   b   b  b  O
X
(t
                                                                                                                                               r
                                                                                                                                               I
o
n
        ooooooooooooooooooooooooooooooooooo
(8
X
•o
                    OD   OD
                             oooooooooooooooooooooooooooooo
                                                                                                                                                       i
a
n
                                                                                                                                                             .

        ooou>wooooooooooooooooooooobbbbbbboo»g.

-------
•»iq»I

-------
                   HC FTP  (g/mi)
    o
    o
E
    o
    o
«   r*
^  y
P-  °
    O
    o
    to


    o
                                     B

                                     O
                   BC FTP (g/mi)
       o
       o
N>

O
    o
    o
                  o

                 —I-
                                      to
                                      o
                               o

                               •H
    ro
    o
    o


    o
    CO
    o

-------
CO  FTP

   i-»
   o       m
                                      K)

                                      O
s
8
    N>
    o
 K>



-H
                                8
                    CO FTP  (g/ai)
                       o       m      o
                                              to
     Ul  •
 8
     to
     o
                                      -I	1
                                                                      8

                                                                      CO
                                                                         I
                                                                         B
                                                   O
                                                   ft

-------
                 MOx FTP  (ff/«i)
I
e
1
    es>  • •
 t  u,  -.
                                        10
                                        •
                                        tn
                                        •H
                                8
                                o
                 MOx rPP (g/ml)
o
u«      M
 t      I
                                         NJ
                           in
                            1
                                  10
                                                              CO
                                                              S
                                                            Sfl
                                                   1
                                                   10

-------
                   Appendix G:
ARCO,  Si«rra,  Environment  Canada Data  Analysis

-------
 i . o
       The objective  of thia  report is to respond to the pilot ASM teat
 programs performed by Sierra Research, Inc., ARCO Products Company, and
 Environment Canada.  Sierra  and ARCO both previously published papers
 praising the capabilities of the ASM, and both concluded that some form of the
 ASM could replace the IM240  as an enhanced  I/M test.

       EPA has concluded that the ARCO and Sierra reports are incorrect in
 claiming the ASM as  equal to the IM240.  Based on a comparison with a similar
 database of IM240 vehicles,  the ASM is inferior to the IM240 at identifying
 excess emissions without committing false failures.  Moreover, a series of
 regressions were run for both the ASM and the IM240 versus the FTP.  The
 scatterplots for these regressions, contained in the Appendix to this report,
 show significant variability for the ASM at predicting FTP values, compared to
 the IM240.

       A contractor for EPA is currently testing a number of vehicles at a
 state  I/M lane in Mesa, Arizona on both the IM240, and a 4 mode steady state
 test,  which includes two ASMs, the ASM2525  and the ASM5015.  A sample of
 vehicles is being recruited  to the contractor's lab for further FTP testing.
 The data from that program will give EPA a  chance to determine, with greater
 confidence, if some  form of  the ASM is as effective as the IM240.

       This report focuses on a small dataaet of vehicles, therefore the
 conclusions made in  this report are subject to change when more data is
 available to EPA.  However,  from the data that has been presented  to EPA to
 date on the ASMS, the ZM240  remains the only enhanced I/M test.

2.0        Database   Peaertpfcian

       There are 31 vehicles  in the ASM database EPA used for this  analysis.
 The data were gathered from  programs performed by three different
 organizations:  Environment  Canada^, Sierra Research^, and ARCO Product s^.
*•  Ballantyne, Vera Jf.  Pr«<». St-^ady St-at-» Tgatirny tUtparfc  and Dafca.
  Environment Canada,  August 28,  1992.

2  Austin,  Thomas C., Sherwood, Larry, P»vi»lnpm*»nr at  Tmprovad Loaded—Moda
  Teafc Praexiufaji  £oy Ingpaeti.Qt^ and Mainfftnanra Prnrrrama. Sierra Research,
  Inc. and California Bureau of Automotive Repair,  SAE Paper No. 891120,
  Government/Industry Meeting and Exposition, May 2-4, 1989.
                                      G-2

-------
 EPA started performing ASM teats in Mesa Arizona on September 10,  1992.   These
 data will be the topic of  a separate analysis.

      A number of vehicles in the ASM database were tested with and without
 implanted defects,  so 51 test configurations were used for this analysis.  All
 the vehicles tested by the three different organizations  received the ASMS015
 and the FTP, but ARCO did  not perform the ASM2525.  This left 39 test
 configurations receiving multiple-mode ASM tests and FTPs.

 2.1       ASM  Vehicle*   Removed  from  Database

    There were originally 55 vehicles tested in the three programs,  resulting
 in  117 teat configurations, broken down as follows:  Environment Canada  (32
 vehicles or 36 configurations); Sierra Research  (18 vehicles or 51
 configurations), and ARCO  Products  (5 vehicles or 30 configurations).
 Vehicles were removed from the database for reasons which are discussed below.

        First, all  pre-1983 vehicles were removed to focus on newer
 technology vehicles.  So 3 Canadian vehicles and 5 Sierra vehicles were
 removed, leaving 29 Canadian vehicles with 33 configurations and 13 Sierra
 vehicles alao with  33 configurations.

    Next all pre-1988 Canadian vehicles were removed.  Canadian vehicle
 standards were not  lowered to 0.41/3.4/1.0 until the 1988 model year, so the
prior model years could not be used.  So 13 Canadian vehicles were removed,
leaving 16 Canadian vehicles with 20 configurations.

    Next,  all ARCO vehicles that were not certified to the 50-state standards
of 0.41/3.4/1.0 were removed.  Three ARCO vehicles were certified to
California-only standards, so they were removed, leaving 2 ARCO vehicles with
12 configurations.

    Finally,   all Sierra configurations that received hot-start FTPS instead
of cold-start FTPs were removed.  Because the normal cold-start FT?  is more
variable than hot-start FTPs, short test comparisons should be mad*  using
cold-start FTPs.  Also, vehicles are certified using cold-start FTPs,  so the
results are more relevant.  So 14 Sierra configurations were removed,  leaving
13 Sierra vehicles with 19 configurations.
   BoelchauS Kenneth L., et al. Evaluation at Enhanearf Tn.ip^efclon Taehnlaruaa on
                   Automobiles.  ARCO  Products  Company  Report,  May 8,1992.
                                      G-3

-------
2.2
Selection  of  IM240  vehicles  Used ia  Database
    In order to compare the ASM to the IM240, the analysis should be performed
on a set of vehicles that have received both tests.   However,  none  of the ASM
vehicles received  the IM240, therefore 39 vehicles  were randomly selected
from the Indiana laboratory IM240 database.  These vehicles  were chosen  from
those used in the IM240 cutpoint table analysis in EPA's I/M Coa<-«.  Benafita.
and Impacta Analyaiar which included 274 vehicles with both  IM240 and FTP
results.  In order to make the IM240 database similar to the ASM database,  the
following process was used.
    First,  the ASM vehicles were categorized by emission  levels according to
the following table:
Table 1
          Miunbar of Vahlelaa In
                              oar Emittant
HC/CO
Category
Normal
Normal
High
Very High
Very High
NOx
Category
Normal
High
Normal
Normal
High

HC Range*
0£HC<0 . 82
0£HC<0 . 82
0.8252KX1.64
1.64£HC<10.0
1.64SHC<10.0

CO Range*
0£CO<10.2
OSCCK10.2
10.2SOX13.6
13.6SOX150
13.65CO<150

NOx Range
0£NOx<2
253IOx<4
O^NOx<2
OSNOx<2
2SNOx<4
f ia
Dataset
29
1
2
4
3
*  These are the same categories as those used in the I/M Technical Support
    Document

    Second,  the Lab IM240 database was broken down into these same categories.
All vehicles were 1983* model years, and only vehicles that received the lab
IM240 after the FTP were kept in the database.  This kept the IM240 database
as similar as possible to the ASM database.  From the remaining vehicles, a
random sample waa chosen from each category  so that both databases had the
same number of vehicles in each category.

    By selecting the same number of vehicles from each emittant range, it
prevents one test from getting an unfair advantage in achieving identification
rates.  For example, if the  ZM240 database included considerably  higher  FTP
scores, it would have identified much more excess emissions, thus making its
Identification Rates (IDRs)  higher.
                                      G-4

-------
 3 . 0       g»lgul«fclner  ASM  Maaa

    Sierra indicated (SAE Paper No. 891120)  that  calculated ASM mass emissions
 correlate better to the FTP than concentration measurements,  so  their method
 of  converting ASM NOx  concentration measurements to "mass" emissions was
 applied to this ASM database for HC and CO, as well as NOx.   This was done by
 multiplying the emission concentrations (ppm for HC and NOx,  and % for CO)  by
 the vehicles' Inertia  Weights  (IW), yielding the following units:  kiloton-ppm
 for HC (IW * ppm/103), ton-% for CO (IW * %), and megaton-ppm for NOx (IW *
 ppm/106) .  These are the values EPA used for the regressions in this report.

 3.1       BPA  Equation*  Versus  Sierra  equations

    In their test program,  Sierra measured the ASM emissions  on  both a
 concentration basis and mass basis.  This allowed them to regress
 Concentration * Inertia Weight  (IW) versus mass emissions for the same test,
 and develop equations  that convert (Concentration * IW] to Mass.  As expected,
 these mass calculations correlated very well with the measured mass emissions.

    Sierra's next step wss to regress the measured steady state  mass emissions
 against the FTP emissions and report r^s for these regressions.   They did not
 actually use the calculated mass emissions to predict FTP scores.  This is
 where EPA's analysis of the ASMS was slightly different.  EPA regressed the
 [Concentration * IW] values against the FTP emissions for each vehicle.  This
 was done because EPA did not have measured mass emissions from all three test
programs compiled in this report.  However, the major benefit of the ASMS,
according to Sierra and ARCO, is the ability to use the less expensive BAR90
type analyzers when measuring the exhaust concentrations.  Sinca this is a
claimed benefit of the ASMS, the readings from these less expensive analyzers
should be used when comparing the ASM to the IM240.
4 . 0       Mnltelpla  t.ln^»g ft*ag«««tona  tor  th*

   Using data from all thre* previously mentioned programs, EPA calculated
the IW * Concentration for each emittant.  Then a multiple  linear regression
was performed, using th« calculated  (Hf*Concentration) ASM2525 and ASM5015
scores as two separate variables vs measured FTP emissions.  Equations were
developed from these regressions that predict an FTP score  from a combination
of the ASM2S25 and ASM5015 concentrations * IW scores:
                                      G-5

-------
Table 2,
Bemafc ieina Developed r.o Pr»rffr!«-  FTP from ASM Modes.
                    Predicted FTP  -  [IW (A*ASM2525 -t- B*ASM5015) + C]
Bmittant
HC (ppm)
CO (%>
NOx (ppm)
A
-3. 96xlO-7
2.64xlO-3
1.13x10-7
B
4.60xlO-7
S.lOxlO-5
1.30x10-7
C
0.523520
4.222840
0.515531
r2
49.2%
43.5%
71.4%
4.1
 Simple  Linear  Regressions
    Aside from those already mentioned,   regressions were also run for each
individual ASM mode vs the FTP, and for the IM240 vs the FTP.   Since the IM240
is a transient test, like the FTP, it correlates much better to the  FTP  than
the ASM modes.

4.1.1     Coefficient  of   Determination  (r2)

    The r2 may be interpreted as the proportion of the total FTP variability
that was predicted by the short test.  For example,  if the r2 equalled 100%,
the short test would have perfectly predicted the FTP scores for these cars.
If the r2  for these vehicles was zero, the short test would not have any
linear relationship to the FTP.

    The r2 data,  listed in table below,  show that the  IM240  is  considerably
better than the ASM tests in predicting FTP HC, CO,  and NOx scores.   For HC
and CO, less than half of the FTP variation is explained by the ASM scores.
Table 3.
St-ar.iar.ieal C.emtetMfiaon of r.h» FTP Versus I/M Tests

r2
EC
IM240
95%
5015
36%
2525
20%
CO
IM240
92%
5015
45%
2525
44%
NOx
IM240
84%
5015
62%
2525
70%
4.2
 Scatterplots
      For an I/M test, more important than the r2 ia the ability to identify
high proportions of dirty cars without falsely failing vehicles.  The  IK240
also has a significant advantage at identifying more of the dirty cars while
failing less of the clean cars.  The scatterplots in the appendix show this
clearly.  When viewing the plots, consider the following chart for a
reference.
                                      6-6

-------
                              Used to Pefina Terma
5 y
4 . .
5 Excess
-*. * Emissions
~ Omitted ^»
E 2" ^^^
Clean Cars
.sip&aa both tests)
^
^/^
^)X^xcess
^^^ Emissions
*r Identified

Errors of Commission
01234
                               Short  Test  Score
      The short test outpoint under consideration is the vertical line,  and
the FTP standard is the horizontal line.  The intersection of these lines
splits the chart into quadrants.  The goal of the short test is to maximize
the number of FTP failing vehicles into the upper right quadrant, while
minimizing the false failures in the lower right quadrant.

      The more vehicles that appear in the upper left quadrant, the less
effective the test becomes, because these are all dirty cars that are not
identified by the short teat.  From this perspective, the advantages of the
IM240 is clear.  Every IM240 chart shows that an x-axis value  (outpoint) can
be selected that clearly places the vast majority of dirty cars in the upper-
right quadrant, without errora of commission.  The ASM tests do not display
this trait nearly as well.  Only the 2-mode ASM tests and the ZM240
scatterplots have the horizontal and vertical lines on them, so the reader can
examine different outpoint scenarios.
                                      G-7

-------
 4.2.1     Scattarplot  Statistics

    Each of the regression scatterplots contains the following information:

      •  The best-fit regression line showing predicted FTP for a continuum of
         short test scores, developed from a regression of the actual data.
      •  'Boundary Lines' at + 2 and - 2  standard error from the predicted
         value.
      •  A horizontal dotted line at the  FTP standard.
      •  A box containing descriptive statistics.

    On each Regression plot,  a box in the upper-left corner provides the
 following statistics:  1)  The equation of the  line  used to predict FTP values
 from the short test's score.  2) r2, discussed  above.  3)  The standard error*
 , a statistic that describes the variability of the  FTP score predicted from
 the selected short test.  The next section discusses standard error in more
 detail.

 4.3        Standard  Krxox  aa  a  Measure of   Variability

      The weakness of the ASM testa regarding r^ and the low proportion of
 cars that can be identified aa dirty while simultaneously  avoiding false
 failures, is related to teat variability.  The  standard error  is  an objective
measurement of test variability.  The following shows  that the ASM tests  are
 significantly more variable than the XM240, using  the  standard error  as an
 objective measure of variability.

 4.3.1     Aaauaptiona  Mad*  for  Using  Standard Error

      The following assumptions were made in order to  use  standard error  as it
 is used in this report:

      • Linear relationship between the FTP and the short  testa.

      • Normally distributed data.

      • Homoscedaatic diatribution  (i.e., the standard deviation of FTP values
         is constant for all abort teat values).
    What is referred to in this report is formally termed standard error of
  estimate, but for convenience purposes, will simply be called standard
  error.
                                      G-8

-------
      The standard error is similar to standard deviation because a bandwidth
 of ±1 std. error includes -68% of the data and ±2 std. error includes -95% of
 the data.

 4.3.2     Example  Using  Standard  Error

      Consider a 3000 Ib. vehicle that emits 1500 ppm NOx on both the ASM5015
 and ASM2525.  Plugging these numbers into the equation for predicting FTP
 values  (Table 2) yields 1.61 g/mi.  However, because the standard error  for
 ASM NOx  (see Table 4) is 0.36 g/mile, roughly 5% of the FTP scores predicted
 by the ASM result will be greater than 2.33 g/mile  (1.61 + 2*0.36) or less
 than 0.89 g/mile (1.61 - 2*0.36).  Since half of these will err on the low
 side, it is probable that -2.5% of the vehicles identified as  failures by an
 ASM cutpoint of 1.61 g/mi would be false failures.
4.3.3     Effect  of  Standard  Error  on  "Safe  FTP  Predictions"

      In order to be confident the false failure rate would be less than 2.5% the
selected cutpoint should predict an FTP value of 2 standard errors greater than
the FTP standard.  This ensures that the low values (FTP prediction - 2 std.
error) are still failing the FTP.

      For example,

           FTP NOx standard is 1.0  g/mi
           The ASM NOx std. error is  0.36 g/mi
           FTPstandard + 2 std. errors -1.72 g/mi

      So, the selected ASM cutpoint should predict an FTP of no less than 1.72
g/mi.  Applying the same logic to the IM240, whose standard error is 0.28 g/mi, a
predicted FTP score of 1.56 g/mi (1.0 + 2*0.28) will also yield an error of
commission rate leaa than  2.5%.  But because the "safe* predicted FTP score is
more stringent, the excess emissions identified will be higher.  The standard
errors and predicted FTP levels that are expected to limit false failures to
approximately 2.5% are compared in Table 4 below.
                                      G-9

-------
Table  4.  Comparison of ASM and TM24Q St-anHard erraca  And Their Effect
           Prodletad FTP  Sfrrlnaanev  at a 2 51 Falsa Failure

1 std. error (g/mi)
Predicted FTP Level
8 -2.5% EC (g/ mi)
HC
IM240
0
0

.24
.89

ASM
0
1

.60
.61

XM2
3.
11

CO
40
8
.0

ASM
4.
13

8
.0

NO*
IM240
0
1

.28
.56

ASM
0.36
1.72

5 . 0       gutpatnf
      Another way to assess the effectiveness of I/M tests is to evaluate  the
following factors, which were discussed in detail in Section 4.2.1 of EPA*s
T/M Coats. Bftnafltsr and Imparts Analysts; excess emission identification
rates, failure rates, error-of-commission rate, the failure rate among
vehicles that pass FTP standards, and the failure rate for so-called "normal
emitters," which may fail an FTP standard (normal emitters are defined as
vehicles whoa* FTP HC < 0.82 g/mi and FTP CO < 10.2 g/mi), but are clean
enough to make  the coat effectiveness of repairs an issue.  These factors are
highly interactive, for example, high IDRa can be achieved with stringent
outpoints, but this will adversely affect failure rates.

      Cutpoint tables for the ASM tests and the IM240 in the appendix allow
these factors to be compared. The cutpoints for the tables were chosen using
an iterative process.  The goal was to select cutpoints that would give
reasonable identification rates while limiting errors of commission.  The goal
was to keep the EC rate at 0% for both procedures.

    For both cutpoint tables, four different cutpoints were selected for each
of the three emittanta, resulting in 64 different cutpoint combinations.  For
the IM240, the "Two Nays to Pass Criteria" was used, as described in Section
4.2.3.2 Of EPA* 8 T/M Coats. Banafits. and Impacts Analysis.  Thia ia a method
of combining the compoaite HC and CO acorea with the bag 2 HC and CO scores in
order to minimize Errors of Commission on vehicles with cold start problems,
while maintaining high Identification Rates.

5 . 1       Selecting ASM  Outpoint*

    Scatterplota were done plotting Measured FTP vs. Calculated FTP from the
ASM scores.  From these scatterplots,  EPA determined a range of cutpoints  to
use for the cutpoint tables.  For example, looking at  Chart  x,  FTP CO vs  ASM
                                     G-10

-------
 Prediction  CO,  it can be determined that an ASM Prediction between  6  and 10
 grams/mile  would identify most dirty vehicles (those above the  standard  of 3.4
 g/mi)  without failing the clean vehicles.  It is also obvious that  a  outpoint
 of  5 would  falsely fail at least one vehicle while achieving no added benefit.
 Consequently, the chosen CO outpoints for the ASM range from 6  to 20.  The
 range  of outpoints for HC and NOx were chosen the same way.

 5.1.1     Using  Standard  error to Predict  Reasonable  Outpoints

      Although  the cutpoint tables do include a wide range of outpoints, there  is
 still a concern that the errors of commission are not representative  of  what they
 might be in a real world scenario.  For this reason, the outpoints  shaded at the
 end of each table were selected using the standard error.

      The ASM cutpoints used are identical to the "safe FTP predictions"  in
 Table 4.  This  is because the values are obtained from calculations using both
 ASM scores.  Each mode has a "sliding scale" of cutpoints, dependent  on  the  other
 mode results.   In other words, no single ASM5015 or ASM2525 value can be used for
 a cutpoint since a vehicle might be clean on one mode and very dirty on  the
 other.  The cutpoints for the IM240,  on the other hand, are direct IM240 scores,
 in grams per mile.  The "safe cutpoints* for the IM240 were determined by
 calculating the IM240 score that would predict the "safe FTP Level -  2.5% EC"
 (see Table 4), using the equations on each respective IM240 scatterplot.

      For example, the IM240 HC FTP Level - 2.5% EC is 0.89 g/mi.   The regression
 equation on the IM240 vs FTP scatterplot is:

                   - 1.429  *  IM240 + 0.04;

      Since we want to predict an FTP of no less than 0.89 g/mi, setting FTPpced.
equal to 0.89 yields an IM240 score of 0.60 g/mi.  This was done to calculate
each "safe cutpoint" for the ZM240.

 5.2      Limitations  of  Cutpoint  Table*

      It ia important to recognize several limitations in  these tables.   Most
 important is that the database ia very small and does not  represent the in use
 fleet.  Additionally, the vehicles were preconditioned by  the FTP before the
ASM test and before th« IM240 testa, so the correlation between these short
 tests will be much better than can be expected  for  vehicles tested in an I/M
 lane, because of all the uncontrolled variables associated with I/M  lane tests
 like temperature, fuel RVP, distance driven prior to the test, catalyst
 temperature, etc.  Because all of these variables were controlled for the
                                     G-ll

-------
vehicles in the ASM and IM240 databases,  the cutpoints  can be very stringent
while still avoiding false failures.  For example,  the  ZM240 table shows that
cutpoints of 0.4/6/1.0 yield IDRs of 97%, 93%,  and 100%,  for HC, CO and NOx,
respectively, without errors of commission.   If cutpoints this  stringent were
used for random vehicles tested in I/M lanes, the error of commission  rate
would be unacceptably high.  Similarly, because of the  introduced
malfunctions, the failure rates are not representative  of the in-use fleet
failure rates for an acceptable I/M program.  So, while it is valid to use
these cutpoint tables to compare the ASM to the IM240,  it is not valid to
assume that the rates are representative of those that  will be  realized  in  a
real I/M program.  The ASM and IM240 testing that EPA is currently sponsoring
in Mesa will provide the actual in-use fleet rates.

5.3        XM240  Identifies  Much  More Excess  Emissions

      Using the cutpoint tables to compare the two procedures,  the IM240 did
considerably better than the Two-Mode ASM at each tests' optimal cutpoints* .
The IM240 identified 97% of excess HC, 931 of excess CO, and 100%  of  excess
NOx at cutpoints of 0.4/6/1.0 (HC/CO/NOx).  The Two-Mode ASM identified 87%,
80%,  and 75% of HC, CO, and NOx, respectively at cutpoints  of  0.6/6/1.50.

      As discussed in the Variability section, using the standard error of
estimate to choose cutpoints that should prevent exceeding an error of
commission rate of 2.5% can help in assessing the performance of I/M tests.
The shaded cutpoints at the end of each test's cutpoint table suggest that the
IM240's performance is significantly better than the two-mode ASMs.   Using the
"safe" cutpoints, the IM240 identifies 92%,  84%, and 71% excess of the excess
HC, CO,  and NOx, respectively - the Two-Mode ASM only identifies 75%, 63%, and
64%.
£.0
      The ASM teats were considerably more variable than the IM240 under
controlled laboratory conditions, aa evidenced by subjective analyses of the
scatter plots and objective measurements using the standard error statistic.
Testing at real-world I/M lanes will add considerably more variability to both
tests, because conditions known to affect emissions such aa temperature,
humidity, and vehicle operating conditions prior to the test.  These
uncontrolled variables are expected to add proportionally more variability  to
* 'Optimal Cutpoints', aa used here,  is the  lowest  cutpoints the test could go
  to and still have  zero errors  of commission.
                                     G-12

-------
a steady state teat like the ASM, but data are not available to evaluate the
validity of the hypothesis.

      On the other hand, the increased variability associated with actual I/M
testing will be somewhat offset for the ASM by adding two additional modes; a
50 mph steady mode at road-load horsepower, and an idle mode.  This four-mode
ASM procedure is now being performed by EPA in a Mesa Arizona I/M lane.

      The result of these offsetting effects on variability will determine the
viability of the ASM as a lower cost substitute for the IM240.  A final
conclusion should be postponed until enough Mesa data can gathered for a valid
evaluation.
                                     G-13

-------
    Appendix A





Outpoint  Tables

-------
to VMM"
          XMMO*
39 19*3+ ?"r*
Inalir tmlmr
IM240 Cvt-Poiot*
CMS) t 1M 2
0.4/15/2.00 +
0.8/15/2.00 +
1.0/15/2.00 +•
1.2/15/2.00 +
0.4/10/2.00 +
0.8/10/2.00 +
1.0/10/2.00 *
1.2/10/2.00 +
0.4/ 6/2.00 *
0.8/ 6/2.00 *
l.O/ 6/2.00 +
1.2/ 6/2.00 «•
0.4/ 5/2.00 *
0.8/ 5/2.00 +
l.O/ 5/2.00 +
1.2/ 5/2.00 +
0.4/15/1.50 +
0.8/15/1.50 +
1.0/15/1.50 +
1.2/15/1.50 *
0.4/10/1.50 +
0.8/10/1.50 *
1.0/10/1. SO +
1.2/10/1.50 +
0.4/ 6/1.50 *
0.8/ 6/1.50 +
l.O/ 6/1.50 +
1.2/ 6/1.50 *
0.4/ 3/1.50 *•
0.8/ 5/1.50 *
l.O/ 5/1.50 *
1.2/ S/1.50 +
0.4/15/1. 23 +
0.8/15/1.23 *.
1.0/13/1.23 *
1.2/15/1.25 +

0.3/12
0.5/12
0.6/12
0.8/12
0.3/ 8
O.S/ 8
0.6/8
0.8/ 8
0.3/4.5
0.5/4.5
0.6/4.5
0.8/4.3
0.3/ 4
O.S/ 4
0.6/ 4
0.8/ 4
0.3/12
0.5/12
0.6/12
0.8/12
0.3/ 8
O.S/ 8
0.6/ 8
O.S/ S
0.3/4.S
0.3/4.3
0.6/4.3
0.8/4.3
0.3/ 4
O.S/ 4
0.6/ 4
0.8/ 4
0.3/12
0.3/12
0.6/12
0.8/12
itexi Vm
^4 •
•
XdMt4flo*t*oa
R*t*
K
96%
91%
91%
91%
96%
91%
91%
91%
97%
94%
94%
94%
97%
94%
94%
94%
96%
91%
91%
91%
96%
91%
91%
91%
97%
94%
94%
94%
97%
94%
94%
94%
96%
91%
91%
91%
CQ
86%
80%
80%
80%
86%
82%
82%
82%
91%
89%
89%
89%
94%
93%
92%
»a%
86%
80%
80%
80%
86%
82%
82%
82%
91%
69%
8»%
89%
M*
M%
m
tat
87%
81%
81%
81%
ate
93%
71%
71%
71%
93%
71%
71%
71%
97%
75%
75%
73%
97%
73%
73%
75%
93%
71%
71%
71%
93%
71%
71%
71%
97%
73%
73%
73%
97%
73%
73%
73%
93%
73%
73%
73%

of
Hilan,
22
17
16
16
22
18
17
17
26
23
22
22
30
27
26
26
22
17
16
16
22
18
17
17
26
23
22
22
30
27
26
26
23
IS
17
17


B*a
48%
37%
35%
35%
48%
39%
37%
37%
57%
50%
48%
48%
65%
59%
57%
57%
48%
37%
35%
35%
48%
39%
37%
37%
37%
50%
48%
48%
65%
59%
57%
57%
50%
39%
37%
37%

of
Xft^
0
0
0
0
0
0
0
0
0
0
0
0
1
1
1
1
0
0
0
0
0
0
0
0
0
0
0
0
1
1
1
1
0
0
0
0


Bftfc*
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
2%
2%
2%
2%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
2%
2%
2%
2%
0%
0%
0%
0%

tailor* 8*to
foe R»
XtUolu
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
10%
10%
10%
10%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
10%
10%
10%
10%
0%
0%
0%
0%

fa&lvr* R*to
for NotMl
teittiag
UfcifllM
16%
3%
3%
3%
16%
6%
6%
6%
29%
23%
23%
23%
42%
33%
35%
33%
16%
3%
3%
3%
16%
«%
«%
6%
29%
23%
23%
23%
42%
33%
33%
33%
19%
6%
6*
«»
A-l

-------
           "Tw« Itay* to *•••"
                                                             40*
39 1983* •— *
«lv a*lma
11840 Cw*-P«iat»
r.tma t IM a
0.4/10/1.25 *
0.8/10/1.25 *
1.0/10/1.25 +
1.2/10/1.25 +
0.4/ 6/1.25 +
0.8/ 6/1.25 *•
l.O/ 6/1.25 +
1.2/ 6/1.25 +
0.4/ 5/1.25 +
0.8/ 5/1.25 *
l.O/ 5/1.25 +
1.2/ 5/1.25 +
0.4/15/1.00 *
0.8/15/1.00 *
1.0/15/1.00 *
1.2/13/1.00 *
0.4/10/1.00 *
0.8/10/1.00 +
1.0/10/1.00 *
1.2/10/1.00 *
0.4/ 6/1.00 +
0.8/ 6/1.00 +
l.O/ 6/1.00 *
1.2/ 6/1.00 +
0.4/ 5/1.00 *
0.8/ 5/1.00 *
l.O/ 5/1.00 *
1.2/ 5/1.00 +
0.6/10/1.60 +

0.3/8
0.5/ 8
0.6/8
0.8/ 8
0.3/4.3
0.5/4.5
0.6/4.5
0.8/4.5
0.3/4
0.5/ 4
0.6/4
0.8/ 4
0.3/12
0.5/12
0.6/12
0.8/12
0.3/ 8
O.S/ 8
0.6/ 8
0.8/ 8
0.3/4.9
0.9/4.9
0.6/4.9
0.8/4.9
0.3/ 4
0.9/ 4
0.6/ 4
0.8/ 4
O.S/ 9
**4 v«
k4««l A*
Xdaatlfiiu
ac
96%
91%
91%
91%
97%
94%
94%
94%
97%
94%
94%
94%
97%
93%
95%
93%
97%
93%
93%
95%
97%
93%
93%
99%
97%
99%
99%
99%
92%
CQ
87%
83%
83%
83%
91%
89%
89%
89%
94%
92%
92%
92%
91%
90%
90%
90%
91%
90%
90%
90%
93%
92%
92%
92%
94%
94%
94%
94%
84%
•
itlea
BOB
93%
73%
73%
73%
97%
75%
75%
75%
97%
75%
75%
75%
100%
93%
93%
93%
100%
93%
93%
93%
100%
93%
93%
93%
100%
93%
93%
93%
71%

ot
laUasM
23
19
18
18
26
23
22
22
30
27
26
26
27
23
24
24
27
23
24
24
29
27
26
26
32
30
29
29
19


!•£•
50%
41%
39%
39%
57%
50%
48%
48%
65%
59%
57%
57%
59%
54%
52%
52%
59%
54%
52%
52%
«3%
59%
57%
57%
70%
65%
63%
63%
41%

e«
IflU*
0
0
0
0
0
0
0
0
1
1
1
1
0
0
0
0
0
0
0
0
0
0
0
0
1
1
1
1
0

1
••
Bate
0%
0%
0%
0%
0%
0%
0%
0%
2%
2%
2%
2%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
2%
2%
2%
2%
0%

ration ««t»
f o* ra
PflLAdkBV
Vftki^A^AiV
0%
0%
0%
.0%
0%
0%
0%
0%
10%
10%
10%
10%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
10%
10%
10%
10%
0%

f 0* M*CMl
tatttla*
S&biaiM
19%
10%
10%
10%
29%
23%
23%
23%
42%
35%
35%
35%
32%
29%
29%
29%
32%
29%
29%
29%
39%
35%
35%
35%
48%
43%
43%
45%
10%

the* ••!*«••*
           thttt
th* •**•*(•) •« O
                                               A-2

-------
    AMB01S
MBlti-MM Mw
Cot-Point*
BC/CO/WO*
0.6/20/2.00
0.8/20/2.00
1.0/20/2.00
1.5/20/2.00
0.6/15/2.00
0.8/15/2.00
1.0/15/2.00
1.5/15/2.00
0.6/ 8/2.00
0.8/ 8/2.00
l.O/ 8/2.00
l.S/ 8/2.00
0.6/ 6/2.00
0.6/ 6/2.00
l.O/ 6/2.00
l.S/ 6/2.00
0.6/20/1.50
0.8/20/1.50
1.0/20/1.50
1.5/20/1.50
0.6/15/1.50
0.8/15/1.50
1.0/15/1.50
1.5/13/1.30
0.6/ 8/1.50
0.8/ 8/1.50
l.O/ 8/1.50
l.S/ 8/1.50
0.6/ 6/1.50
0.6/ 6/1.50
l.O/ 6/1.30
l.S/ 6/1. SO
0.6/20/1.23
0.8/20/1.23
1.0/20/1.23
1.5/20/1.23
to XdM
ac

75%
74%
74%
8«%
73%
75%
75%
88%
76%
76%
76%
8«%
87%
87%
87%
•8%
75%
74%
74%
80%
73%
73%
73%
a.%
76%
76%
76%
M*
87%
87%
87%
88%
73%
74%
74%
•klfioatioa Wiirti*
R»*« of rwLlac*
Sfi
77%
62%
60%
60%
77%
62%
62%
62%
81%
67%
67%
67%
82%
80%
80%
80%
78%
63%
61%
61%
78%
63%
63%
63%
82%
67%
67%
67%
83%
80%
80%
80%
78%
63%
61%
61%
BBS
68%
49%
43%
43%
68%
49%
49%
49%
68%
49%
49%
49%
68%
59%
59%
39%
83%
64%
38%
58%
83%
64%
64%
64%
83%
64%
64%
64%
83%
73%
73%
73%
•3%
(4%
M%
M%
'ili'inrM
13
7
6
6
13
7
7
7
14
8
a
8
13
11
11
11
14
8
7
7
14
8
8
8
IS
9
9
9
X,
12
12
12
14
9
8
8
St»/km
33%
18%
15%
15%
33%
18%
18%
18%
36%
21%
21%
21%
38%
28%
28%
28%
36%
21%
18%
18%
36%
21%
21%
21%
38%
23%
23%
23%
41%
31%
31%
31%
36%
23%
21%
21%
of
ta'm
1
0
0
0
1
0
0
0
1
0
0
0
1
0
0
0
1
0
0
0
1
0
0
0
1
0
0
0
1
0
0
0
1
1
1
1.
Co
ttafea
3%
0%
0%
0%
3%
0%
0%
0%
3%
0%
0%
0%
3%
0%
0%
0%
3%
0%
0%
0%
3%
0%
0%
0%
3%
0%
0%
0%
3%
0%
0%
0%
3%
3%
3%
3%
railoM B*t«
f o* IT»
ftmmlay
¥feUfllU
9%
0%
0%
0%
9%
0%
0%
0%
9%
0%
0%
0%
9%
0%
0%
0%
9%
0%
0%
0%
9%
0%
0%
0%
9%
0%
0%
0%
9%
0%
0%
0%
9%
9%
9%
9%
r«Uaz* tat*
fa* Mom*!
talttla?
Xfcfciallf
17%
3%
0%
0%
17%
3%
3%
3%
20%
7%
7%
7%
23%
10%
10%
10%
20%
7%
3%
3%
20%
7%
7%
7%
23%
10%
10%
10%
27%
13%
13%
13%
20%
10%
7%
7%
A-3

-------
                                                                    923
39 Voaiel**
Molti-ASM Mod
Cat-Poiato
•e /eo/MQx
0.6/13/1.23
0.8/15/1.23
1.0/15/1.25
1.5/15/1.25
0.6/ 8/1.25
0.8/ 8/1.25
l.O/ 8/1.23
1.5/ 8/1.25
0.6/ 6/1.25
0.6/ 6/1.25
l.O/ 6/1.25
1.5/ 6/1.25
0.6/20/1.00
0.8/20/1.00
1.0/20/1.00
1.5/20/1.00
0.6/15/1.00
0.8/15/1.00
1.0/13/1.00
1.5/13/1.00
0.6/ 8/1.00
0.8/ 8/1.00
l.O/ 8/1.00
1.5/ 8/1.00
0.6/ 6/1.00
0.6/ 6/1.00
l.O/ 6/1.00
l.S/ 6/1.00
1.6/13/1.72
to XdM
K
84%
75%
75%
75%
84%
76%
76%
76%
84%
87%
87%
87%
94%
82%
80%
80%
94%
82%
82%
82%
99%
82%
82%
82%
98%
94%
94%
94%
73%
*lflo«4
CQ
78%
63%
63%
63%
82%
67%
67%
67%
83%
80%
80%
80%
85%
72%
70%
70%
89%
72%
72%
72%
90%
74%
76%
74%
91%
90%
90%
90%
63%
elo*
02*
83%
64%
44%
64%
93%
44%
44%
44*%
43%
79%
79%
79%
94%
44%
40%
40%
94%
44%
44%
84%
•4%
44%
44%
44%
94%
•4%
94%
94%
44%
of
F.tlt..^
14
9
9
9
IS
10
10
10
16
13
13
13
18
14
13
13
14
14
14
14
19
IS
IS
IS
20
14
19
14
4
railan
ana
36%
23%
23%
23%
38%
26%
26%
26%
41%
33%
33%
33%
46%
36%
33%
33%
44%
34%
36%
34%
49%
38%
38%
34%
51%
44%
44%
44%
21%
of
(all
i
i
i
i
i
i
i
i
i
i
i
i
2
2
2
2
2
2
2
0
2
2
2
2
2
2
2
2
0
la
Bift«
3%
3%
3%
3%
3%
3%
3%
3%
3%
3%
3%
3%
5%
5%
3%
5%
5%
5%
5%
0%
5%
5%
5%
5%
5%
3%
3%
3%
0%
r«ii«M ««*•
f o» n»
PaMiaff
jEafeialAC
9%
9%
9%
9%
9%
9%
9%
9%
9%
9%
9%
9%
18%
18%
18%
18%
18%
18%
18%
0%
18%
18%
18%
19%
18%
18%
18%
19%
0%
railan (Uto
to* NonoJ.
teLttiag
YatAalat
20%
10%
10%
10%
23%
13%
13%
13%
27%
17%
17%
17%
30%
23%
20%
20%
30%
23%
23%
23%
33%
27%
27%
27%
37%
30%
30%
30%
7%
1
XdoBti.«io«fcloa
owkpoiat *•*
                                                A-4

-------
  Appendix B




Scatttrplot*

-------
A


I
?:«
o 25
                                                             0
                                                             M
                                                             u
                                                             U
                                                            •a
                                                            u
                                                            en

                                                            i

                                                             M
                                                             O
                                                             M
                                                             M
                                                             U
                                                             •a
                                                             4J
                                                    n
                                                    

-------
                                 o
                                 Ul
                                 Ul
                                 u
                             8
                             O I
                             * I

                             11
                                 M
                                 O
                                 u
                                 w
                                 u
                                 •o
                                 si
n
41
3
                                  •O  i
                                  v  ;
                                  4J
                                  O  i
                                  4)
                                  U
                                  o>
                                  (ft
                                  V
                                  3
                                  O4
OO 4U

-------
                                                                                                                                                                           1-4

                                                                                                                                                                           o
                                                                                                                                                                           M
                                                                                                                                                                           M

                                                                                                                                                                           Ct]
    i
                                                                                                                                                                               04    j

-------
                                                                                          o
                                                                                          o
                                                                                           o
                                                                                           o
u
1
o
           to

           o
               o
               in
               r-
               o
               i
               U
               ul
               ei
               ut
               r-     O*
               o
               I     O
               u     vo
             •  *  I
            f»  
-------
                                                                                                        o
                                                                                                        o
                                                                                                        o
                                                                                                        o
                                                                \
m
 i
a
I
o
in



I
in
O
 I
u
                                    01
                                    ul
                                    o

                                    u
                                    VO  *  I
                                     •  m
                                    CM   •  M
                                    — 
                                       v   *J

                                    X  U  OT
                                                                                                         O
                                                                                                         O
                                                                                                         o
                                                                                                         o
                                                            e
                                                            o
                                                           ••4
                                                           4*
                                                            U

                                                           1
                                                            to

                                                            8

                                                            5
                                                                      o
                                                                      o
                                        I-        I
                             o
                             o
                             m
    o
    o
o
o
                                                  u)

                                                  
-------
                                                                                                                                                  o
                                                                                                                                                  o
•

i
£
3
                                       CM
                                       in

                                       o
in
iH
o
in
X
CO
at
«wn»

I-
O


U


O
en
                                       r-      vo
                                       o      <*»
                                        i
                                       U      o
                                         •  f  u
                                        f-l    •  O
                                        — r-»  W
                                        «   r-  u
                                        X       ul
                                        M   I
                                            <    tl
                                             U   O)

                                                                                                                                                   O
                                                                                                                                                   O
                                                                                                                                                           §
                                                                                                                                                    o
                                                                                                                                                    in
                                                                                                                                                     o
                                                                                                                                                     o
                                                                                                                                                     o
                                                                                                                                                     in
                                                                                                                                                      o
                                                                                                                                                      o
                                         o
                                         o
                                                    o
                                                    o
                                                     •
                                                    CM
                                                     O
                                                     m
o
o
o
in
o
o

-------
I
A

S
                   (J
                   1
ocl
                                                                                        4-
HC Emission;
Vehicles -
u
|
m
OB
9k
Ok
*O




*•
d
a
in
0»
4*
••*
*










^
4J
0>
in r*
o
1
CO * M
0 • 0
0 vo U
o on u
U
0 1
1 o< T»
                               o
                               o
o
o
                                                 I
                                                o
                                                                                            o
                                                                                            o
                                                                                            o
                                                                                            o
                                                                                            o
                                                                                            o
                                                                                            o
                                                                                            a\
                                                  o
                                                  o
                                                  o
                                                  oo
                                                  o
                                                  o
                                                  o
                                                  r-
                                                            M
                                                            O
                                                            U


                                                            U



                                                            T3
                                                                                                 m  i
                                                                                                       o
                                                                                                       u
                                                                                                       M
                                                                                                       U
                                                                                                       •o
                                                                                                       4J


o
0
o
in



o
1
0.
1




U : n
a «
3
in
o
lA
^

-------
A

!
en
«a
                   X
                   in
                   o
                   o
                                             o

                                             in

                                             I

                                             u
                                             o
                                             u
                                             u



                                             •o

                                             (O
            o
            o
            oo
                                        I -


                                        o

                                        o
                                                                                        0
                                                                                        U
                                                                                        U

                                                                                        U
                                                                                        •a

                                                                                        Cfl

                                                                                         i
                                                                                         0
                                                                                         u


                                                                                         •o
                                                                                         r-l
                                                                                         fl



                                                                                         •o
                                                                                         V
                                                                                         *J
                                                                                          V
                                                                                          M
                                                                                          cu
                                                                                          n
                                                                                          v
                                                                                          3
                                                                                          Ot
o
o
                            o
                            VO
o
o

o
m
o
o
o
o
                                                                       o
                                                                       CO
o
o
                               o
                               04
o
o
o
o
                                   (TV/6)  00

-------
                                                                                                                     o
                                                                                                                     o
at
 i
I
••4
                        «o
                        Ok
O     >£>



O     O

+•      I
    .*»


(O   •  0
vn  cs  w
M  VO  U
       u
O  I

 I  CM -O

 >i U ttl

                                                                                                                                   M
                                                                                                                                   0
                                                                                                                                   u
                                                                                                                                   u
                                                                                                                                  u
                                                                                                                     o
                                                                                                                     o
                                                                                                                     o
                                                                                                                     o
                                                                                                                      o
                                                                                                                      a
                                                                                                                            m  i
                                                                                                                           vo
                                                                                                                      o
                                                                      O
                                                                      o
                                                                                                                      o
                                                                                                                      o
                                                                                                                       o
                                                                                                                       o
                                                                                                                       o
                                                                                                                       o
                                                                                                                       O
                                                                                                                       o
                                                                                                                       o
                                                                                                                       o
                                                                                                                                   u
                                                                                                                                   0
                                                                                                                                   u
                                                                                                                                   M
                                                                                                                                   U
                                                                                                                                   •a
                                                                                                                                   +j
                                                                                                                                   V)
                                                                                    
-------
i
                                               01
                                               in     in
                                                      r»
                                               O
                                                      O
                                               X *
                                               <•> (<«  U
                                               o   •  o
                                               o o  u

                                               °. N  u
                                               O  I

                                               I  ci  -d
                                                  <    *i
                                               >i  U  (0
                                       o
                                       o
                                               I

                                              o
o
o
o
in
                                       If)      f      V
o
o
 •
m
o
m
                                                                             CM
                                                                              OH

-------
I
A
                              O
                              , M
                         o
                         o
                         o
                         00
a
o
o
a
      o
      vo
O
o
o
o
o
a
                                           (tm/6) CO

-------
                                                                                                                                 o
                                                                                                                                 o
                                                                                                                                                M
                                                                                                                                                0
                                                                                                                                                u
                                                                                                                                                U
                                                                                                                                                hi
A

i
                                                   o      o

                                                   +•       I
                                                       d»
                                                    W  en  U
                                                   O    •  O
                                                   in  o  u
                                                   vN  Is-  14

                                                   o   .  U

                                                    I  04 TJ

                                                    >t  W 0)
                                                             O
                                                             in

-------
               Appendix  B:
Estimated  Coat of  High-T«ch  I/M Tasting

-------
    5.2.1     General  Methodology

         EPA's estimates of the costs of high-tech test procedures  are driven
by a number of assumptions.  Costs in conventional centralized and
decentralized test-and-repair programs were derived using current  inspection
costs in  I/M programs as they are reported to EPA as the starting  point.  For
decentralized test-only networks costs are modelled in a manner  similar to
centralized programs, since all current test-only programs  are centralized,
however,  costs are estimated using a range of test volumes  and a higher level
of state  oversight is assumed since the network is composed of independent
operators and may have a higher number of test sites than  in centralized
programs.

        Another key assumption is that adding the new tests will increase
inspection costs in programs that are now efficiently designed and operated.
In programs that are not now well designed, current costs  are likely to be
higher than necessary and the cost increase less if efficiency  improvements
are made  simultaneously.  In order to perform the high-tech tests  new
equipment will have to be acquired and additional inspector time will  be
required  for some test procedures.  The amount of the cost increase will  be
determined to a large degree by the costs of acquiring new equipment and the
impact of the longer test on throughput in a high volume operation.  Average
test volume in decentralized programs is low enough to easily absorb the
additional test time involved (although at a cost in labor time).   Equipment
costs are analyzed in terms of the additional cost to equip each inspection
site (i.e., each inspection lane in centralized inspection networks, and each
licensed  inspection station in decentralized networks).

        By focusing on the inspection lane or station  as the basic unit of
analysis, the resulting cost estimates are equally applicable in  large
programs, with many subject vehicles and inspection sites, or small programs,
with few  subject vehicles and inspection sites.  Previous EPA analyses of
costs in  I/M programs have found that the major determinants of inspection
costs are test volume and the level of sophistication  of the inspection
equipment.  Costs of operating programs were not found to be measurably
affected by the size of the program  (for further  information the  reader may
refer to  EPA's report entitled, "I/M Network Type:  Effects on Emission
Reductions, Cost, and Convenience").  Figures on  inspection volumes at
inspection stations and lanes are available from  I/M program operating data.
This information enables the equipment cost per vehicle and the additional
staff cost per vehicle to be calculated for each  test  procedure.
                                      H-2

-------
         The equipment  cost  figures presented in this paper are based on the
 costs  of the equipment EPA  believes is best suited for high-tech testing.
 They are current prices quoted by manufacturers,  and do not reflect what the
 per unit prices might be if this equipment were purchased in volume.   Staff
 costs  are based on prevailing wage rates for inspectors in both types of
 programs as reported in conversations with state I/M program personnel.
 Construction costs in centralized programs are based on estimates supplied by
 centralized contractors.  Other site costs and management overhead in
 centralized programs are back calculated from current inspection costs.  For
 decentralized networks, it  is assumed that longer test times could be absorbed
 with no  increase in sites.  The current average volume in decentralized
 stations is 1,025 vehicles  per year (between 3 and 4 vehicles per day,
 depending upon the number of days per year the station is open).
 Consequently, increasing the length of the test,  to the degree that the new
 procedures  would, is not expected to impact the number of inspections that can
 be  performed.

    5.2.2      Equipment  Need*  and  Costs

         A pressure metering system, composed of a cylinder of nitrogen gas
 with a regulator, and hoses connecting the tank to a pressure meter, and to
 the vehicle's evaporative system is needed to perform evaporative system
 pressure testing.  Hardware to interface the metering system with a
 computerized analyzer is also needed and is included in the cost estimate.
 Purge testing can be performed by adding a flow sensor with a computer
 interface,   a dynamometer, and a Video Driver's Aid.  With the further addition
 of a Constant Volume Sampler (CVS) and a flame ionization detector  (FID) for
 HC analysis, two nondispersive infrared (NDIR) analyzers for CO and carbon
monoxide (C02>, and a chemilumineacent  (CD analyzer for NOX, transient
 testing can be performed.

         The analyzers used  for the transient test are  laboratory grade
equipment.   They are designed to higher accuracy and repeatability
 specifications than the NDIR analyzers used to perform the current I/M  tests.
Table 5-4 shows the estimated cost of equipment for conducting high-tech
tests.   This quality of technology is essential for accurate  instantaneous
measurements of low concentration mass emission levels.
                                      H-3

-------
                                   Table  5-4
                        Equipment  Coats  for  Hew  Testa

                  Test        Equipment              Price
                  Pressure    Metering System        $600
                  Purge       Flow Sensor            $500
                              Dynamometer            $45,000
                              Video Drivers Aid      $3,000
                  Transient   CVS 6 Analyzers        $95,000
                              TOTAL                  $144,100

        The figures in Table 5-4 do not include the costs of  expendable
materials.  Nitrogen gas is used up in performing the pressure test.
Additionally, the FID burns hydrogen fuel.  Calibration gases are needed for
each of the analyzers used in the transient test.  Because the analyzers used
in the transient test are designed to more stringent specifications than the
analyzers currently used in the field, bi-blends, gaseous mixtures composed of
one interest gas in a diluent (usually nitrogen)  are used to  calibrate them.
Multi-blend gases, such as are typically used to calibrate current I/M
equipment, are not suitable.  Current estimates for expendables  are shown in
Table 5-5.  The replacement intervals are estimated baaed on  the usage rates
observed in the EPA Indiana pilot program and typical inspection volumes as
presented later invthis section.  Calculations of per vehicle equipment costs
presented throughout this report include per vehicle costs of these
expendables as well.
                                    Table  5-5
                          Kxpendablea  for  New  Teata
                                          Replacement Interval
   Test           Material     Coat    Centralized     Decentralized
   Pressure       N2  Gaa       930      250 tests        250 teata

   Transient      H2 Fuel       $60      2 months         1000 tests
                 HC  Cal Gaa     $€0      2 months         1000 teats
                 CO  Cal Gaa     $60      2 months         1000 testa
                C02 Cal Gaa     $$0      2 months         1000 tests

        Staff costs have been found to vary between centralized and
decentralized programs, as does the effect  on the number of sites  in the
network infrastructure.  Therefore, the following sections are  devoted to
separate cost analyses for each network type.
                                      H-4

-------
    5.2.3       Coat  to  Upgrade  Centralized  Network*

    5.2.3.1   Basic  Aaaumptiona

         The starting point in this analysis is the  current average per
 vehicle  inspection cost in centralized programs.  A figure of $8.50 was used
 based upon data from operating programs.   This figure includes the cost of one
 or more  retests and  network oversight costs.   The key variables to consider in
 estimating the costs in centralized networks  are throughput, equipment, and
 staff costs.  Data on these variables were obtained by  contacting program
 managers in a number of these programs, and by surveying program contracts and
 Requests for Proposal.

         Throughput refers to the number of vehicles per hour that can be
 tested in a lane.  The higher the throughput  rate,  the  greater the number of
 vehicles over which  costs are spread, and the lower the per  vehicle cost.  EPA
 contacted program managers and consulted the  contracts  in  a  number of
 centralized programs to determine peak period throughput rates in the
 different systems.   Rates were as reported in Table 5-6.

                                    Table  5-6
        Peak  Period  Throughput  Rates  ia  Centralised  Z/M  Programs
       Program                             Vehicles Tested per Hour
       Arizona                             20
       Connecticut                         25-30
       Illinois                             25
       Maryland                             25-35
       Wisconsin                           25-30

        On the basis of this information, 25 vehicles per hour was assumed to
represent the typical peak period throughput  rate  or  design capacity in
centralized X/M programs.  During off-peak hours and  days, throughput is  lower
since there is not a constant stream of arriving vehicles.  Conversations with
individuals in the centralized inspection service  industry indicate that
inspectors start at minimum wag* or slightly higher,  that  by the end of the
first year they earn $5.50 to $€ per hour/ and that they generally stay with
the job for one to three) years.  Thus, $6 per hour was used to estimate the
average inspector's hourly wag*.

        Estimates of the costs of adding pressure testing, purge testing/ and
transient tailpipe testing were derived by taking  the current coats for the
new equipment to perform the new tests, dividing it by the number of
inspections expected to be performed in the lane over a five year period and
                                      H-5

-------
 adding it to the current $8.50 per vehicle cost,  with a  further adjustment foe
 the impact of test time on throughput,  and thus on the number of sites and
 site costs.  The same is done to estimate additional  personnel costs
 associated with adding the new tests,  when independent  programs were surveyed
 to determine the length of a typical contract,  it was discovered that
 Illinois, Florida, and Minnesota all have five year contracts, Arizona has a
 seven year contract, and the program in the State of  Washington is operating
 under a three year contract, resulting in an average  contract length of  five
 years among the five programs surveyed.  Five years was  therefore chosen as
 the typical contract length.

        The number of inspections expected to be performed over the  five year
 contract period was derived by calculating the total  number of  hours of  lane
 operation, estimating the average number of vehicles  per lane and multiplying
 the two.  A lane is assumed to operate for 60 hours a week (lane  operation
 times were found to vary from 54 to 64 hours per week),  52 weeks  a year  for
 five years for a total of 15,600 hours.  Lanes are assumed to have a peak
 throughput capacity of 25 vehicles per hour.  Modern  centralized inspection
 networks are designed so that they can accommodate peak demand  periods with
 all lanes operating at this throughput rate.  Networks are usually designed  so
 that average throughput is 50-65% of peak capacity or 13-15 vehicles  per hour.
 When operating for 15,600 hours over the life of a contract, a  centralized
 inspection lane is estimated to perform a total of 195,000 inspections,  or
 about 39,000 per year.

    5.2.3.2      The  Kffeet  of  Changing  Throughput

        The addition of evaporative  system pressure  testing to a centralized
program would result in a slight decrease  in the throughput capacity.  The
 addition of purge and transient testing, along with pressure testing,  would
 result in a further decrease.

        Assuming the sane test  frequency (i.e.,  annual  or biennial) the
 reduced throughput rat* means that the number of  lanes  needed to test a given
 number of vehicles would increase accordingly, as  would the size of the
 network infrastructure needed to support the test  program.  The result  is an
 increase in the cost per vehicle.  Actual  consumer cost depends on the  test
 frequency; EPA would encourage  states to adopt biennial programs to reduce the
 costs and imposition of the program.  Less frequent  testing only slightly
 reduces the emission reduction  benefits  while  cutting test costs almost in
 half.
                                      H-6

-------
         One way to estimate  the cost would be to simulate an actual network
 of  stations and lanes in a given city.  One could attempt to assess land
 coats, building coats, staff and equipment costs,  costs for all necessary
 support  systems, and other cost factors.  However,  this approach would be very
 time consuming  and would rely on information which is proprietary to the
 private  contractors that operate the programs and is, therefore,  unavailable.
 Instead, the cost  of the increased number of lanes and stations is derived by
 analyzing current  costs and  subtracting out equipment, direct personnel,
 construction, and  state agency oversight costs.  The remainder is adjusted by
 the change  in throughput in the new system.  Then,  new estimates of equipment,
 personnel,  construction, and oversight costs are added back in to obtain the
 estimated total cost.

         As  discussed previously, the typical high volume station can test 25
 vehicles per hour,  performing (in most cases) a test consisting of 30 seconds
 of  high  speed preconditioning or testing, followed by 30 seconds of idle
 testing.  In addition, a short time is spent getting the vehicle into position
 and preparing it for testing.  This leads to a two to three minute test time
 on  average,  depending upon what short test is performed.  EPA recently issued
 alternative test procedures  for steady-state tests that reduce various
 problems associated with those tests, especially false failures, but at a cost
 of  longer average per test time.

         Current costs were estimated by contacting operating program
personnel, equipment vendors and contractors.  The most sophisticated
equipment installation (i.e., the equipment for loaded steady-state testing)
was used to  estimate current equipment costs.

         The cost to acquire  and install a single curve dynamometer and an
analyzer in  existing networks ia about $40,000 or 21$ per vehicle  using the
basic teat volume assumptions.  As indicated previously, a staff person is
assumed to earn 36.00 per hour.  When this figure is multiplied by 15,600
total contract  hours and divided by 195,000 vehicles, direct staff costs  are
estimated at 48$ per vehicle.  Existing centralized  networks typically have
two staff per Ian*.  Thus, total staff costs work out to 96$ p«r vehicle.
Total average construction costs are estimated at 9800,000 for a  five  lane
 station, yielding an average p«r vehicle cost of 826.  In this analysis a
figure of $1.25 is  used to estimate the amount of the state retainer.   This
 reflects EPA*s  best estimate of the per vehicle expense for a good state
quality assurance program in a centralized network.   Equipment,  staff,
construction, and state costs add up to $3.24 per vehicle.  Subtracting this
amount from the current average of $8.50 leaves $5.26 in infrastructure costs
and other overhead  expenses  including employee benefits and employer taxes  as
                                      H-7

-------
 shown  in Table 5-7.  This amount is then factored by the change in the
 throughput rate and the equipment, oversight,  and staff costs for the new
 tests  are then added.

                                   Table  5-7
                            Current  Program Costs
                                                  Total Cost Less
          Increments           Per Vehicle Cost   Increments
          Current                                     $8.50
          Equipment                $0.21              $8.29
          Staff                    $0.96              $7.33
          Construction             $0.82              $6.51
          State Retainer           $1.25              $5.26

    5.2.3.2     Costs  of New Tests

        Most centralized programs use a two position test queue;  emission
test are done in one position while emission control devices are checked in
the other, along with other functions such as fee  collection.   In  this  type of
system the throughput rate is determined by the length of time required to
perform the longest step in the sequence, not by length of the entire test
sequence.  The new tests would likely be performed in a three position  test
queue,  with one position dedicated to fee collection and other administrative
functions, one to performing the pressure test, and the third to  performing
the transient and purge tests.  The transient/purge test is a longer test
procedure than the ones currently used in most I/M programs and is the  longest
single procedure in the whole inspection process.   Thus,  it is the determining
factor in lane throughput and will therefore influence the number of test
sites required.

        The transient test takes  a maxinmm of  four minutes to perform.  An
additional minute is assumed to prepare the vehicle for testing,  for a maximum
total of five minutes.  The pressure test would take approximately two
minutes, and could be shortened through such potential strategies as
computerized monitoring of the rate of pressure drop.  EPA is in the process
of looking at potential fast-pass and fast-fail strategies, and'preliminary
results suggest that roughly 33% of the vehicles tested could be fast passed
or failed based upon analysis of data gathered during the first 93 seconds of
the IM240 (i.e.. Bag 1) using separate fast-pass and fast-fail cutpoints.
Hence,  EPA estimates that the average total test time could be shortened  to at
least four minutes per vehicle.   This translates into a throughput capacity of
15 vehicles per hour.  To accommodate peak demand periods and maintain  short
wait times, a design throughput rate of half  of capacity  is assumed, for  a
                                      H-8

-------
 typical throughput  rate  of  7.5 vehicles p«r hour.  Assuming the same number of
 hours of lane operation  as  previously, the total number of tests per lane in a
 transient lane is estimated to be 117,000 over the five year contract period.

         State quality assurance  program costs would increase given the
 complexity and diversity of the  test system; an estimate of an additional 50$
 is used here  but the  amount could vary depending upon the intensity of the
 oversight function  the state chooses.  Staff costs per vehicle are calculated
 using the same assumptions  for wages and hours of operation as shown in Table
 5-7;  however,  the cost is spread over 117,000 tests over the life of the
 contract rather than  195,000.  The result is staff costs of 80* per staff per
 vehicle.   Three staff per lane are assumed to perform the tests.  The
 additional tasks performed  by inspectors in conducting the new tests - i.e.,
 disconnecting vapor lines and connecting them to analytical equipment for the
 evaporative tests and driving the vehicle through the transient driving cycle
 -  do  not  require that inspectors have higher levels of skill than they do
 presently.  Rather, these tasks  can be performed by comparably skilled
 individuals trained to these specific tasks.  Total staff costs work out to
 $2.40 per vehicle.  Equipment costs for each test procedure are derived by
 taking  the equipment  coats  from  Table 5-4 and calculating the costs of five
 years worth of expendables  using the figures in Table 5-5 and dividing by
 117,000.   Construction costs for a five lane station are assumed to rise to
 $1,000,000.   This is  due; to the  fact that slightly longer lanes may be needed
 in order  to accommodate  test equipment and facilitate faster throughput.
 Dividing  this  figure  by  117,000  vehicles per lane yields a per vehicle cost  of
 $1.71.  The resulting costs estimates are shown in Table 5-8.  Table 5-8 shows
 the result  of  factoring  the figure of 35.26, from Table 5-7, by the change  in
 the throughput  rate and  adding in the equipment, staff, construction and state
 costs associated with the new test procedures.  The figure of $5.26 is
multiplied by 12.5/7.3,  i.e., the ratio of the design throughput  rate in the
 current program to  the- design throughput rate in a program conducting pressure
purge and transient testing.
                                      H-9

-------
                                   Table  5-8
           Coats  to  Add  Proposed  Tests  to  Centralised  Programs

                                                     Running Total
    Increments                Per Vehicle  Cost        Cost per Vehicle
    Adjust for Throughput     $5.26 *  12.5/7.5        $9.12
    Staff                     $2.40                  $11.52
    Construction              $1.71                  $13.23
    Oversight                 $1.75                  $14.98
    Pressure Test              $0.13                  $15.11
    Purge Test                $0.41                  $15.52
    Transient Test            $0.87                  $16.40

        Thus, the cost of adding the new tests  to centralized networks is
found to be about double the current average cost.   The cost of centralized
test systems has been dropping in the past few  years as a result of
competitive pressures and efficiency improvements.   These factors may drive
down the costs of the new tests as well,  especially as they relate to
equipment costs.  Given that conservative assumptions were made regarding
equipment costs of $144,000 per lane,  and low throughput rates, the cost
estimate presented here can be fairly viewed as a worst case assumption.  As
discussed earlier, the important issue is the quality of the test, not the
frequency/ so doing these tests on a biennial basis would offset the increased
per test cost.

    5.2.4     Coat  to  Upgrade  Decentralized  Programa

    5.2.4.1      Baaie   Assumptions

        The methodology used to estimate costs in decentralized programs is
similar to that described above for centralized programs.   Equipment  and labor
costs are key variables aa they were in determining costs  for  centralized
programs.  However, estimates of costs for decentralized programs  presented
here do not include estimates of land costs and overhead.   While inspections
in decentralized programs are generally conducted in pre-existing facilities
rather than newly built ones, there are nonetheless a variety of overhead
expenses as well aa opportunity costs associated with making space available
for inspections in a facility that provides a number of other services as
well.   Data on these costs are not available and they cannot be deduced from
reported inspection fees since, in most programs, fees are capped by law and,
hence, do not reflect the actual cost of providing an inspection.
                                     H-10

-------
         Total test volume rather than throughput and test time are the
 critical factors  affecting cost in decentralized programs.   Licensed
 inspection stations at present only perform, on the average,  about 1,025
 inspections per year, as  shown in Table 5-9 (note that this number is  a
 station-weighted  average).  Test volumes among stations in a single program
 can vary widely as shown  in Section 7.0.  It should also be noted that all
 decentralized programs in enhanced I/M areas,  except for California, Virginia,
 and Colorado (which tests vehicles five years old and newer biennially, and
 vehicles older than five  years annually) are annual programs.  In this
 analysis the effect on per vehicle costs of switching from an annual
 inspection frequency to biennial, as well the effect of varying inspection
 volume,  will be examined.
                                   Table  5-9
            Inspection Volumes  in  Licensed  Inspection  Stations

    Program                 Vehicles  per Year        Vehicles  per Station
    California              6,180,093               799
    Colorado                1,655,897               1,104
    Dallas/Ft. Worth        1,948,333               1,624
    Bl Paso                 278,540                 1,161
    Georgia                 1,118,448               1,729
    Houston                 1,482,349               1,348
    Louisiana               145,175                 1,037
    Massachusetts           3,700,000               1,321
    Nevada                  523,098                 1,260
    New Hampshire           137,137                 564
    New York                4,605,158               1,071
    Pennsylvania            3,202,450               834
    Rhode Island            650,000                 684
    Virginia                481,305                 1,301
    Weighted Average                                1,025

         Annual tests of 1,025 vehicles per  station  is  equivalent  to between
three and four inspections per day depending upon the  number of days per week
the facility  is open and  inspections are available.  This is far below the 75
inspections per day projected in a multi-position high volume lane with three
inspectors conducting high-tech tests, and  significantly below the 16
inspections per day that  could be don* in a single position inspection bay
with only one inspector (the derivation of  this  figure is detailed below).
Two conclusions can be drawn from this.  The first  is  that the additional time
requirements of the new tests will not force a reduction in the total number
of inspections that most  stations can perform.   The second is that, because
costs are spread over a smaller number of vehicles  than in the case of high-
                                     H-ll

-------
 volume, centralized stations, the coat per vehicle for the new teats will be
 larger in this type of inspection network.

        The  higher costs for high-tech testing equipment have prompted
 questions of whether all current inspection stations would choose  to stay in
 the inspection business with the implementation of an enhanced program,  and
 how high a drop-out rate programs would experience if some did not.  EPA knows
 of no data or reasonable assumptions by which a station drop-out rate  could be
 reliably estimated.  In this analysis inspection costs for high-tech testing
 are estimated for three scenarios: one where all stations remain in the
 inspection business, one where 50% of the stations drop out,  and one where
 enough stations drop out such that those that remain are operating at  maximum
 possible volume assuming that each has one inspection bay which has not been
 improved for high throughput and one inspector performing all parts of the
 inspection.  In all three scenarios a biennial inspection frequency is
 assumed.

        The current average test fee for vehicle inspection in decentralized
 programs is about $17.70 (again, the derivation of this figure can be found in
 EPA's technical information document, "I/M Network Type: Effects on Emission
 Reductions, Cost, and Convenience").  Note that this figure may substantially
 underestimate actual costs since most states limit the inspection fee that a
 station may charge.  In many cases, the actual fee is likely to be below cost;
 stations presumably obtain sufficient revenue to stay in business by providing
 other services, which may include repair.  It should also be noted that the
 intensity of the inspection and the sophistication and cost of the analyzer
 vary significantly among programs.  Average inspection costs and revenues by
program,  taking these factors into account, are estimated in Section 7.4.1.

        The costs for adding high-tech tests are derived by estimating the
per vehicle costs of the key components: labor; equipment, including
expendables; and support, i.e., service contracts and annual updates.   Per
vehicle costs are estimated by deriving total costs  for each component and
dividing by the number of vehicle inspections expected to be performed in a
 year,  again, taking into account variations in  inspection volumes  and changes
 in frequency.  Equipment: costs are spread over  the useful life of  the
 equipment.  While a piece of equipment's useful life can vary considerably in
 actual practice, a five year equipment life is  assumed.

        While large businesses,  such as  dealerships, may be  able to afford to
 purchase current analyzer equipment  outright,  the  smaller gas stations and
 garages typically have to finance these purchases  (although  in some cases  they
 may lease equipment).  The higher cost of the  equipment  needed to perform
                                      H-12

-------
 purge and transient testing  ($144,000 for the dynamometer,  CVS,  analyzers,
 etc.,  as  opposed to $12,000 to $15,000 for the most sophisticated of the
 current NDXR-based analyzers) makes it even more likely that these purchases
 will have to be financed for most inspection stations.   Equipment costs are
 amortized over five years at 12% interest in the analysis in this report.

        Program personnel in decentralized programs were contacted to
 determine inspector wage rates.  In many cases, inspectors are professional
 mechanics earning about $25 per hour.  However, most states do not require
 inspectors to be mechanics, and inspections may be performed by less skilled
 individuals who typically earn $6 or $7 per hour.  The prevalence of different
 wage rates among inspectors is unknown.  Therefore, EPA assumed an average
 wage of $15 per hour for this analysis.  An overhead rate of 40% is assumed,
 for  a  total labor cost of $21 an hour.

    5.2.4.3  Coat  Components  and  Scenarios

        The full test, including data entry on the computer, preparing the
 vehicle for the different steps in the test procedure and conducting them,  is
 estimated to take 30 minutes with only one inspector performing all tasks in a
 repair bay that is not configured specifically for inspection throughput.
 With labor costs at $21 per hour, as described above, this works out to $11.50
 per vehicle.  Equipment costa arc taken from Table 5-4 and are amortized over
 a five year period at 12 percent annual interest  (changing the assumed
 interest  rate does not significantly affect the total per vehicle cost).  This
 brings the total cost for the equipment package over the five year period to
 $192,325.  These costs arc divided by five years worth of inspections.  The
 costs of expendables from Table 5-5 are added in according to the usage rates
 assumed for decentralized programs.  Two other expenses typically encountered
 in decentralized programs arc service contracts and software updates.  Based
 on information from states, service contracts are estimated at $200 per month
 and annual software updates are assumed to cost $1,500.

        Per vehicle costa arc estimated for three scenarios, biennial  testing
 is assumed in all three.  la the first, all stations remain in the inspection
program.   la the second> 50 percent of the stations drop out of the program,
 and in third there are only the minimum number of  stations in the program to
enable each to inspect at full volume with one inspector performing all parts
 of the inspection and a service station bay that  has not been improved for
 high throughput.

        In the first scenario, the switch to biennial  would mean that  annual
volume is  cut in half, or 513 vehicles per year.   In the second scenario the
                                     H-13

-------
 SO percent reduction in the number of stations brings the annual inspection
 volume back to 1,025.  In the fourth scenario, it is assumed that each station
 inspects at maximum capacity, i.e., one vehicle every thirty minutes, and that
 an inspector is available 50 hours per week.  This results in an annual volume
 of 5,200 vehicles.

                                   Table  5-10
      Cost*  to  Conduct High-Tech  Testing  in  Decentralized Program*

    Scenario                Annual Volume           Cost per vehicle
    No Drop-out              513                     $106
    50% Drop-out              1,025                   $58
    72% Drop-out              5,200                   $32
 (Maximum volume)

        Note that while reducing inspection frequency to biennial cuts
motorists' costs in centralized programs, in decentralized programs  such cost
reductions are only achieved by reducing opportunities  for stations  to
participate.  In the scenario in which 50 percent of the stations drop out  and
testing is biennial, annual station volume is the same  as if testing were
annual and no stations dropped out.  Hence, the  estimated per  vehicle cost  in
a biennial program with a 50 percent station drop-out rate is  the sane as
would be derived for an annual program with no  stations dropping out.
Reducing inspection frequency to biennial, while maintaining the same number
of stations, has the effect of almost doubling  the  per vehicle cost  since
operating costs are spread over half as many vehicles.   Note also that the  per
vehicle cost far exceeds the per vehicle cost in centralized programs except
in the scenario where 72 percent of the stations drop-out.

   5.3     Costs)  of  four-Mode,  Purge  and  Pressure  Testing

        It has been proposed that  a  series of simpler, loaded mode and other
steady-state tests would provide equivalent emission reductions to the IM240
at a lower cost.  The emission reduction potential of this approach is
currently being evaluated at EPA's test lane in Phoenix, Arizona.  The
information needed to do a cost analysis can be approximated at this time
based upon the test process.

        The test procedure being evaluated is a series of emission tests
referred to as the four-mode test: A 40 second 5015 mode  (IS mph at a load
equivalent to ETN / 250), a  40 second 2525 mode (25 mph at load equivalent to
ETW / 300), a 40 second mode at 50 mph and normal road load,  and a 40 second
idle mode.  EPA anticipates a 30-60  second preconditioning mode would be
                                     H-14

-------
 needed to  insure proper warm-up and canister purge down.  Allowing also for
 necessary  tin* to transition between test modes (5-10  seconds), the four-mode
 test  would require a total of approximately four minutes.  As with the IM240-
 based test scenario, purge testing is assumed to occur simultaneously with the
 tailpipe test and pressure testing would be done separately.  It should be
 noted,  however, that some vehicles may not purge during this test and may
 require a  short transient retest to activate purge.

    5.3.1  equipment  and  Expendables

        The equipment used for the four-mode test is simpler than  for  the
 IM240  test.  The dynamometer may not need inertia weights, and  a raw gas
 analyzer,  like the ones used in the current I/M testa, is upgraded with a NOx
 analyzer and an anemometer, to enable mass concentration calculations, for
 this test.  The equipment for the purge and pressure test are the  sane as
 described  previously.  The estimated costs are shown in Table 5-11.

                                   Table  5-11
                    equipment  and  Cost*  foe  the ASM  Test
            Pressure System             $600
            Flow Sensor                 $500
            Dynamometer                 $20,000
            Anemometer                  $2,000
            BAR90 w/NOx Analyzer        $16,900
            Total                       $40,000

        Expendables for this test are nitrogen gas for the pressure teat and
calibration gases for the analyzer.  The cost of nitrogen gaa ia the same as
in the previous analysis oa XM240 coata (the pressure teat  procedure ia the
same regardless of the type of tailpipe test used).  Current calibration gases
are multi-blends consisting of propane, CO, and CO2.  A coat of 945 per bottle
is used here.  In thia analysis, it ia assumed that multi-blend gases that
include NO will b» available at the same cost.  Alternatively,  one could
assume that two bottle* of calibration gaa, one current standard multi-blend
and a bottle of NO will be needed, however, the additional coat per teat ia
insignificant (leas than 5$, even in a low volume situation).

    5.3.2   Centralized  Vrogxaaa

        The total teat time per vehicle would be  about 11 minutea, including
administrative processing in an efficiently run testing lane.  In a multi-
position lane the throughput would be governed by teat time at the longest
position,  which would be four minutea.  Thia translates into a peak throughput
                                     H-15

-------
 rate of 15 vehicles per hour and, using the standard design criteria for
 centralized programs described earlier, an average throughput of 7.5 vehicles
 per hour.  Using the lane operation assumptions detailed earlier, this
 translates into 23,400 vehicles per lane per year and 117,000 vehicles over an
 assumed five year contract period.  Three staff per lane would be needed to
 perform the entire test sequence including inputting vehicle identification
 information, conducting the tests and presenting and explaining the results to
 the motorist.
        The per vehicle cost of the four-mode test in centralized programs is
 estimated by the same methodology as was used to estimate  IM240 costs.
 Current costs for test equipment, staff, state oversight,  and construction are
 subtracted from the current average per vehicle cost, this amount is factored
 by the change in throughput, and estimated costs for equipment, staff,
 construction, and state oversight in a four-mode test program are added to
 obtain an estimated total cost.

                                   Table  5-12
           Coats  to  Add  Proposed Tests  to  Centralised Programs
                                                     Running Total
    Increments                Per Vehicle Cost       Cost per Vehicle
    Adjust for Throughput     S5.26 * 12.5/7.5       $9.12
    Staff                      $2.40                  $11.52
    Construction              $1.71                  $13.23
    Oversight                 $1.75                  $14.98
    Pressure Test              $0.13                  $15.11
    Purge  Test                $0.18                  $15.29
    Four-mode Test            $0.35                  $15.64

    5.3.3   Decentrallied  Programs

        The same methodology used to estimate costs of IM240 testing is used
here.   Most assumptions are unchanged.  Total test time is thirty minutes,
equipment is amortized over a five year period.   Two parameters are changed in
this analysis: equipment costs total $40,000 instead of $144,100,  and state
costs include a coat for state mass emission testing.
                                   Table  5-13
    Cost*  to  Conduct roue-Mod*  Testing  in  Decentralized  Programs
    Scenario                Annual Volume           Cost per Vehicle
    No Drop-out             513                     $51
    50% Drop-out            1,025                   $31
    72% Drop-out            5,200                   $25
                                     H-16

-------
                   Appendix  I:
ASM  and IM240  Credits  for  State Implementation
             Plans  With MOBILES  Runs

-------
en
j



14
Hi
^5
n
ll
nt
                                II
                                        H «


                                        9*


                                    ....3"
                                «««»(<*SjN
                                OOOXXXXZ'tt
                                ...      $
                                <4> « W      -4
                                « A 0>      u

                                         o.
                                    333   : -
                                    111  .14 t
                                    353  3fl
                                    >t >• >l •• O I •• •• O
                                    -IH-I • al •• a pi •
                                    a a s •* • 41 • I S »4


                                    3>iii«!9N!

                                    :::,iis*4si
                                    XXX      i   c2 4i'
Ck tlH   r- •  ~ -
                                    O O O
                                        H
                                     •
                                     Cb •
o 004 u a
MMM lj 0 4
I I I • 41 g
Ki-liH-H O I
r»r-i-ri-* H
IK at«  M •. a
•**•** u u 4
                                        !:s:
                                        • u

                                    539 •"« * |
                                  a sssf-a::
                                    ..  i^^f
                                                     i

                                                     H
                                                   b


                                                   O
                                                         O fH * iH

                                                         O 10 O M
                                                         O in M m
                                             333
                                             '    f
                                             ^::0:
                                             0 X X X
                                                 w z ••
                                                 [»$ j
                                         M O
                                         0 O
                                         U O
                                         0 M
                                               fei
                                               bi
                                               bi
                                               b
                                                         O «0 CM r4 O O O *• *
                                               k
                     o  MM



                     ^5
                     « Z n r» <

                     OOrffHOOOOl
B

                                                         o  i-ii-i
                                                                           H

-------
     s*»
    134
     3821
     ^0.0.41,

     •M *  •*

     0^93
     S  Y8:
<3\
0 O
410
001
« „
        I *
        I o  m A 1-4
        Ka o  in eo n

        IA o  m -4 rt
        M
             <0  00
             O  N 0«
             0  OO
             o  « «
             o  r^ rf
             o  ww

             o  0 O
                    A «


                    f-i A
b
            I H

      n r« 6  two  wmr-vwMwn


      a, ji    |S °   ""    5 *
                rt rtOO OO«i-l
     •     n  I  co  «»M«MWr-»
     J)     Fl  MO  



                               O
                                     o  tf>Mm
                    on    I  CM
                    j   > I  r- •»»
                    ....  § H« --
                    g^  s B0 MN
                             -H 3



                             iSlj
fe
                                 o » »
                                 o r- r-

                                 o o o
                             > I  o 4O w m

                                                  IA d  N i4 d e d d » «
                                                  M           -<





                                                  I  SL.H	


                                                  |«0 _r*MOOOO»-«
                                           u w
                                           OO
                                           41 O
                                           O N
                                                B
                           U
                          A£
                          A OA«OA«»«A
                                                             o  iiooooono
                           I
                           i
                          r,
                                                               I -
                                                i
                                                                          I

                                                                          H

-------
CO
vo
               «MMMMOM«0


               MMOOOO»M
      O
      I O

      !«
      nu
      M
• in w
 i^d
r:

]!i
       I  S „,.
     u h° •»«•«
     » |M° ""*'

     > I  r» MM
     g to o MM
     H WJ O MM



     Ik


     ik


     Iki
                      n « r-
                      * r- r»

                      o « o
                       M A
                       N N
« o in « <<» «
        k!
         N I  «  «MMin»»rtA

            • O  O«MNMOr-»
             « X « « -•« »<
             r^v,IOOr-l Am o n M CD
                        • &B>  H K«IM~XIO ooiMMono
                         95 5 -o o  •••	
                         uuiq MIOOMMOOOOVM
                        T Qi O. U J M  —        M
                           MZ -
                                        » IM r- •
                                         O O «
 §  u B   i  «w
 i  t  o> 1  <» o o> w o> 01 » IM r-
   •H
-------
en
en
vo


>t
                                                              I
 o£ -U
  -j    i  *
 *•••  H I  O  t» r»      »H
 b   g to o  o  I  o  mm      Of
 innoKoowA      «n

 r- r-  3 IAO  o o      -!'*      HW


 ••••  • IAO  WNHOOOA«
 »J    M           tc!*r:

                                                         s.*J*a9B   P"*      5—
                                                        5W«'!i
                                                         3 u u
5  ;

=s
                                                                I S«o.
                                                                ^.&:°.
                                                                P°.—

-------
en
at
                tMt-tOOOO-Oi-l
             O  rt 01 t^   (H      I n

         I  hs  °°
     §1  *  p°  ""•
     I
£*  §  [•?  «3
*"   9  lei o  o o
in»     H

-d
> «•


^  >  I O  M^

!•?  a  H:°  **
. r-  3|«o  o o
>M     M
            fe
     n:

     4J Ol

    134
    U -H 41
     n
     4J »
     34
     •H 41

    fi
    H4 • •

    OXX!
     'iis;   i  <»;
     iir(-"-
              in m o w •* 0 o i^ <•
              O w r* i^ ^ w »-4 »-4 m
           to <• A « CKW «l~





           04>4OOOOr>i-l
            I  «
            I  o» o«M• o n <•


            M0 **'*oooeg'<
     ...  1
     XkCk 4>>

    «4X  -H
     0*83
     M ' P
       S4J R
       I 01
       •HWI
     •  4» «(

     s  a5-


     5-

     2§
            1  2  »        «

            n  •i.'^ ."I*"J"!®* .
            po^^oooon-
•   si
    I
K,   ;.    1  .-

I   li    :.«
                                      o m »-»«n   o « o
                              *§
                              Q«
 II

 13
 ~a
                              in »


                              i>d
 intn

 PT>
                             I «<•
                        r- o
                        mn
 n
i.i
is*
     i  ^
     I  O  » »      n r4
     M o  r- r>      « n

     Ud  o d      ^t-i





V.    I  M
   > I  o  <• <•      CD o>
nQMOinin      «t*

i-  3 latd  d d      rf-«
M    p










     fe
         n  nwnx«««o
         o  r>N<»«inoo«
                                      O  «Nf«OOO««
                               i
          A n t^ N *H o 


          *4 KO OO O <• •
                                    I  ""
                                  2 LS  53S55SS

                                 ,9 B°  «^«««»°»
                        ill-
                        } Ot 0> .

                        S^j'

                        sMr§b. i  o«»»««,«.,»
                        •  i'SS r'T2 p


                                                     »x  .-5
                         5  i1
                         on



                         li   I
                                                                                     in

                                                                                     I

                                                                                     H

-------
ro
vo
               fi-t.-l~  % Uo
             o n 01 *   n «?-
             o « I  n •»<0o«or-w««
U I  « g « o o « c» n n o

   4J*q p • u	
   qiti mo M^ooooni-i
   +    \* a        H
o o
4J O
o N

t~
Typ«:
s   ii
     ^3
     o +
                                I
                                    O mr4(n   o •* o
 n m
 t» i»
  o
 • M
k!

bi

fe

b
                                                                                H
                                    O l-> -H
                                    O m m

                                    o o O
                                    o o » w » N o n «o

                                    O n^oooootn
                       •      I  S r-«eoin»nr4r.
                       d    H wo o in •* M ^ o » «


                       Mi3o9M          -i"4
MX -4     •
o* i a     y
 Mil     O
s  *• 5^ i  «*•
4  I O> I  « O » »i « « in 
-------
  PI  *O O * «» *
  CD *  I 'H
     •«
      OI
                  :s.
                      288S
               g i i  i i
1  b •  U UOI • d  O I      «IO>..
M  3  a! Q.r-i o o NO H>IHM> •» • • •
Si  3rfg«8~"- g-HH-H 6>Q H H U H g «* 41 41
o  >I  S u M--4 o o g S3 S S H v»fl-J-}4
  ttfs-Hii"!*1
  SS3333I333*
  01 » h J * * O H * H M
                                         i >l •• 0 •  •• •• O

osi^lhri3
    t>  >••»>  a
4J 41 41 0 13 • O  04 0
:ssii..«"
HHH M-4 » « .
      -^ *••
                                                  I
                                                    o  n«f
                                                    o  « 4
                            AMNNNOV^
                                                         ^F4oooomF4
                                                     k
                                                     H

                                                     ki
                                              !|   M3?"
                                              2x   • «S• • *
                                                    09 r-co r-«•«•«>• r»
                                                    O n»e	

                                                    O N>4<
                                                    AZNWNmp|iifAr«
                                                    ^•XOONNNOVt
                                                    o a-*** o o o o •••*

-------
n
CM



H

•H

M
                 oiwninor-ao
       £


      ,§
             I w
             I O  n A i*   m onr i
             ho o  « CD M   •» r-1


             IAO  mr-tm   o«c
             l-l           IN
   SS>    I  A
     >  I  w  o o
     a  koo  c««i

 ai §  Ud    I  O  rtr-l
 m n a  fa o  w w
  . . Q  I . .   .  .
 r- r- «3  MI o  o o
 CD N    M





£2 >  I  s  .-
u« »
                         » at


                         H n
                         mo
                         nn
    133
    41 -H 41
                 I~ n |H OT N W A «
             I W

     Jl     H  (WO  M » M M0« O O W


     ««!o3  B°  NMOOOO«rt
     ox>.£
        s:   , .5
      bUUi*  1 A X »4 M
            ,  .
            1  A X »4 MO no w o r«
            f« H S f O P» N «• O » «•
            \  ° ' M  ........
            mo Oi-iiHoooof*i<-i
            M  ->         rt
      ..93   ,   8
     •  41 M   I fl 41
     •  I 0>  I r-l O »« W A A W A O
       ^•MU  IWW «in«Hr-INOini-t
     •  •"JH  G •••	
     s  a"  |2°g'«>°°oo3-
     l-)Ch»4l.


       I ?  »»
                                   Q  koo  -<-i


                               §•8     P°  NM
                               I"
                               c 41






                               n w


                               CD (<•
        k
                                       o  » »
                                       o  r- r-


                                       O  O O
      >  I  O w w      «g>
 innQlwoinin      » M

 r>r>3|Aooo      *•*






      >  I  n r«c>«A
      U  wo n*«««Aoom

        JA   nMHOooo*







ill  s
                              OHM:
                                      I S


                                      IAO  i-i -idddd w «
                                 I Aznnw

                                 Ht^*!*" ..-.-.  :
                                 kjoOi-iooooonH



                          5j|      I

                          M3fi^  i wS

                            J,^R  l.S2SS32?S^S
                            34liQ  I • • b	•
                            5aa  ICJO^OOOOOMM



                         Sn         3
                         00         •


                         S8
                                             Utitttt
                                     .
                                 k
                                                                   •HOOOOONr*
                                                                W
                                                                o  n A iH
                                                                o  m IB IN
                          i*

                          §•3
                          H 9
                                                        t> O

                                                        CD N
                                                       U V."x


                                                       1-5-
                                                       o MX;

                                                       ^J i* ••
                         3221!
                         ^ o»e> 41.

                         « *  --1
                         I^Sjg

                         '  Sjf
                            2a!
                                                       si
  i  ^
H I  O  -•-
S k«o  r-r-

3 IAO  o o






  I  ™
> I  O  -IrH
O Ko O  in n

3 Ud  do
CO


I


H
                                                                           4




   P°  Hrt<'0002'<



      T


   I Axon«nn»o-4




   |S  —rt       ^






   |«OW»^AA»«A



   Ud  riddddddd
   H  d        »^
                                                       35    .  «p
                                                       ;.   -8   -J:

-------
CM


H
•r»
M


a
           3
                    rt p- » p- It CM ••<
               o  « rt   I  f  Ml-t
      Q  KO O  <-l •*
      Q  I '  •   • •
      8  IAO  CMCM







         IS  A A
      S  MO  ««

      9  |jd  dd
       ;«  i LI  ss
       '*  Q I * *   *  '
       •r-  3 IAO  oo
       i CM    M
                            •-•A
                            CM CM


                            r4t~
      H 9

      JiJ
      m -io
     o^S


     41 ..
         I  S  »HH  ««(••
d     S  Ue  V 14 fl M d O <• «

44 ••  8  Uo  HHadddiA^
»• • • 5 3  P            "


            "J
            ^
         I  A -4
         |AX«niiini>««OM


         jOJOO^MOOOOOCMrt





         ,  J
         I  A oonOA

         ff^atriit0.':"
         IAO  dooooooo
     1 QiO. 41


      I  to  o o      cow

....  § H°  "^      "•

§1  8 12°  NN      2*


S3

 H    I  «n
— 3  N I  O  AA      COM
b    o|«o I  O  rf t««
                                                   55
                                                                                                 I

                                                                                                H

-------
m
 (M
      « m

      p-
      W

 •'    MV.^

 • r-  • • m  »
         o M v o a .0»
         -|o**S
           CMO OM«or-1

           cnnoooow><
       I «
       I eg  o nonww <»r>
       MO  «n«N
                                          j  X lai o  m IH m
                          0 in     1 at
                          J   >  I «  o o


                          e •  9  ki o  N N

                          8]     B



                           -t     i -^
                          -~4  H  I O  00
                          b   Q  K» O  CD 00
                          —   Q  I . .   •  •
                              3  mo  o o
                          mte     |)-<
 in m

 r- r-
 (D ««««n


                                       to  ««o"ooo^
                                       IS             m
                           S4» M     n 41
                            I o>  I  <-> 0 « »P- « •» « 01 o
                           -l-uG  M« • « •(N.-.IOO A«
                           3««2   •  • c.	
                            qi23  loio  N>-I o o o o « <-•
                           *     Ft  d         H
                                                        i
                                                                          o  n a* M   n » r-
                                                               m a M   vr- r-

                                                            o  in *4 m   o •
                                                                  c J
                                                                  si
                                                              >  I  r-  •»•
                                                              *  liO
                                                                                            I

                                                                                            H
                                                                             Sw w
                                                                             n in
                                                           i  *
                                                           I  n  »or-Hi-i««<»
                                                           MO  r~io«»«oo«
                                                           mo  lortt-iooon^
                                                           H            •»
                                                                 §M A m o *> irt
                                                                 « N is m o n
                               N  I  « mmom*inf-«

                               Q  mo nr«oooOr4"<
                            5 o3  M            n
                          J 9 Z 4


                          '33L  i  sL,.«««MO
                          ^§••98  hrS"*^"?"*
                          1 " **  9  B°S.N« •
                                                   •     4    >9  3 o • q q 3 s
                                                   • •    •   •   Si  3 is H ^ ^ <
                                                    SH    >   .4   f" ' --  ^ - ' '


                                                    \       >   \
                                                   o o    o +    o

-------
CO
O\
<3\
CM
CM

H
•r|
M
          I
 i  *
>§
--8
si  i
13
**  s
      «>••»
     • m w
     >« • •
      i» o
     r:
     B«
 4J W
134
^U
o J S
             o  n A >4   n «r-
             o  in o r<   •* r* r-
             o  « •*!•!   o « o
             o  nn
             d  do
       I  n  on«o«r-mo

       M          n
     f|
     i    N I CD  r-<-iainA«A«
     41    H laid  nrtdddddei
     w 0 0 O 3 M          n
R5j-
 M 9
!  J-
      I * 4*

      bd^^dddoS5
      JAO  NHOOOOAr4
     0 O
      O
                          I
                                   O  «i-* m
                             o
                            I O
                            r- >

                            13
                                   n
                                     §A A
                                     <0«
                            t* O
                            co n
                         > i  o  o o
                     mm  Q V* O  in«
                      " "  Q LJ *   * *
                     r* t*  M W* o  o o
                     « C4    M
                     r"
                     ^-j    i«
                     r- o  > I
                     «
iu
sli
                                        O O O O V *4
                           -5i3
                             I 3  5 01 ik *< a  H —       (M
                     X
                     •4
                     0

                     •M!

                     i«
                     b Oi Ik
                     MX  •
                     O^J
S Y8> I AOOW^A^n«0
3 3^^ |^d^-;dddo«-;
                                                                       I
                                                                      H

-------
                    Appendix J:
Emissions  Analyzer Price  Information from  Horiba

-------
HORIBA
HORIBA INSTRUMENTS INCORPORATED
3901 Varsity Drive. Ann Arbor, Michigan 48108
Telepnone-1(800) 3 HORIBA. In Mich 1(800) 624-0899 or (313) 973-2171
Fax (313)973-7868
                                                          April 7, 1993
          Environmental Protection Agency
          2565 Plymouth Road
          Ann Arbor, Ml  48105

          Attn:   Mr. Bill Pidgeon

          Re:     IM 240 Analyzer Information

          Dear Mr. Pidgeon:

                 This letter is a follow-up to prior discussions we've had regarding the list price of IM 240
          Analyzer Systems.  We  would like to thank you for the opportunity of discussing our equipment and
          market with you and your staff.

                 We would like to make a clarification in reference to the IM  240 pricing.  Horiba is actively
          working with six of the seven centralized contractors.  Four of these contractors currently have IM 240
          analyzers installed. The current list price for the Analyzer/CVS System is $75.515. It should be noted
          that this price does not include a blower or external sample line.  As you can understand, this is a
          "single unit price" and does not reflect discounting  for  quantity  orders.  For long-term  pricing
          considerations, it should be recognized that we also anticipate price reductions following improvements
          in manufacturing efficiencies.

                 Horiba's analytical system can be supplied with other options, such as; a driver's aid, purge
          and pressure equipment, data collection and processing capabilities.


                                      IM 240 Analyzer System

                                      HC - FID
                                      CO • NOIR
                                      CO, - NOIR
                                      NOx - Chemiluminescent
                                      CVS - 500-700 CFM

                                      Total:         $75,515

-------
EPA                                                                 Page 2
Mr. Bill Pidgeon


       We feel that our forte' is in the analytical and sample handling portion of the testing lane. For
this reason, we are providing you with analytical system pricing only.  Most of our customers have
sourced or built the other components themselves.

       If you should have any additional comments or questions, please feel free to contact me at 1-
800-3HORIBA.
                                                 Sincerely, /
                                               /  "   ^ ,;•:/•  V
                                                 Kenneth W. Thomas
                                                 Marketing Manager,
                                                 IM Systems
KWT/pm

cc:     Neal Harvey
       Andy Marko
kt041 .Itr

-------
             Appendix  K:
 Centrifugal  Blower  Price  Quotation
from Combined Fluid Products  Company

-------
                   COMBINED
                   FLUID
                   PRODUCTS
                   COMPANY
   TO:    Environmental Protection Agency
         2565 Plymouth Road
         Ann Arbor,  Michigan   48105
                                    QUOTATION

                                    ISSUED FROM

                                    G  80S Oakwood Rd., Late Zurich, IL 60047
                                        Phone (708) 5404054          FAX (708) 540-0513
                                    G  125 N. Executive Dr., Brookfield, Wl 53005
                                        Phone (414) 258-7770          FAX (414) 821-1492
                                    G  P-O. Box 216, 24 S. Given St., Brownatounj, IN 46112
                                        Phone (317) 852-3961          FAX (317) 852-2337
                                    %  5025 Venture Or., Ann Arbor, Ml 48108
                                        Phone (313) 930-2024          FAX (313) 747-7040

                                    Please Note:
                                    A/R Dept. is located in Lake Zurich. Illinois
   ATTENTION:
 Mr.  Dan Sampson
   EFFECTIVE DATE:    January  27,  1993
                                    REFERENCE:
                                                    QUOTATION NO.:
                                                       AA408
   EXPIRATION DATE:    February 27,  1993

                         in Compliance With Your Request. We Are Pleased to Quote You As Follows:
   QUANTITY
                         DESCRIPTION
    PRICE
    1-10
Paxton Centrifugal  Blower Model RM-87, including:  10 HP
electric motor running at 3*600 RPM on 230/460 volt
vacuum*  three-phase/  60 Hz/ TEFC motor.

- Inlet Filter/Silencer
$3,320.00
 Per Unit Net
  11-100
                                                                  $2,656.00
                                                                   Per  Unit Net
DELIVERY    Six to  eight weaka P.O.B. Santa Monica*  California      (Delivery Subject to Prior Saiei

TERMS.   Net 30 Days. Subject To Credit Approval
                                                              >. Corrunker* Sales Engineer
                  This quotation subject to standard terms and conditions o> sale aa stated on reverse trie.

-------
                  Appendix  L:
       Average IM240  Test Time  Utilizing
Preliminary  Fast-Pass and  Fast-Fail Algorithms

-------
                      Average IM240 Test Time Utilizing
                 Preliminary Fast-Pass and Fast-Fail Algorithms

    The objective of this analysis was  to estimate the average IM240 test time
using algorithms that allow vehicles with very low emissions to  fast-pass  and
vehicles with very high emissions to fast-fail.  This reduces the average  time
required for the IM240, allowing higher throughput,  which reduces the number
of inspection lanes required.  The reduced number of lanes lowers equipment
and personnel costs, having the potential to significantly improve the cost
effectiveness of the I/M program.

    This analysis describes the fast-pass and fast-fail algorithms used to
estimate the average IM240 test time.  The results are preliminary,
representing what could be achieved in time to comply with the court ordered
deadline for this rulemaking.  Developing these algorithms requires using
second-by-second data for HC, CO, NOx,  and purge, which is very time
consuming, given the huge amount of data per vehicle.

    The ideal fast-pass/fast-fail algorithm consists of two continuous
functions.  One function represents emission levels at each second of the
IM240 that reliably predict a passing result while the other function
represents emission levels that reliably predict a failing result.  Because
this requires evaluating the results at each second of the test for each of
the vehicles, we determined that this could not be achieved under the time
constraint.  Instead, we evaluated nine segments (modes)  of the IM240, which
significantly reduces the burden, but gives a less than optimal result.

    So,  additional fast-pass and fast-fail  algorithms will be evaluated in the
future, and additional vehicles will be available for those analyses, so these
results should be regarded as preliminary.   For example,  very low emitters or
extremely high emitters can be fast-passed or fast-failed early in the IM240
cycle,  while vehicles near the certification emission levels will require more
time to accurately predict a passing or failing result.  The emission
reduction benefits, obtained from repairing vehicles whose emission levels are
slightly dirtier than their certification standards, are not very cost
effective.  Similarly, it also may not be cost effective to run the full IM240
as required to accurately distinguish marginal emitters that pass the full
IM240 from marginal emitters that fail.  This can be evaluated by comparing
IDRs, failure rates, and error of commission rates for each second of the
IM240 to determine the best tradeoff.

    Another consideration is the IM240  reversed.   The IM240 was  designed as a
two-mode test.  The second mode includes the maximum speed of 56.7 mph.  The
                                      L-2

-------
IM240-reversed starts with this high speed mode,  then is  followed by the low
speed mode.  This may further reduce the average  test time  required to
distinguish malfunctioning cars from properly functioning cars.  It should be
especially helpful in rapidly determining whether the purge system is
performing adequately.

    The algorithm used in this analysis  was comparatively crude due to time
and data handling constraints.  Several discrete  modes of the IM240 were
selected for determining passing and failing emission levels.  These modes
were selected to avoid ending the test during an  acceleration or deceleration
and to provide a reasonable duration for each of  the nine modes.  The average
IM240 test time was calculated as the average of  the selected mode times
weighted by the number of vehicles passing or failing at  each mode.  A  more
detailed description of the data and methodology  used as  well as the results
are included in the following sections.

    The database used for this analysis  conformed to the model I/M program, so
it was limited to 1986 and newer vehicles with second-by-second IM240  results
- 494 vehicles.  These vehicles were tested between June  4, 1992 and August  4,
1992.  Data were only used if the composite results calculated from the
second-by-second data had passed EPA's quality control measures.  Due  to  the
volume of second-by-second data and the time constraints  involved,  the  second-
by-second data were not QC'd separately.

    The following nine modes were selected for pass/fail determinations:
Modes For Evaluating Fast-Pass And Fast-Fail

Mode
(#)
1
2
3
4
5
6
7
8
9

IM240Mode
(sees.)
0-34
0-60
0-74
0-93
0-113
0-154
0-173
0-206
0-239
IM240 Speed
@ End of Mode
(mph)
22.6
30.4
29.8
0.0
27.2
26.0
47.2
51.6
0.0
    To determine the passing and failing emission levels for each mode/  the
sample was divided into passing and failing vehicles.  The pass/fail
determination was made based on the "two ways to pass" criteria with 0.8 g/mi
                                      L-3

-------
HC,  15.0 g/mi CO and 2.0 g/mi NOx as composite IM240 outpoints and, 0.5 g/mi
HC and 12.0 g/mi CO bag 2 outpoints.  One liter of purge volume was used as
the  outpoint for purge flow.  These criteria are illustrated below.

            Pass/Fail Decisions Based On Two-Ways-To-Pass-Criteria
Decision
IM240
HC
g/mi
Fail

Fail

Fail
>

5

5
0.8

0.8

0.8
IM240
CO
g/mi
515.0

>15.0

5 15.0
Bag 2
HC
g/mi
> 0.5

50.5

50.5
Bag 2
CO
g/mi
5 12.0

>12.0

5 12.0
IM240
NOx
g/mi
5

5

>
2.0

2.0

2.0
Purge
liters
51.0

51.0

51.0
Comments

Must fail HC
Composite &

on both
Bag 2 to


fail
Must fail CO on both
Composite &
Bag 2 to
fail
Only 1 way to Pass:
Composite NOx 5 2.0

Fail
Pass
Pass
Pass
Pass
Pass
Pass
Pass

5
5
>
5
>
5
5
5

0.8
0.8
0.8
0.8
0.8
0.8
0.8
0.8

515.0
5 15.0
>15.0
515.0
5 15.0
5 15.0
>15.0
5 15.0

50.5
50.5
50.5
> 0.5
50.5
> 0.5
50.5
50.5

512.0
5 12.0
5 12.0
>12.0
5 12.0
5 12.0
512.0
>12.0

5
5
5
5
5
5
5
5

2.0
2.0
2.0
2.0
2.0
2.0
2.0
2.0

>1.0
51.0
51.0
51.0
51.0
51.0
51.0
51.0
pass.

















to









    The minimum emission levels and maximum purge volume for failing vehicles
at each mode were used as fast-pass outpoints.  Conversely,  the maximum
emission levels for passing vehicles at each mode were used  as fast-fail
outpoints.  Vehicles were not  fast-failed based on purge  results  since many
vehicles purge late in the IM240 cycle.  As mentioned, the IM240-reversed may
help rapidly determine if the  purge system is  functioning adequately.

    The modal outpoint levels, the number of vehicles fast-passing or  fast-
failing at each mode and the average IM240 test time  as a result  of  the
application of this fast-pass/fast-fail algorithm are displayed in the
following table.
                                      L-4

-------
                               Number
              Fast-pass Outpoints    of
        Time    Purge
2

3

4

5

6

7

8

9

0-34
0-60

0-74

0-93

0-113

0-154

0-173

0-206

0-239






<0.479/1. 02/0.99
<0.487/0.89/0.99
>0.3
<0.429/0.929/0.90
>0.3
<0.377/0.921/0.84
>0.4
<0.460/0.932/0.89
>0.5
<0.567/1.088/0.96
>0.6
<0.697/3.52/1.33
>0.7
<0.916/14.99/1.77
>0.8
50.805/15.05/2.05
Weighted Sum with
Fast-pass Only
Average IM240 Test
Time with Fast-pass
Only-
passing
16
2

1

0

3

3

65

210

45

102410


207 see
                                                             Number   Time *
                                                     Number    of     Number
                                                       of    Vehicles     of
                                                     Vehicle   Fast-    Vehicles
                              Fast-    Fast-fail Outpoints  s Fast-  passing  with Fast
                                       >HC/CO/NOx    failing  and Fast-   Result
                                                              failing
                                     >3.405/56.72/7.30    15      31      1054
                                        > 1.891/47.30/4.63     22
                                        > 1.648/38.09/3.58
                                        >1.536/41.09/3.19
                                        > 1.518/36.78/3.02
                                        > 1.296/30.34/2.57     11
                                        >1.120/25.22/2.65     11
                                        >0.915/18.06/2.33     35
                                        >0.805/15.05/2.05     33
24
8
9
9
14
76
245
78
1440
592
837
1017
2156
13148
50470
18642
                                                               Weighted
                                                                 Sum
                                                                Average
                                                                IM240
                                                               Test Time
                                                                       89356
                                                                         180 sec
    These  results indicate that the test time  for the IM240  can be reduced by
25% when fast-pass/fast-fail  criteria are  applied and a  reduction  of over half
a minute occurs when only fast-pass criteria are applied.   Using only fast-
pass criteria allows for the  collection  of diagnostic data so that failing
cars may be repaired more effectively.

    Because Hammond  cars with second-by-second data were typically shut off
for 10  minutes, catalyst cool down could have caused high emissions during the
early parts of the  test and adversely affected fast-pass and fast-fail.
Similarly, vehicles  that drive  a short distance to an  I/M station  may not be
fully warmed up when they start the test.   Therefore,  additional analyses were
                                         L-5

-------
performed  without integrating  over the first part of  the IM240.  In effect,
utilizing  the first segment  of the IM240 as preconditioning.  Three different
integration  starting points  were used.  Since the accelerations contribute  the
most toward  catalyst light-off,   these starting points  follow the first three
accelerations of the IM240 cycle.   The integrations begin after 17, 35 and  47
seconds of the test.  The results of these analyses are displayed here.
                             Number
             Fast-pass Outpoints    of
                                                 Number    Time*
                                        Number     of     Number of
                                           of     Vehicles   Vehicles

ode*


1
2

3

4

5

6

7

8

9
Tune
> (sec)


17-34
17-60

17-74

17-93

17-113

17-154

17-173

17-206

17-239
Purge


<0.525/0.95/1.33
^/\ t
•^ v» A
<0.504/0.54/1.10
>0.3
<0.465/0.90/0.96
>0.3
<0.400/0.90/0.88
>0.4
<0.486/0.91/0.93
>0.5
<0.593/1.09/1.00
>0.6
<0.641/3.08/1.38
>0.7
<0.826/15.33/1.82
>0.8
£0.805/15.05/2.05
Vehicles
Fast-
passing

11
1

4

0

5

3

56

217

48

Fast-fail Outpoints
>HC/CO/NOx

>2.643/76.94/10.33
> 1.892/53. 86/5. 11

>1. 615/41. 40/3.77

> 1.498/45.64/3.27

>1. 484/40. 16/3.08

>1.265/32.27/2.66

>1.080/26.48/2.71

>0.936/18.44/2.32

>0.805/15.05/2.05
Vehicles
Fast-
failing

19
11

11

10

7

10

10

37

34
Fast-
passing
and Fast-
failing
30
12

15

10

12

13

66

254

82
with Fast
Result


1020
720

1110

930

1356

2002

11418

52324

19598
Weighted Sum with
  Fast-pass Only    103230
  Average IM240
  Test Time with
  Fast-pass Only    209 sec
                                                             Weighted
                                                               Sum      90478
                                                              Avenge
                                                              IM240
                                                             Test Time    183 sec
                                        L-6

-------
                                   Number
                Fast-pass Outpoints     of
         Tune     Purge         Fast-
                                   passing
                  Number
                    of
                  Vehicles
Fast-fail Outpoints   Fast-
  >HC/CO/NOx     failing
Number    Time *
   of     Number of
Vehicles    Vehicles
  Fast-    with Fast
passing    Result
and Fast-
 failing
1
2

3

4

5

6

7

8

9






N/A
35-60

35-74

35-93

35-113

35-154

35-173

35-206

35-239






N/A
<0.493/0.79/0.90
>0.3
<0.403/0.73/0.79
>0.3
<0.340/0.69/0.75
>0.4
<0.454/0.9 1/0.82
>0.5
<0.585/1.10/0.93
>0.6
<0.575/2.85/1.37
>0.7
<0.795/15.17/1.84
>0.8
SO.805/15.05/2.05
£1.0
Weighted Sum with
Fast-pass Only
Average IM240
Test Time with
Fast-pass Only
N/A
19

4

2

5

2

48

221

44

102452



207 sec
N/A
>1.983/41.71/3.71

>1.499/31.32/3.08

> 1. 450/55.7 1/3.09

>1.406/47.21/3.07

>1.299/35.99/2.59

>1.061/28.83/2.81

>0.966/19.48/2.37

>0.805/15.05/2.05






N/A
41

8

5

3

7

7

35

43






N/A
60

12

7

8

9

55

256

87

Weighted
Sum
Average
IM240
Test Tune
N/A
3600

888

651

904

1386

9515

52736

20793

90473



183 sec
                                                L-7

-------
              Fast-pass Outpoints
        Tune    Purge
   1
   2
 N/A
47-60
3
4
5
6
7
8
9
47-74
47-93
47-113
47-154
47-173
47-206
47-239
      N/A
 <0.458/0.40/1.05
      >0.3
 <0.375J0.46/0.83
      >0.3
 <0.310/0.52/0.76
      >0.4
 <0.434/0.94/0.85
      >0.5
 <0.594/1.14/0.96
      >0.6
 <0.550/2.88/1.43
      >0.7
<0.751/14.82/1.91
      >0.8
SO.805/15.05/2.05
      £1.0
Weighted Sum with
  Fast-pass Only
  Average IM240
  Test Tune with
  Fast-pass Only
                                Number    Time *
Number                  Number     of    Number of
  of                      of     Vehicles   Vehicles
Vehicles                  Vehicles    Fast-    with Fast
 Fast-   Fast-fail Outpoints   Fast-    passing    Result
passing   >HC/CO/NOx    failing   and Fast-
                                 failing
 N/A        N/A         N/A     N/A      N/A
  6     >2.089/37.20/3.67    41       47       2820
                               102119
                               207 sec
                                       >1.282/33.21/2.99    14
                                       >l.737/67.22/3.10
                                 15    > 1.619/54.68/3.08
                                       >1.355/39.55/2.55    8
                                 48    >1.095/30.98/2.82
                                220    >1.004/20.39/2.42    35
                                 38    >0.805/15.05/2.05    45
                                                         23
                                           1702
                                                                  465
17
12
52
255
1921
1848
8996
52530
                                                         83
                                           19837
                                                       Weighted
                                                        Sum     90119
                                                       Average
                                                        IM240
                                                       Test Time   182 sec
    These  results  indicate,  that for the data used in this  analysis,
preconditioning has little effect  on the average test time of the fast-
pass/fast-fail  algorithm used.   In spite of  this, these estimates are
considered conservative  for several reasons.   First, older cars  are excluded
from  the analysis.  Since most grossly emitting vehicles are older vehicles,
the inclusion of these cars would  be expected to increase the number of  fast-
failing vehicles and reduce the  test time further.  However, this reduction
may be offset by a reduction in  the percentage of vehicles fast-passing.  More
important than  the vehicle sample  is the algorithm used.  If a continuous
function were used, actual test  times could  be used to  calculate the average.
This  should lead to significant  time savings compared to using the last  second
of a  particular mode as  the required test time for all  vehicles  that pass or
                                         L-8

-------
fail during that mode,  it is unlikely that all the vehicles failing or
passing a particular mode would have required the full mode to determine their
outcome.  Therefore, average test times for vehicles passing the IM240 at
second 60 would be significantly less than 60 seconds.  Likewise,  this would
be true for each mode.  On-going analyses are being performed to investigate
this and  other alternatives such as the IM240-reversed.  Finally, EPA will
continue to develop alternative algorithms which are also expected to reduce
the average test time for the IM240.
                                      L-9

-------