United States
Environmental Protection
Agency
November
1992
Air
I/M Costs, Benefits, and Impacts

-------
Table of Contents
Page
List of Tables	vi
1.0 INTRODUCTION	1
2.0 GLOSSARY OF KEY TERMINOLOGY	3
3.0 I/M PERFORMANCE STANDARDS	6
3.1	Enhanced I/M Performance Standard	6
3.2	Recommended Enhanced I/M Program Design	7
3.3	Basic I/M Performance Standard	7
4.0 EMISSION REDUCTIONS FROM I/M PROGRAMS	8
4.1	Recent I/M Test Programs	8
4.2	FTP HC/CO Correlation Comparison Between the	IM240 and
the Second-chance 2500 rpm/ldle Test	12
4.2.1	I/M Test Assessment Criteria Overview	15
4.2.2	Detailed Discussion of Correlation and	Test
Assessment	18
4.2.3	Two-Ways-To-Pass Criteria	20
4.3	Evaporative Test Errors of Commission	23
4.4	Approval of Alternative Tests	26
4.5	Transient Testing Fast-Pass/Fast-Fail Strategies	26
4.6	Estimating I/M Testing Credits for MOBILE4.1	28
4.6.1	Tech4.1 Background and Assumptions	29
4.6.2	Evaporative and Running Loss Modeling,	and the
Effectiveness of Purge/Pressure Testing	31
4.6.3	Benefits of IM240 NOx Inspections	34
5.0 REGULATORY IMPACT ANALYSIS - ESTIMATING COST AND COST
EFFECTIVENESS	41
5.1 Cost of Conventional I/M Testing	41
5.1.1 Inspection and Administration Costs	42

-------
5.2	Estimated Cost of High-Tech I/M Testing	47
5.2.1	General Methodology	47
5.2.2	Equipment Needs and Costs	48
5.2.3	Cost to Upgrade Centralized Networks	50
5.2.4	Cost to Upgrade Decentralized Programs	55
5.3	Costs of Four-Mode, Purge and Pressure Testing	59
5.3.1	Equipment and Expendables	60
5.3.2	Centralized Programs	60
5.3.3	Decentralized Programs	61
5.4	Repair Costs	61
5.4.1	HC and CO Exhaust Repair Costs and Methodology	61
5.4.2	NOx Repair Costs and Methodology	63
5.4.3	Evaporative System Repair Costs and Methodology 65
5.5	Fuel Economy Benefits	67
5.5.1	Fuel Economy Benefits of Evaporative System
Repairs	67
5.5.2	Fuel Economy Benefits of IM240 Repairs	68
5.5.3	Fuel Economy Benefit for the 2500 rpm/Idle Test	69
5.6	Recurring Failure and Repair Rates	71
5.7	Method for Estimating Cost Effectiveness of I/M Programs73
5.7.1	Inspection Costs	74
5.7.2	Repair Costs	75
5.7.3	Fuel Economy Cost Benefits	75
6.0 REGULATORY IMPACT ANALYSIS - COSTS AND BENEFITS OF ENHANCED
I/M 77
6.1	Emission Reduction Benefits	77
6.2	Cost Effectiveness Estimates	79
6.2.1 Assumptions and Inputs	7 9
6.2.3	Cost-Effectiveness Calculations	79
6.2.4	National Cost of Choosing Less Stringent I/M	81
6.3	National Costs and Benefits	82
6.3.1	Emission Reductions	82
6.3.2	Economic Costs to Motorists	83

-------
6.4 Motorist Inconvenience Costs	85
7.0 REGULATORY FLEXIBILITY ANALYSIS	86
7.1	Regulatory Flexibility Act Requirements	86
7.1.1	The Universe of Affected Entities	87
7.2	Types of Economic Impacts of Concern	88
7.3	Changes in Repair Activity	88
7.3.2	Repair Activity in Future I/M Programs	90
7.4	Changes in Emission Testing Activity in I/M Areas	91
7.4.1	The Existing Market in Centralized and
Decentralized Programs	91
7.4.2	Future Market in Enhanced I/M Programs	97
7.4.3	Centralized Programs	97
7.4.4	Decentralized Programs	98
7.4.5	Impact on Jobs in Decentralized Programs	102
7.4.6	National Impact on Jobs	105
7.5	Mitigating the Impact of Enhanced I/M on Existing
Stations	106
7.6	Public Comment	107
8.0 ONBOARD DIAGNOSTICS AND ON-ROAD TESTING	110
8.1	Onboard Diagnostics, Interim Provisions	110
8.2	On-road Testing, Interim Provisions	110
9.0 ALTERNATIVE TESTS	113
9.1	Status of Alternative Exhaust Tests	113
9.2	Current Analysis of Available Data on ASM Tests	114
9.3	Alternative Purge Tests	117
9.4	Alternative NOx Testing	122
9.5	Repair Grade IM240 Testing	127

-------
Append;^
A	Evaporative Emissions and Running Loss Emission Factor
Derivation
B	Purge and Pressure Test Effectiveness Figures and
Spreadsheet
C	Exhaust Short Test Accuracy
D	M0BILE4.1 Technology Distribution and Emission Group
Rates and Emission Levels
E	Regression Analyses and Scatter Plots for Fuel Injected
1983 and Later Vehicles
F	IM240 Cutpoint Table Analysis
G	Evaporative System Purge and Pressure Diagrams
H	Evaporative System Failures and Repairs
I	M0BILE4.1 Performance Standard Analyses, By Option
J	Identifying Excess Emitters with a Remote Sensing Device
K	Model Year Failure Rates by Test Type
L	Comparative Purge Flow Data

-------
1
2
3
4
5
6
7
8
9
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
List of Tables
Page
IM240 Selection Standards for Stratified FTP Recruitment	9
Weighting Factors for Correcting Recruitment Biases	10
Evaporative Test Results	23
IM240 Bag-1 Fast-Pass/Fast-Fail Analysis	27
Short Test Identification Rates	32
Short Test Repair Effectiveness	33
Lane IM240 Based Emission Factor Levels with IM240 NOx
Cu'tpoints	36
Side Effects of I/M on NOx Emissions	38
Lane IM240 Based Emission Factors with IM240 Outpoints	40
I/M Program Inspection Fees	43
Quality Assurance Functions and Costs in Decentralized
Programs	4 6
Quality Assurance Functions and Costs in Centralized
Programs	47
Equipment Costs for New Tests	49
Expendables for New Tests	50
Peak Period Throughput Rates in Centralized I/M Programs	50
Current Program Costs	53
Costs to Add Proposed Tests to Centralized Programs	55
Inspection Volumes in Licensed Inspection Stations	56
Costs to Conduct High-Tech Testing in Decentralized Programs59
Equipment and Costs for the ASM Test	60
Costs to Add Proposed Tests to Centralized Programs	61
Costs to Conduct Four-Mode Testing in Decentralized Programs61
Average Cost of Repairing Emission Control Components	63
NOx Repair Costs	65
Average Repair Costs and Fuel Economy Benefits	66
Zero Improvement Vehicle Sample Size Adjustments	69
Adjusted Zero FE Benefit Vehicle Sample Size	70
Exhaust Test Failure Rates	71
Default Inspection Costs in CEM4.1	74
Default Repair Cost in CEM4.1	75
Fuel Economy Benefits in CEM4.1	76
MOBILE4.1 Inputs for the High-Tech Enhanced Model Program 78
Benefits of I/M Programs Options	78
Total Annual Program Cost	7 9
Cost per Ton Allocating All Costs to VOC	80
VOC Cost per Ton Accounting for NOx and CO Benefit	81
Total Cost and Benefits of I/M Options	81
Excess Cost of Choosing Low Option I/M	82
National Benefits of I/M	83
Program Costs and Economic Benefits	85
Costs of the Biennial High Option including Inconvenience 86
Affected Businesses	88
Repair Expenses in Enhanced I/M Programs	90
Number of Inspection Stations by State	92
Inspection Stations by Category	93
Inspection Station Volumes and Incomes	94

-------
95
96
97
99
100
102
103
104
105
118
124
128
Average Inspection Station Revenues, Costs, and Profits
Inspection Volumes in California
Station Revenues and Profits by Volume
Assumed Station Distributions
Revenues and Profits for Low and Medium Volume Stations
Numbers of Inspectors per Station by State
Estimated Inspection FTE
Summary of FTE Gains and Losses
Impact on Jobs of I/M Proposal
Purge Vehicle Descriptions
NOx Vehicle Description*
Estimated Costs for Repair-Grade IM240 Emission System

-------
1.0 INTRODUCTION
Despite having the best vehicle control program in the world,
many areas in the United States continue to measure unhealthful
levels of air pollution, approximately half of which can be
attributed to motor vehicles. As a result, in addition to tighter
standards on new vehicles and their fuels, the Clean Air Act
Amendments of 1990 (Act) require the implementation of vehicle
Inspection and Maintenance (I/M) programs in areas that have been
designated as nonattainment for ozone or carbon monoxide (CO). A
total of 181 such areas currently exist in the United States, 56
of which do not presently operate I/M programs. Depending upon
the severity of the nonattainment problem, these areas will have
to implement either a basic I/M program (required in areas with
moderate ozone nonattainment, and in marginal areas with existing
I/M programs) or an enhanced I/M program (required in most
serious, severe, and extreme ozone areas, as well as most CO areas
registering levels greater than 12.7 parts per million (ppm)).
Eighty-three of the 181 nonattainment areas currently designated
will require the implementation of an enhanced I/M program.
The Environmental Protection Agency (EPA) has had oversight
and policy development responsibility for I/M programs since the
passage of the Clean Air Act in 1970, which included I/M as an
option for improving air quality. The first such I/M program in
the United States was begun in New Jersey in 1974, and the program
elements which made up this program's design (i.e., a centralized,
annual, idle test of all light-duty gasoline vehicles, with no
waivers or tampering checks) still constitute those design
features upon which the basic I/M performance standard is based.
However, many advances have been made in vehicle technology since
the time of that first I/M program, and while the idle test in use
in many current programs works well enough when it comes to
detecting emission problems in older, low-tech vehicles, its
effectiveness as a testing strategy rapidly drops off as we begin
testing newer, more sophisticated, computer-controlled vehicles.
High-tech vehicles need high-tech testing which more closely
simulates real-world driving conditions and the sort of test to
which vehicles are originally certified - a loaded, transient
test, which requires driving the vehicle through a prescribed
pattern of accelerations and decelerations on a dynamometer.
Much has also been learned since 1974 about the many ways
vehicles contribute to the problem of air pollution. Previously,
it was thought that the majority of the air pollution problem
attributable to mobile sources was the result of exhaust
emissions; it is now understood that emissions in the form of
evaporative and running losses are also major contributors. The
gasoline evaporating in the tank of a vehicle and escaping into
the environment is as much a source of volatile organic compound
(VOC) emissions as are the exhaust gases emitted from the
tailpipe. Vapor recovery and recirculation mechanisms have been
-1-

-------
installed on vehicles since 1971, but these systems can
deteriorate with time and are often rendered useless as a result
of wear, tampering, and design defects. Cost effective tests have
been developed to detect evaporative system failures of this sort,
including the evaporative system purge test and the evaporative
system pressure test.
Under the terms of the Clean Air Act Amendments of 1990, EPA
is required to establish minimum performance standards for I/M
programs. The Act further specifies that the standard for
enhanced I/M shall be based upon a program that employs an annual
cycle of automated emissions analysis, performed at a centralized
test-only site, and enforced through the denial of registration.
EPA has developed these standards and has formalized them as part
of the I/M rulemaking.
In the past, the model program used to establish the
performance standard assumed a model program along the lines of
the original New Jersey program - a standard which remains
essentially unchanged for basic I/M programs. For the enhanced
I/M performance standard, however, EPA has developed a model
program based on loaded, transient testing, in conjunction with
evaporative system purge and pressure tests. Using EPA's
MOBILE4.1 computer model, a high-tech I/M program such as that
included in the enhanced I/M performance standard is expected- to
achieve emission reductions from mobile sources on the order of
approximately 31% for ozone-forming hydrocarbons (HC) and 34% for
CO (compared to 5% HC and 16% CO emission reductions from the
basic I/M performance standard program design).
Given the potentially significant economic impact of this
decision, it is necessary to assess the costs and benefits of
enhanced I/M performance standards. This report provides the
technical background information supporting EPA's cost and benefit
pro jections.
In assessing the costs and benefits of enhanced I/M, we will
detail the findings of recent research and development on test
procedures and vehicle emissions, the basis for the computer
models used to establish emission benefits and program cost-
effectiveness, the differences in cost-effectiveness among
programs based upon network and test types, as well as projections
of the average per vehicle cost for inspection and repairs, and
the cost offset of the fuel economy benefit achieved by making
such repairs. Graphic and tabular support data are attached to
this report as appendices.
It should be noted that in finalizing this document, EPA
continues to base its estimates on the M0BILE4.1 emission factor
model, primarily because the latest model - MOBILE5 - is still in
the process of development and revision and is not ready for final
release.
-2-

-------
2.0 GLOSSARY OF KEY TERMINOLOGY
Throughout this report several key terms will be used with
which the reader may not be immediately familiar. To facilitate a
better understanding of the issues involved, the following
glossary is provided.
"Concentration"	Versus "Mass	Emissions" Tests: Mass emissions
tests provide a much better indication of vehicle emission levels
than concentration tests. A concentration reading of 200 ppm HC
from a subcompact car and the same 200 ppm reading from a large
truck (which is entirely possible) suggest that the two vehicles
pollute equally. However, this is incorrect. The truck will have
a much higher volume of exhaust. So, over a given one-mile drive,
the subcompact car may only emit 50 cubic feet of exhaust gases,
whereas the truck may emit 500 cubic feet. With both vehicles
emitting 200 ppm HC over the mile, the total amount of HC emitted
by the truck will be 10 times greater than the amount emitted by
the small car. A mass emissions test allows the total emissions
per mile to be measured; a concentration test does not. All
currently approved I/M tests are concentration tests. The Federal
Test Procedure and the IM240 test, however, are mass emissions
tests.
Decentralized Test-Only -Network: A program design in which
multiple participants are contracted to perform I/M testing (as
opposed to a single contractor). To establish equivalency with
traditional centralized programs and to avoid the decentralized
discount incorporated in EPA's MOBILE model, participants must
operate test-only facilities and are barred from making repairs,
selling replacement parts, making referrals, or otherwise engaging
in activities that would violate the intention of the test-only
requirement (i.e., the avoidance of conflict-of-interest).
Error-of-Commission (Ec): On the basis of an emissions test, the
false failure of a vehicle as "dirty" (i.e., emitting high enough
that repair and a retest are required) when the vehicle, in fact,
meets EPA new car standards, based upon the Federal Test Procedure
(see definition below) . Usually, HC and CO Ec's are defined
without regard to NOx emissions, and vice versa.
Error-of-Omission: To falsely pass as clean a vehicle which, in
fact, exceeds EPA new car standards, based upon the results of the
Federal Test Procedure.
Federal	Test	Procedure: The Federal Test Procedure (FTP) is a
mass emissions test created to determine whether prototype
vehicles comply with EPA standards, thus allowing production
vehicles to be 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
-3-

-------
"in-use" vehicles. The FTP is too costly to use for I/M because
vehicles must be maintained in a closely controlled environment
for over 13 hours. The FTP is based on a 20 minute trip, driven
once when the engine is cold, and again when it is hot.
Hiah-Tech Vehicles: Vehicles with computerized control of the
engine and emission control system, especially 1983 or newer
vehicles employing fuel injection (either port fuel injection
(PFI) or throttle-body injection (TBI)) as opposed to carburetion
as a fuel metering methodology.
Idle Test: A concentration-type emission test to measure the
percentage of CO and ppm HC in the exhaust stream of a gasoline-
powered vehicle operating at idle. The nondispersive infrared
detector (NDIR) equipment normally used gives a less accurate
measure of HC than does the flame ionization detector (FID)
equipment used in the FTP and IM240 tests.
IM240 Exhaust Test1: A mass emissions (as opposed to
concentration), transient short test run on an inertial and power-
absorbing dynamometer using a 240 second driving cycle loosely
based upon the LA4 cycle used in the FTP. EPA originally divided
the driving cycle into 2 parts or "bags" with separate emissions
determinations, but recently has begun integrated analysis of
emissions on a second-by-second basis. Unlike the idle test which
is conducted at a single speed and expresses emissions in terms of
percentages and ppm, the IM240 is conducted at a range of
accelerations and decelerations and provides emissions
measurements in terms of grams per mile (gpm) . The IM240 has
proved particularly effective in accurately identifying high
emitting, newer technology vehicles.
Preconditioning: Operation of a vehicle at a specific speed, load
(including no load), and time to ensure that a vehicle is properly
warmed up prior to testing. For the purpose of transient testing,
a period of operation prior to testing to avoid errors of
commission as a result of evaporative system purging into the
sample. Under the two-ways-to-pass criteria (see section 4.2.3
for a more detailed discussion) this goal is achieved by
establishing two sets of cutpoints, a set of cutpoints for the
composite results, as well as cutpoints for Bag-2 results (with
the first 93 seconds - or Bag-1 - being used as the
preconditioning mode).
Pressure Test: A test whereby inert gas is injected into a
vehicle's evaporative system to establish the system's integrity
Pidgeon, W. and Dobie, N., "The IM240 Transient I/M Dynamometer Driving
Schedule and The Composite I/M Test Procedure," U.S. EPA Technical Report
Number EPA-AA-TSS-91-1, January 1991.
-4-

-------
by indicating the presence of a leak or by confirming the system's
ability to hold pressure.
Purge Test.: A test to determine whether a vehicle's evaporative
emissions system recycles the gasoline vapors adsorbed on the
charcoal in the evaporative canister (i.e., whether or not the
canister purges vapors to the engine to be combusted). To provide
representative operation and opportunity for the, purge control
system to demonstrate its proper working order, the purge- test is
conducted on a dynamometer using the same 240-second transient
driving cycle as the IM240 exhaust gas test. The test is
conducted simultaneously with the tailpipe emission test.
2500 rpm/ldle Test: A two-speed, steady-state, concentration-type
test in which emissions are sampled at both idle and 2500 rpm. To
be considered a pass, a vehicle must pass at both speeds. The
two-speed test has a better identification rate for high emitting
vehicles than does the standard idle test.
-5-

-------
3.0 I/M PERFORMANCE STANDARDS
3 . 1 Enhanced I/M Performance Standard
Under the Act, EPA is required to establish a performance
standard for enhanced I/M programs including, at a minimum,
centralized, annual, automated emission testing of light-duty
vehicles and trucks, including a tampering check for emission
control devices, a misfueling check, and provisions for including
on-road emission testing and inspection of onboard diagnostic
devices (OBD). The performance standard is defined by completely
specifying the design of a model or benchmark I/M program. While
enhanced I/M programs need not match the performance standard's
model program element by element, such programs must be designed
and implemented to meet or exceed the minimum emission reductions
achieved by the performance standard. Any deviations from the
performance standard's program design that may lead to emission
reduction losses must be made up by strengthening other aspects of
the program. For example, while the Act constrains the
performance standard for enhanced I/M programs to be based on an
annual program, it is clear that a biennial program is more cost-
effective and results in relatively small emission reduction
losses over those achieved by an annual program. The emission
reduction losses resulting from a decision to test vehicles
biennially as opposed to annually can be made up, for example, by
extending transient exhaust testing and purge testing to cover
earlier model years than those specified in the performance
standard. This specific example will be discussed in more detail
in Section 3.2 below.
EPA's enhanced I/M performance standard is based on
centralized, annual testing of light-duty vehicles (LDVs) and
light-duty trucks (LDTs) rated to 8, 500 pounds Gross Vehicle
Weight Rating (GVWR) using the transient IM240 exhaust test
incorporating NOx cutpoints, and purge testing of the evaporative
control system of 1986 and later vehicles (using cutpoints of 0.8
to 0.7 gpm HC, 20 gpm CO, and 1.4 to 3.0 gpm NOx> depending upon
the age and weight rating of the vehicle). Two-speed testing is
to be performed on 1981-1985 model year vehicles (using cutpoints
of 1.2% CO, 6% C02, and 220 ppm HC) while idle testing is to be
used on pre-1981 vehicles. Idle test cutpoints for older vehicles
must yield a 20% failure rate. The performance standard also
includes visual inspection of the catalyst and fuel inlet
restrictor on all 1984 and later vehicles and evaporativfe system
integrity (pressure) testing of 1983 and later vehicles. Using
EPA's mobile source emission model, MOBILE4.1, this performance
standard is estimated to yield a 28% reduction in VOCs, a 31%
reduction in CO, and a 9% reduction in NOx by the year 2000 over a
non-l/M scenario.
-6-

-------
3.2 Recommended Enhanced I/M Program Design
The Act requires EPA to establish a performance standard
based on an annual test program. States, however, are free to
implement alternative program designs, including a biennial
program, provided the emission reductions achieved meet or exceed
those achieved by the model program. This demonstration is made
using EPA's mobile source emission model which includes biennial
and annual program credits. Given the added convenience and cost-
effectiveness of a biennial program, EPA recommends that states
adopt a biennial program that can meet the performance standard,
through, for example, increased vehicle coverage.
3.3 Basic I/M Performance Standard
The basic I/M performance standard is based upon the program
design of the original New Jersey program and remains essentially
unchanged as a result of EPA's proposed action. The basic I/M
performance standard is estimated to yield a 5% reduction in
mobile source VOC emissions and a 16% reduction in CO. The
performance standard includes annual, centralized idle testing of
model year 1968 and later light-duty vehicles. The pre-1981
failure rate is assumed to be 20%, with 0% waivers and 100%
compliance. The basic I/M performance standard does not include
testing of light-duty trucks; neither does it include visual
inspections of any emission control components.
-7-

-------
4.0 EMISSION REDUCTIONS FROM T/M PROGRAMS
4.1 Recent I/M Test Programs
The data used by EPA to assess the benefits of high-tech I/M
testing concepts, including the IM240, and evaporative system purge
and pressure testing, have been obtained as a result of two
special testing programs performed under contract to EPA. The
first testing program - an IM240 transient test pilot study - was
conducted as part of a cooperative project with the State of
Maryland in 1989, and utilized one of the state's I/M stations for
testing and recruiting vehicles. This was the first attempt to
perform transient emissions tests on consumer vehicles in a high
throughput system. More extensive programs are currently being
run in Indiana and Arizona, although data from Arizona is still
too new for incorporation in this report. The Maryland pilot
study began testing in August 1989, and continued through December
of that year, testing a total of approximately 600 vehicles for an
average of approximately 120 vehicles per month. The larger-scale
Indiana program began testing in February 1990. As of November 1,
1991, approximately 8,300 vehicles had been tested as part of the
Indiana program, with an average of approximately 120 vehicles per
week. As such, the database produced by this test program is the
largest of its kind ever assembled to assess I/M testing. The
Arizona program began testing vehicles on June 8, 1992 and has
tested over 1,500 vehicles so far. EPA has not had time to
quality assure the Arizona data, however, and it therefore has not
been used in compiling the figures in this report.
The Indiana testing contracts include two test facilities, a
laboratory in New Carlisle (a few miles west of South Bend) , and
an I/M station in Hammond. The laboratory is owned by Automotive
Testing Laboratories, Inc. (ATL), a contractor to EPA, and the I/M
station is owned by the Indiana Vocational-Technical College,
which operates the I/M program for the State of Indiana. The I/M
station includes four lanes, with ATL running one of the four.
EPA has three separate testing contracts in Indiana that
utilize the two facilities: Emission factor (EF), I/M, and running
loss testing. Reformulated fuels testing is being performed under
the EF contract. The three contracts use vehicles that are
selected at the I/M station. The selection criteria for follow-up
laboratory testing include model year, fuel metering type, and
results from the following tests: The IM240, canister purge flow
measurement, and evaporative control system pressure tests.
The goal at the I/M station originally was to test a random
sample of 1976 and newer light-duty vehicles. On May 15, 1991,
the recruitment goal changed to randomly sample 1983 and newer
vehicles, to increase the number of fuel-in >ected vehicles
represented in the database. This change was made to reflect the
-8-

-------
fact that fuel injection is rapidly replacing carburetion as the
preferred fuel-metering method for new vehicles, and the
percentage of carbureted vehicles in the in-use fleet will become
insignificant in the future.
Choosing cars for further laboratory testing is driven by the
overriding 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, given that the
majority of excess emissions comes 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 is employed to deliberately over-recruit dirty cars, based on
the results of IM240, purge and pressure tests. Actually, two
recruitment and lab testing programs operate simultaneously. In
one, a nominally 50/50 mi:-: of IM240-clean and IM240-dirty vehicles
is recruited for FTP exhaust testing. In actual practice, more
clean cars than dirty have been recruited rather than allow lab
testing slots to be idle while waiting for a dirty car to be
recruited. The Hammond I/M lane vehicles were categorized as
clean or dirty using the IM240 standards listed in Table 4-1. In
the other lab-testing recruitment effort, a sample even more
heavily weighted toward purge and pressure test failures is
recruited for evaporative and running loss emissions testing.
Table 4-1
IM240 Selection Standards for Stratified FTP Recruitment
Model Years
1986+ *
1983-85
Selection Standards
(grams per mile)
>1.10	>15.0
>1.20	>16.0
* The 1986+ standards were set to be more stringent than 1983-
1985 standards to improve recruitment of high emitters and
to balance the failure rates between model year groups.
The FTP database that results from EPA's recruitment targets
must be corrected to represent the clean/dirty vehicle ratio in
the in-use fleet to correctly determine excess emission
identification rates (IDR), error-of-commission rates (Ec) and
failure rates (all important criteria for assessing the overall
effectiveness of I/M testing strategies) . The database was
corrected using the weighting factors presented in Table 4-2.
-9-

-------
Table 4-2
Weighting Factors for Correcting Recruitment Biases
Fuel
Lane

Lab

# Lab Veh
# Normals
Metering
IM24 0
Lane
Sample
Weighting
Passing
Failing
Svstem
Results
Count
Count
Factor
FTP
FTP
PFI
Clean
1505
55
27.36
23
24

Dirtv
97
19
5 .11
1
2

Total
1602
74

24
26
TBI
Clean
1555
73
21. 30
25
32

Dirty
166
35
4 . 74
4
6

Total
1721
108

29
38
Weighting factors are used as follows: If the 19 dirty
vehicles that received FTP tests in the PFI vehicle sample had
excess HC emissions which totaled 100 gpm, the database would be
corrected in this case by multiplying 100 by the 5.11 weighting
factor, resulting in a corrected excess emission rate of 511 gpm
for the dirty vehicles (excess emissions are those FTP-measured
emissions that exceed the certification emission standards for the
vehicle under consideration; an I/M test's identification rate for
excess emissions represents one of the important criteria for
assessing an I/M test's effectiveness, as detailed in Section 4.2
of this report). In comparison, the excess emissions of the IM240
clean vehicles have to be multiplied by 27.36 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 large sample of 55 clean cars in
this sample provides confidence in conclusions about a test's
relative tendency to avoid failing clean cars.
Appendix F provides additional information on adjustments to
make the FTP database representative of the Hammond lane fleet's
ratio of clean and dirty vehicles. Appendix F also includes
tables that allow a comparison of outpoint effects on IDR, I/M
failure rates, Ec rates, and I/M failure rates for FTP-passing
vehicles.
At the Hammond I/M station, in addition to the IM240,
technicians perform the official Indiana I/M test (2500 rpm/ldle)
and an additional second-chance 2500 rpm/ldle test for those that
fail the first chance test. Vehicles that require a second-chance
test first receive 3 minutes of preconditioning. The combination
of this "enhanced" steady-state testing, along with the IM240 and
purge/pressure tests allows for direct comparison of these
alternative I/M procedures. Section 4.2 of this report provides a
more detailed di'scussion of the results of comparing the degree to
-10-

-------
which the IM240 and the second-chance 2500 rpm/ldle test correlate
with the FTP.
In addition to assessing the IM240 for correlation with the
FTP, several other issues are addressed as part of the Hammond
study. Since dirty vehicles are repaired at the lab, the repair
effectiveness can be evaluated. The running loss tests allow EPA
to characterize the air quality impact of vehicles failing
pressure and purge tests and the effectiveness of repairing these
vehicles. The transient short test developed by the Colorado
Department of Health (CDH-226) as well as a variety of steady-
state tests are performed at the lab and can be evaluated as
potential I/M tests. Additionally, one of the IM240s performed at
the lab was restricted to inertia weight settings of 2,500 pounds
or 3,500 pounds. This restriction allowed EPA to evaluate the FTP
correlation effect of a more economical dynamometer (with fewer
inertia weight settings). We found that an inertia weight range
of 2,000 to 5,500 pounds using four inertia wheels (500, 1,000,
and 2,000 pounds with a fixed wheel of 2,000 pounds) is worth the
moderate additional cost.
The evidence displayed in Section 4.2 (see below) and
Appendices C and E of this report graphically and quantitatively
shows the advantage of the high-tech IM240 test for the sample of
vehicles tested in Indiana in 1990 and 1991. The actual
calculations of the exhaust emission reductions of the several
short tests are more detailed in order to best reflect the actual
characteristics of the fleet as it ages and changes in technology
mix. A computer model called Tech4.1 is used to calculate
technology- and age-specific adjustment factors that represent the
effect of I/M programs of different types (the so-called "I/M
credit"), and these factors are built into the mobile source
emissions model MOBILE4.1. Section 4.6.1 of this document
contains details on the Tech4.1 model.
Finally, the Indiana testing program has revealed the true
seriousness of evaporative emission control system malfunctions
that develop during real world operation. Previous EPA testing
programs (i.e., those conducted during the last 10 years or so)
that did not make use of an operating I/M lane to screen and
recruit vehicles for more thorough laboratory testing have focused
mostly on vehicles that were about 5 years old or younger, in
order to most quickly obtain information on the latest generation
of new technology vehicles. When special efforts were made to
recruit high mileage vehicles, they tended to be vehicles that had
accumulated unusually high mileage for their age, for example
vehicles from owners with long commutes or who used their vehicles
for business during the day. EPA staff have been concerned for
some time that testing such vehicles was not giving a true picture
of evaporative emission problems, which may develop more as a
function of passing time than of miles driven; for example,
deterioration of rubber and plastic components would be more time-
than mileage-based. Also, the recruitment practices in the test
-11-

-------
programs prior to the Indiana I/M lane program relied on owner
response to letters and phone calls. There has been concern that
this resulted in a different sample of vehicles, probably a sample
biased towards better maintenance condition than would be found if
owners could be solicited face-to-face, as they are in the Hammond
study (where the level of motorist participation has been
sufficiently high to ameliorate these concerns).	These
differences in study design explain why the Indiana program has
produced results very different from previous estimates of in-use
evaporative emissions.	EPA's interest in the high-tech
evaporative purge and pressure tests has been in response to these
findings.
Because of the extensive detail of the evaporative emissions
findings from Indiana, the results of the testing are presented in
Appendix A, rather than illustrated with figures and tables here.
Briefly stated, the Indiana program showed that by 13 years of
age, nearly one-half of all vehicles will experience an
evaporative system failure that renders the control system
virtually ineffective, causing evaporative and running loss
emissions to increase by factors of up to 10 times. Nearly all of
these failures can be detected by the combination of the pressure
and purge tests,. Use of only one of these tests finds at least
some of the problem vehicles. The problems can be repaired, and
vehicles will then pass a re-inspection using the pressure and/or
purge test. Appropriate repairs reduce emissions back to normal
levels. Of course, the purge and pressure tests cannot overcome
the limited control capacity designed into vehicles by their
manufacturers, so under certain conditions of temperature and fuel
volatility, both passing and repaired vehicles will fail to meet
the certification emission standard.
4  2 ftpHC/CO Correlation Comparison	Between the	IM24Q	and the
Second-chance 2 500 rpm/ldle Test
This section focuses on the comparison of the IM240 transient
test (using cutpoints of 0.8 gpm HC and 15 gpm CO for the results
over the full 240 seconds, with a provision that a vehicle also
may pass by having emissions during the last 14 7 seconds of the
test less than or equal to 0.5 gpm HC and 12 gpm CO - see Section
4.2.3 for a more detailed explanation of the two-ways-to-pass
criteria) to EPA's currently recommended second-chance 2500
rpm/ldle test procedure2, and details the evaluation criteria upon
which the comparison is based. This comparison shows how an I/M
program based on one of the better currently used (non-
dynamometer) I/M tests (second-chance 2500 rpm/ldle) can be
improved upon by changing to the IM240 test, which has a much
better classical correlation with the FTP than the idle or 2500
2 Tierney, E., H'erzog, E. and Snapp, L. "Recommended I/M Short Test
Procedures For the 1990s: Six Alternatives", U.S. EPA Technical Report
Number EPA-AA-TSS-90-3, January 1991.	
-12-

-------
rpm/Idle test for matched pollutants (see the regression analyses
including R-squared values and scatter plots in Appendix E for an
illustration of this better correlation).
For the sake of the correlation analysis illustrated in
Appendix E, only 1983 and newer vehicles equipped with fuel
injection were considered3. The vehicles in this sample received
both the second-chance 2500 rpm/Idle test and the IM240 at the
Hammond test site. At this time, most I/M programs have not
adopted second-chance testing and the test algorithms recommended
in EPA's Alternative Test Procedure report, which calls for an
immediate second-chance test for vehicles that initially fail the
emission standards. Under the recommended procedures, vehicles
are preconditioned in a non-loaded state for three minutes at 2500
rpm prior to the second test. Second-chance testing was devised
to reduce, to the extent possible, the problem of falsely failing
vehicles. For the purposes of this comparison and to enable
analyses of the effectiveness of more stringent standards, second-
chance tests were performed on 1983 and newer fuel-injected
vehicles if their emissions exceeded 100 ppm HC or 0.5% CO on
their initial 2500 rpm/Idle tests. Note that these standards are
substantially tighter than the standards of 220 ppm HC and 1.2% CO
used in nearly all I/M programs on 1981 and later vehicles.
One of the central concerns in developing a new I/M short
test was to devise a test that would pass vehicles that would pass
the FTP and fail those that would fail the FTP. With that in
mind, the IM240 was devised by truncating, splicing, and otherwise
augmenting the first two hills of the FTP driving cycle. One of
the goals of the pilot program was to assess how well the IM240
correlates with the FTP. Since performing the FTP in the Indiana
lane was not a practical alternative, both IM240s and FTPs were
conducted in the lab after the vehicles were recruited in the I/M
lane. The lab results of the IM240 and the FTP showed excellent
correlation. One can conclude that the IM240 is an excellent
measurement of the true emissions of the vehicle at the time and
place it is performed, given the fuel being used at the time.
Comparing lab FTP and lane IM240 results is problematic for
several reasons, but still shows good correlation. Since the lab
tests are performed at a different time from the lane IM240s,
intervening factors, such as intermittent problems or changes in
the vehicle, may affect the results. For example, exhaust systems
are often repaired, when needed, prior to the lab tests. Another
major problem making lab and lane comparisons difficult is the
3 The emission reduction benefits presented in Section 6, however, do reflect
the application of the IM240 to carbureted vehicles as well as fuel-
injected vehicles; the comparisons of IDR, Ec rate, and failure rate for
the various I/M testa presented in Appendices G and H also address
carbureted vehicles.
-13-

-------
fact that FTP tests are all done on Indolene fuel4 while lane tests
are done on the fuel in the tank of the vehicle as received.
Also, the equipment used in the lanes measured a lower maximum
emission value than the lab equipment; for example, a car would
have pegged the lane instrument for hydrocarbons at 13 gpm while
in the lab it actually measured 25 gpm. Temperature and
preconditioning at the lane were also often different than at the
lab. For these reasons, lab/lane comparisons say less about the
actual performance of the test and more about the influence real
world differences make on vehicle emissions. Nevertheless, both
sets of comparisons are presented in Appendix E of this report.
One of the conclusions evident from the data collected as
part of the Hammond study is that for fuel injected vehicles in
particular, the high-tech IM240 test has a better correlation with
the FTP than the conventional idle or 2500 rpm/Idle test. This
section and Appendix E present some illustrations of this better
correlation.
For example, one indication of better correlation is
demonstrated by higher R-squared values from least-squares
regressions with FTP emissions as the dependent variable and short
test emissions as the independent variable. Statistics for these
regressions are given in the regression analyses tables in
Appendix E.
The better correlation of the IM240 test also can be seen
visually in the scatter plots of emissions results from vehicles
which received all four tests (Appendix E). Separate plots of FTP
versus short test results are included for each type of fuel
injection (whether PFI or TBI), pollutant (HC, CO, and NOx) , and
each short test type (except for idle and 2500 rpm/Idle for NOx,
since representative in-use NOx emissions cannot be measured on
these tests) . Because of the wide range of the data, the graphs
showing all the data contain a lump of points near the origin. To
allow examination of the correlation for vehicles emitting in this
range, an enlargement of the data in this range is also provided
for each of the graphs in Appendix E.
The above two indications (R-squared values and scatter
plots) of better correlation do not directly enter the calculation
of the emission reduction advantages of the IM240. In an I/M
program, predicting the absolute level of a vehicle's FTP
emissions is not as important as identifying a large majority of
the vehicles whose emissions are likely to be high enough to merit
repair (which are, themselves, a minority of the overall in-use
fleet) . Also, the short test should pass vehicles that are not
Indolene is a special test fuel whose properties are held constant. This
is necessary because the normal changes in fuel properties of commercial
fuel can change a car's emissions results even if all of the other test
procedure variables and vehicle variables did not change between tests.
-14-

-------
malfunctioning, in order to avoid impacting owners of vehicles
which have emissions low enough to not merit repair. The figures
in Appendix C, which are discussed in the following sections,
graphically demonstrate the differences between the second-chance
2500 rpm/Idle test and IM240 in regard to these objectives.
The I/M test must also do a good job of ensuring that
vehicles that have shown emission reductions from repairs large
enough to pass re-inspection on the short test have also achieved
sizeable FTP reductions. Better performance of one short test
versus another in identifying vehicles as generally clean or dirty
will also ensure that fewer vehicles can pass reinspection without
achieving real FTP reductions. Therefore, it is clear that the
IM240 test will be the better enforcer of good repairs. Analysis
of data from vehicles in Indiana that were repaired at the
laboratory and retested on both the FTP and IM240 shows that
reductions measured by the two tests are highly correlated, even
better than the correlation discussed above. Figures and
statistics to illustrate this are also included in Appendix E.
4.2.1	I/M Test Assessment Criteria Overview
In assessing the overall effectiveness of an I/M testing
procedure, it is important to determine the test's effectiveness
in measuring and determining a variety of factors, including the
IDR, the failure rate, the 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
but are clean enough to make it an issue whether they will benefit
much from normal repair procedures. Each of these is discussed,
in turn, below. Section 4.2.2 provides a more detailed discussion
of the same topics.
4.2.1.1 Excess Emission Identification Rate	(IDR)
EPA commonly uses the rate of excess emissions identified
during an I/M test to objectively and quantitatively compare I/M
test procedures. As mentioned earlier, 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 gpm HC standard that failed the
second-chance 2500 rpm/Idle I/M test with an FTP result of 2.00
gpm, would have excess emissions equalling 1.59 gpm (i.e., 2.00 -
0.41 = 1.59).
The excess emissions identification rate (IDR) equals the sum
of the excess emissions for the 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/M passing
vehicles also have excess emissions which are used for calculating
the total excess emissions) . Thus, assuming an I/M area that
tests 1000 vehicles, 100 of which are emitting 1.59 gpm excess
-15-

-------
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:
80 failinq vehicles * 1.59 qpm excess per vehicle
		:	, . e-	 	~		 * 100 = 80% IDR
100 vehicles * 1.59 gpm excess per vehicle
As can be seen in Figures 1 and 4 in Appendix C, the IM240 using
two-mode criteria has been shown to identify more excess emissions
among the cars tested at the Indiana lane than the second-chance
2500 rpm/Idle test with current I/M program cutpoints.
4.2.1.2 Failure Rate
As the IDR increases, 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 is calculated by dividing the number of failing
vehicles by the number of vehicles tested. For example:
50 vehicles failed I/M ^	_ , _ . ,
	:	 * 100 = 5% I/M failure rate
1000 vehicles tested
The ideal I/M test is one that fails all of 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 utilizes resources. As the figures in
Appendix C show, tightening the cutpoints on the idle test to
achieve IDRs comparable to the IM240's results in increasing the
failure rate well beyond that of the IM240. For example, for 1983
and newer, PFI vehicles, the failure rate rose from 12% to 38%
when second-chance, two-speed cutpoints were tightened to 100 ppm
for HC and 0.5% for CO, even though the two-speed test's IDRs for
HC and CO were only 77% and 82% respectively (compared to the
IM240's 82% and 85% IDRs for HC and CO, and its 14% failure rate).
The remaining figures in Appendix C illustrate a similar
relationship between IDR and failure rate for tighter two-speed
cutpoints for both TBI and carbureted vehicles . For a more
specific, model year breakdown of failure rates among the vehicles
in the Hammond lane sample, by test type, see Appendix K, ."Model
Year Failure Rates by Test Type."
4.2.1.3 Error-of-Commission (Ec) Rate
Properly functioning vehicles which pass FTP standards
sometimes fail the 2500 rpm/Idle test; these are referred to as
-16-

-------
false failures or errors of commission (Ecs). When error-of-
commission vehicles are sent to repair shops, no emission control
system malfunctions exist. Often, the repair shop finds that the
vehicle now passes the test without any changes. These false
failures waste resources, annoy vehicle owners, and may lead to
emissions increases as a result of unnecessary and possibly
detrimental "repairs." Motor vehicle manufacturers see this as a
significant problem, since it can contribute to customer
dissatisfaction and increased warranty costs. 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-commission rate is calculated, assume
an I/M area which tests 1000 vehicles, of which 100 fail the I/M
test, although only 50 of those 100 failing vehicles also exceed
their FTP standard for HC or CO. The error-of-commission rate
equals the number of vehicles that fail the I/M test while passing
the FTP for HC and CO, divided by the total number of vehicles
which were I/M tested:
50 vehicles failed I/M but passed FTP HC and CO . 	 __ _ *
	rrrr	e	;	 * lOO = 5% Ec rate
1000 vehicles tested
*Error-of-commission
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.
4.2.1.4 Failure Rate Among FTP-Passina Vehicles
The risk of failing an I/M test with a clean vehicle is not
expressed very clearly, however, by stating fleet error-of-
commission rates. Fleet rates tend to be very low, but the impact
on any individual motorist can be very significant. A more
informative statistic than error-of-commission rate is the failure
rate among all inspected vehicles which still pass their FTP
standard. This indicates the risk to the owner of having a clean
vehicle failed. For the IM240 using the two-ways-to-pass
criteria, only one vehicle out of 274 (i.e., 0.4%) failed the
IM240 while passing the FTP (see Appendix C, as well as the
discussion under Section 4.2.3 "Errors of Commission Under the
Two-Ways-To-Pass Criteria"). While the false failure rate for the
second-chance two-speed test is initially comparable to the IM240
using the two-speed outpoints in current use, tightening these
cutpoints to improve IDR has the effect of increasing the false
failure rate for the steady-state test. For example, as
illustrated in Figures 1 through 3 of Appendix C, for 1983 and
newer PFI vehicles, tightening the steady-state cutpoints from 220
-17-

-------
ppm HC and 1.2% CO (the outpoints most commonly used in current
I/M programs) to 100 ppm HC and 0.5% CO has the effect of
increasing the test's false failure rate from 0% to 13% - this,
even though the two-speed test's IDRs for both HC and CO still
fall appreciably below that of the IM240. For 1983 and newer TBI
vehicles, the same tightening of cutpoints achieves HC and CO IDRs
for the steady-state test that actually exceed those of the IM240
by a percentage point or two, but this at the cost of a false
failure rate of 20% compared to one, debatably "false" failure for
the IM240 (Figures 4-6, Appendix C).
Even when the two-ways-to-pass criteria are not used for the
IM240, the false failure rate for the vehicles in EPA's sample was
only 0.8%, representing a total of 5 Ec vehicles - still much
lower than the false failure rate for the steady-state test with
comparable IDRs. Since any number of false failures is
unexpected, given the IM240's similarity to the FTP and the
looseness of the 0.8/15 cutpoints compared to the 0.41/3.4 new car
standards, Section 4.2.3 is included to discuss this false failure
in depth.
4.2.1.5 "Normal Emitter" Failure Rate
The IM240 failure rate for normal emitters will also be
lower. For the purposes of this discussion, "Normal"'emitters are
defined as those vehicles that emit less than twice the FTP HC
standard and less than three times the FTP CO standard. Normal
emitters include those vehicles that pass the FTP. Repairs on
such vehicles usually do not produce large emission reductions (at
least short of catalyst replacement, which EPA generally avoids in
its emission repair evaluations due to cost and because testing
after a new catalyst is installed would not necessarily indicate
what emissions will be after the catalyst "wears in"), their
emissions are sometimes increased by inept repairs, and they
account for little of the total excess emissions. Therefore,
normal emitters are not the most cost-effective to identify for
repairs. These vehicles often lack overt defects. Those that
fall above one of the FTP standards obviously have some problem,
but may only have suffered catalyst deterioration (which is
difficult to diagnose) or may have been either poorly designed or
built in the first place. Thus, the marginal costs of identifying
and effectively repairing these vehicles may not always be worth
the marginal benefits that could be expected.
4.2.2	Detailed Discussion of Correlation and Test Assessment
The following analysis shows that the IM240 test using the
two-ways-to-pass criteria is considerably more powerful as an I/M
test than the second-chance 2500 rpm/Idle test for all technology
type vehicles, but especially newer technology, fuel-injected
vehicles. The analysis presumes that the IM240 is implemented to
achieve higher IDRs. Given that rationale, IM240 standards of 0.8
-18-

-------
gpm HC and 15 gpm CO for the full test and 0.5 gpm HC and 12 gpm
CO for the last 147 seconds were selected for this analysis.
These IM240 standards achieve IDRs that are significantly higher
than for the present second-chance 2500 rpm/ldle standards, while
maintaining a false failure rate of zero.
This discussion is limited to PFI vehicles, as this is the
most commonly used fuel metering system on new vehicles. Throttle
body injection, which is less sophisticated, may also be used on a
significant proportion of the future fleet, though less than for
PFI. Therefore, although analogous figures and tables are
included in Appendices C and F for both TBI and carbureted
vehicles, they are not formally discussed.
Figure 1 in Appendix C provides a comparison of the present
second-chance 2500 rpm/ldle test using current standards (220 ppm
HC and 1.2% CO) to the more effective, high-tech IM240 test using
the two-ways-to-pass criteria. Note the following:
	The FTP excess emissions identification rates are 19% higher
for HC and 13% higher for CO with the IM240 as compared to
the second-chance 2500 rpm/ldle test using the 1.2%/220 ppm
standards.
	Neither test failed FTP-passing vehicles.
	The IM240 increases the failure rate to 13% from 10% for the
preconditioned, second-chance 2500 rpm/ldle test.
Figure 2 in Appendix C illustrates the power of the IM240
test compared to the 2500 rpm/ldle test using the more stringent
idle standards currently in use in California. I/M programs might
consider California idle standards because the emission reduction
from the program can be increased and the cost of implementation
is relatively small.
California uses standards of 1.0% CO and 100 ppm HC for the
idle mode, while using 1.2% CO and 220 ppm HC for the 2500 mode.
In Figure 2, only the stringency of the 2500 rpm/ldle test is
increased, while the IM240 standards are the same as those used in
Figure 1 (see Appendix C for both figures). Note the following:
	The IDRs are still 8% higher for HC and 5% higher for CO with
the IM240 as compared to the second-chance 2500 rpm/ldle test
with more stringent standards.
	The second-chance 2500 rpm/ldle test failure rate using
California standards is 29% compared to only 13% for the
IM240. So even with the IM240's higher IDRs, significantly
fewer vehicles will need to be repaired.
	Twelve percent of the FTP-passing vehicles fail the second-
chance 2500 rpm/ldle test, while none fail the IM240.
-19-

-------
Sending this many cars for unnecessary repairs, while also
identifying less excess emissions, wastes resources.
 The normal emitter failure rate is only 2.5% for the IM240
versus 22% for the second-chance 2500 rpm/Idle test. This
means that the vehicles identified for repairs by the IM240
are more likely to achieve significant emission reductions.
In Appendix C, Figure 3 compares the same IM240 standard to
the more stringent standards of 0.5% CO and 100 ppm HC for both
modes of the second-chance 2500 rpm/Idle test for PFI vehicles,
while Figures 4, 5, and 6 present data analogous to the first
three figures, but this time for TBI vehicles, and Figures 7-9
present this information for 1981 and newer carbureted vehicles.
Second-chance testing was only performed on 1983 and newer
vehicles, however, so Figures 7, 8, and 9 only include second-
chance results for 1983 and newer vehicles, not for 1981 and 1982
vehicles.
4.2.3	Two-Wavs-To-Pass Criteria
The theory behind the two-ways-to-pass criteria is as
follows. Assuming that the 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 fuel vapors and that the vapors 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 from the canister continually decreases during
IM240 operation. The decreasing fuel vapor concentration results
in decreasing HC and CO emissions. So, Bag-2 results should be
lower than the composite results, on a gram per mile basis. On
the other hand, if the vehicle is actually malfunctioning, Bag-2
emissions should remain high. For this reason, second chance
tests after preconditioning, as shown for the current 2500
rpm/Idle test, should be less influenced by canister purge.
Catalyst temperature can also effect 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 cause the vehicle to register
as a failure, when, in fact, it should be passed. It is this
problem of catalyst cool down that has lead EPA to recommend
preconditioning as a means for avoiding false failures. Under the
two-ways-to-pass criteria, Bag-1 acts as a preconditioning mode,
thus providing insurance against this particular variety, of false
failure.
-20-

-------
4.2.3.1 Errors of Commission Under the Two-Wavs-To-Pass Criteria
I/M test procedures and standards that cause low emitting
vehicles to fail an I/M test are obviously undesirable. Because
the emissions of properly functioning vehicles are known to vary
in a predictable manner with changing test conditions, the FTP
controls variables such as test temperature, vehicle temperature,
humidity, vehicle prior operation, and fuel characteristics (by
using a special test fuel), as well as other variables to help
achieve repeatable results on a given vehicle. Since many of the
variables known to affect vehicle emissions cannot be controlled
in an I/M program, EPA is forced to relax the stringency of its
pass/fail standards to allow properly- functioning vehicles to
pass, even when such variables "stack up" or otherwise conspire to
produce seemingly high emissions readings.	EPA is also
constrained by cost-effectiveness disbenefits that attend relaxed
standards. As the standards are loosened, the percentage of high
emitting malfunctioning vehicles not identified for repairs
increases. On the other hand, forcing properly functioning cars
to be diagnosed by a mechanic also hurts cost-effectiveness along
with other obvious undesirable effects.
The model program uses IM240 two-ways-to-pass standards of
0.8/15.0/2.0 composite results and 0.5/15.0 for Bag-2. The
Appendix F cutpoint tables show that the error-of-commission rate
is zero for PFI and carbureted vehicles, but is 1.2% for TBI
vehicles. The purpose of-this section is to discuss whether the
error-of-commission rate of 1.2% indicates that the IM240
standards are too stringent.
A false failure resulted on only one^ of the 274 1983 and
newer vehicles that received FTP tests. It is surprising that any
FTP-passing cars failed the IM240 two-ways-to-pass standards,
however, since the IM240 driving schedule is taken from the FTP
and is a hot start test at the Indiana lane.
Vehicle number 1724 failed the IM240 HC standard in its
Hammond lane test, with a score of 0.96 gpm, but passed the FTP
with a score of 0.31 gpm. Vehicles that pass the FTP are normally
considered properly functioning .vehicles, but the mechanic's
inspection identified the following problems:
This vehicle was actually tested at the laboratory. As explained in
Section 4 1, the database was corrected to accurately represent the in-use
fleet distribution, so the error of commission vehicles discussed in
previous sections were from the corrected database. This section only
discusses the vehicles that were actually tested, so the single error of
commission PFI vehicle becomes 4 vehicles after the weighting factor
discussed in Section 4.1 is used. Similarly, the three actually-tested TBI
error of commission vehicles become 17 vehicles in the corrected database.
This section is unique in discussing only the actually-tested vehicles.
-21-

-------
Checklist Comments
	Idle Mixture: Rich
	Fuel Injection Components: Meters excessive fuel
	Distributor Assembly: Cap and rotor dirty
	Initial Timing: Specification = 8 BTDC, actual = 3
Retarded
	Spark Plugs and Wires: Plugs worn, wires arcing
	Catalyst: Poor performance
Narrative Comments
Injection meters excessive fuel.
Cap & Rotor dirty.
Wires Arcing
Plugs worn
Timing -5 [Slight disagreement with checklist which
indicated -3.]
These are hardly results that would be expected of a prope-rly
functioning vehicle, so the question is not: "Why did this vehicle
fail the IM240?" The more appropriate question is: "Why did this
car, considering these problems, pass the FTP?" Thevanswer seems
to be that the car passed the FTP due to several interactive
variables. High HC emissions are frequently caused by ignition
system problems which cause a vehicle to misfire. If misfiring is
only an intermittent problem, it is possible that a vehicle that
fails one test, will register as a pass when tested later.
The worn spark plugs, arcing spark plug wires, and dirty cap
and rotor all can contribute to intermittent misfire. If bad
enough, any of these problems can lead to steady misfiring, but
since the vehicle passed the FTP, the presumption is that the
engine was misfiring more during the IM240 at the inspection
station than during its FTP test at the lab. Additionally, the
dynamometer inertia weight setting at the lane was 3,000 pounds,
whereas it was only 2,875 pounds for the FTP. While not a large
difference, the voltage required to fire the spark plugs increases
with increasing load. With a marginal ignition system, the
voltage available at the spark plug may be less than the voltage
required to fire the spark plug, so logically, more misfire should
be expected with the higher loading this vehicle was subjected to
during the IM240. Also, the vehicle received its IM240 test on
July 30, 1991, but ATL did not receive the vehicle from the owner
until August 12, 1991 and it did not receive its FTP test until
August 15, 1991. The fact that the owner retained possession of
the vehicle for nearly two weeks between the lane IM240 test, and
the FTP test is important because spark plugs that are misfiring
one day, usually due to carbon deposits, can clean themselves
under high temperature operation, and have less or no apparent
misfire on a different day. Also, given the proverbial problem of
malfunctions that do not exhibit themselves when the mechanic is
in the car, only to reappear during the trip home, most people can
-22-

-------
easily relate to such intermittent problems. This vehicle's
passing FTP score is probably because the intermittent misfire
that occurred during the IM240, occurred to a lesser degree during
the FTP. This brings us back to the question; "Should the IM240
standards be relaxed to avoid false failures?" The evidence
suggests that this car was correctly identified as needing repairs
by the 0.8/15.0/2.0 and 0.5/15.0 two-ways-to-pass standard, and
only passed the FTP by a fluke. Therefore, EPA's judgement is
that the debatably "false" failure of this vehicle is insufficient
justification for relaxing the standard.
4.3 Evaporative Test Errors of Commission
In its submission to the I/M docket, Toyota commented that
some vehicles that failed the purge or pressure test appeared to
be passing the current certification evaporative SHED test with
combined diurnal and hot soak emissions of less than 2 grams.
Toyota expressed concern that the existence of false evaporative
failures would make them responsible for a more stringent, post-
certification regulatory requirement that denies them "due
process." EPA is also concerned about the possibility of
evaporative test false failures, and has identified five vehicles
which were potential evaporative test false failures from a list
of 20 failing vehicles. The test results from these five vehicles
are shown in Table 4-3.
The majority of the apparent false failures had serious
mechanical problems, or evaporative system leaks. In addition,
these apparent evaporative false failures can be categorized into
those that may have occurred due to errors in performing the test,
and those that were due to an intermittent malfunction of the
vehicle.
Table 4-3
Evaporative Test Results






As Recv
Aft Rep






Running
Running




Diurnal
Hot Soak
Losses*
Losses*
Veh
Pres
Pura
Malfunction
(a/tst)
(a/tst)
(a/tst)
(a/tst)
1596
F
P
Loose Gas Cap
0 .10
0 . 39

	
1689
P
F
Purge TVS
0 . 74
0 . 68
78 . 6
4.4
1704
F
P
Gas Cap Seal
0 . 96
0 . 73


1712
F
P
Vent Line Leak
0 . 53
0 . 45
190.9
175.0
1714
F
P
Gas Cap Seal
1. 09
0.4 6
	
	
~Running loss emissions are based on the Modified LA4 Running Loss Test at 95
F. The test consists of three consecutive LA4 driving schedules conducted in
an enclosed SHED.
4.3.1	Vehicle; 1689
This vehicle, a 1985 Mercury Marquis, was the only apparent
false purge failure. It received two purge tests. The first test
-23-

-------
was at the lane, and a second confirmatory test was done at the
contractor's lab. As Table 4-3 notes, the vehicle was shown to
have zero purge during the purge test. Diagnosis of the vehicle
identified a stuck thermal vacuum switch controlling the
evaporative purge. Six subsequent running loss tests showed this
vehicle to be a gross emitter. However, during two of the six
running loss tests (each test was three consecutive LA4 cycles),
the vehicle's purge system operated intermittently, and provided a
total purge of 38 liters during one test, and 28 liters during the
other test (these are low levels of purge for a running loss
test) . During the remaining four running loss tests the purge
flow was zero. Therefore, this vehicle demonstrated that it could
purge occasionally, but that in general it was not operating as
designed and should be considered a failure. In addition, after
repair of the thermal vacuum switch which controls purge flow,
this vehicle showed a dramatic reduction in running loss emissions
from 78.6 grams HC/test to 4.4 grams HC/test while its purge flow
was increased to a relat ively consistent 85 liters per running
loss test.
Because the failure mode of this vehicle was of an
intermittent nature, it is possible that sufficient purge occurred
randomly on this vehicle. Thus, adequate purge may have occurred
prior to the hot soak and diurnal enabling the vehicle to pass
these tests. In any case, the data supports the fact that a
critical emission control component malfunctioned on this vehicle.
4.3.2	Vehicles 1596. 1714 and 1704
Vehicle 1596, a 1990 Chevrolet, was initially found to be a
pressure failure at the lane when the system would not hold any
pressure. This type of failure requires a substantial leak which
is usually readily apparent. Nevertheless, the technician could
not identify the source of the leak even after attempting two
pressure retests, both of which the vehicle passed. As a result
of the lane pressure initial test failure, the vehicle was
recruited to the lab for running loss testing and repair. Here,
it was again retested and found to clearly pass the pressure test.
Although, the actual reason for the initial failure is unknown, it
is believed to be the result of improper testing technique or
equipment malfunction which was apparently recognized by the
inspector (thus explaining why the vehicle was retested at the
lane) . Thus, based on the retest results in the lane, this
vehicle should never have been recorded as a failure.
Vehicle 1714, a 1986 Chevrolet, was also diagnosed as a
pressure failure at the lane. The lane inspector identified a
leak near the gas cap or filler neck. This vehicle was then
retested at the contractor's lab, and passed. However, the final
passing pressure was just over the standard of 8 inches of water
after two minutes (below 8 inches of water is a failure). Thus, a
very small leak might have been present depending on the tightness
of the gas cap, and the quality of the seal between the gas cap
-24-

-------
and filler neck. At the time, the lane pressure test procedure
did not call for tightening the gas cap prior to the test.
However, the lane procedure did call for removing the gas cap at
the end of the test to check for pressure in the tank. Following
removal, the inspector would then reinstate and properly tighten
the gas cap.
At the laboratory, gas caps have always been tightened prior
to conducting the pressure test. In addition, the lane procedure
has been changed so that gas caps are tightened prior to the
pressure test.
Vehicle 1704, a 1983 Toyota, was also diagnosed as a pressure
failure at the lane due to a leak identified near the gas cap.
However, after recruitment to the lab, the vehicle marginally
passed two pressure tests, and was not recruited for running loss
testing. Like vehicle 1714, it is probable that this vehicle had
a very small leak due to the condition of the gas-cap/filler-neck
seal. The test procedure changes are expected to eliminate
failures such as these.
4.3.3	Vehicle 1712
Vehicle 1712, a 1987 Chevrolet, was found to have a leak in
the vent line at the connection between the rubber hose and the
steel line between the canister and the fuel tank. This leak was
found after several pressure test failures at the lane and the
lab. Modified LA4 running loss tests (three consecutive LA4
cycles at 95 F) produced evaporative emission levels of more than
190 grams over the 22' mile test. Likewise, modified high
temperature (95 F) diurnals produced emission levels of 44 grams
(hot soak) and 10 grams (diurnal) . An after-repair running loss
test was also conducted resulting in running loss emissions of 175
grams per test .
EPA views I/M false failures as a significant problem, and is
committed to investigating and implementing strategies to prevent
their occurrence. For the evaporative system tests, such
strategies include tightening gas caps prior to the pressure test,
automation and computerized control of the test, test algorithms
that insure all sequence are properly performed, and refined
procedures to eliminate the possibility of technician testing
errors. It is not advantageous to falsely fail, and attempt
repairs on vehicles which are passing the certification standards
and operating as designed. However, EPA also feels that
malfunctions that cause excessive evaporative emissions from
vehicles in-use such as leaking gas caps, leaking fuel tanks,
broken fuel tank vent lines, and malfunctioning purge controllers
should be identified and repaired. It has been shown that both
the pressure and purge tests are effective at identifying vehicles
with these problems, while minimizing the identification of the
vehicles without such problems.
-25-

-------
4 . 4 Approval of Alternative Testis
Although the IM240, purge, and pressure tests represent EPA's
current trio of recommended high-tech tests, we do not rule out
the possibility of future, valid alternatives to these tests,
including fast-pass and fast-fail transient testing strategies
(see Section 4.5, "Transient Testing Fast-Pass/Fast-Fail
Strategies"). States may seek approval of such strategies,
contingent upon the state's demonstrating to EPA's satisfaction
that such strategies are at least as effective as EPA's
recommended tests at identifying excess emissions while
maintaining a comparably low error-of-commission rate. As the
sheer number of analyses contained in this report can attest, EPA
does not promulgate new testing strategies capriciously. Before
proposing the IM240, purge, and pressure tests, EPA amassed a
compelling body of data on each through pilot programs conducted
in Maryland and Indiana (see Section 4.1) for further discussion
of these pilot studies). Rigorous evaluations of each were
conducted to determine their effectiveness at identifying excess
emissions while maintaining low error-of-commission rates.
Economic analyses were also conducted to assess the cost-
effectiveness of the tests, as no degree of technical excellence
will justify a testing strategy that is exorbitant in its overall
cost. For example, the FTP is the hallmark against which I/M
testing strategies are measured, but cannot itself be used as an
I/M test, given its cost.
A more detailed discussion of several currently proposed
high-tech testing alternatives is included in Section 9.0 of this
report.
4.5 Transient Testing Fast-Pass/Fast-Fail Strategies
Among the alternative testing strategies that make
environmental and economic sense, the potential for fast-pass and
fast-fail transient testing ranks the highest. 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.
In evaluating potential fast-fail criteria, EPA looked at a
sample of 4,158 1983 and newer vehicles tested at the Hammond
IM240 lane described in Section 4.1, 1,033 (or 24.8%) of which
failed the IM240. 298 (or 28.8%) of the 1,033 vehicles that
failed would have failed within the first 93 seconds of the test
if Bag-1 cutpoints of 2.5 gpm HC, 50 gpm CO, and 5.0 gpm NOx were
used; there were no errors-of-commission. Although stricter Bag-1
cutpoints could be used to increase the percentage of fast-failed
vehicles, the error-of-commission (Ec) rate would also rise. In
turn, when fast-pass Bag-1 cutpoints of 0.41/3.4/1.0 were used,
1,074 (or 34.4%) of the 3,125 vehicles that passed overall passed
-26-

-------
within the first 93 seconds of the test. Seven additional false
passes were also recorded, resulting in an error-of-omission rate
of 0.7%. Tightening the fast-pass outpoints to 0.25/1.5/1.0
eliminates the false passes but also reduces the fast-pass rate to
13.2%. Table 4-4 provides further details on the Bag-1 outpoints
looked at in this analysis. While more development of fast-pass
and fast-fail criteria is needed, it is reasonable to conclude
that criteria can be developed to accurately pass and fail about
one third of all vehicles tested after only 93 seconds rather than
the full 240 seconds. Furthermore, EPA has begun collecting
second-by-second IM240 data. This will allow the development of
algorithms that will permit especially clean cars to pass well
before 93 seconds, and others to pass after 93 seconds, but well
before 240 seconds. Once the algorithms are developed, only
vehicles that are close to the cutpoints are expected to continue
for the full 240 seconds to ensure that they are not falsely
failed.
Fast Fail
0.8/15/2.5
2.0/40/4.0
2.5/50/5.0
Table 4-4
IM240 Baa-1 Fast-Pass/Fast-Fail Analysis
Fail
Fail

Fail
Fast-

IM240
F ast-Fail
Fail
Fast-Fail
Fail

Total
Total
Both
Only
ID rate
Ec Rate
1033
1297
902
395
87.3%
30.5%
1033
450
445
5
43.1%
1.1%
1033
298
298
0
28 . 8%
0 . 0%
Fast Pass
0.8/15/2.5
0.41/3.4/1. 0
0.25/1.5/1.0
Pass
Pass

Pass
Fast-
False-
IM2 4 0
F ast-Pass
Pass
Fast-Pass
Pass
Pass
Total
Total
eoth
Only
IP Rate
Rate
3125
2861
2730
131
87 . 4%
12.7%
3152
1081
1074
7
34 . 4%
0 . 7%
3125
413
413
0
13.2%
0 . 0%
Another area that EPA is investigating is the possibility
that the overall test time may be reduced. The IM240 is itself an
FTP-like short test based upon a modified and condensed driving
cycle that takes as its reference the LA4 cycle used in the FTP.
EPA is currently investigating the possibility of further
abbreviating the test by comparing how well data from either of
the two hills of the IM240 driving cycle (i.e., Bag-1 and Bag-2)
taken separately correlate with the current two-mode IM240.
Preliminary results based upon a sample of 188 1983 and newer
fuel-injected vehicles which were recruited at the Indiana I/M
lane and subsequently retested under lab conditions (which
included each vehicle receiving an FTP) suggest that analysis of
Bag-2 (i.e., emissions sampled during the second hill of the IM240
driving cycle) may be about as good as the full IM240 when it
comes to identifying vehicles that would pass or fail on the basis
of the full test. Using Bag-2 cutpoints of 0.60/12 for HC and CO
-27-

-------
respectively, and looking at Bag-2 results only, 90% of the excess
HC emissions and 84% of the excess CO emissions were identified,
with an Ec rate of 0.7%, as compared to the full IM240 using the
0.8/15 cutpoints only (i.e., no Bag-2 outpoints), which identified
82% and 85% of the excess HC and CO emissions, respectively, with
an Ec rate of 0.8%. These findings come with the caveat that they
are based upon a Bag-2 sample which followed the Bag-1 portion of
the driving cycle, meaning that Bag-2's high degree of correlation
with the IM240 may be the result of preconditioning occuring
during the Bag-1 phase. Even if such is, in fact, the case, the
prospect of a shorter overall test time still seems- good since
adequate preconditioning for Bag-2 could probably be obtained in
less than 93 seconds by modifying Bag-1 to use a higher speed over
less time.
To determine whether or not preconditioning is a factor, EPA
has begun testing a sample of vehicles using what is, in effect, a
three bag test, beginning with the second hill of the IM240
driving cycle up front (hence no possibility for "Bag-1"
preconditioning) followed by a regular IM240. Once this data is
analyzed, it should help EPA determine (1) whether or not,
preconditioning is a factor in Bag-2's high degree of correlation
with the full test and (2) whether preconditioning would improve
the correlation between Bag-1 and the full test. In addition, as
mentioned above, EPA has also begun collecting second-by-second
data, which will allow us to determine whether or not there is
some point in the testing cycle by which time if vehicle X is
emitting at a rate Y, it will clearly pass or fail.
4.6 Estimating I/M Testing Credits for MOBILE4.1
As stated earlier, the data from the Indiana program were
analyzed and re-assembled in a manner which allows a comparison of
I/M program designs over a wide range of time frames and
conditions, rather than just for the particular sample of vehicles
tested in Indiana. This method for estimating the effect of I/M
program options on exhaust emissions (i.e., the I/M credit) is
fairly simple. Using the emission factor database, the fraction
of total vehicle FTP emissions which is identified by a particular
short test is determined for each of four strata of vehicles based
on FTP emission level. Using a subsample of vehicles which have
been repaired, the emission reductions attributable to these I/M-
triggered repairs is estimated for each strata. The Tech4.1 model
is used to calculate the emissions impact of a given short test by
reducing the total FTP emissions identified at each age by the
estimated emission reductions resulting from I/M repairs. When
the fleet average emission rates are recalculated by considering
the strata, the difference between the I/M and non-I/M case is
stored as an I/M credit for use in MOBILE4.1.
-28-

-------
4.6.1	Tech4 . 1 Background and Assumptions
The Tech4.1 model divides the 1981 and newer light-duty
gasoline vehicle (LDGV) sample into several groups. The 1981 and
1982 model years are kept separate from the 1983+ model years. In
each model year group, the vehicles are divided by technology type
into closed-loop port fuel injection (PFI), closed-loop throttle-
body fuel injection (TBI), closed-loop carbureted (Carb) and all
(carbureted and fuel injected) open-loop (Oplp). Further, each of
these groups are divided into emission levels for Normal, High,
Very High and Super emitters. Table D-l in Appendix D provides
details on national fleet averages for passenger vehicle
distributions by model year and technology type; Tables D-2 and D-
3 provide data on emitter groups by model year group, technology
type, emission levels and rates, and mileage accumulation.
The model allows a separate IDR and repair effectiveness
estimate for each of these divisions of the data by I/M test type,
as illustrated in Table 4-5. It should be noted that the IDRs
listed in Table 4-5 for the traditional I/M tests (i.e., the idle
and 2500 rpm/Idle tests) are based upon historical emission factor
data gathered at EPA's National Vehicle and Fuel Emissions Lab
(NVFEL) in Ann Arbor, Michigan, as well as elsewhere, and not at
the Hammond, Indiana test lane. The IDRs mentioned elsewhere in
this report (Appendices C and F, for example) were derived as part
of the Hammond study, and are not divided by emitter group, as is
the case in Table 4-5.
In practice, because of small sample sizes, several of the
divisions represented in Table 4-5 share information. In
particular, the small amount of steady-state Loaded/Idle testing
required that all vehicles without Loaded/Idle testing be assumed
to have the same short test result for Loaded/Idle testing as they
had for the 2500 rpm/Idle test for the purpose of determining the
IDR for the Loaded/Idle test.
For Super emitters (vehicles over 10 gpm HC or 150 gpm CO) ,
the IDR is the same for all technologies, but is separate for
1981-82 and 1983+ vehicles. Most 1981-82 vehicles are carbureted.
Most 1983+ vehicles are fuel injected. There are no Super open-
loop vehicles in the sample.
The two fuel injection groups in the 1981-82 grouping use the
same IDRs for Very High emitters (vehicles over 1 . 64 gpm HC or
13.6 gpm CO), High emitters (vehicles over 0.82 gpm HC or 10.2 gpm
CO) and Normals. In some cases, such as the. High emitters, the
1983+ open-loop and carbureted technologies were combined.
Repair effectiveness (Table 4-6) was determined by dividing
the repaired sample by technology into PFI, TBI and Carb. Model
year grouping was not used. To be eligible for the repair
effectiveness analysis, a repaired vehicle must" first fail the
short test of interest before repairs, and then after repairs,
-29-

-------
must pass the same short test. Thus, different samples of
repaired vehicles were used for each short test . The sample was
then ranked by before repair emission level and divided into four
equal-sized subgroups of increasingly more severe emissions
failure. The before and after repair emission levels of each
subgroup were then determined.
When plotted, before repair emission level versus after repair
emission level, these four emission failure points represent a
technology specific function used to determine repair effectiveness.
Generally, the vehicles with higher before repair emission levels
get larger absolute emission reductions from repairs, but do not
reach as clean a level after repairs as vehicles which began with a
milder degree of emission failure. Before repair emission levels of
High, Very High and Super emitters in many cases will fall between
the calculated points, and so had their after repair emission levels
determined by interpolation. Before repair emission levels lower
than the lowest point were interpolated between the low point and
zero. Before repair emission levels above the highest point were
assumed to be the same as the highest point.
Since few of the repaired vehicles had Loaded/Idle or IM240
testing data, it was assumed that vehicles repaired using a
Loaded/Idle test and the IM240 test would use the same before and
after repair curve as the 2500 rpm/ldle testing. EPA is being
conservative in assuming that vehicles failing the Loaded/Idle test
or the IM240 test, after repair, will have the same after repair
emission level as we estimate for the 2500 rpm/ldle test vehicles.
However, since the failure rates of vehicles in the high emitter
groups are larger for the Loaded/Idle test and the IM240 transient
test than for the 2500 rpm/ldle test, the total emission reduction
due to repairs will be larger.
As an example, the zero mile HC emission level of Very High
emitters for 1983+ PFI vehicles is 2.019 gpm. and their slope is
taken to be the same slope as the Normals (i.e., 0.0115 gpm/10,000
miles) (see Table D-2) . At 5 years old, the average mileage of
these vehicles will be 60,829 miles. The non-I/M emission level is
therefore:
2.019 + .0115*6.0829 = 2.089 gpm
Assuming a 2500 rpm/ldle test is done, the HC IDR (see Table
4-5) for this group is 0.6187, or nearly 62% of the total emissions
from these vehicles is identified by failing vehicles using the 2500
rpm/ldle test. Table 4-6 shows the results of a data analysis
indicating the predicted average after repair levels given the
before repair emission level. The series of points in the table are
used to predict the after repair emission levels for all emitter
groupings, only dependent on the average before repair emission
level for that group. The before repair emission level falls
between the two emission levels 1.9846 and 3.9314. The after repair
levels for these emissions are 0.59231 and 1.0271 respectively.
-30-

-------
Interpolating, the after repair level for the 2.089 gpm before
repair emission level is:
0.59231+((2.08 9-1.984 6)/(3.9314-1.984 6))*(1.0271-0.59231)=0.6153
Therefore the after repair HC emission level for 5 year old,
1983+ PFI vehicles tested on the 2500 rpm/Idle test is:
0.6187*0.6153 + (1-0.6187)*2.089 = 1.1772 gpm
Comparing the I/M and non-l/M cases indicates the "I/M benefit"
among Very High emitters.
(2.089-1.1772)/2.089 = 43.6%
In the Tech4.1 model, the technologies and emission
categories are combined before an average I/M benefit for the
model year is calculated.
4.6.2 Evaporative and Running Loss Modeling, and the
Effectiveness of Purge/Pressure Testing
A large part of the additional emission reduction available
through the use of high-tech I/M tests is the result of the
evaporative and running loss emission reductions achieved by the
repair of vehicles which fail the new evaporative system pressure
and purge tests. The effectiveness of evaporative system pressure
and purge checks in reducing the rate of pressure and purge
problems was calculated assuming that programs with these checks
would detect 100% of all problems detected by the EPA checks run
in the Hammond I/M program. This assumes that the program will
use methods similar to the procedures used in Indiana. Although
all of the pressure and purge problems are assumed to be detected,
since some problems will re-occur with time, the average rate of
problems over the inspection cycle will not be zero.
For purposes of determination of program effectiveness, the
combined evaporative system pressure and purge failure rates from
over 2,400 vehicles tested in Indiana were used. The resulting
effectiveness estimates were then used for application of pressure
checks, purge checks and combined pressure and purge checks in the
MOBILE4.1 model.
The average reduction in the rate of failure is calculated by
determining the rate of failure at the midpoint between two
vehicle ages. The effect of inspection can be visualized by
plotting the non-program rate over age with the calculated before
and after repairs failure rate estimates assuming inspection (see
figure in Appendix B). At each age, vehicles due for inspection
-31-

-------
4-5
Short Test Identification Rates*
Super Emitters
Test
Model
PFI
PFI
TBI
TBI
Carb
Carb
Oplp
Oplp

Xaaca





QQ

QQ
Idle Test
81-82
0.6048
0.6968
0.6048
0.6968
0.6048
0.6968
0.0000
0.0000
Idle Test
83 +
0.8978
0.9656
0.8978
0 . 9656
0 .8978
0.9656
0.0000
0.0000
2500/Idle
81-82
0.6523
0.8577
0.6523
0.8577
0.6523
0.8577
0.0000
0.0000
2500/Idle
83 +
0.8978
0.9656
0.8978
0.9656
0.8978
0.9656
0.0000
0.0000
Load/Idle
81-82
0.6523
0.8577
0.6523
0.8577
0.6523
0.8577
0.0000
0.0000
Load/Idle
83 +
0.8978
0.9656
0.8978
0.9656
0.8978
0.9656
0.0000
0.0000
IM240
81-82
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
0.0000
0.0000
IM24 0
83 +
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
0.0000
0.0000



Verv
Hiah Emitters




Idle Test
81-82
0.2736
0.3231
0.2736
0.3231
0 . 3858
0.4108
0.4568
0.5194
Idle Test
83 +
0.5676
0.6129
0.2651
0.2695
0 .3640
0.3180
0.3640
0.3180
2500/Idle
81-82
0.2736
0.3231
0 . 2736
0.3231
0 . 4789
0.5331
0.6197
0.6162
2500/Idle
83 +
0.6187
0.7465
0.3616
0.4206
0 .5684
0.6832
0.5684
0.6832
Load/Idle
81-82
0.2736
0.3231
0.2736
0.3231
0.5476
0.6037
0.6197
0.6162
Load/Idle
83 +
0.6187
0.7465
0.3904
0.4337
0.5684
0.6832
0.5684
0.6832
IM240
81-82
0.8920
0.9460
0.8770
0.8750
0.8760
0.8680
0.8760
0.8680
IM240
83 +
0.8800
0.9400
0.8600
0.8600
0.9400
0.8300
0.9400
0.8300
Hioh Emitters
Idle Test
81-82
0.0506
0.1135
0.0506
0.1135
0.0563
0.0492
0 .2274
0.1522
Idle Test
83 +
0.2507
0.2208
0.0336
0.0613
0.0694
0.0415
0.0694
0.0415
2500/Idle
81-82
0.0506
0.1135
0.0506
0.1135
0.0898
0.0834
0.2274
0.1522
2500/Idle
83 +
0.3436
0.3501
0.1924
0.1532
0.0694
0.0415
0.0694
0.0415
Load/Idle
81-82
0 . 0506
0.1135
0.0506
0.1135
0.0910
0.0896
0.2274
0.1522
Load/Idle
83 +
0.3866
0.3937
0.1924
0.1532
0.0694
0.0415
0.0694
0.0415
IM240
81-82
0.0930
0.0600
0.5080
0.4190
0.1820
0.2060
0.1820
0.2060
IM2 4 0
83 +
0.1300
0.0800
0.5100
0.4200
0.1800
0.2200
0.1800
0.2200
Normal Emitters
Idle Test
81-82
0.0556
0.0774
0.0139
0.0139
0.0188
0.0204
0.0093
0.0131
Idle Test
83+
0.0360
0.0414
0.0425
0.0436
0.0023
0.0078
0.0023
0.0078
2500/Idle
81-82
0.0556
0.0774
0.0139
0.0139
0.0371
0.0427
0.0201
0.0317
2500/Idle
83+
0.0575
0.0694
0.0476
0.0514
0.0140
0.0156
0.0065
0.0208
Load/Idle
81-82
0.0556
0.0774
0.0139
0.0139
0.0371
0.0427
0.0201
0.0317
Load/Idle
83 +
0.0907
0.1023
0.0712
0.0739
0.0140
0.0156
0.0231
0.0403
IM240
81-82
0.0450
0.0560
0.0970
0.0750
0.1340
0.1200
0.1340
0.1200
IM240
83 +
0.0500
0.0600
0.1000
0.0800
0.2400
0.2100
0.2400
0.2100
* Identification Rate (IDR) is the fraction of the total sample emissions
from vehicles failing the short test.
-32-

-------
Table 4-6
Short Test Repair Effectiveness
PFI/TBI	Carb/Qplp




Before
After
Before
After




Repair
Repair
Repair
Repair
Idle
Test
HC
0 . 7400
0 . 4108
0.9677
0. 6224
Idle
Test
HC
1.9223
0.6062
2 . 0226
1.1894
Idle
Test
HC
3.9023
1.0769
3 . 1063
1.3254
Idle
Test
HC
14.2820
1.3808
8.5543
1 . 5286
Idle
Test
CO
9.2708
4.9900
10.4870
9.8624
Idle
Test
CO
28.0310
9.4669
29.5500
12.9690
Idle
Test
CO
90.0380
12.1480
53.5200
17.4340
Idle
Test
CO
190 . 6600
20.6200
134.7500
18.2810
2500
rpm/Idle*
HC
0 . 8267
0. 4075
0.9303
0.5764
2500
rpm/Idle*
HC
1 . 9846
0.5923
1.9431
1 .0349
2500
rpm/Idle*
HC
3 . 9314
1.0271
2.9862
1 . 1413
2500
rpm/Idle*
HC
14 . 2820
1.3808
8.2523
1 . 4141
2500
rpm/Idle*
CO
10.3340
4.8950
10 . 6220
9.2808
2500
rpm/Idle*
CO
35.5180
9.8631
29 .0530
12.4890
2500
rpm/Idle*
CO
104 . 5000
11.9250
54.2820
13.1900
2500
rpm/Idle*
CO
190.6600
20.6200
136.9700
13.5960
* Also used for Loaded/Idle and IM240 repair effects.
are checked and necessary repairs made. Between inspections, the
rate of failures increases until the vehicles are due for
inspection again. The slope of this failure rate line between
inspections is assumed to be equal to the slope of the non-program
line for that vehicle age. This creates a rising and falling
pattern of rates resembling a saw blade. The average reduction in
rates is then the average value of the "saw teeth" compared with
the non-program case.
With an inspection program, at age zero, when the calendar
year equals the model year, no vehicles are yet one year old and
due for inspection; therefore, no reductions are made. Assuming
an annual inspection, at age one, 25% of the model year is one
year old or older. Therefore, the rate at one year is reduced by
25% to reflect repairs on the vehicles due for inspection. By the
second year, all vehicles are inspected each year and the after
repair rate is always zero. The failure rate after a check is
always zero, since the detection rate is 100%. Therefore, the
midpoint failure rate is half the number of failures that occur in
that year, once inspections begin. In the biennial case, vehicles
are inspected every other year and the rate of failures
accumulates in the years between inspections.
-33-

-------
This method was used in a computer spreadsheet to calculate
the reduction in failures from evaporative system pressure and
purge checks used in the M0BILE4.1 model. The spreadsheet is
shown in Appendix B with and without formulas. The spreadsheet
originally contained errors which caused the benefits used in the
M0BILE4.1 model to be smaller than the estimates reported in this
document (which are based upon the corrected spreadsheet). The
version of MOBILE4.1 released to the public does not yet reflect
these changes, although they will be incorporated into the next
MOBILE release.
4.6,3	Benefits of IM240 NO:-: Inspections
None of the existing I/M program models or the M0BILE4.1
model itself are designed to estimate the effect of N0X emission
inspection as part of an I/M program. Therefore, to estimate the
effect of an IM240-based N0X inspection, a simple model was
developed.
A sample of over 3,200 1983 and newer model year vehicles,
tested in Hammond, Indiana using the IM240 test procedure, was
analyzed. The sample was divided into three technology groups:
multi-point fuel injection vehicles, throttle-body fuel injection
vehicles and carbureted vehicles . Two NOx cutpoint cases were
examined for each technology, one with a 10% failure rate and one
with a 20% failure rate.
Using an emission correlation mapping between IM240 NOx
measurements and NOx measured on the FTP, an FTP NOx emission
level was estimated for each vehicle in the sample. A linear
least-square regression was run for estimated FTP N0X emissions
versus mileage for each technology for two model year groups: 1983
through 1985 model year vehicles and 1986 and newer model year
vehicles. The regressions were then run again excluding vehicles
which fail the IM240 N0X inspection first using the 10% failure
rate cutpoints and then the 20% failure rate cutpoints . The
exclusion of the higher NOx emitters was intended to represent
their deletion from the fleet through repairs.
Using the technology mix used in MOBILE4.1, the regressions
were weighted together to produce emission factor zero mile levels
and deterioration rates for each model year from 1983 through
1992. The difference in the emission levels between the cases
with and without NOx failures removed is assumed to be the benefit
from the IM240 NOx emission test with only NOx-related repairs
performed. Results are shown in Table 4-7.
Since it is expected that most NOx emission testing will be
done along with testing for HC and CO emissions, the side effect
of HC and CO repairs on NOx emissions should also be accounted
for. This effect is ignored in the standard MOBILE4.1 model.
-34-

-------
Typically, NOx emissions will increase, on average, when HC and CO
emission repairs are performed. The extent of this NOx emission
disbenefit was determined by calculating average NOx emission
levels corresponding, to the Normal, High, Very High and Super
HC/CO emitter categories used in the M0BILE4 . 1 Tech4 model.
Using the post-repair emission levels of the same vehicles
used to calculate the after repair emission levels for HC and CO
emissions, the N0X emission levels of these vehicles after repairs
were determined. These N0X emission levels are not the result of
NOv-related repairs, but a by-product of HC and CO emission
repairs. Using the standard HC/CO 2500 rpm/Idle test IDRs along
with the repair effects on NOx and the NOx emission rates by
emitter group in the Tech4 model, the eff.ect of NOx disbenefits
was determined for each age of each model year (see Table.4-8).
The NOx disbenefits, as a percent change, are applied to the
emission levels estimated from the regression equations at each
age. The resulting NOx emission levels by age are regressed
versus mileage for each model year to give the final emission
factor equation for NOx. Comparing the emission factor results of
the baseline case with the cases with 10% or 20% NOx emission
testing failure rates was done to estimate the benefits, in tons,
of the IM240 NOx emission test. Results are shown in Table 4-9.
For example, at age 5 and mean mileage of 60, 829 miles, the "20%
fail" IM240 NOx cutpoints will reduce 1992 model year NOx from
0.887 to 0.710 gpm, a reduction of 20%.
The final emission factors were used as alternate input to
the MOBILE4 . 1 model and, in combination with the CEM4.1 model,
used to calculate the tons of NOx emission benefit from use of
IM240 NOx cutpoints. These benefits were used in applying the
cost credit . It should be noted that since both the cases with
and without the IM240 N0X inspection cutpoints should include the
disbenefits of HC/CO repairs, the disbenefits do not effect the
calculation of incremental NOx reduction from IM240 cutpoints.
For simplicity and consistency, therefore, the disbenefits were
not applied to the I/M scenarios involving only HC/CO cutpoints.
-35-

-------
Table 4-7
Lane IM240 Based	Emission Factor Levels with IM240	NO-^ Outpoints
Age	0 1	2	3	4	5	6	7	8	9	10	11	12
Miles	0	1.3118	2.6058	3.8298	4.9876	6.0829	7.119	8.0991	9.0262	9.9031	10.7326	11.5172	12.2594
Year	Base
1983	1.146	1.197	1.248	1.296	1.341	1.384	1.425	1.463	1.499	1.533	1.566	1.597	1.626
1984	1.048	1.115	1.181	1.243	1.303	1.359	1.412	1.462	1.509	1.554	1.596	1.636	1.674
1985	0.983	1.049	1.114	1.175	1.233	1.288	1.340	1.389	1.436	1.480	1.521	1.561	1.598
1986	0.608	0.683	0.758	0.828	0.895	0.958	1.017	1.074	1.127	1.177	1.225	1.270	1.313
1987	0.593	0.669	0.744	0.815	0.882	0.946	1.006	1.063	1.1 17	1.168	1.216	1.262	1.305
1988	0.561	0.633	0.705	0.773	0.837	0.897	0.954	1.009	1.060	1.108	1.154	1.198	1.239
1989	0.570	0.639	0.707	0.772	0.833	0.890	0.945	0.997	1.046	1.092	1.136	1.177	1.216
1990	0.550	0.614	0.677	0.737	0.794	0.847	0.898	0.946	0.991	1.034	1.075	1.113	1.149
1991	0.547	0.610	0.672	0.731	0.787	0.840	0.889	0.936	0.981	1.023	1.063	1.101	1.136
1992	0.547	0.610	0.671	0.729	0.784	0.837	0.886	0.932	0.977	1.018	1.058	1.095	1.130
1983	1.078
1984	1.036
1985	0.970
1986	0.600
1987	0.591
1988	0.559
1989	0.556
1990	0.529
1991	0.525
1992	0.523
1983	0.985
1984	0.955
1985	0.911
1986	0.591
1987	0.582
1988	0.551
1989	0.550
1990	0.525
1991	0.521
1992	0.520
1.092	1.106
1:05 1	1.066
0.984	0.998
0.650	0.699
0.640	0.688
0.603	0.647
0.600	0.643
0.569	0.608
0.564	0.602
0.562	0.600
0.992	0.999
0.962	0.968
0.917	0.922
0.629	0.666
0.619	0.656
0.586	0.621
0.585	0.619
0.558	0.590
0.553	0.585
0.552	0.584
1.120	1.132
1.080	1.093
1.011	1.023
0.746	0.790
0.734	0.777
0.688	0.727
0.684	0.723
0.645	0.681
0.639	0.673
0.636	0.670
1.006	1-.012
0.974	0.979
0.927	0.932
0.702	0.735
0.691	0.724
0.654	0.685
0.652	0.683
0.621	0.650
0.616	0.644
0.614	0.642
10% Fail
1.144	1.155
1.105	1.117
1.034	1.045
0.831	0.871
0.817	0.856
0.764	0.800
0.759	0.794
0.714	0.745
0.706	0.737
0.703	0.733
20% Fail
1.018	1.023
0.984	0.989
0.937	0.941
0.766	0.796
0.755	0.785
0.715	0.743
0.712	0.739
0.677	0.703
0.671	0.697
0.669	0.695
1.166	1.176
1.128	1.139
1.056	1.065
0.908	0.943
0.893	0.927
0.833	0.864
0.827	0.858
0.775	0.803
0.766	0.793
0.762	0.790
1.028	1.033
0.994	0.999
0.945	0.949
0.825	0.851
0.813	0.839
0.769	0.794
*0.765	0.790
0.728	0.751
0.721	0.744
0.719	0.741
1.185	1.194
1.149	1.158
1.074	1.083
0.977	1.008
0.960	0.991
0.894	0.922
0.887	0.914
0.830	0.855
0.820	0.844
0.816	0.840
1.038	1.042
1.003	1.007
0.953	0.956
0.877	0.901
0.864	0.888
0.818	0.840
0.813	0.835
0.773	0.793
0.766	0.786
0.763	0.783
1.203	1.21 1
1.167	1.176
1.091	1.099
1.038	1.067
1.020	1.047
0.948	0.973
0.941	0.965
0.879	0.901
0.868	0.890
0.864	0.88 5
1.046	1.050
1.011	1.014
0.959	0.962
0.923	0.945
0.910	0.932
0.861	0.881
0.856	0.875
0.813	0.832
0.805	0.824
0.803	0.821
-36-

-------
Table 4-7
- continued -
13
14
15
16
17
18
19
20
21
22
23
24
25 Age
Regression
12.9615 13.6237 14.234 M.MM 13.4104 13.0-421 16.4431 16.9209 17.3711 17.7969 H.I997 l*_306 l.94l Miles ZML DET
Base
1.653
1.710
1.633
1.353
1.345
1.278
1.253
1.183
1.170
1.164
1.679
1.744
1.666
1.392
1.384
1.314
1.288
1.216
1.202
1.195
1.704
1.776
1.698
1.428
1.420
1.349
1.321
1.247
1.232
1.225
1.727
1.807
1.728
1.462
1.455
1.382
1.353
1.276
1.261
1.254
1.749
1.835
1.756
1.494
1.488
1.413
1.382
1.303
1.288
1.280
1.770
1.863
1.782
1.525
1.519
1.442
1.410
1.329
1.313
1.306
1.790
1.888
1.808
1.554
1.548
1.470
1.437
1.354
1.338
1.329
1.808
1.913
1.831
1.581
1.575
1.497
1.462
1.377
1.360
1.352
1.826
1.936
1.854
1.607
1.602
1.521
1.486
1.399
1.382
1.374
1.842
1.957
1.875
1.632
1.626
1.545
1.508
1.420
1.402
1.394
1.858
1.978
1.896
1.655
1.650
1.567
1.530
1.439
1.422
1.413
1.873
1.998
1.915
1.677
1.672
1.588
1.550
1.458
1.440
1.431
1.887
2.016
1.933
1.698
1.693
1.608
1.569
1.476
1.457
1.448
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1.146
1.048
0.983
0.608
0.593
0.561
0.570
0.550
0.547
0.547
0.0391
0.0511
0.0501
0.0575
0.0581
0.0553
0.0527
0.0489
0.0481
0.0476
10% Fail
1.218
1.184
1.107
1.093
1.074
0.997
0.989
0.923
0.911
0.906
1.226
1.191
1.114
1.119
1.098
1.020
1.011
0.943
0.931
0.926
1.232
1.198
1.120
1.143
1.122
1.041
1.032
0.962
0.949
0.944
1.239
1.205
1.126
1.165
1.144
1.061
1.052
0.980
0.967
0.962
1.245
1.211
1.132
1.187
1.165
1.080
1.07 1
0.997
0.984
0.979
1.251
1.218
1.138
1.207
1.185
1.098
1.088
1.01 3
1.000
0.994
1.256
1.223
1.143
1.226
1.203
1.115
1.105
1.028
1.015
1.009
1.261
1.229
1.148
1.244
1.221
1.131
1.121
1.043
1.029
1.023
1.266
1.234
1.153
1.261
1.238
1.146
1.136
1.056
1.042
1.037
1.271
1.239
1.157
1.277
1.254
1.161
1.150
1.069
1.055
1.049
1.275
1.243
1.162
1.293
1.269
1.174
1.164
1.081
1.067
1.061
1.279
1.248
1.166
1.307
1.283
1.187
1.176
1.093
1.078
1.072
1.283
1.252
1.169
1.321
1.296
1.199
1.188
1.104
1.089
1.083
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1.078
1.036
0.970
0.600
0.591
0.559
0.556
0.529
0.525
0.523
0.0108
0.0114
0.0105
0.0381
0.0372
0.0338
0.0334
0.0303
0.0298
0.0296
20% Fail
1.054
1.018
0.965
0.965
0.952
0.900
0.894
0.849
0.841
0.838
1.058
1.021
0.968
0.984
0.971
0.918
0.912
0.866
0.858
0.854
1.061
1.024
0.971
1.002
0.988
0.935
0.928
0.882
0.873
0.870
1.064
1.027
0.973
1.019
1.005
0.951
0.944
0.896
0.888
0.885
1.067
1.029
0.976
1.035
1.021
0.966
0.959
0.910
0.902
0.898
1.070
1.032
0.978
1.051
1.037
0.980
0.973
0.924
0.915
0.911
1.073
1.034
0.980
1.065
1.051
0.994
0.987
0.936
0.927
0.924
1.075
1.037
0.982
1.079
1.065
1.006
0.999
0.948
0.939
0.935
1.077
1.039
0.984
1.092
1.077
1.019
1.011
0.959
0.950
0.947
1.080
1.041
0.986
1.104
1.090
1.030
1.022
0.970
0.961
0.957
1.082
1.043
0.987
1.116
1.101
1.041
1.033
0.980
0.971
0.967
1.084
1.045
0.989
1.127
1.112
1.051
1.043
0.990
0.980
0.976
1.086
1983
0.985
0.0053
1.046
1984
0.955
0.0048
0.990
1985
0.911
0.0042
1.137
1986
0.591
0.0288
1.122
1987
0.582
0.0285
1.061
1988
0.551
0.0269
1.053
1989
0.550
0.0265
0.999
1990
0.525
0.0250
0.989
1991
0.521
0.0247
0.985
1992
0.520
0.0246
-37-

-------
Table 4-8
Side Effects of I/M on NOx Emissions




(Disbenefit of
HC/CO repairs)



Age
0
1
2
3
4
5
6
7
8
9
10
11
12
Miles
0
1.31 18
2.6058
3.8298
4.9876
6.0829
7.1 19
8.0991
9.0262
9.9031
10.7326
1 1.5172
12.2594
Year






Base






1983
0.61
0.67
0.74
0.80
0.86
0.91
0.98
1.04
1.10
1.15
1.20
1.25
1.30
1984
0.63
0.69
0.75
0.81
0.87
0.92
0.99
1.05
1.11
1.17
1.22
1.27
1.32
1985
0.63
0.69
0.74
0.79
0.84
0.89
0.96
1.02
1.09
1.14
1.20
1.25
1.30
1986
0.51
0.56
0.62
0.67
0.72
0.76
0.81
0.85
0.89
0.92
0.96
0.99
1.02
1987
0.50
0.55
0.60
0.64
0.69
0.73
0.77
0.81
0.84
0.88
0.91
0.94
0.97
1988
0.47
0.52
0.56
0.60
0.64
0.68
0.72
0.75
0.78
0.8 1
0.84
0.87
0.89
1989
0.47
0.52
0.57
0.62
0.66
0.70
0.74
0.78
0.81
0.85
0.88
0.91
0.94
1990
0.46
0.50
0.55
0.59
0.64
0.68
0.71
0.75
0.78
0.82
0.85
0.88
0.90
1991
0.45
0.50
0.55
0.59
0.63
0.67
0.71
0.75
0.78
0.81
0.85
0.87
0.90
1992
0.45
0.50
0.55
0.59
0.63
0.67
0.71
0.75
0.78
0.82
0.85
0.88
0.90







Iille






1983
0.61
0.69
0.76
0.82
0.88
0.93
0.99
1.05
1.10
1.15
1.20
1.25
1.29
1984
0.63
0.71
0.77
0.83
0.88
0.93
1.00
1.06
1.12
1.17
1.22
1.27
1.31
1985
0.63
0.70
0.76
0.81
0.86
0.90
0.97
1.04
1.09
1.15
1.20
1.25
1.30
1986
0.51
0.58
0.64
0.69
0.74
0.78
0.83
0.87
0.91
0.95
0.98
1.02
1.05
1987
0.50
0.56
0.61
0.66
0.71
0.75
0.80
0.84
0.87
0.91
0.94
0.97
1.00
1988
0.47
0.53
0.58
0.63
0.67
0.7 1
0.75
0.78
0.82
0.85
0.88
0.90
0.93
1989
0.47
0.54
0.59
0.64
0.69
0.73
0.77
0.8 1
0.85
0.88
0.91
0.94
0.97
1990
0.46
0.52
0.57
0.62
0.66
0.70
0.74
0.78
0.82
0.85
0.88
0.91
0.94
1991
0.45
0.52
0.57
0.62
0.66
0.70
0.74
0.78
0.82
0.85
0.88
0.91
0.94
1992
0.45
0.52
0.57
0.62
0.66
0.70
0.74
0.78
0.82
0.8 5
0.88
0.91
0.94
Two Speed
198 3
0.61
0.69
0.76
0.83
0.88
0.93
1.00
1.05
1.10
1.15
1.20
1.25
1.29
1984
0.63
0.71
0.78
0.83
0.88
0.93
1.00
1.06
1.12
1.17
1.22
1.27
1.31
1985
0.63
0.7 1
0.76
0.82
0.86
0.91
0,97
1.04
1.10
1.15
1.20
1.25
1.30
1986
0.51
0.58
0.64
0.70
0.75
0.79
0.84
0.88
0.92
0.96
1.00
1.03
1.06
1987
0.50
0.57
.0.62
0.67
0.72
0.76
0.81
0.85
0.89
0.92
0.96
0.99
1.02
1988
0.47
0.54
0.59
0.63
0.68
0.72
0.76
0.80
0.83
0.87
0.90
0.93
0.95
1989
0.47
0.55
0.60
0.65
0.70
0.74
0.78
0.82
0.86
0.90
0.93
0.96
0.99
1990
0.46
0.53
0.58
0.63
0.67
0.72
0.76
0.80
0.84
0.87
0.90
0.93
0.96
1991
0.45
0.53
0.58
0.63
0.67
0.71
0.76
0.80
0.83
0.87
0.90
0.93
0.96
1992
0.45
0.53
0.58
0.63
0.67
0.72
0.76
0.80
0.84
0.87
0.90
0.93
0.96
-38-

-------
Table 4-8
- continued -
13
14
15
16
17
18
19
20
21
22
23
24
25
Model
Regression
12.9615
13.6257
14.254
14.ft433
15.4104
15.9421
16.4451
I6.92J9
17.3712
17.7969
IB. 1997
18.5806
18.941
Year
ZML
DET







Base










1.34
1.38
1.41
1.44
1.47
1.49
1.51
1.53
1.55
1
56
1.58
1
60
1.61
1983
0.601
0.0550
1.36
1.40
1.44
1.46
1.49
1.51
1.53
1.54
1.56
1
58
1.59
1
60
1.62
1984
0.617
0.0549
1.35
1.39
1.41
1.44
1.45
1.47
1.48
1.50
1.51
1
53
1.54
1
55
1.56
1985
0.614
0.0527
1.05
1.08
1. 10
1.13
1.15
1.17
1.19
1.21
1.23
I
25
1.26
1
28
1.29
1986
0.511
0.0415
1.00
1.02
1.04
1.07
1.09
1. 11
1.13
1.15
1.16
1
18
1.19
1
21
1.22
1987
0.497
0.0383
0.92
0.94
0.96
0.98
1.00
1.02
1.04
1.05
1.07
I
08
1.10
1
11
1.12
1988
0.470
0.0345
0.96
0.99
1.01
1.03
1.05
1.07
1.09
1.11
1.13
1
14
1.16
1
17
1.18
1989
0.475
0.0375
0.93
0.95
0.98
1.00
1.02
1.04
1.05
1.07
1.09
1
10
1.12
1
13
1.14
1990
0.455
0.0364
0.93
0.95
0.97
1.00
1.02
1.03
1.05
1.07
1.09
1
10
1.12
1
13
1.14
1991
0.453
0.0365
0.93
0.95
0.98
1.00
1.02
1.04
1.06
1.07
1.09
1
11
1.12
1
13
1.15
1992
0.452
0.0367







Idle










1.33
1.37
1.40
1.42
1.44
1.46
1.48
1.50
1.52
1
53
1.54
1
56
1.57
1983
0.628
0.0519
1.36
1.39
1.43
1.45
1.47
1.49
1.50
1.52
1.53
1
55
1.56
1
57
1,58
1984
0.640
0.0524
1.34
1.38
1.41
1.42
1.44
1.46
1.47
1.48
1.49
1
50
1.51
1
52
1.53
1985
0.637
0.0505
1.07
1.10
1.13
1.15
1.17
1.19
1.21
1.23
1.25
1
26
1.28
1
29
1.30
1986
0.530
0.0414
1.02
1.05
1.07
1.09
1.11
1.13
1.15
1.17
1.18
1
20
1.21
1
22
1.24
1987
0.516
0.0385
0.95
0.97
0.99
1.01
1.03
1.05
1.07
1.08
1.10
1
11
1.12
1
13
1.14
1988
0.491
0.0350
0.99
1.02
1.04
1.06
1.08
1.10
1.12
1.13
1.15
1
16
1.17
1
19
1.20
1989
0.497
0.0377
0.96
0.99
1.01
1.03
1.05
1.06
1.08
1.10
1.11
1
12
1.14
1
15
1.16
1990
0.479
0.0367
0.96
0.98
1.01
1.03
1.04
1.06
1.08
1.09
1.1 1
1
12
1.14
1
15
1.16
1991
0.477
0.0367
0.96
0.99
1.01
1.03
1.05
1.07
1.08
1.10
1.11
1
13
1.14
1
15
1.16
1992
0.476
0.0369







Two Speed









1.33
1.36
1.39
1.42
1.44
1.46
1.47
1.49
1.51
1
52
1.53
1
55
1.56
1983
0.637
0.0509
1.35
1.39
1.42
1.45
1.46
1.48
1.50
1.51
1.53
1
54
1.55
1
56
1.57
1984
0.645
0.0516
1.34
1.38
1.40
1.42
1.44
1.45
1.46
1.48
1.49
1
50
1.51
1
52
1.53
1985
0.642
0.0499
1.09
1.12
1.14
1.17
1.19
1.21
1.23
1.24
1.26
1
28
1.29
1
30
1.32
1986
0.535
0.0421
1.04
1.07
1.09
1.1 1
1.13
1.15
1.17
1.19
1.20
1
22
1.23
1
24
1.25
1987
0.521
0.0395
0.98
1.00
1.02
1.04
1.06
1.08
1.09
1.11
1.12
1
13
1.14
1
16
1.17
1988
0.497
0.0362
1.01
1.04
1.06
1.08
1.10
1.12
1.14
1.15
1.17
1
18
1.19
1
21
1.22
1989
0.503
0.0385
0.98
1.01
1.03
1.05
1.07
1.09
1.10
1.12
1.13
1
14
1.16
1
17
1.18
1990
0.487
0.0374
0.98
1.00
1.03
1.05
1.07
1.08
1.10
1.11
1.13
1
14
1.15
1
17
1.18
1991
0.485
0.0374
0.98
1.01
1.03
1.05
1.07
1.09
1.10
1.12
1.13
1
15
1.16
1
17
1.18
1992
0.485
0.0376
-39-

-------
Table 4-9
Lane IM240 Based Emission Factors with IM240 Cutpoints
with Disbenefits of HC/CO Repairs Included
Age	0
Miles	0
Year
1983	1.146
1984	1.048
1985	0.983
1986	0.608
1987	0.593
1988	0.561
1989	0.570
1990	0.550
1991	0.547
1992	0.547
1983	1.078
1984	1.036
1985	0.970
1986	0.600
1987	0.591
1988	0.559
1989	0:556
1990	0.529
1991	0.525
1992	0.523
1983	0.985
1984	0.955
1985	0.911
1986	0.591
1987	0.582
1988	0.551
1989	0.550
1990	0.525
1991	0.521
1992	0.520
1	2
1.3118 2.6058
1.235	1.288
1.145	1.214
1.081	1.147
0.707	0.787
0.695	0.775
0.662	0.740
0.667	0.742
0.646	0.714
0.642	0.710
0.641	0.709
1.126	1.142
1.080	1.095
1.014	1.027
0.673	0.726
0.664	0.717
0.630	0.679
0.627	0.675
0.599	0.642
0.593	0.636
0.591	0.634
1.023	1.032
0.988	0.995
0.945	0.950
0.651	0.692
0.643	0.684
0.613	0.652
0.611	0.650
0.587	0.623
0.582	0.618
0.580	0.617
3	4
3.8298 4.987.6
1.336	1.374
1.271	1.325
1.208	1.260
0.861	0.930
0.851	0.922
0.814	0.884
0.8 12	0.877
0.781	0.841
0.775	0.835
0.774	0.833
1.154	1.160
1.104	1.112
1.039	1.045
0.776	0.821
0.766	0.812
0.725	0.768
0.720	0.761
0.684	0.721
0.676	0.715
0.675	0.712
1.037	1.037
0,995	0.996
0.953	0.952
0.730	0.764
0.721	0.757
0.689	0.724
0.686	0.719
0.658	0.689
0.652	0.684
0.651	0.682
5	6
6.0829	7.119
Base
1.4 13	1.448
1.378	1.427
1.313	1.361
0.995	1.059
0.989	1.053
0.949	1.013
0.939	0.999
0.899	0.956
0.891	0.947
0.887	0.943
10% Fail
1.168	1.174
1.121	1.130
1.054	1.062
0.864	0.907
0.854	0.896
0.808	0.849
0.801	0.839
0.757	0.793
0.749	0.784
0.746	0.781
20% Fail
1.039	1.040
0.998	1.001
0.955	0.956
0.797	0.829
0.790	0.822
0.756	0.788
0.750	0.781
0.718	0.749
0.712	0.742
0.710	0.739
7	8
8.0991 9.0262
1.480	1.509
1.473	1.517
1.406	1.448
1.119	1.175
1.114	1.172
1.072	1.128
1.054	1.106
1.006	1.056
0.998	1.045
0.993	1.040
1.179	1.183
1.137	1.145
1.068	1.074
0.947	0.984
0.936	0.973
0.885	0.919
0.874	0.907
0.825	0.855
0.816	0.845
0.812	0.841
1.040	1.040
1.002	1.004
0.956	0.957
0.860	0.888
0.852	0.881
0.817	0.845
0.809	0.835
0.774	0.800
0.768	0.792
0.766	0.790
9	10
9.9031 10.7326
1.536	1.565
1.558	1.595
1.486	1.526
1.227	1.275
1.226	1.278
1.181	1.230
1.155	1.200
1.101	1.143
1.091	1.131
1.084	1.125
1.187	1.193
1.152	1.157
1.079	1.087
1.018	1.050
1.008	1.041
0.952	0.982
0.938	0.967
0.884	0.909
0.874	0.898
0.869	0.894
1.040	1.041
1.005	1.006
0.957	0.959
0.914	0.937
0.908	0.933
0.871	0.895
0.860	0.883
0.823	0.844
0.816	0.836
0.813	0.833
11	12
11.5172 12.2594
1.590	1.614
1.631	1.666
1.561	1.594
1.323	1.365
1.325	1.368
1.275	1.319
1.243	1.283
1.183	1.219
1.170	1.206
1.164	1.199
1.198	1.202
1.164	1.169
1.091	1.097
1.081	1.108
1.071	1.098
1.009	1.036
0.993	1.018
0.934	0.956
0.922	0.944
0.918	0.939
1.042	1.043
1.007	1.009
0.959	0.960
0.961	0.982
0.956	0.977
0.916	0.938
0.904	0.923
0.864	0.882
0.856	0.874
0.853	0.871
-40-

-------
Table 4-9
- continued -
[3
14
15
16
17
18
19
20
21
22
23
24
25 Model Regression
12.9615 13.6237 14.234 14.84H 13.4104 15.9421 16.4431 16.9209 17.3712 17.7969 U.I997 IB.3S06 11.941 Year ZML
DET
Base
1.637
1.699
1.626
1.405
1.409
1.357
1.320
1.252
1.240
1.231
1.660
1.728
1.655
1.443
1.448
1.395
1.354
1.285
1.269
1.262
1.680
1.757
1.685
1.478
1.484
1.429
1.387
1.314
1.299
1.290
1.698
1.783
1.711
1.510
1.519
1.462
1.417
1.342
1.327
1.3 18
1.716
1.807
1.735
1.541
1.549
1.493
1.445
1.367
1.351
1.342
1.733
1.832
1.761
1.571
1.579
1.522
1.471
1.392
1.376
1.366
1.748
1.854
1.783
1.597
1.607
1.547
1.496
1.414
1.396
1.387
1.763
1.875
1.803
1.623
1.632
1.570
1.519
1.4 35
1.4 16
1.406
1.779
1.895
1.823
1.645
1.655
1.594
1.539
1.454
1.435
1.425
1.791
1.913
1.842
1.667
1.679
1.616
1.560
1.473
1.453
1.444
1.803
1.931
1.860
1.689
1.699
1.634
1.578
1.488
I .470
1.460
1.816
1..948
1.878
1.707
1.719
1.653
1.596
1.506
1.486
1.475
1.825
1.962
1.893
1.726
1.738
1.671
1.612
1.520
1.501
1.490
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1.196
1.087
1.022
0.638
0.623
0.594
0.605
0.589
0.587
0.587
0.0338
0.0468
0.0463
0.0584
0.0599
0.0580
0.0543
0.0503
0.0494
0.0488
1.207
1. 176
1.102
1. 135
1.124
1.059
1.041
0.976
0.965
0.959
1.21 I
1.180
1.106
1.160
1.149
1.082
1.062
0.996
0.982
0.977
1.215
1.185
1.111
1.183
1.172
1.102
1.083
1.014
1.001
0.995
1.218
1.189
1.115
1.203
1.194
1.122
1.102
1.031
1.017
1.011
1.221
1.193
1.119
1.224
1.213
1.141
1.119
1.046
1.032
1.026
1.225
1.197
1.124
1.243
1.232
1.158
1.135
1.061
1.047
1.040
1.227
1.201
1.128
1.260
1.249
1.173
1.151
1.074
1.059
1.053
10% Fail
1.230 1.234
1.205
1.131
1.277
1.265
1.187
1.164
1.086
1.071
1.064
1.208
1.134
1.291
1.279
1.201
1.176
1.098
1.082
1.076
1.235
1.210
1.137
1.305
1.294
1.214
1.189
1.109
1.093
1.087
1.237
1.214
1.140
1.319
1.307
1.225
1.201
1.118
1.103
1.096
1.240
1.216
1.143
1.331
1.319
1.235
1.211
1.129
1.113
1.106
1.241
1.218
1.145
1.343
1.331
1.246
1.221
1.137
1.121
1.114
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1.118
1.067
1.001
0.628
0.619
0.590
0.588
0.564
0.560
0.558
0.0068
0.0082
0.0077
0.0385
0.0384
0.0355
0.0343
0.0311
0.0305
0.0302
20% Fail
1.044
1.011
0.961
1.002
0.997
0.956
0.942
0.899
0.891
0.887
1.045
1.011
0.962
1.021
1.015
0.974
0.958
0.915
0.905
0.902
1.046
1.012
0.963
1.038
1.033
0.990
0.974
0.929
0.921
0.916
1.046
1.013
0.964
1.053
1.050
1.006
0.989
0.943
0.934
0.930
1.047
1.014
0.964
1.068
1.064
1.021
1.003
0.955
0.946
0.942
1.048
1.015
0.966
1.082
1.078
1.034
1.015
0.967
0.958
0.954
1.048
1.015
0.967
1.095
1.091
1.045
1.027
0.978
0.968
0.964
1.048
1.017
0.967
1.107
1.103
1.056
1.038
0.988
0.978
0.973
1.050
1.017
0.967
1.118
1.114
1.067
1.047
0.997
0.987
0.982
1.049
1.017
0.968
1.128
1.125
1.078
1.057
1.006
0.996
0.992
1.050
1.018
0.969
1.139
1.134
1.085
1.066
1.013
1.004
0.999
1.051
1.019
0.970
1.147
1.143
1.094
1.074
1.022
1.011
1.006
1.050
1.019
0.970
1.157
1.153
1.102
1.082
1.028
1.018
1.013
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1.021
0.982
0.939
0.617
0.608
0.581
0.581
0.559
0.555
0.554
0.0018
0.0021
0.0017
0.0291
0.0294
0.0283
0.0272
0.0256
0.0252
0.0250
5.0	REGULATORY IMPACT ANALYSIS - ESTIMATING COST AND COST
EFFECTIVENESS
5.1	Cost of Conventional I/M Testing
EPA has collected and analyzed cost data from all operating
I/M programs that could provide the information. EPA has analyzed
-41-

-------
per vehicle costs in I/M programs based upon four basic pieces of
information: The I/M program agency budget, number of initial
tests, the fee for each test, and the portion returned to the
state or local government. This discussion will deal with three
aspects of I/M cost: Inspection costs, oversight costs, and repair
costs. Costs are analyzed for three different types of programs:
conventional centralized and decentralized test-and-repair, and
decentralized test-only.
5.1.1	Inspection and Administration Costs
Inspection fees are set in one of three ways: By a bid
process for a contract to supply inspection services, by
legislation or regulation establishing a maximum fee, or by market
forces. Ideally the fee is scaled to cover the cost of providing
the inspection, cover the fee to the state for oversight and
management, and to provide a reasonable profit to the operator
(except in government-run programs).
This ideal is not always met in actual I/M programs. In some
programs the inspection fee does not include a share for the
state's oversight costs, so these must be derived from the general
fund, with the result that oversight efforts are often
significantly underfunded. In many decentralized programs the
maximum fee is set below the actual cost (with profit) for the
test, so providers must make up for that cost by providing other
goods and services.
The economies of scale and efficiency of operation in high
volume test-only inspection networks enable motorists in these
programs to enjoy lower average inspection fees than in low volume
decentralized programs. Based upon 1989 I/M audits (which
collected information from all I,/M programs), and taking into
account both inspection and oversight costs, decentralized
programs using computerized analyzers have the highest costs,
averaging about $17.70 per vehicle; centralized contractor-run
programs average $8.42 per vehicle (recently gathered 1990 data
show slightly different numbers, although these have not greatly
affected the overall averages). Table 5-1 shows the estimated
cost of the I/M program on a per vehicle basis, including
inspection and oversight costs.
-42-

-------
Table 5-1
I/M Program Inspection Fees
Decentralized Programs	Centralized Programs
Proaram
Coat Per Vehicle
Program
Ari zona
Anchorage
$32.00
Fairbanks
$29.00
Connecticut
California
$48.39
Florida
Colorado
$11.20
Illinois
Georgia
$10.68
Louisville
Massachusetts
$17.18
Maryland
Michigan
$10.87
Minnesota
Missouri
$9.00
Nashville
North Carolina
$15.40
Washington
New Mexico
$16.00
Wisconsin
Nevada
$21.26

New York
$19.92

Pennsylvania
$9.01

Dallas
$17 .25

El Paso
$17.25

Davis County
$9.00

Utah County
$9.71

Salt Lake City
$11.49

Virginia
$13.50

Cost E'er Vehicle
$6.00
$10.00
$10.00
$8 .07
$6.00
$8.53
$8.00
$6.00
$9. 00
$8 . 73
In a centralized contractor-run program the contractor bears
the cost of acquiring land, constructing and equipping inspection
facilities, hiring and training staff, collecting and processing
data, conducting public information campaigns, as well as doing
the routine testing work. The state's role in this case is to
make sure the contractor meets its obligations and to study the
outcomes of the program to make sure it is meeting the goal.
In a decentralized program, individual firms and small
businesses are licensed to perform the inspection. In this case,
the state takes primary responsibility for many of the day to day
functions, such as data collection and processing, public
information, and inspector training, which are performed by the
contractor in a centralized program. The fact that inspections
are performed by many business entities instead of one, and that
there are more inspection sites means that state oversight and
program evaluation activities need to be more intensive in this
type of program.
5.1.2.1 Quality Assurance in Decentralized Programs
Costs of quality assurance (QA) measures vary among programs
depending upon the comprehensiveness of the QA program and are
not well documented in most state programs. The estimates in this
section are based on EPA's proposed requirements for QA; i.e.,
they are more representative of costs that would be incurred in a
good QA program than of QA programs as they currently exist. Cost
-43-

-------
information was obtained from the some I/M programs, principally
California, and from various industry sources.
Performance audits are conducted to ensure that records are
properly kept, that document security is adequate, that required
inspection equipment is present and properly maintained, that
inspectors have followed the rules, and to assess the general
state of operations. There are two types of performance audits:
overt and covert.
EPA's proposal requires overt audits of all test lanes or
bays at least twice per year. For those stations where problems
are discovered, either through administrative auditing or through
covert auditing or other oversight functions, follow-up audits are
needed to verify that the problems are resolved. Station visits
would also have to be conducted to perform monthly record audits
if such audits could not be performed via electronic link. In
this analysis it is assumed, given all these factors, that an
average of si:-: station visits per year will be performed.
Based upon information from California and New York, EPA
estimates that overt audits cost approximately $89 per audit.
Staff time is estimated at $80.80 per audit based on the
assumptions that an audit takes a total of three hours and that
staff are paid $35,000 per year with overhead at 60 percent.
Travel costs are estimated at $8.00 per audit based upon an
average round trip distance of .25 miles and operating costs of 32C
per mile based upon MVMA estimates. Hence, the annual cost per
station is estimated at $534.
EPA's proposal requires at least one covert audit per year
per inspector using vehicles set to fail the inspection. This
requirement would establish a minimum level of activity, although
it would not necessarily require that each inspector be covertly
audited. Additionally, in test-and-repair programs, the proposal
requires that each station receive one covert audit annually that
includes the purchase of repairs. Follow-up audits would be
performed.at stations where problems are discovered.
California estimates that its covert auditing program costs
about $1,000 per audit, on average. A number of different types
of costs are incurred in performing covert audits. The vehicle
has to be induced to fail the inspection and the inducement has to
be documented so that the improper testing can be proven in court
if necessary. The staff time and travel costs to perform and
document the audit also contribute to the overall cost . In
addition vehicles have to be acquired and should either be
replaced or have their appearance altered through repainting in
order to avoid recognition. The costs of pursuing a case through
the administrative legal system in those instances where improper
testing is discovered are also included in the overall $1,000 per
audit figure. EPA's proposal also requires that repairs be
purchased in at least one covert audit per station per year.
-44-

-------
In this analysis the overall level of activity is estimated
at three covert audits per station per year with repairs purchased
in one. The estimated annual cost per station is therefore
$3,250.
As indicated previously, station and inspector records are to
be reviewed or screened at least once per month to assess station
performance and identify problems that may indicate potential
fraud or incompetence. The preferable way to do this would be for
the state to obtain station records in a computerized format via a
direct data link (required in enhanced I/M programs) to the
inspection station and review and analyze them. Failing that,
monthly visits would be made to any test stations not connected by
the electronic data link and to review any records not recovered
via this link. In addition, data analyses would be needed to
track motorist compliance and to compile periodic reports.
California reportedly spends $1.8 million per year for data
analysis staff. Its staffing level is estimated at about one FTE
per 250 stations. As shown in Table 7-5 California has 8,752
stations, yielding an annual cost of $205.67 for data analysis
activities. This figure does not include the cost of acquiring
computer equipment for this purpose which some states may need to
do.
Referee stations are needed to process waiver requests and to
resolve consumer complaints of improper testing. In California
the referee system costs $36 per vehicle for those vehicle that
use it. (The California referee system is operated by a
contractor, the State estimates the per vehicle cost would be
roughly the same if the referee system were operated by the
State.) The referee system is designed to accommodate three
percent of the subject vehicle population. Tighter waiver limits
to be imposed in enhanced I/M programs are likely to increase the
pressure on referee stations. The cost estimates used here assume
a five percent utilization rate for the referee stations.
In enhanced I/M programs where the regular tailpipe test is
something other than the IM240, a facility to conduct transient
tests on 0.1 percent of subject vehicles would be needed.
There are a number of different ways the state could obtain
such a facility. Most likely a pre-existing garage or warehouse
would be acquired that could be easily converted to a testing
facility with only the addition of the necessary equipment. The
equipment package to perform IM240, purge, and pressure testing
costs an estimated $144,100. While building acquisition and
operating expenses can vary considerably, in this analysis, these
expenses are assumed to total $1 million over a five year period.
Testing volume is conservatively estimated at four vehicles per
day for a total of 1040 per year, and again, the total number of
vehicles tested represent 0.1 percent of the subject fleet. Using
-45-

-------
these very general assumptions the cost of the state testing
function is estimated at 220 per vehicle.
Inspector training courses have to be continually updated in
order to stay abreast of new developments.	Inspector
certification tests also have to be updated in order to keep them
from becoming too easy. California spends approximately $65 per
station per year on these efforts.
In states where I/M responsibilities are divided between the
environmental agency and the motor vehicle agency there is a cost
associated with transfer of data between the two agencies. While
such costs are difficult to estimate, California estimates that it
would cost 50C per vehicle per year
This analysis does not cover all costs that would be incurred
in overseeing a decentralized I/M program. As mentioned
previously, the cost of acquiring computer equipment is not
considered here . Some states may not be able to use existing
equipment. This analysis does not cover costs associated with
enforcement activities against non-complying motorists, nor are
estimates for conducting on-road testing provided. The costs of
these functions would have to be priced out and divided by the
number of subject vehicles. Table 5-2 details the per vehicle
costs of a quality assurance program consisting of those functions
analyzed here.
Table 5-2
Quality Assurance Functions and Costs in Decentralized Programs
Component	Cost per Station	Cost per Vehicle
Administrative Audits	$534.00	$0.52
Covert Audits	$3,250.00	$3.17
Referee Station	$1,845.00	$1.80
Data Staff	$205.67	$0.20
Training	$65.00	$0.06
Inter-agency Costs	$0.50
State Testing	$0.22
Total without State	$6.25
Testing
Total with State	$6.47
Testing
Per vehicle costs for most of these components are derived by
dividing the per station costs by 1,025, the average number of
vehicles tested per station in decentralized I/M programs.
Programs with lower vehicle to station ratios will incur higher
per vehicle costs. The per vehicle costs can be reduced by
limiting the number of stations to maintain a high vehicle to
station ratio.
-46-

-------
5.1.2.2 Quality Assurance in Centralized Basic I/M Programs
The same activities needed in decentralized programs are
performed to quality assure inspections in centralized programs
with some differences. Referee stations may be replaced by a
full time state referee at each facility. Auditing frequencies
are assumed to be three times a year per lane for administrative
audits and four times a year per lane for coverts (assuming one
per inspector). Data analysis costs are estimated based on the
assumption that the state's level of effort is tied to the number
of vehicles. Hence, the vehicle per station figure used for
decentralized programs is factored by the increased traffic at a
centralized lane. The number of vehicles per lane is estimated to
be 39, 000 per year, based on a peak capacity of 25 vehicle per
hour, and typical rate of 13 vehicles per hour (the derivation of
these estimates is detailed in the next section).
Table 5-3
Quality Assurance Functions and Costs in Centralized Programs
Component	Cost per Lane	Cost per Vehicle
Administrative Audits	$267.00	$0.01
Covert Audits	$4,000.00	$0.10
Data Staff	$10,942.88	$0.28
State Referee	$14,040	$0.36
Inter-agency Costs	$0.50
Total	$1.25
5 . 2 Estimated Cost of Hiah-Tech I/M Testing
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
-47-

-------
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.
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 Meeds 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
-48-

-------
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 (CO2) < and a chemiluminescent (CI)
analyzer for N0X, 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.
Table 5-4
Equipment Costs for New Tests
Test
Pressure
Purge
Transient
Equipment
Metering System
Flow Sensor
Dynamometer
Video Drivers Aid
CVS & Analyzers
TOTAL
Price
$600
$500
$45,000
$3,000
$95.000
$144,100
The figures in
expendable materials.
Table 5-4 do not include the costs of
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 based on the usage rates observed in the
EPA Indiana pilot program and typical inspection volumes as
presented later in this section. Calculations of per vehicle
presented throughout this report
equipment costs
vehicle costs of these expendables as well.
include per
-49-

-------
Table 5-5
Expendables for New Tests
Replacement Interval
Test
Material
Cost
Centralized
Pecentralized
Pressure
N2 Gas
$30
250 tests
250 tests
Transient
H2 Fuel
$60
2 months
1000 tests

HC Cal Gas
$60
2 months
1000 tests

CO Cal Gas
$60
2 months
1000 tests

CO2 Cal Gas
$60
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.
5.2.3	Cost to Upgrade Centralized Networks
5.2.3.1 Basic Assumptions
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 in Centralized I/M Programs
Program	Vehicles Tested per Hour
Arizona	20
Connecticut	25-30
Illinois	25
Maryland	25-35
Wisconsin	25-30
-50-

-------
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 I/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 wage or slightly higher, that by the
end of the first year they earn $5.50 to $6 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 wage.
Estimates of the costs of adding pressure testing, purge
testing, and transient tailpipe testing were derived by taking the
current costs 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 adding it to the current
$8.50 per vehicle cost, with a further adjustment for 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 Effect 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 same test frequency (i.e., annual or biennial)
the reduced throughput rate means that the number of lanes needed
-51-

-------
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.
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 costs, building costs, 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 is about $40,000 or 21$ per
vehicle using the basic test volume assumptions. As indicated
previously, a staff person is assumed to earn $6.00 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
48C per vehicle. Existing centralized networks typically have two
staff per lane. Thus, total staff costs work out to 96C per
vehicle. Total average construction costs are estimated at
$800,000 for a five lane station, yielding an average per vehicle
cost of 82$ . In this analysis a figure of $1.25 is used to
-52-

-------
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 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
Increments
Current
Equipment
Staff
Construction
State Retainer
Per Vehicle
Cost
$0
$0
$0
$1
21
96
82
25
Total Cost Less
Increments
$8 . 50
$8.29
$7 . 33
$6. 51
$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 maximum 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 outpoints. Hence, EPA estimates that the
-53-

-------
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 typical throughput rate of "7.5 vehicles
per 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 50C 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 BOC
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 costs 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 $5.26, 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 12.5/7.5, 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.
-54-

-------
Table 5-8
Costs to Add Proposed Tests to Centralized Programs
Increments
Adjust for Throughput
Staff
Construction
Oversight
Pressure Test
Purge Test
Transient Test
Per Vehicle Cost
$5.26 * 12.5/7.5
$2 . 40
$1. 71
$1 . 75
$0. 13
$0 . 41
$0 . 87
Running Total
Cost per Vehicle
$9.12
$11.52
$13.23
$14 . 98
$15.11
$15.52
$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
Cost to Upgrade Decentralized Programs
5.2.4.1 Basic 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 as 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 as 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.
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
-55-

-------
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
Procrram
Vehicles per Year
Vehicles oer ;
California
6,180,093
799
Colorado
1,655,897
1, 104
Dallas/Ft. Worth
1,948,333
1, 624
El 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 done 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-volume, centralized stations, the cost per
vehicle for the new tests 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
-56-

-------
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 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 NDIR-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
-57-

-------
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 Cost 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 costs are 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 are 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
are 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 costs are estimated for three scenarios, biennial
testing is assumed in all three. In the first, all stations
remain in the inspection program. In 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 50 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.
-58-

-------
Table 5-10
Costs to Conduct High-Tech Testing in Decentralized Programs
Scenario	fl.nnual 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
same 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 (15
mph at xxx load), a 40 second 2525 mode (25 mph at xxx load), 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 needed to insure proper warm-up and canister purge down.
Allowing also for necessary time 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.
-59-

-------
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 tests,
is upgraded with a NOx analyzer and an anamometer, to enable mass
concentration calculations, for this test. The equipment for the
purge and pressure test are the same as described previously. The
estimated costs are shown in Table 5-11.
Table 5-11
Equipment and Costs for the ASM Test
Pressure System	$600
Flow Sensor	$500
Dynamometer	$20,000
Anamometer	$2,000
BAR90 w/NOx Analyzer	$16,90Q
Total	$40,000
Expendables for this test are nitrogen gas for the pressure
test and calibration gases for the analyzer. The cost of nitrogen
gas is the same as in the previous analysis on IM240 costs (the
pressure test procedure is the same regardless of the type of
tailpipe test used). Current calibration gases are multi-blends
consisting of propane, CO, and C02 . A cost of $45 per bottle is
used here. In this analysis, it is assumed that multi-blend gases
that include NO will be available at the same cost.
Alternatively, one could assume that two bottles of calibration
gas, one current standard multi-blend and a bottle of NO will be
needed, however, the additional cost per test is insignificant
(less than 5C, even in a low volume situation).
5.3.2	Centralized Programs
The total test time per vehicle would be about 11 minutes,
including administrative processing in an efficiently run testing
lane. In a multi-position lane the throughput would be governed
by test time at the longest position, which would be four minutes.
This translates into a peak throughput 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.
-60-

-------
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
Costs to Add Proposed Tests to Centralized Programs
Increments
Adjust for Throughput
Staff
Construction
Oversight
Pressure Test
Purge Test
Four-mode Test
Running Total
Per Vehicle Cost	Cost per Vehicle
$5.26 * 12.5/7 .5	$9.12
$2.40	$11.52
$1.71	$13.23
$1.75	$14.98
$0.13	$15.11
$0.18	$15.29
$0.35	$15.64
5.3.3
Decentralized 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 cost for
state mass emission testing.
Table 5-13
Costs to Conduct Four-Mode 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
5.4 Repair Costs
5.4.1	HC and CO Exhaust Repair Costs and Methodology
The repair costs for HC and CO e:-:haust emission repairs are
split into two elements. One addresses the repair costs due to
failure of a tailpipe test, such as the 2500 rpm/ldle idle test or
the loaded transient test. The other element addresses the repair
costs of correcting tampering identified as a result of the visual
inspection for the presence and connection of emission control
components such as the catalyst (also known as "ATP failures")
-61-

-------
5.4.1.1 Tailpipe Emission Test. Failures
Based on current information from I/M programs which collect
repair cost information, the average cost to repair a 1981 or
newer vehicle failing the 2500 rpm/ldle test is approximately $75,
including parts and labor. For example, 1989 repair data from the
Louisville, Kentucky I/M program shows the average cost to be $54
for all model year vehicles if only commercial repairs are
included. The overall average cost drops to $42 per repaired
vehicle if the cost of self repairs (repairs performed by the
individual vehicle owner) are also included6. In addition, the $75
average repair cost figure is further supported by the findings
from an I/M repair study conducted in California which showed the
average repair cost to be $72 for 1980 and later model year
vehicles1. In this study, 500 vehicles that failed the California
I/M test were recruited, tested, and repaired at independent
commercial garages to pass I/M. Finally, a study of repair costs
conducted by the Oregon I/M program in 1985 and 1986 found the
average repair cost to be about $50 per failure.8
The average cost to repair a vehicle which fails both the
IM240 and the 2500 rpm/ldle test is also assumed to be $75. This
figure is based on the fact that these cars are likely to receive
on average the same types of repairs as are received by vehicles
failing only the 2500 rpm/ldle test. For the vehicles which fail
only the IM240 emission test, the average repair cost is assumed
to be $150, or twice as much. This higher repair cost accounts
for the additional and more thorough diagnosis needed to identify
the causes of the IM240 failures. In addition, it allows for the
possibility of more costly engine parts being required to repair
the IM240 failures. Therefore, blending the $75 cost of repairing
combined IM240 and 2500 rpm/ldle failures with the $150 cost of
repairing IM240-only failures, and assuming (based on observations
in Indiana) that there are slightly more 2500 rpm/Idle/IM240
failures than IM240-only failures, yields an average cost of $120.
5.4.1.2 Emission Control Inspection Failures
The average cost (separated by model year group) to repair
emission control components identified as needing repair or
replacement by a visual inspection are shown in Table 5-14.
These costs were estimated several years ago, based on
average retail parts and labor costs. For example, the average
air pump repair cost reflects the cost of replacing a broken air
6	"1989 Annual Report Vehicle Exhaust Testing Program Jefferson County,
Kentucky", April, 1990
7	"I/M Evaluation Program Series II", Summary from the California Air
Resources Board's I/M Evaluation Program, October 25, 1991.
8	Jasper, W. P. "A Discussion of Reported Maintenance and Repair Expenses
in an I/M Program", SAE Paper 861547
-62-

-------
pump belt or reconnecting an air or vacuum line. This cost was
based on the assumption that most air pump tampering or
malmaintenance will focus on disabling the unit by disconnecting
the belt or line rather than removing the entire unit. If this is
the case, then the repairs will be relatively simple. The average
catalyst replacement cost was based on the retail cost of an
aftermarket converter. The misfueled catalyst replacement
reflected the cost of the converter plus an additional amount to
replace the poisoned oxygen sensor. The evaporative system repair
is the average cost of reconnecting a vapor or vacuum line after a
visual inspection of the system. The PCV and gas cap repairs are
the average cost of replacing these components.
Table 5-14
Average Cost of Repairing Emission Control Components
Component
Pre-81
1981 +
Air Pump
$15
$15
Catalyst Replacement
$150
$165
Misfueled Catalyst Replace
$175
$190
Evaporative System Repair
$5
$5
PCV System Repair
$5
$5
Gas Cap Replacement
$5
$5
Repair of intentional tampering failures will contribute
relatively little to the overall cost of repairing I/M-failed
vehicles in the 1990s, due to decreasing tampering rates. The
estimated costs per vehicle, therefore, were not revisited.
5.4.2	N0X Repair Costs and Methodology
Repair costs for N0X reduction, and the supporting analysis
are discussed separately from the HC and CO repair cost analysis
because repairs targeted to reduce HC and CO emissions often have
no effect on N0X emissions. Moreover, the Indiana data showed that
the HC/CO failures and the N0X failures were essentially separate
sets of vehicles9. For example, many vehicles requiring repairs to
correct high HC or CO emissions frequently have fairly low N0X
emissions, and consequently do not require NOx repairs.
Furthermore, for those vehicles which are high N0X emitters, the
most common repair is to the EGR system, and this often has little
impact on HC or CO emissions. In other words, the vehicles with
excessive HC and CO emissions usually need different types of
repairs than those with excessive N0X emissions. Thus, their
repair costs were analyzed separately.
9 November 1991, EPA memorandum from E. Glover to C. Harvey, "Average
Repair Costs and Benefits from Repairing High N0X Hlmitters.n
-63-

-------
The data used to calculate the average cost and benefit of
performing vehicle N0X repairs was collected in the on-going EPA
Emission Factor test program at the EPA's National Vehicle and
Fuels Emissions Lab (NVFEL) in Ann Arbor, Michigan, as well as at
the ATL facility in South Bend, Indiana. In this program, large
numbers of in-use vehicles were recruited for testing and repair
to better characterize the emissions of the fleet. However, for
the analysis of N0X costs, the overall database was restricted to
1983 and later model year vehicles which had received an FTP test
both before and after repair, and had been tested in the last few
years. As a result, data from 169 1983+ model year vehicles with
repair data were obtained.
Most of the 169 vehicles were high emitters of HC or CO, and
had repairs aimed at those pollutants, since EPA had given the
testing contractor instructions to focus on HC and CO emissions.
In order to more accurately characterize the cost of effective N0X
repairs, criteria were used to further select vehicles which
clearly had high N0X emissions before repair, but had achieved
lower NOx emissions as the result of the repair. These criteria
were: Before repair FTP emissions had to exceed 2.0 gpm NOx, and
after repair FTP emissions could not exceed 1.25 gpm. As a result
of these criteria, 10 cars out of 169 were selected, and 9. were
used in the final cost analysis. Examining the individual vehicle
repairs of these 9 vehicles (see Table 5-15) shows that all of
them' needed EGR repairs to lower the NOx emissions to levels which
could meet the criteria. On 6 of these vehicles, the EGR was
replaced, while on the other 3 the EGR passage was cleaned, or the
delay valve was replaced.
The tenth car (683), a Chevrolet Chevette, was removed from
the cost analysis because the repair it received was not targeted
toward NOx reduction. Instead, N0X emissions decreased primarily
due to an ineffective HC/CO repair, which caused the engine to go
to a rich air/fuel mixture as evidenced by a very large CO
emission increase (10 to 30 gpm).
The repair costs of the 9 individual vehicles as well as the
overall averages are shown in Table 5-15. For example, the price
of the repair parts averaged $44, using Mitchell's Summer
Collision Estimating Guide. The labor cost averaged $34, based on
0.68 hours at $50 per hour. These labor hours were determined
using Mitchell's 1991 Mechanic's Labor Estimating Guide. In
addition, each car was assumed to require 0.5 hours of diagnostic
time at the labor rate of $50/hour for an average cost of $25.
Summing these costs puts the total average cost of an effective N0X
repair at $103. For input into subsequent cost-effectiveness
models this overall cost was rounded to $100.
-64-

-------
Table 5-15
NOjr Repair Costs







Parts
Diag-





NOx Repair
Labor
Labor
Cost
nostic
Total
Veh
MX
Make
Model
Pescciptien
Hours
Cost
Retail
Cost
Cost
861
87
MERC
COUGAR
Replaced EGR Valve
0.80
$40
$42
$25
$107
94
86
FORD
THtJNDERBIRD
Install EGR Vacuum
Line
0. 30
$15
$0
$25
$40
803
86
CHRY
NEW YORKER
Replaced EGR Valve
Clean EGR Passage
0.80
$40
$41
$25
$106
1095
89
PONT
GRAND PRIX
Replaced EGR Valvo
Assembly
0. 80
$40
$161
$25
$226
23
87
FORD
TAOROS
Clean EGR Passage
1.00
$50
$0
$25
$75
545
84
CADI
SEVILLE
Clean EGR Passage
0.70
$35
$0
$25
$60
1131
86
DODG
W150
Replaced EGR Delay
Valve
0.30
$15
$22
$25
$62
1657
83
CHEV
CELEBRITY
Replaced EGR Valve
Clean EGR Passage
0.70
$35
$65
$25
$125
41
85
CHEV
S-10
Replaced EGR Valve
0.70
$35
$67
$25
$127
683
85
CHEV
CHEVETTE
02 Sensor, Coolant
Temperature Sensor
Rebuilt Carburetor
4 . 60
$230
$39
$25
$294
AVERAGE
with #683

1.07
$53
$43
$25
$122
AVERAGE
without
#683

0. 68
$34
$44
$25
$103
5.4.3	Evaporative System Repair Costs and Methodology
The repair and cost data used to calculate the average
evaporative system repair costs and subsequent fuel economy
improvements were collected during an EPA running loss test
program conducted at ATL during the Spring of 1991 in which
failing vehicles were repaired and retested. All comparisons were
done with data obtained from running loss tests at 95 F using a
9.0 RVP emission test fuel, and 3 consecutive LA4 test cycles (the
first LA4 being a cold start).
The cost-benefit calculation was based upon a sample of 25
vehicles which failed either the I/M purge or pressure test in
this test program, and for which evaporative system repair cost
information was available.10 Only 24 vehicles (vehicle 1667 was
not available) were used to calculate the average fuel economy
cost savings resulting from evaporative system repair. The
results are shown in Table 5-16.
10 July 26, 1991, EPA memorandum from E, Glover to C. Harvey, "Average
Repair Coats and Benefits from Repairing Purge and Pressure Failures."
-65-

-------
Table 5-16
Average Repair Costs and Fuel Economy Benefits
Total
Fuel
Test
Pressure
Purge
Total Fuel	Economy	Average	Average	Average
Economy	Savings Parts	Labor	Total
Improvement	aal/mi Cost	Hour	Cost*
7.87 gpm 6.1%	0.0034 $15.03	0.45	$37.76
8.26 gpm 5.7%	0.0035 $21.89	0.96	$70.10
* Labor costs are computed from labor time using a labor rate (including
California) of $50 per hour
The evaporative repair costs, excluding gas caps, are based
on parts costs as invoiced by ATL. If the cost of a repair part
for a particular vehicle was not available, then the average cost
from the other vehicles which also received that repair was used.
For example, in the analysis, the value of $29.46, obtained from
vehicle (1563) was used as an estimate of purge solenoid
replacement cost on two other vehicles (1525 and 1552) which
received that repair, but did not have invoiced repair costs. The
ATL invoiced gas cap replacement cost was available on only two
vehicles (1532 and 1542). For the other vehicles which required
this repair, the gas cap cost was based on auto dealer retail
prices for an OEM part . Typically, the gas cap OEM retail price
was around $7. In addition, repair parts such as evaporative
hoses, or inexpensive in-line tees were assumed to cost nothing,
except as overhead in the labor cost of fixing them.
The time of repair is generally based on individual
diagnostic and repair durations provided by ATL. Typically, they
include both the time to diagnose the problem and replace or
reattach the parts. For example, vehicle 1548 required 6 hours of
diagnosis to discover the cause of the purge problem and replace
the defective part. Most of the time was spent in diagnosis,
though this length was unusual since most diagnoses and repairs
were completed in a half an hour or less.
In some cases, actual labor times were not available to
diagnose or replace a particular part. In these cases estimates
were made regarding the duration of a typical repair. For
example, gas cap replacement (including diagnosis) duration was
not usually itemized and, therefore, was estimated to be 15
minutes. In other cases, repair times from similar repairs on
other cars were used. However, for the sake of clarity, both the
parts and labor cost basis of each vehicle's repair are noted in
Table 2 of reference 8.
-66-

-------
5 . 5 Fuel Economy Benefits
5.5.1	Fuel Economy Benefits of Evaporative System Repairs
The analysis of the data shows a substantial fuel economy
benefit under the 95 F test conditions as the result of
evaporative system repair. This fuel economy benefit is
attributed to two factors. The first is increased performance and
efficiency of the vehicle's engine following, an evaporative repair
such as reconnection of a vacuum line or a TVS repair. This
increase in efficiency was directly measured by the CVS equipment,
and it was found that the measured fuel economy increased by an
average 3.2% for vehicles failing the pressure test, and an
average 2.8% for vehicles failing the purge test. The fuel
economy improvement is calculated by dividing the fuel consumption
reduction by the total fuel consumption, as illustrated in the
following calculations:
4.13 grams fuel/mile/128.3 grama fuel/mile = 3.2% Pressure failures
4.05 grams fuel/mile/143.75 grams fuel/mile = 2.8% Purge failures
The second factor involved in the fuel economy benefit
calculation is the utilization of the captured HC vapor which
would have otherwise been lost as running loss emissions. In a
properly designed closed-loop vehicle the engine should
effectively substitute these vapors for liquid tank fuel, and
reduce the vehicle's real fuel consumption. These vapor fuel
flows from the engine and the evaporative canister are not
measured during the running loss test.
Since actual fuel flow data were not measured, it was assumed
that 100% of the captured running loss emissions (i.e., the
difference between before and after repair levels) can be
effectively utilized as fuel. This assumption may be slightly
high given the fact that on average exhaust CO emissions increased
somewhat as the result of evaporative repairs, indicating that
some of the extra fuel was not fully combusted. However, such an
error (i.e., using an ' R.' factor of 1.0) is probably small, and
its effect should not be large considering that the running loss
reductions are not large in comparison to total vehicle fuel
consumption.
The running loss vapors from pressure failures were converted
to liquid fuel, using an R Factor of 1.0, the standard density of
Emission Test Fuel, and a carbon weight factor of 0.83 for the
fuel.
3.74 gpm running loss CH2.33 * R Factor = 3.74 gpm liquid fuel (CH1.85)
3.74 g C/mi * (lcm^/ 0.745 g Fuel) * (1 g Fuel / 0.83 g C) *
(1.0 liter/ 1000 cm3) * (1.0 gal/3.79 liter) = 0.0016 gal/mi
-67-

-------
The analogous running loss vapors from purge failures were
converted to liquid fuel, and the percentage fuel economy
improvement was calculated in a similar manner.
4.21 gpm running loss CH2.33 * R Factor = 4.21 gpm liquid fuel (CHI.85)
4.21 g C/mi * (1 cm3/ 0.745 g Fuel) * (1 g Fuel / 0.83 g C) *
(1.0 liter / 1000 cm3) * (1.0 gal/3.79 liter) = 0.0018 gal/mi
The measured fuel economy improvements from better engine
operation were combined with the measured running loss reductions
to produce evaporative repair fuel economy benefits of 6.1% for
pressure failures and 5.7% for purge failures. Averaging these
together produced an overall fuel economy benefit from evaporative
repair of 5.9%.
5.5.2	Fuel Economy Benefits of IM240 Repairs
The fuel economy benefit for repairing a vehicle that has
been identified as failing the 0.8 gpm HC cutpoint or the 15 gpm
CO cutpoint on the IM240 test has been estimated as an increase of
12.6% in overall fuel economy, after repairs. This compares to an
8.0% fuel economy benefit realized by identifying and repairing
vehicles using the 2500 rpm/Idle test as a yardstick. These
percentages are derived from data gathered from the IM240 test
site in Hammond, Indiana, and are based upon an average difference
in fuel economy before and after repairs.
The 12.6% fuel economy benefit assessed for identifying and
repairing vehicles on the basis of the IM240 test lane results is
based upon two groups of 1983 and newer vehicles recruited at the
Hammond test site. The first group included those vehicles that
failed the emissions cutpoints of 0.8 gpm for HC and/or 15.0 gpm
for CO, which were subsequently FTP-tested, repaired and retested
at the ATL facility in Indiana (a total of 42 vehicles) . The
second group consisted of those vehicles that failed the emissions
cutpoints, were FTP-tested, but were not repaired (a total of 10
vehicles). Unrepaired vehicles were assumed to represent a fuel
economy benefit of zero, with the net effect that the overall fuel
economy benefit calculation is conservative.
The 10 IM240-failed vehicles mentioned above were not
repaired and retested because the original design of the testing
program sought to conserve testing slots by applying a criteria
that only vehicles with an FTP result twice the certification
standards for the vehicle would receive repairs and be retested.
These unrepaired vehicles were included in the analysis to
represent that fraction of vehicles (i.e., 19%, or 10 out of 52)
expected to fail the IM240 .(in a future I/M program) but which
have only a marginal emissions problem and presumably only a
marginal fuel economy loss (if any), thus requiring only minimal
repairs which will not result in improved fuel economy. The
averaged fuel economy benefit represents a harmonic average of the
-68-

-------
FTP fuel economy before and after repairs for the 52 vehicle
sample group.
Table 5-17 shows that 17 vehicles which failed at the Hammond
lane were not repaired, which raises the issue of why only 10
vehicles were used to represent the "no improvement" vehicles.
The logic and assumptions were as follows. Of the 72 vehicles
that were repaired, only 42 (58.3%) had all the necessary data to
do the calculations. Assuming the same attrition rate (due to
incomplete data) for the vehicles that did not receive repairs
(i.e., 58.3% of the 17 unrepaired vehicles) yields a total of 10
vehicles. The net effect of assuming this fraction of "zero
improvement vehicles" is a lower fuel economy benefit for the
IM240 (12.6% instead of a potential 15.7%) .
Table 5-17
Zero Improvement Vehicle Sample Size Adjustments
Original #	Remaining
of Veha Description of Data Used and Removed	Vehicles
98	1983+ Failed lane 0.8 & 15 & received FTPs	98
17	were less than twice standard & not repaired	81
9	were greater than twice the standard, but were not	72
repaired due to test schedule or coat (engine
rebuild or catalyst)
4	had no as-received IM240s at ATL (IM240-based fuel	68
economy benefits were initially evaluated, so this
test was required. In retrospect, they should have
been added back into the database for the FTP-based
FE improvement)
1	had no after-repair IM240 #1643 (to verify repair	67
success)
25	Failed after-repair IM240 (incomplete ATL repairs)	42
% of 72 repaired that can be included in analysis =	58.3%
58% of 17 <2 x standard & not repaired included as
zero improvement =	10
5.5.3	Fuel Economy Benefit for the 2500 rpm/Idle Test
The 8.0% fuel economy benefit assessed for identifying and
repairing vehicles on the basis of the 2500 rpm/Idle test is based
upon two groups of 1983 and newer vehicles recruited at the
Hammond test site. The first group consisted of 6 vehicles that
failed the 220 ppm HC and/or 1.2% CO cutpoints on their initial
2500 rpm/Idle I/M test, received an IM240 before and after
commercial repairs, and received passing scores on the retest.
The second group consisted of those vehicles that returned to the
Hammond lane after commercial repairs, but again failed the 2500
rpm/Idle test. These latter 6 retest failures are considered to
be the result of incomplete repairs, which would be corrected in
an enhanced I/M program.
-69-

-------
The before and after IM240 fuel economy data was "corrected"
to reflect FTP fuel economy by employing a correction factor of
1.0925, reflecting the fact that, oh average, FTP fuel economy
varies from IM240 measured fuel economy by 9.25%. This variance
reflects the fact that the IM240 and FTP are, after all, different
tests, using different driving cycles, etc. Still, the two tests
show a high degree of correlation, and, in the area of fuel
economy, the variance between the two tests is a relatively
constant difference of 9.25%. Therefore, multiplying IM240 fuel
economy readings by 1.0925 yields a reliable estimate of FTP fuel
economy.
After successful repairs (i.e., those resulting in a passing
retest) , some marginal vehicles will fail to realize a noticeable
fuel economy improvement. Using a database of 48 cars, it was
determined that 4 of the vehicles that failed the 2500 rpm/Idle
I/M test were not repaired because their FTP emissions scores were
less than twice their certification standards, leading to the
conclusion that, had these vehicles been repaired, their fuel
economy benefit would be zero. These 4 vehicles represent 8.3% of
the 48 database vehicles for which all the necessary data was
available. Assuming that 8.3% of the 6 vehicles that were still
failing I/M would not get a fuel economy benefit after repairs
yields a figure of 0.48 vehicles that will show no noticeable fuel
economy benefit. Given that half of a vehicle cannot be added to
the database, each of the other 6 vehicles that did pass after
repairs were duplicated yielding twelve vehicles, and 1 vehicle
was added to represent the "no fuel economy improvement" case.
Adding the single "zero improvement vehicle" lowered the fuel
economy benefit of the 2500 rpm/Idle test from 8.6% to 8.0%.
Table 5-18 further details how these numbers were arrived at.
Table 5-18
Ad-iusted Zero FE Benefit Vehicle Sample Size
Original # of

Remaining
Vehicles
DeacriDtion of Data Used and Rsmoved
Vehicles
312
1983+ Failed IN I/M at lane
312
256
not recruited to lab
56
8
missing data
48
44
dirty enough to expect an FE benefit
4

% that failed IN I/M but too clean for a FE
8.3% .

benefit (4 of 48)


8% of 6 commercially repaired included as
0.48

zero improvement

While 6 vehicles may seem like a slim database, we did not
want to assume too low a fuel economy benefit for the conventional
2500 rpm/Idle test and risk overestimating the incremental benefit
of the IM240 test. A mid-1980s study with actual or simulated
commercial repairs of older technology 1981-83 vehicles showed
only a 3.5% improvement. This has not been shown to be applicable
-70-

-------
to newer technology vehicles. We also did not want to claim too
much benefit. We did not rely on the ATL-performed repairs (as we
did for the fuel economy benefit when using IM240 cutpoints)
because the ATL mechanics were instructed to repair all known
malfunctions that would likely affect FTP HC and CO. Therefore,
the emissions and fuel economy benefits would likely exceed what
would actually occur with real world repairs that stop as soon as
the 2500 rpm/Idle test cutpoints are met. In contrast, we judged
that because the IM240 is a mass emissions test that correlates
well with the FTP, real world repairs aimed at making vehicles
pass the fairly, stringent IM240 cutpoints would not be so
different from those made by the ATL mechanics. The fact that 25
of the 67 ATL-repaired vehicles still failed the IM240 suggests
that ATL mechanics in general did not go too far.
5 . 6 Recurring Failure and Repair	Ratesand Fraction	af	Fleet
Affected by Fuel Economy Benefits
The rates at which vehicles recurrently fail tailpipe tests
and emission control inspections in an ongoing I/M program (i.e.,
the percentage of failing vehicles in a program that has been
established for a few years) are used within the Cost
Effectiveness Model (CEM) for determining repair costs. Fuel
economy credits for repairs resulting from tailpipe tests are
based on the hypothetical failure rates that would occur in the-
first cycle of the I/M program if it were just starting. These
hypothetical rates in effect represent vehicles that have been and
remain affected by the I/M program that has in fact been
operating.
The exhaust- test failure rates for calculation of repair
costs in CEM are in the form of a zero-mile failure rate and a
deterioration rate, such that the fraction of failing vehicles for
a given test type is calculated by multiplying the deterioration
rate by the average mileage and adding that result to the zero-
mile failure rate. Table 5-19 shows the zero-mile and
deterioration rates found in the BLOCK DATA section of CEM.
Table 5-19
Exhaust Test Failure Rates
(fraction)
Test
Zero-Mile
Deterioration
Type
Idle
2-Speed
Loaded
IM240
NOx
0.0252
0 . 00
0.032936
0 . 00
0 . 00
(per 10K miles)
0.01
0.01
0.01190
0.0373
0.0084805
(recurring)
(recurring)
(recurring)
(first-cycle)
(recurring)
-71-

-------
These numbers are based on regressions of emission test data
from the IM240 lane in Indiana. In 1990 and 1991, Indiana had
just revitalized its moribund I/M program and hence can be
considered to represent a hypothetical I/M program in its first
cycle of inspections) . For the IM240 the first-cycle HC/CO
failure rate per 10,000 miles was 0.0373 at an average of 50,000
miles observed among 3, 436 model year 1983 and newer cars in
Indiana. The above recurring rates include adjustment of the
first-cycle rates by a factor of 1/1.87 (e.g., 0.01 =
0.0187/1.87) . This adjustment factor is the recurring initial
failure ratio for idle testing, derived by comparing the Indiana
failure rates with failure rates from other operating I/M programs
with longer histories.
The recurring zero-mile rate used by the model for the IM240
is half of the first-cycle deterioration rate (0.0373/2 =
0.01995). The recurring deterioration rate used by the model for
the IM240 is half of 1/1.87 times the first-cycle failure rate.
This method represents a 50-50 compromise between the following
two assumptions, either of which would be reasonably plausible:
(.a) The IM240 test will require vehicle repairs sufficient to
return the emission control systems to like-new condition thus
yielding a constant failure rate equal to the rate found for the
first 10,000 miles of operation (0.0373), and (b) IM240 repairs
will deteriorate similarly to idle and 2-speed test repairs, which
would yield a deterioration rate of 0.0373/1.87 = 0.01995).
These failure rates assume cutpoints of 1.2% CO and 220 ppm
HC for the idle and 2-speed tests, and 0.8/15 gpm for the IM240
test. For NOx, separate cutpoints of 1.69 for PFI, 2.50 for TBI,
and 3.99 gpm for carbureted vehicles are used resulting in .an
overall nominal failure rate of about 10% on the IM240.
In the case of ATP emission control component inspections CEM
calculates recurring repair rates for the first year a vehicle is
inspected from the difference in tampering rates given by
MOBILE4.1 for the no-program case and the with-ATP case. There is
also a small residual repair rate assumed for latter years, with a
very minor cost impact.
In the case of purge and pressure test failures MOBILE4.1
uses a lookup table which has different malfunction rates for each
vehicle age up to 13 years, and older vehicles are assigned the
rates of the 13 year old vehicles. The malfunction rates range
from roughly 4% to 33% for purge or pressure malfunctions, and 8%
to 50% for the combination of purge and pressure malfunctions.
This lookup table can be found as the EFFECT array at the
beginning of the FAIL function in CEM. (NOTE: The CEM program
listing in can be found in Appendix A of the draft version of this
report). After appropriately weighting together these purge and
pressure failure rates, MOBILE4.1 uses them in its calculation of
evaporative and running loss emission factors in the absence of an
evaporative I/M program. These malfunction rates would become the
-72-

-------
first-cycle failure rates for a new I/M program rather than
recurring failure rates.
CEM assumes these same initial failure rates in determining
fuel economy benefits of purge and pressure tests, since the fuel
economy effect of an I/M program in a given year depends on the
difference between the number of failures that would exist in a
no-program case and the near-zero number present with the I/M
program in operation. The fuel economy benefit calculation using
these failure rates is described in Section 5.5.
To determine purge and pressure repair costs, CEM requires
recurring failure rates corresponding to an ongoing I/M program
wherein the failure rate would be lower than the initial failure
rate observed in Indiana's first cycle and used in M0BILE4.1 and
in the fuel economy benefit calculation to represent the no-
program case. The recurring purge and pressure failure rates used
for this purpose are:
Recurring Purge test failure rate: 3.0%
Recurring Pressure test failure rate:	2.5%
Recurring total Purge/Pressure failure rate: 5.0%
The exact use of these rates can be seen in the FAIL function of
the CEM program listing (see previous note).
These recurring purge and pressure test failure, rates were
derived from the initial rates of M0BILE4.1. As an example, the
5% total failure rate is. based on roughly a 50% failure rate for
ten year old vehicles indicating that roughly 5% went bad each
year on average. For an analysis that did not treat age
explicitly this was an assumption that could be used for all ages,
and would definitely not underestimate costs, since much of the
rise to the 50% failure rate happens at higher mileages when there
are fewer cars still in use.
5.7 Method for- Estimating Cost Effectiveness of I/M Programs
The cost of an I/M program is determined by summing the
estimated inspection fee costs, the estimated repair costs, and
the negative cost of estimated fuel economy benefits (gallons of
fuel saved * $/gallon). The emission benefits of an I/M program
are determined by subtracting the estimated emissions with the
program from the emissions with no I/M program. CEM does the
emissions calculation by making multiple runs of MOBILE4.1 and
manipulating the results of the various runs. Since M0BILE4.1
does not include the necessary cost components, CEM itself
calculates costs by combining the previously discussed information
on per vehicle costs and fuel economy benefits with the estimates
of failure rates.
-73-

-------
Since M0BILE4.1 calculates the emission levels, tampering
rates, and misfueling rates for January 1st of each calendar year,
CEM performs two two consecutive sets of ' MOBILE4 . 1 runs and
interpolates between them to get an annual average emission rate
which is then converted into a ton per year value using the fleet
vehicle miles travelled (VMT) data contained in MOBILES.1. In
order to separate out costs and benefits associated with various
portions of an I/M -program, two intermediate M0BILE4.1 runs are
done between the full program and no-program runs. Therefore,
each CEM run performs a total of eight M0BILE4.1 runs as follows.
1)	Full I/M & ATP program {as requested)
2)	Run 1 minus any ATP and evap testing
3)	Run 2 minus any tailpipe I/M, but with tampering
deterrence effect of I/M
4)	Baseline, no program benefits at all)
5)	Run 1 for next calendar year
6)	Run 2 for next calendar year
7)	Run 3 for next calendar year
8)	Run 4 for next calendar year
5.1.1 Inspection Costs
Inspection costs are determined by multiplying user-input
inspection costs by the number of vehicles adjusted for compliance
rate (percentage of vehicles that fail to get inspected) .
Separate costs are input for tailpipe emission tests, emission
control checks, purge test, and pressure test. If a program tails
for biennial rather than annual inspections, the inspection costs
per year are divided in half. All default costs are found in the
SETUP routine of the CEM program listing. Default inspection
costs are shown in Table 5-20. Note that the cost of performing
the purge test overlaps many of the costs associated with
transient testing, including the cost of a dynamometer, video
driver's aid (VDA), and the throughput adjustment associated with
the longer test time. If purge testing is assumed, the
incremental cost of including the transient test is relatively
minor, including the cost of a constant volume sampler (CVS) and
the analyzers necessary to perform mass emissions testing.
Table 5-20
Default Inspection Costs in CEM4.1
Test
Steady-state Tailpipe Test
Emission Control Checks
Pressure Test
Purge Test
Transient Emission Test
Cost
$10
25C-1.75
69C
$6 . 53
67
Comments
$12 if biennial
Depends on checks done
Includes dyno, adjusted thruput
Increment over purge cost
-74-

-------
5.7.2	Repair Costs
Calculating total repair costs is performed similarly to the
inspection costs, except that the costs are only applied to the
percentage of vehicles estimated to fail a given I/M test. It is
further adjusted for the percentage of vehicles that do not get
repaired because they require repairs costing more than the
applicable cost waiver limit. Default repair costs are as
follows.
Table
5-21-

Default Repair
Cost i n CEM4.1

Failure Triggerincr Repair
Pre-81
fil
Idle or 2500 rpm/Idle Test
$50
$75
Transient Test (IM240)
N/A
$150
Air Pump
$15
$15
Catalyst
$150
$165
Misfueled Catalyst Cost
$175
$190
Evaporative System
$5
$5
PCV System
$5
$5
Gas Cap
$5
$5
Purge Test
$70
$70
Pressure Test
$38
$38
NOx
Not
$100

Estimated

In the case of transient exhaust testing, the fraction
failing vehicles that would have failed a 2500 rpm/Idle test
assigned the repair cost for the 2500 rpm/Idle test, while
remainder is assigned the higher transient test repair cost.
5.7.3	Fuel Economy Cost Benefits
Fuel economy benefits are based on cumulative repairs made to
vehicles that fail an I/M tailpipe test and/or an evaporative
system pressure test. As described in Section 5.5, the repair
rate used is the first-cycle failure rate corresponding to
inspection of vehicles that have not previously been subject to an
I/M program. The percentage improvement in fuel economy depends
on the type of test that was failed. The following benefits are
from the BLOCK DATA section of the CEM program listing.
of
is
the
-75-

-------
Table 5-22
Fuel Economy Benefits in CEM4.1
Test	FE Benefit
2500 rpm/ldle (pre-81)	0.0%
2500 rpm/ldle (81+)	8.0%
IM240 (83+)	12.6%
Purge/Pressure	5.9%
The model converts these percent MPG benefits into dollar benefits
using the VMT information from MOBILE4.1, fleet average fuel
economies for appropriate model years from CEM and a user-input
gasoline cost from CEM, which defaults to $1-. 25 per gallon.
-76-

-------
6.0	REGULATORY IMPACT ANALYSTS - COSTS AND BENEFITS OF ENHANCED
IlU
6.1	Emission Reduction Benefits
Gram per mile emission factors were calculated using
M0BILE4.1 for the high-tech enhanced program. The design elements
and proposed performance standard inputs are detailed below in
Table 6-1. These inputs include annual, centralized testing of
1968 and later light-duty vehicles and light-duty trucks, as
required by section 182 (c) (3) (B) of the Clean Air Act Amendments
of 1990. Other inputs reflect national default values assumed in
MOBILE4.1. It should be noted that these inputs are substantially
similar to those that appeared in the draft version of this'
document, with the exception of assumed waiver and compliance
rates, which have been loosened to reflect more realistically
achievable levels. Nevertheless, the emission reductions
projected for the enhanced I/M performance standard are within a
percentage point of those previously reported.
The gram per mile emission factors for various I/M scenarios
and the emission reduction benefit as a percentage of the no-I/M
case in the calendar year 2000 are shown in Table 6-2. The no-I/.M
factors were calculated assuming the same RVP, ambient
temperatures, maximum and minimum temperatures, operating modes,
altitude, vehicle speeds, and VMT mix variables as assumed for the
I/M scenarios. Stage II and on-board vapor recovery system
effects were not modeled in either the I/M or no-I/M cases.
Emission benefits from basic I/M (the current performance
standard) and from the biennial high-tech program (which EPA
recommends) are also shown. Note that the proposed enhanced I/M
performance standard listed below in Table 6-2 is an annual
program, as required by the Act. Note further that emission
reductions are expressed as a percentage of total highway mobile
source emissions. Many other mobile source programs are described
based on light-duty vehicles; doing so here would show a much
higher percent benefit.
The results shown in Table 6-2 are our best estimates at this
time, but our test programs and data analyses are continuing and
we anticipate refining the numbers as time goes on.
-77-

-------
Table 6-1
MOBILE4.1 Inputs for the High-Tech Enhanced Model Program
Elag
(Standard Inputs)
Pre-1981 Stringency
Idle
2500 rpm/Idle
Pressure
Purge
Transient
tWaiver Rate
tCompliance Rate
*Network Type
*Test Frequency
*Vehicle Coverage
ATP MY coverage
Catalyst
Fuel Inlet
Air Pump
Tailpipe Lead Test
Evap Disablement
PCV Disablement
Gas Cap
(Local Inputs)
Altitude
Period 1 RVP
Period 2 RVP
Period 2 Start Year
Minimum Temperature
Maximum Temperature
Ambient Temperature
Operating Mode
Onboard Controls
Stage II Control
Vehicle Speeds
VMT Mix
Input
20%
1968-1980
1981-1985
1983+
1986+
1986+
3%
96%
central
annual
LDV/ LDT1/LDT2
1984 +
Yes
Yes
No
No
No
No
No
500 feet
11. 5
8 . 7
1992
7 2F
92F
87 . 5F
20.6/27.3/20.6
no
no
19.6 mph
MOB4.1 default
t	These percentages may not be realistic for some programs, in which case
the program will have to be "over designed" to make up the performance loss.
*	Clean Air Act Amendments require these inputs as elements of the
performance standard.
Table 6-2
Benefits of I/M Programs Options*
Scenario
Base - No I/M
Basic I/M
Biennial High-Tech
Program
Proposed Enhanced
Performance Standard
VOC Emission Effects
Emission
Factor
[gpm)
2.084
1 . 971
1. 495
1.503
Percent
5. 4%
28.3%
27.9%
CO Emission Effects
Emission
Factor
(gpm)
11.874
10.021
8 .223
8.230
Percent
Reduction
15 . 6%
30 . 7%
30.7%
Total Highway Mobile Source Emissions in 2000
-78-

-------
6.2 Cost Effectiveness Estimates
6.2.1
Assumptions and Inputs
EPA' s estimates of the cost-effectiveness of I/M scenarios
are based upon modeling with M0BILE4.1 and CEM4.1 with assessments
done for calendar year 2000. These are compared with a modeling
scenario in which no I/M program is assumed.
The assumed cost for an I/M inspection, including a visual
check of emission control devices, is $8.50. The incremental cost
of adding the evaporative system pressure test is $1.94. The
incremental costs of adding the purge and transient tests are
$5.19 and $0.87, respectively. As indicated in section 5.6.1, the
cost of the purge test includes the cost of a dynamometer and VDA,
and also reflects a throughput adjustment to accommodate the
longer test; adding transient testing to the purge test requires
the addition of a CVS and the necessary emissions analyzers. In
addition, gasoline is assumed to cost $1.25 per gallon. The
average repair costs shown in Table 5-17 were assumed. It should
be further noted that the incremental costs of adding purge and
transient testing to a decentralized network ($12.40 and $24.97,
respectively) are larger than in a centralized network because of
the assumption these addi-tional costs will be spread out over a
smaller test volume (i.e., it is assumed that the average number
of vehicles tested per station in a decentralized network will not
change).
6.2.3	Cost-Effectiveness Calculations
Total annual program costs per million vehicles, as
calculated by CEM4.1, are presented in Table 6-3, including
inspection costs, repair cost and fuel economy benefits, shown on
an annual basis. Note that the total cost (on a per million
vehicle basis) of a biennial enhanced program is less than either
the annual enhanced program or the basic I/M program. These
results make it clear that biennial testing should be a top
priority.
The next step is to calculate cost-effectiveness ratios,, or
the annual cost per ton of emission reductions. For areas that
are required to do enhanced I/M due to ozone nonattainment (the
majority of enhanced I/M areas), the ratios could be calculated by
dividing the annual program costs, from Table 6-3, and dividing
Table 6-3
Total Annual Program Cost
Scenario
Basic I/M
Annual Enhanced
Biennial Enhanced
Cost
$6,412,000
$11,390,000
$5, 429, 000
-79-

-------
them by the annual tons of hydrocarbon reductions. The results
are shown' in Table 6-4. Unlike the total costs in Table 6-3, the
cost per ton decreases with program stringency. This is because a
major part of the cost is the inspection and the small marginal
cost of doing a more effective test is overwhelmed by the large
marginal benefit. This is a critical factor to keep in mind when
choosing among various different ozone control strategies.
Since the I/M program yields CO benefits as well as VOC
benefits and some areas need reductions in both, it makes sense to
split the cost among pollutants. High-tech I/M can also obtain
significant N0X benefits and many ozone areas may need NOx control
as well to bring ozone levels into compliance with EPA standards.
To estimate the cost of only the VOC portion of the I/M benefit,
one can assess what the cost would have been to obtain the CO and
N0X reductions by other strategies. If all the program costs were
allocated to N0X reductions (which only occur in the high option
program), then the cost per ton for the annual enhanced, high-tech
I/M program would be $6,298 per ton and for the biennial high-tech
program $3,267 per ton of N0X benefit. Alternative costs for N0X
reductions are estimated using cost per ton figures to obtain
stationary source N0X reductions through the use of more efficient
burners, estimated at $300 per ton. Allocating all of the program
costs to CO yields a cost per ton of about $143 for the biennial
high-tech program. Costs for other control programs range from
roughly $100-225 (without fuel economy benefits) for cold
temperature CO standards. Oxygenated fuels programs range from
about $200-400 per ton. A conservative, alternative cost per ton
figure of $125 was chosen for this analysis. These alternative
cost per ton figures are then multiplied by the annual ton
reductions attributable to the various program scenarios. Other
assumptions about the cost of alternate CO or NOx programs would
change the cost remaining to allocate to VOC. Higher costs would
leave less to assign to VOC and vice-versa.
Since CO reductions are not needed in all areas, and only
about 44% of the vehicles that will be subject to enhanced I/M are
in CO areas, costs are not assigned in all areas. This is done by
reducing the tons of emission reduction to 44% of full benefit and
using that result to calculate the alternative cost per ton.
Table 6-4
Cost per Ton Allocating All Costs to VOC
Scenario
Basic I/M
Annual Enhanced I/M
Biennial Enhanced I/M
Costs per Ton
$5,410
$1,694
$879
-80-

-------
The results are shown in Table 6-5. As expected the costs
are lower in all cases, and the biennial high-tech program is
about $461 per ton.
Table 6-5
VOC Cost per Ton Accounting for NO^ and CO Benefit
Scenario
Basic I/M
Annual Enhanced I/M
Biennial Enhanced I/M
C.nst Per Ton
$4,518
$1,271
$461
6.2.4
National Cost of Choosing Less Stringent I/M
The Clean Air Act requires nonattainment areas to meet
specific milestones of 15% reduction in VOC emissions by 1996 and
a 3% reduction per year thereafter. There are two ways for states
to achieve these goals: impose additional controls on stationary
sources (i.e., those beyond RACT requirements) or additional
controls on mobile sources. The question is: What is the cost of
doing a less stringent I/M program and getting additional
reductions from stationary sources, instead?
Adopting a weak performance standard for I/M means fewer tons
of VOC reductions than EPA's proposed high-tech program, as shown
in Table 6-6. The low-tech "enhanced" program listed in Table 6-6
is essentially the basic I/M performance standard with light-duty
trucks included along with visual inspection of the catalyst and
inlet restrictor. This less stringent standard, even when
implemented in a centralized network, costs more per ton than the
high-tech approach. Thus, if states choose to implement a weak
I/M program there is a direct cost to the nation because of the
higher expense. In addition to the direct cost, there is also an
indirect cost. As more and more controls are imposed on
stationary sources, the law of diminishing returns would predict
that the cost per ton will rise. It is estimated that the cost of
these marginal controls will likely exceed $5,000 per ton.
To estimate the total cost of implementing an only marginally
"enhanced" program (i.e., the low-tech program mentioned above) it
was assumed that of the 56 million vehicles subject to enhanced
I/M 42 million vehicles would be in a decentralized system and 14
million would be centralized. This reflects the current mix of
Table 6-6
Total Cost and Benefits of I/M Options
Per Million Vehicles
High-Tech Enhanced I/M
Centralized Low-Tech I/M
Decentralized Low-Tech I/M
Tons
6, 724
2,245
2,245
Total Cost
$8,544,000
$8,204,000
$17,062,000
-81-

-------
programs in the affected areas. It was also assumed that each ton
not obtained from I/M would be gotten from stationary source
controls at $5,000 per ton. The results are shown in Table 6-7.
The extra direct cost of the low-tech option would be about $353
million while the indirect cost of the more expensive stationary
source controls amounts to about $1,254 million, for a total of
about $1.6 billion in excess cost.
Table 6-7
Excess Cost of Choosing Low Option I/M

Vehicles
Benefits
Cost

millions
tons
millions
High-Tech I/M
56
376, 529
$479
Low-Tech Centralized
14
31,426
$115
Low-Tech Decentralized
42
94,279
$717
Total Low-Tech
56
125,705
$832
High-Tech - Low-Tech

250,824
$353
Stationary Cost @ $5000/ton


$1,254
Total Excess Cost


$1,607
6.3 National Costs and Benefits
6.3.1	Emission Reductions
Estimates of the total costs and emission reduction benefits
of current and future I/M programs were obtained using CEM4.1.
Because average costs and effectiveness vary between centralized
and decentralized programs11 the costs and reductions were modeled
d,ifferently for each program type. The MOBILE4 . 1 output showing
the scenarios used are in Appendix I. Vehicle population figures
are needed in order to calculate total costs and emission
reductions. Because figures obtained from the states vary in
reliability, estimates were derived based upon Census data for
each area.
As shown in Table 6-8 below, current I/M programs obtain
estimated total annual emission reductions of 116,000 tons of VOC
and 1,566,000 tons of CO. Implementation of a biennial high-tech
program would yield estimated annual- emission reductions of
384,000 tons of VOC and 2,345,000 tons of CO from enhanced I/M
programs, and 36,000 tons of VOC and 500,000 tons of CO from basic
programs. Enhanced high-tech I/M programs would also reduce NO;.:
emissions. The transient test with N0:,: cutpoints designed to fail
11Tierney, E.,J. "I/M Network Type: Effects on Emission
Reductions, Cost, and Convenience," U.S. EPA Technical Information
Document, number EPA-AA-TSS-I/M-89-2, January 1991
-82-

-------
10% to 20% of the vehicles would yield estimated NOx reductions of
9% relative to emission levels with no program in place.
Table 6-8
National Benefits of I/M
(tons of emissions reduced annually)
VOC
CQ
Reductions from Continuing I/M Unchanged
Centralized Areas	55,540	775,228
Decentralized Areas 60.476	731,167
Current Total	116,016	1,566,395
Expected Reductions fcom Proposal
Enhanced Areas
Basic Areas
Centralized
Decentralized
384,130
2,345,278
23,289
12,996
326,290
174,186
Basic .Total
36,285
500.476
Total Future Benefits 420.415
2,845,754
Thus, enhanced I/M and improvements to existing and new I/M
programs will result in national emission reductions substantially
greater than current I/M programs.
EPA has developed estimates of inspection and repair costs in
a high-tech I/M program. The derivation of these estimates is
detailed in section 5.0. A conventional steady-state I/M test
including ATP currently costs about $8.50 per vehicle on average
in a centralized program, and $17.70 per vehicle on average in a
decentralized program. A complete high-tech test, including
transient, purge, and pressure testing, is expected to cost
approximately $17 per vehicle in an efficiently run high-volume
centralized program. In a program where 1984 and later vehicles
received th^ high-tech test, and older vehicles received a steady-
state test and ATP, and the inspection were performed biennially,
the estimated annual per vehicle cost would be about $9. The cost
is sensitive to whether test equipment and personnel face a steady
stream of vehicles or have idle periods. Therefore the cost would
be somewhat higher in a test-only multi-participant system if the
inspection network had more excess capacity than a typical
centralized program. Test-only stations may also not be as
proficient in testing each vehicle quickly, adding somewhat to
costs.
6.3.2
Economic Costs to Motorists
-83-

-------
The overall average repair cost for transient failures is
estimated to be $120. Average repair costs for pressure and purge
test failures are estimated to be $38 and $70, respectively.
Repairs for N0>: failures are estimated to cost approximately $100
per vehicle. Data from the Hammond test program indicate that it
would be very rare for one vehicle to need all three of these
repair costs.
These repairs have been found to produce fuel economy
benefits that will at least partially offset the cost of repairs.
Fuel economy . improvements of 6.1% for pressure test failures and
5.7% for purge test failures were observed. Vehicles that failed
the transient short test at the proposed cutpoints were found to
enjoy a fuel economy improvement of 12.6% as a result of repairs.
Fuel economy improvements persist beyond the year of the test.
Currently, there are an estimated 63,550,000 vehicles subject
to I/M nationwide. Of these, 23,574,000 are in centralized
programs and 39,976,000 are in decentralized programs (see
Appendix I) . Inspection fees currently total an estimated $747
million annually, $182 million in centralized programs, and $565
million in decentralized programs. Repair costs are estimated at
$392 million, $140 million in centralized programs, and $252
million in decentralized programs. Current fuel economy benefits
are estimated at $245 million, $92 million in centralized
programs, and $153 million in decentralized programs.
As shown in Table 6-9 below, estimates using EPA1s cost-
effectiveness model show that total inspection costs in the year
2000 in enhanced I/M programs accounting for growth in the size of
the vehicle fleet are expected to be $451 million, with repairs
totaling $710 million assuming that programs are biennial. Fuel
economy benefits are expected to total $825 million, with $617
million attributable to the tailpipe emissions test and $208
million due to the functional evaporative tests.
In basic I/M programs, total annual inspection costs in the
year 2000 are estimated at $162 million, and repair costs are
expected to be approximately $113 million.
Thus, despite significant increases in repair expenditures as
a result of the program, the switch to biennial testing and the
improved fuel economy benefits from programs will result in a
lower national annual cost of the inspection program.
-84-

-------
Table 6-9
Program Costs and Economic Benefits
(millions of dollars)


Emission

Emission
Evap



Test
Evap
Test Fuel
Fuel


Test
Repair
Repair
Economy
Economy
Net

Cost
C.St
Cost
Savinas
Savinas
Cost
Costs and
Economic
Benefits
of Continuincr I/M Unchanaed

Central
$182
$140
na
($92)
na
$230
Decentral
$565
$252
na
($153)
na
$664
Total
$747
$392

($245)

$894
Expected 1
Costs and
Economic
Benefits
From ProDOsal

Enhanced
$451
$489
$221
($617)
($208)
$336
Basic






Central
$67
$60
na
($39)
na
$88
Decentral
$95
51
na
($31)
aa
117
ICtaJL
$162


($70)

$?.Q;7
Grand
$613
$60 2
$221
($687)
($208)
$541
Total






*	Net cost is derived by adding inspection and repair costs and
subtracting fuel economy benefits.
6. 4 Motocis.t Inco.nve.ni.ence. Costs
There is an additional cost factor associated with I/M, the
cost of the time spent by vehicle owners in complying with the
inspection requirement. This cost was estimated by assuming that
motorists' leisure time is worth about $20 per hour. The amount
,of time spent getting an inspection can vary considerably as well
and very little data on this subject is available. For the
purpose of this analysis, it was assumed that motorists typically
spend roughly 45 minutes travelling to the test site, getting
tested, and returning in an efficiently designed high volume test
program.
EPA calculated the cost-effectiveness of a biennial high-tech
program with this additional cost included. Table 6-10 below
shows the estimated total program cost per million vehicles, the
cost per ton with all costs allocated to VOC reduction, and the
adjusted cost per ton of VOC with costs allocated among pollutants
as discussed previously.
-85-

-------
Table 6-10
Costs of the Biennial High Option including Inconvenience
Total CQSt	$12,254,000
Cost per Ton
All costs to VOC	$1,983
Cost per Ton
Adjusted VOC Cost $1,566
Comparing these figures with those in Tables 6-4 and 6-5
shows that a biennial high-tech program, even with motorist
inconvenience costs included, is still more cost-effective than a
weak, low-tech program without those costs considered.
7.0	REGULATORY FLEXIBILITY ANALYSIS
7.1	Regulatory Flexibility Act Requirements
The Regulatory Flexibility Act recognizes three kinds of
small entities and defines them as follows:
	Small business - any business which is independently owned
and operated and not dominant in its field as defined by
Small Business Administration regulations under Section 3 of
the Small Business Act.
	Small organization - any not-for-profit enterprise that is
independently owned and operated and not dominant in its
field (e.g., private hospitals and educational institutions).
	Small governmental jurisdiction - any government of a
district with a population of less than 50,000.
Small governmental jurisdictions, as defined above, are
exempted from the requirements of this regulation. There are no
private non-profit organizations involved in the operation of I/M
programs. Consequently this analysis will be limited to the
affects on certain small businesses, namely providers of
inspection and repair services and of inspection equipment.
There is a significant impact on small entities whenever the
following criteria are satisfied:
	Annual compliance costs (annualized capital, operating,
reporting, etc.) increase total costs of production for small
entities for the relevant process or product by more than 5%
-86-

-------
	Compliance costs as a percent of sales for small entities are
at least 10% higher than compliance costs as a percent of
sales for large entities
	Capital costs of compliance represent a significant portion
of capital available to small entities, considering internal
cash flow plus external financing capabilities
	The requirements of the regulation are likely to result in
closures of small entities
The enhanced I/M performance standard contained in the
proposed action includes new "high-tech" test procedures for newer
vehicles and enables states to obtain significantly higher
emission reductions from their I/M programs than they have
previously. This performance standard will affect different types
of businesses differently. Test providers will need to invest in
new equipment. Repair providers will be repairing more vehicles
for more types of inspection failures. The enhanced performance
standard will also affect different types of inspection networks
differently.
7.1.1	The Universe of Affected Entities
The Regulatory Flexibility Act's definition of "small
business" is based on the Small Business Administration's (SBA)
definitions. These are listed in 13 CFR Part 121 by Standard
Industrial Code (SIC) categories. The types of businesses that
have either been licensed to perform inspections or have been
involved in I/M in some other way, such as by selling inspection
equipment, and their SIC categories are listed in Table 7-1, along
with the size cutoffs used by SBA to define small business for
each. Size cutoffs are defined either in terms of the number of
employees or gross annual revenue, expressed in millions of
dollars.
-87-

-------
Table 7-1
SIC
5013
5511
5521
5531
5541
7531
7534
7535
7538
7539
7549
Affected Businesses
Description	CutQff
Automotive Part and Supply Wholesalers	100
(i.e., -auto engine testing equipment,	employees
electrical)
Motor Vehicle Dealers (New and Used)	$11.5 M
Motor Vehicle Dealers (Used)	$11.5 M
Auto and Home Supply Stores	$3.5 M
Gasoline Service Stations	$4.5
Top and Body Repair Shops	$3.5
Tire Retreading and Repair Shops	$7.0
Paint Shops	$3.5
General Automotive Repair Shops	$3.5
Auto Repair, Not Elsewhere Classified,	$3.5
(e.g., radiator shops muffler shops,
transmission shops, etc.)
Automotive Services, Except Repair and	$3.5 M
Car Washes (e.g., diagnostic centers,
inspection centers, towing etc.)
M
M
M
M
M
M
Note that although all analyzer manufacturers are "affected,"
the size cutoff of 100 employees prevents them from meeting the
definition of "small business."
7.2 Types of Economic Impacts of Concern
This analysis looks at the types of impacts that inspection
and repair providers in existing programs will experience as a
result of the requirements of EPA's rulemaking. Since the
requirements for basic I/M programs will remain essentially the
same as the current I/M requirements, significant impacts are not
expected in these programs. Hence, this analysis will focus on
existing I/M programs that will have to become enhanced. This
analysis assumes that the enhanced program implemented will a
high-tech I/M program on the basis that this would represent a
"worst case" scenario (i.e., that with the greatest economic
impact potential).
7 . 3 Changes in Repair Activity
The repair industry in enhanced areas that currently have I/M
programs, will enjoy a significant increase in repair revenues.
The repair industry consists of motor vehicle dealers (SICs 5511
and 5521), general automotive repair shops (SIC 7538) and some
gasoline service stations (SIC 5541).
7.3.1	Repair Activity in Current I/M Programs
Reliable data do not exist on the number of repair facilities
in I/M program areas that do I/M repairs. However, repair
-88-

-------
revenues that accrue to the industry as a whole can be estimated
using vehicle population data. EPA estimates that there are 64
million vehicles in current I/M program areas, 24 million of which
are in areas with centralized programs. Of these, an estimated 15
million are in areas that will become enhanced. There are an
estimated 40 million vehicles in decentralized programs. Of
these, about 33 million are in areas that must implement enhanced
I/M.
Repair cost information is generally not collected by the
states except when a motorist applies for a waiver. However, as
described in Section 5.6, estimates of total repair costs can be
made using CEM4.1. EPA estimates that $392 million worth of
repair business would be generated by current I/M programs in the
year 2000 if these programs continued unchanged, $302 million in
areas that will go enhanced. Of this latter figure, an estimated
$89 million would be performed in areas that currently operate
centralized programs and $213 million in areas with decentralized
programs.
-89-

-------
7.3.2	Repair Activity in Future I/M Programs
The transient test, with its superior ability to identify
excess emissions, is expected to generate more repairs than the
steady-state tests, while the purge and pressure tests will enable
I/M programs to identify excess evaporative emissions for the
first time. Estimates using CEM4.1 indicate that an additional
$100 million in annual repair business will be generated in areas
that currently operate centralized programs, and an additional
$212 million in areas that currently operate decentralized
programs as a result of the requirements proposed in this action.
The additional emission repairs identified by the transient test
are expected to generate an additional $41 million in areas that
currently have centralized programs and $79 million in areas that
currently have decentralized programs. The addition of purge and
pressure testing is expected to generate an additional $59 million
in areas that currently have centralized programs, and $132
million in areas that currently have decentralized programs. Thus
the repair industry in these areas is estimated to receive an
additional $312 million, and a total of $613 million annually as a
result of the proposed action, as summarized in Table 7-2.
Table 7-2
Repair Expenses in Enhanced I/M Programs
(millions of dollars!
Current
Additional
Transient Repairs
Evaporative Repairs
Total New
Total
Centralized
$89
$41
$100
$18-9
Decentralized
$213
$79
$132
$211
$424
All Programs
$302
$120
$191
$311
$613
The $311 million in extra repair expenditures is estimated to
comprise about 40% parts cost and the remainder for labor, profit,
and overhead. The automotive parts industry estimates that 20,000
jobs are created for every $1 billion spent on parts. Hence, the
additional parts demand ($125 million) will create 750 jobs in
parts manufacturing as well as additional business for retailers
and distributors, and is likely to create more jobs for clerks and
delivery employees. The remaining 60% is estimated to comprise
about 50% profit and overhead at the repair shop and 50% labor.
Hence, mechanics will earn an additional $93 million over all
program areas. At an average pay rate of $25 per hour, this
translates into 1,800 full time equivalents (FTE) over all program
areas.
Firms that pursue this repair business may need to upgrade
repair technician skills and obtain additional diagnostic and
other equipment to perform effective repairs on new technology
-90-

-------
vehicles. Inspection stations in decentralized programs, as well
as many repair shops in centralized programs, possess emission
analyzers. These will be useful in testing those vehicles still
subject to steady-state tests and may be used to diagnose vehicles
failing the transient test and to assess repair success. BAR90
analyzers, in particular, are designed to function as a platform
for a variety of engine diagnostic functions and to download OBD
fault codes.
7. 4 Changes in Emission Testing Activity in I/M Areas
7.4.1	The	Existing Market in	Centralized	and,	Decentralized
Programs
A number of different types of entities, are involved in
providing inspections. The centralized programs in the states of
New Jersey, Delaware, Oregon, and Indiana are operated by the
state, those in the cities of Memphis, Tennessee, and Washington,
D.C. are operated by the local government. These programs cover
approximately 6 million vehicles. All of these programs except
Oregon and Memphis will be subject to the enhanced I/M
requirement. Therefore, 5 million vehicles in government operated
programs will be covered by this requirement. The remaining 18
million vehicles are in programs operated by private contractors
(SIC 7549), of which 10 million vehicles are in areas covered by
the enhanced I/M requirement. Both the government agencies, and
the private contractors exceed the cutoffs for small entities.
Inspection providers in decentralized programs fall into all
SIC categories in Table 7-1 except 5013 - Automotive Part and
Supply Wholesalers. However, the prevalence of the different
categories among licensed inspection stations varies. The total
number of inspection stations in decentralized areas covered by
the enhanced I/M requirement are listed in Table 7-3.
-91-

-------
Table 7-3
Number of Inspection Stations by State
State
Stations
Cali fornia
8, 752
Colorado
1, 500
Georgia
647
Houston
1, 100
Louisiana
140
Massachusetts
2, 800
Nevada
415
New Hampshire
243
New York
4, 300
Pennsylvania
3,838
Rhode Island
950
Virginia
370
Total
25,055
Data on the distribution of inspection stations among the
different categories are not collected by most states, neither is
data on the number of stations that fall below the cutoffs for
small entities listed in Table 7-1. However, listings of
inspection stations were obtained from California and Pennsylvania
and stations were broken down into the following categories:
Service Stations, gas stations that also perform repairs (5541);
Dealerships (5511 and 5521); Independent Repair Shops (7538); Non-
Engine Repair Shops, such as tire shops, body shops, or
transmission shops (7531, 7534, 7535, and 7539); Retailers (5531);
and Test-Only Stations (7549). The California data is based on an
analysis of the entire station population. The Pennsylvania data
is based on an analysis of a 10% random sample of licensed
stations.
-92-

-------
Table 7-4
Inspection Stations by Category
California
Station Type	Number	Percentage
Service Stations	2,183	27
Dealerships	1,361	17
Independent Repair Shops	3,272	41
Non-Engine Repair Shops	7 34	9
Retailers	276	3
Test-Only Stations	131	2
Total	7978
Pennsylvania
Station Type	Number	Percentage
Service Stations	124	36
Dealerships	95	27
Independent Repair Shops	67	19
Non-Engine Repair.Shops	46	13
Retailers	16	5
Test-Only Stations	0	0
Total	348
Information on the number of subject vehicles in each I/M
program, and the inspection fee and the portion of the fee
returned to the state in each program is readily available. EPA
also gathers data on the number of licensed stations in
decentralized programs. With this information, inspection station
revenue in decentralized programs can be estimated. These
estimates for programs in enhanced I/M areas are presented in
Table 7-5.
-93-

-------
Table 7-5
Inspection Station Volumes and Incomes


Vehicles
Vehicles

State
Net
Program
Stations
per Year
/Station
Fee
Share
Revenue
California 12
8, 752
6,426,636
734
$48.3913
$6.00
$31,127
Colorado
1, 500
1,655,897
1, 104
$9.00
$1.50
$8,279
Georgia
647
1,118,44B
1, 729
$10.00
$0.50
$16,422
Houston14
1,100
1,482,349
1,348
$11.25
$3.50
$10,444
Louisiana13
140
145,175
1, 037
$10.00
$5 .25
$4,926
Massachusetts
2, 800
3,700,000
1, 321
$15.00
$2.50
$16,518
Nevada
415
523,098
1, 260
$16.00
$3.00
$16,386
New Hampshire
243
137,137
564
$14.00
$1.25
$7,195
New Yorkt
4, 300
4,605,158
1, 071
$17.00
$1.25
$16,868
Pennsylvania
3, 838
3,202,450
834
$8. 48
$0.48
$6,675
Rhode Island
950
650,000
684
$12.00
-0-
$8,211
Virainia
22n
481.305
1. 3 .01
$12.50
$1.10
$14,829
Total
25055
24.127.653




Averages weighted
2,088*
2,010,638*
963
$15.39
$3.35
$18,914
by # of stations
* Simple averages (i.e., non-weighted)
The costs incurred by inspection stations are driven by a
number of factors. Labor (i.e., the amount of time required to
perform the inspection and the inspector's hourly wage) appears to
be the largest component of cost. The cost of the analyzer is the
second largest component. PC-based (BAR90) analyzers are the
latest generation of analyzers used in decentralized programs.
Their cost can vary from $13,000 to $20,000. The most common
price appears to be approximately $15,000 each. A number of
service station based programs in areas required to implement
enhanced I/M are currently using BAR84 analyzers. These cost
approximately $5,000 each. Many stations in the older BAR84
programs have paid off the cost of their analyzers, which in turn
12	BAR 90 analyzers are used in these programs. All others currently use BAR
84 except Houston, Louisiana, and Rhode I3land.
13	This figure was supplied to EPA by the State in October of 1991 and
represents an estimate based upon data from calendar year 1990. In its
Third Report to the Legislature (December 1991), the I/M Review Committee
reported an average cost per inspection of $36.23. This number is based
upon a survey conducted in September 1991, and includes only the cost of
the inspection (not the $6 fee for the certificate). The resulting figure
of $42.23 suggests that, at least during September 1991, the average fee
charged to motorists may have dipped slightly.
14	Current X/M inspection is anti-tampering only. Station, vehicle, and
income data may change with the addition of tailpipe emissions testing.
-94-

-------
decreases their annual inspection expenses. Analyzer, service
contracts and calibration gas add lesser increments to the total
cost.
Estimates were made of the typical costs incurred by
inspection stations, net profits were estimated and the results
presented in Table 7-6. While large businesses may be able to
afford to purchase current analyzer equipment outright, the
smaller entities, with which this analysis is concerned, often
have to finance these purchases. Analyzers are assumed to be
purchased and paid off over a five-year period at a 12% rate of
interest. Conversations with program personnel in decentralized
programs indicated that inspectors are paid about $15 per hour.
Overhead (employers taxes, benefits, etc.) is assumed to be 40%,
for a total labor cost of $21 per hour.
Some cost factors are subject to regional variability. Local
data, as reported by state program officials and EPA Regional
offices, is used for such parameters as number of vehicles per
station per year, average length of test, and cost of service
contracts. Labor and equipment costs are estimated as described
previously. In programs where the equipment specification is more
than five years old, the analyzers are assumed to be paid off.
This, in turn, increases the stations' profits. The results are
listed in Table 7.-6.
Table 7-6
Average Inspection Station Revenues.	Costs,	and Profits

Vehicles

Net
Annual

State
/Station
Fee
Revenue
Cost
Net Profit
California11
734
$48.39
$31,127
$11,899
$19, 228
Colorado
1, 104
$9. 00
$8,279
$5,202
$3,078
Georgia
1, 729
$10.00
$16,422
$9,320
$7, 102
Houston13
1,348
$11.25
$10,444
$7,075
$3, 369
Louisiana13
1, 037
$10.00
$4,926
$5,444
($518)
Massachusetts15
1,321
$15.00
$16,518
$13,498
$3,020
Nevada
1, 260
$16.00
$16,386
$7,681
$8,705
New Hampshire
564
$14.00
$7,195
$4,257
$2,938
New York.11
1, 071
$17.00
$16,868
$20,268
($3,400)
Pennsylvania14
834
$8.50
$6,675
$2,811
$3,864
Rhode Island14
684
$12.00
$8,211
$2,653
$5,557
Virginia
1,301
$13.50
$14,829
$5,546
$9,283
Averaae
963
$15.39
$lfl.914
$10,818
SH.096
Averaae w/o CA
1,086
$12.39
$12.357
$10,238
$2,120
Average w/o CA & NY
1,091
$11.93
$10,741
$6,645
$4,097
15 Due to the age of the state analyzer specification, analyzer costs are
assumed to be paid off in stations in these programs.
-95-

-------
This analysis revealed anomalies in the California and New
York programs relative to the others. California has a much
higher average fee than the other programs, and estimated average
profit is nearly twice that of the next highest program. The
estimate for New York reflects an unusually long test duration
(see Table 7-11) and shows the average station operating at a
loss; this estimate is supported by reports that station operators
have sued the state to be allowed to charge a higher fee.
Therefore, average revenues and profits were also calculated with
data from those states omitted.
These figures, based on the average inspection volumes for
each state, show that inspection services, by themselves, do not
yield significant profit to the average inspection station. While
the average profit is low, the amount of revenue and profit can
vary a great deal among inspection stations since inspection
volumes vary considerably as well. The best available data on
station volumes was obtained from the California program. The
data covers a three month time period and is shown in Table 7-7.
Table 7-7
Inspection Volumes in California
Tests
Stations
% Total.
% Active Stations
0
1, 958
22
NA
1-100
1, 156
13
17
101-200
1, 676
19
25
201-300
1, 178
13
17
301-400
754
9
11
401-500
469
5
7
501 +
1. 571
JJ.
21
Total
8. 752


Total Active
6,794


EPA analyzed revenues and profits for inspection stations at
different volumes; the results are presented in Table 7-8.
Revenues, costs and profits are calculated as in Tables 7-5 and 7-
6. California has a market-based inspection fee (i.e., stations
charge what the market will bear, since the state does not
regulate the fee) . Conversations with California program
officials indicate that higher volume stations charge lower fees
than the average. The fees assumed for 1, 200- and 2,000-
inspection-per-year cases are based on figures suggested by the
state.
-96-

-------
Table 7-8
Station Revenues and Profits bv Volume
Veh/Qtr
Veh/Year	Fee
Net Revenue Annual Cost Net Profit
0
100
300
500
0	$48.39
400	$48.39
1200	$42.00
2000	$32.00
$0	$5,474	($5,474)
$16,956	$8,974	$7,982
$43,200	$15,974	$27,226
$52,000	$22,974	$29,026
These figures indicate that inspections can be profitable if
volume is high, however, relatively few stations have high
inspection volumes. Based on the data in Table 7-7, 22% of the
licensed stations perform no inspections and therefore are losing
money invested in equipment, licensing, and training (only
equipment costs are estimated here). An additional 32% perform
800 inspections per year or less, and therefore appear to be
earning only a modest level of profit. 22% perform from 800 to
1,600 inspections per year, and an additional 23% perform more
than 1,600 inspections per year. Profitability is higher in these
latter two categories.
7.4.2	Future Market in Enhanced I/M Programs
Test providers will be required to invest in new equipment
for that portion of the subject vehicle fleet that will undergo
transient, purge, and pressure testing. The total cost to re-
equip an existing inspection site to perform the new tests is
estimated at about $144,000. EPA based this estimate on
conversations with equipment manufacturers over the past year;
more recent information indicates that a lower figure is likely.
7.4.3	Centralized Programs
As indicated in Section 5.0, throughput rates would be lower
in centralized lanes performing transient, purge, and pressure
testing than in inspection lanes performing the current test
procedures. Since programs will be able to switch from an annual
inspection frequency to biennial at the same time they implement
the high-tech tests, EPA does not anticipate that a significant
number of new inspection lanes will need to be built in
centralized programs in order to satisfy the proposed requirements
and maintain waiting times at minimal levels.
-97-

-------
7.4.4	Decentralized Programs
Enhanced areas that currently have decentralized programs
will have two options in meeting the requirements of the proposed
action: they can institute either a multi-participant test-only
network, or a single operator centralized system.
If a program were to switch to a multi-participant, test-only
system, stations that currently participate in the test and repair
network would have a choice between concentrating on inspections,
and becoming test-only stations, or concentrating on repairs.
That choice would likely be driven by the station's current
inspection volume and the degree to which its prospective income
is expected to be derived from inspection as opposed to repair and
other services. This analysis utilizes the simplifying assumption
that stations that perform a large volume of inspections, and that
currently derive more income from inspection than from repair or
other services, would be likely to become test-only stations. By
the same reasoning, stations that are more oriented toward repair
would focus on the additional repair business generated by the
inspections conducted elsewhere.
Data correlating average inspection volume with station type
are not available. However, survey data of motorists in I/M
programs point to the fact that stations that currently focus on
repair work and that do a steady volume of repairs are often
unable to make facilities available to provide inspections
promptly on request16. 27% of motorists in decentralized programs
reported being asked to bring their vehicles back for testing
another time. 20% reported having to take their vehicles to more
than one station to obtain a test. Nearly one out of three had to
leave their vehicles for inspection. On the average, the vehicles
had to be left for five hours. These data suggest that a focus on
repair leads to reduced opportunities to perform inspections and
probably to lower inspection volumes as a result.
The converse appears also to be true. Stations that are
readily able to provide inspections are often either unable, or
simply have not chosen to perform repairs. 53% of motorists
reported taking their vehicle to another station, other than the
one where the inspection was performed, for repairs.
Based on the data from Pennsylvania and California, the
following distribution of station types is assumed for this
analysis:
"Attitudes and Opinions Regarding Vehicle Emission Testing," Riter
Research. September, 1991
-98-

-------
Table 7-9
Assumed Station Distributions
Station Type
Service Stations
Dealerships
Independent Repair Shops
Non-Engine Repair Shops
Retailers
Test Only Stations
Some stations, such as dealerships and independent repair
shops, would be likely to concentrate on I/M repairs since their
business already has a decided orientation toward engine repairs.
Together, these constitute 52% of the assumed station population.
Because of their focus on repair, it is likely that these stations
tend to have lower inspection volumes, as discussed above, and
some of them are likely to be among the 22% of stations that
report no testing activity. For the purposes of this analysis, it
is assumed that half of the inactive inspection stations are in
this repair-oriented group.
These repair-oriented stations will likely get the majority,
though not all, of the additional repair business estimated
previously at $211 million among all decentralized programs. If
these stations ultimately get 85% of this business (allowing for
15% of the repair stations to come from other categories, mainly
service stations) it will amount to annual revenues of roughly
$13,000 per year. This would offset inspection losses of $10,000
to $12,000 per year (Table 7-6).
The stations that have higher inspection volumes than average
are likely to be deriving a substantial portion of their current
profit from the inspection business and relatively little or none
from repair. Based on the California data, it is assumed that the
23% of the stations that have inspection volumes of approximately
200% of the program average or more would be likely to opt to
become test-only stations. Test-only stations, in those
decentralized programs where they exist, would, of course, be in
this group.
Some stations in this high volume group may be repair-
oriented stations, such as dealerships, independent repair shops,
and some service stations, and may prefer to opt out of the
inspection business for more profitable repair business. This
would create opportunities for other businesses to enter the test-
only market, including stations whose current inspection volume is
somewhat lower.
Current repair revenues in decentralized enhanced programs
are estimated at $213 million. If this 23% segment of the
-99-

-------
stations had been getting 23% of this business (based on the
foregoing discussion, they have probably been getting less) , then
they are giving up current annual revenues of $8,500 each in order
to pursue the inspection market.
The remaining .25% that do not have a clear orientation toward
engine repair, and that do not perform a high volume of
inspections, are a mix of service stations, whose business is a
mix of gasoline sales and, in some cases, engine repairs including
I/M repairs on some portion of the vehicles they test; non-engine
repair shops, such as tire shops, muffler shops, transmission
shops, etc.; and retailers. Members of this group are assumed to
make up the other half of the 22% of stations that do no
inspections. These stations would not be adversely affected by
this rulemaking since they are currently deriving no income from
the inspection business.
This leaves 14% of the population of licensed inspection
stations that do not have a clear orientation toward engine repair
and derive some income from inspections. Since they are not high
volume stations, stations in this group do not derive high profits
from inspections on the average. Table 7-10 shows the projected
current revenues and profits for these stations assuming that they
are evenly distributed among the four low to medium groups in
Table 7-7 (those doing 1 to 400 inspections per quarter), assuming
that all stations charge the average fee of $48.39. Note also
that the numbers of inspections in each category represent the
mid-points of the ranges presented in Table 7-7. The column
entitled "% Avg Profit" shows the estimated profit for each
category as a percentage of the program average profit for
California in Table 7-6.
Given that the average profit in California is almost double
that for the next most profitable program, the profits calculated
based on California data were adjusted to reflect projected
national profits for stations with inspection volumes ranging from
about 25% to 200% of the average for the program. The national
average profits are based on the figure of $4,097 obtained as the
average net profit without data from California and New York.



Table "
1-10

Revenues and
Profits
for Low
and Medium Volume Stations

% Avg
% Total
Net
Net % Profit Based
Veh/Qt
r Vol.
Stations
Revenue
Profit Aver. Profit on Nat ' 1 Avq
50
27
3.36
$8,478
$1,254 6.5 $266
150
82
4 . 90
$25,434
$14,710 76.5 $3,134
250
136
3.36
$42,390
$28,166 146.0 $5,982
350
191
2.38
$59,346
$41,622 216.0 $8,849
The first two categories, representing 8% of the total number
of stations, appear to earn 77% of the program average profit or
-100-

-------
less. The two higher volume categories, representing roughly 6%
of the total station population, derive substantial profits from
the inspection business (these estimates are based on data from
California which has the most profitable inspection program;
profits in other states probably do not increase with increasing
test volume as steeply as this analysis suggests, while revenues,
on the other hand, do increase in direct proportion to volume).
Data on the relative contribution of inspection revenue, compared
to other types of business are not available . Some of these
stations may be service stations that are currently doing a
profitable business in engine repairs, and would continue to do
so. Others, such as the 2.38% earning an estimated 216% of the
average profit might still opt into the test-only business where a
high volume station has opted out, as discussed previously.
Others, such as the non-engine repair shops and the retailers have
primary lines of business unrelated to I/M.
However, it may be that some of those stations earning 200%
or more of the average revenue would be unable to recoup this loss
any other way, and would be forced to close. The average revenue
loss for these stations would be $37,828 nationally, and $21,482
outside California and New York. It may also be that some of the
stations in the lower profit categories are so marginally
profitable that loss of inspection business would result in
closure as well. If 10% of this group of stations without clear
I/M-related alternatives (14% of the total) were to close it would
amount to a total of roughly 350 stations nationwide.
If a single contractor centralized program were instituted in
an area where a decentralized program is currently operating, the
option to pursue the test-only business would not be available to
the 23% of the station population that would be likely to pursue
it. Based on the foregoing analysis, these stations have current
inspection volumes of 200% or more of the program average, and may
have average profits of roughly 220% or more of the program
average . Members of this group without profitable alternatives
would also face the risk of closure.
The likelihood of closure would depend upon the fraction of
income derived from inspections. Data on this is not available.
Since many of these stations have other lines of business, such as
gasoline sales, auto parts sales, or various types of vehicle
repair and servicing, the loss of business will not necessarily
mean closure. The fraction of these stations that would be unable
to recoup this loss and face closure is difficult to estimate
given the paucity of data. However, if, as before, 10% of these
stations were to close as a result of a switch to a single-
contractor centralized system, as well as 10% of the 14% of
stations identified previously as being at risk, then 927 stations
might close nationwide if all decentralized programs in enhanced
I/M areas switched to centralized, single-contractor systems. If
the areas containing half of the current inspection stations were
-101-

-------
to switch to single-contractor, centralized systems, then
potential closures would number about 464.
The most severely impacted would be the test-only stations,
which in California comprise 2% of the test stations. Given that
they have no other lines of business to compensate for the loss of
inspection revenue, these stations would almost certainly close if
the area were to switch to a centralized, single-contractor
system, unless these stations were able to win the contract (some
of these businesses have indicated to EPA they they would try to
do so).
7.4.5	Impact on Jobs in Decentralized Programs
Table 7-11 shows the number of inspectors in each program,
and the average number of inspectors per station for all
decentralized enhanced programs except Rhode Island, for which
data on the number of inspectors is unavailable. The national
weighted average number of inspectors per station excludes the
highest and lowest averages in the set, those from New York
(program officials in this state have indicated that the total
number of licensed inspectors is likely to include individuals no
longer working as inspectors) and Massachusetts.
Table 7-11
Numbers of Inspectors per Station by State
State
Stations
Inspectors
Average
Time per
California
8, 752
18,000
2 . 06
25
Colorado
1, 500
2, 930
1 . 95
5
Georgia
647
2, 845
4 . 40
10
Houston
1, 100
2, 645
2 . 40
15
Louisiana
140
513
3.66
15
Massachusetts
2,.800
1, 208
0 . 43
25
Nevada
415
1,249
3 . 01
10
New Hampshire
243
933
3. 84
5
New York
4, 300
21,640
5 . 03
40
Pennsylvania
3, 838
19,221
5. 01
3
Virginia
370
1, 114
3.01
5
National Weighted Average

2 . 05
20
Average station volumes are low (Tables 7-5 and 7-6) - about
four per day. Given that there are, on the average, two
inspectors per station, and that the average inspection takes
twenty minutes to perform, it follows that the average inspector
spends 40 minutes per day performing inspections. This works out
to 0.08 of an FTE (i.e., inspections take about three hours and
twenty minutes out of a forty-hour work week). Hence, inspectors
are generally individuals employed primarily for other jobs (in
-102-

-------
most cases as mechanics) who spend a small amount, of their time on
inspections. Communications with program officials in these
states and EPA's experience in auditing these programs support
this conclusion. Table 7-12 shows the estimated total number of
FTE devoted to inspections in the different station categories
developed in this analysis, using the volume assumptions developed
previously.
Table 7-12
Estimated Inspection FTE
Station Type
1
Number
Tests/Day
FTE
Repair Oriented
52%
13,029
3
1, 612
Inspection Oriented
23%
5, 763
8
1, 902
No Inspections
11%
2, 756
0
0
Remainder
14%
3. 508
A
579
Total
4, 093
In most cases, the time spent on inspections could be easily
re-oriented toward other tasks if inspection business were to
cease, however, some stations might experience some contractions
as a result of losing inspection business, and some might close,
as estimated previously. For the sake of analysis, all FTEs
currently devoted to inspections in decentralized enhanced
programs, as shown in Table 7-12, are counted as lost. Estimates
are also made of additional FTEs lost as a result of potential
station closures.
If a decentralized test-only program were instituted, it was
estimated that 10% of the 14% of stations that have some
inspection business, and are not clearly positioned to pursue
either the inspection or repair markets, might potentially close.
Assuming that these stations have two FTEs in addition to
inspector FTEs, total job losses would amount to an additional 700
FTEs .
In the event of a switch, to a single-contractor centralized
system, 10% of the 23% of stations that would otherwise have
pursued the test-only option would also be at risk of closing.
Potential closures are estimated to total 927. The average number
of non-inspection FTE per station in this case is assumed to be
2 . 5 since some larger stations would be included in the risk
group. In this case, losses could total an additional 2,318 FTEs.
New jobs would be created by the test-only program, and the
increased repair business that would offset these potential losses
to the small business community and to labor.
EPA estimates that in a high volume enhanced I/M lane,
testing an average of 7.5 vehicles per hour, 3-4 inspectors would
be needed per lane instead of the 1-2 typically employed in
-103-

-------
current high volume systems. Using an industry estimate of 267
FTE per million vehicles, and assuming a 20% retest rate, 5,340
FTEs are required to test the 33 million vehicles in currently
decentralized programs on a biennial basis (this estimate is based
on the assumptions and methodology developed in Section 5.2 of
this report, "Estimated Cost of High-Tech I/M Testing").
In a decentralized test-only system volume would likely be
lower. This analysis estimates that 4,200 inspections per year,
or about 16 per day would be likely. Therefore, two or three
inspectors per lane would be adequate. If two inspectors per lane
were employed, 11,525 FTEs would be created if all current
decentralized areas adopted a decentralized test-only system.
Additional, jobs that would be created in the repair sector
were estimated previously in this analysis. Approximately 1,217
mechanic FTEs, and 506 FTEs in auto parts manufacturing would be
created, in addition to clerical, delivery and other support
personnel. The results are summarized in Table 7-13.
Some new inspection facilities would be constructed whether
programs adopted decentralized test-only networks or single
contractor networks, also creating jobs. FTE estimates are based
on an industry estimate that construction of an inspection station
requires 4.79 man-years of construction and 5.1 man-years of
subcontracting. An average station is assumed to have 2.4 lanes.
The number of lanes required to inspect the fleet is based on the
assumptions of biennial inspections and a 20% retest rate. FTE
calculations are based on the assumption that total effort, i.e.,
modification of existing structures in those areas adopting
decentralized test-only programs and construction of new
facilities in those areas adopting single-contractor programs, is
equal to that needed to construct lanes for half of the vehicles
in decentralized enhanced areas. The results are summarized in
Table 7-13.
Table 7-13
Summary of FTE Gains and Losses
(in currently decentralized areas required to do enhanced I/M)
Losses
#
Gains
#
Current Inspection FTE
4,093


Station Closures

New Inspector FTE

Multiple Independent
700
Multiple Independent
11,252
Contractor
2, 318
Contractor
5, 340


New Reoair FTE



Mechanic
1, 217


Parts Manufacture
506


Construction

Net Gain



Multiple Independent


8, 769
Contractor


1, 239
-104-

-------
7.4.6	National Impact on Jobs
EPA. has estimated the total FTE in current I/M programs and
the projected changes in FTE nationwide as a result of the
proposed changes. These are summarized in Table 7-14. Note that
Table 7-14 includes areas which will be starting enhanced or basic
programs from scratch, while earlier tallies included only areas
already operating I/M programs.
Table 7-14
Impact, on Jobs.Qf I/M Proposal
Current Test and Repair Jobs
Inspector Jobs
Repair Jobs
FTE
Decentralized Programs	6,600
Centralized Programs	2,500
Decentralized Programs	800
Centralized Programs	1, 5Q0
Total Current Jobs	11,400
Future Test and Repair Jobs
Enhanced I/M Programs
Inspector Jobs
Multiple Independent Supplier	10,500
Single Contractor.	2. 7Q0
Inspector Job Subtotal	2,700 - 10,500
Repair Jobs	5,500
Basic I/M Programs
Inspector Jobs	2,700
Repair Jobs	700
Total Future Inspection and Repair Jobs	11,600 - 19,400
Other Job Gains
Parts Manufacturing	1,034
Construction	1,800
Small Business Services	800
Total Net Gain in Jobs	3,800 - 11,600
Small Business Services are estimated by assuming 15
additional FTEs per urbanized area. The 800 FTEs presented in the
table represent the jobs generated in the 52 urbanized areas that
-105-

-------
do not have I/M programs now, but will be implementing them as a
result of the proposed action.
Whether programs adopt a decentralized test-only network or a
single-contractor centralized one there will be shifts in job
opportunities with some net gain in either case. Hence, the shift
to high-tech enhanced I/M may cause significant shifts in both
business and job opportunities. Small businesses that currently
do both inspections and repairs in decentralized I/M programs will
have to choose between the two. Significant new opportunities
will exist in these areas for small businesses to continue to
participate. EPA believes there are ways states can help test
stations make the transition to an enhanced I/M program.
7.5 Mitigating the Impact of Enhanced I/M on Existing Stations
Three potential approaches, to helping test stations make the
transition are presented here. The first approach would provide
direct assistance to stations that might be adversely affected by
the transition to a high-tech system. The second would be to
design the enhanced program to include transitional mechanisms to
soften the impacts of the new system. The third would be for
states to establish programs to assist stations and inspectors
through retraining and retooling programs. The previous section
discussed various strategies to assist repair technicians in the
retest process, including free retests and priority access to
retest lanes, as well as diagnostic and repair assistance.
In some states that are currently decentralized and will have
to implement enhanced I/M, analyzers have been in use for 10 years
or more and are fully amortized. In states that upgraded to BAR90
equipment (California and New York), the equipment was purchased
since 1990, and has years of useful life left. A number of other
states upgraded their equipment to BAR84 in the period from 1987
to 1990. Stations in these areas are likely to still be paying
for their equipment (see the footnote to Table 7-6). One means by
which the state could provide direct assistance to current test
stations would be to set up some type of state-supported analyzer
buy-back program for stations that were no longer going to
participate in either the test or. repair business, possibly using
funds obtained from inspection fees. BAR90 analyzers would be
needed in the repair business both for diagnostic and repair work
as well as to check whether repairs on old technology vehicles
were effective. BAR90 analyzers could also be used to test older
technology vehicles in test-only stations. This concept would
allow stations that were planning to leave the I/M business to
recover all or part of their capital investment for equipment that
could not be used for diagnostics and repair. Such a buy-back
program might allow a fairer transition to test-only status.
A related strategy would be for EPA, the states, and industry
to support the development of new and improved uses for BAR90
analyzers so that current as well as future analyzer owners can
-106-

-------
use this technology more effectively in the repair process. In
particular, it was California's intent in developing the BAR90
specification for the computer in the analyzer, which is an IBM
386 DOS-based system, to become a platform for vehicle diagnosis
and repair. EPA, the states, and industry could potentially
provide technical and financial support to speed the development
of such software. This would not only make better use of the
equipment in the field but would serve as an excellent mechanism
for providing critical technical assistance and training to the
repair community.
A second strategy to mitigate the impacts is to design
transitional features into the program. One approach would be to
allow test and repair shops to continue to do testing on vehicles
not subject to the t rans ient/purge test for some transitional
period (note that EPA's recommended enhanced program would require
biennial, transient/purge test's on 1984 and later model year
vehicles, and biennial steady-state tests on older vehicles). EPA
is proposing to permit a phase-out of the decentralized test-and-
repair portion of the program such that all vehicles would be
inspected in test-only stations starting January 1, 1996. This
would allow these decentralized stations to continue to obtain
revenue to recover the investment made in testing equipment and
would allov; additional time to plan other strategies to replace
the income to be lost from testing.
A related approach is to allow vehicles that have failed
initial inspections in test-only stations to be retested in
existing test and repair stations using conventional test
techniques during the first inspection cycle. This would allow
those stations to attract customers, conduct testing and perform
repairs, with the added benefit of sparing the customer from
returning to the test-only station for the retest.
A third strategy would be to provide targeted assistance to
stations to assure they were able to provide high-tech repair
services. This would require pre-program start-up training to
bring repair technicians in these stations up to speed on the
high-tech tests, vehicle diagnosis, and engine repair. It might
mean tuition grants or other financial assistance. This dovetails
with stronger repair technician training programs which EPA
envisions as being part of future I/M requirements, but differs in
terms of funding, timing, and intensity. This approach might also
include financial assistance to stations for the purchase of
equipment to perform sophisticated diagnosis and repair on new
technology vehicles or to upgrade tools and equipment for more
sophisticated diagnosis and repair.
7 . 6 Public Comment
Two independent analyses of job impacts were conducted by the
Coalition for Safer, Cleaner Vehicles (CSCV) -and EPA's Office of
Policy Planning and Evaluation (OPPE). Both projected an increase
-107-

-------
in employment opportunities as a result, of the implementation of
enhanced I/M. The magnitude of the estimated increase varies
between the two studies and the estimates discussed above. The
OPPE study projects an increase of 1,300-1, 400 FTE in the areas
that currently have decentralized test-and-repair programs as a
result of the implementation of enhanced I/M, while CSCV's study
projects an increase of 4,670 FTE in those areas, and a total
increase of 8,420 FTE in all enhanced areas. Hence, there is
general agreement among the parties that- have tried to quantify
the overall employment impacts of the proposal that employment
opportunities will increase, although the magnitude of the
projected increases varies.
The National Automobile Dealers Association (NADA) submitted
comment questioning the conclusion that there will be a net
increase in emission, control employment as a result of the
implementation of enhanced I/M. However, NADA offered no analysis
of its own on employment affects, nor did it critique EPA's
analysis in any detail.
Some test-and-repair station owners commented that the
inspection business generates $7,000 per month in revenues. This
figure appears to include repair revenues as well as inspection
revenues. The previous analyses indicate that inspection revenues
average about $10,000 annually per station, or less than $1,000 a
month. These stations would still be able to pursue emission
repair business in a test-only program and there would be a
considerable increase in this business. Many of these commenters
appeared to be under the impression that, in the event of a switch
to a test-only system, they would be barred from doing repairs as
well as inspections. This is not the case.
The comment was made that the profit margin on gasoline sales
is low and that service station dealers depend on ancillary sales,
such as inspections and repairs. . The foregoing analysis shows,
and independent analyses confirm that repair business will
increase significantly with the implementation of enhanced I/M,
and that service stations with a strong orientation toward engine
repair will have an opportunity to increase profits. EPA's
analyses indicate that inspections do not generate large profits
for the average station, hence, the loss of this business will not
necessarily result in significant losses for other service
stations that do not have a strong orientation toward engine
repair.
The New Hampshire Department of Environmental Services and
the Texas Automobile Dealers Association were both supportive of
the concept of buying back old test equipment, but were concerned
about how such a program might be funded. New Hampshire suggested
that EPA recommend a means to fund such a program without
increasing the cost of emission testing. States are encouraged to
consider these measures, but they are not mandated. A wide
variety of funding mechanisms besides a surcharge on the
-108-

-------
inspection fee could be found to fund such a program. What means
might be available and appropriate are likely to vary from state
to state.
Virtually all commenters supported allowing transitional
mechanisms such as phase-in of test-only and high-tech testing,
and the final rule allows for these transitional mechanisms. No
specific comments were received on the targeted re-training
assistance concept, although the comments reflected overwhelming
support for technician training in general.
-109-

-------
8.0 ONBOARD DIAGNOSTICS AND ON ROAD TF.STTNCi
8 . 1 QnbQflgd Piagnostj.es,	Interim Provisions
EPA is required to issue onboard diagnostic (OBD) regulations
by May 15, 1992, while I/M programs will begin OBD checks two
years after the regulation has been issued. OBD checks are hot
currently a part of EPA's performance standard and no credit has
been assessed for such checks in the M0BILE4.1 model; such will be
determined after formal issuance of OBD regulations. For the
purpose of this cost-benefit analysis, the impact of OBD has not
been addressed. The impact of OBD will be relatively minor up
until the attainment deadline for serious areas, in November 1999.
EPA will certainly revisit the issue once OBD regulations are
final and as their implementation clarifies the potential of this
strategy in an I/M setting.
8.2 Qn-road Testing,	Interim Provisions
Section 182(c) (3) (B) (i) of the Act requires EPA to establish
a performance standard for enhanced I/M "including on-road
emission testing." The Act does not specify how programs or EPA
are to address the "on-road testing" requirement, and neither is
on-road testing defined within the Act itself. While potentially
a fruitful suppleraentail testing strategy, it is clear from the
legislative history of the 1990 Amendments that on-road testing
was not viewed as a potential replacement for I/M programs, as has
been suggested by some. Under the section addressing enhanced I/M
programs, the legislative history notes:
On-road emission testing is to be a part of the emission
testing system, but is to be a complement to testing
otherwise required since on-road testing is not intended
to replace such testing. On-road emission testing may
not be practical in every season or for every vehicle,
and is not required. However, it should play some role
in the state program. It is the Committee's intention
that states should take into consideration that the
results of on-road emission testing, when used, have not
been shown to be consistent with Federal emission
testing procedures. [Emphasis added]
EPA has specified that on-road testing be defined as "the
measurement of HC, CO, NOx, and/or CO2 emissions on any road or
roadside in the nonattainment area or the I/M program, " and that
it be required in enhanced programs and an option for basic' I/M
areas. Minimally, the on-road testing effort must evaluate the
emission performance of at least 0.5% of the subject fleet each
year. EPA believes that the on-road testing requirement can be
fulfilled by a range of approaches, including, but not limited to:
remote sensing devices (RSD), random road-side pull-overs using
-110-

-------
tailpipe tests and emission control device checks, or road-side
pull-overs of vehicles with high RSD readings, as well as through
the use of portable analyzers that can be placed on the vehicle
prior to on-road driving.
Of the above approaches, RSD has gained the most public
attention and has generated considerable interest. The objective
of RSD is to remotely measure the concentration of emissions from
vehicles as they are operated on public roads, and in this aim,
RSD fully meets the definition of an on-road testing strategy. In
its current" version, RSD works by. focusing a beam, or, in some
cases, multiple beams, of infrared light across the roadway into
an infrared detector. The concentration of certain pollutants in
the exhaust stream are then determined by measuring the amount of
infrared light absorbed at specific wavelengths as it passes
through the exhaust in much the same way that astronomers study
stellar atmospheres by analyzing specific portions of a star's
spectrum. The analysis is tied to a vehicle through the use of a
video camera which records the vehicle's license plate as it
passes through the beam(s).
Given its non-intrusive nature and potentially high
throughput capabilities, RSD warranted further investigation. EPA
has conducted a preliminary analysis of RSD (see Appendix J,
"Identifying Excess Emitters with a Remote Sensing Device: A
Preliminary Analysis") that investigated the comparability of the
results obtained to those in the 2500 rpm/Idle test. EPA found
that, under controlled conditions and using stringent cutpoints,
RSD's performance in measuring CO emissions was comparable to the
2500 rpm/Idle test. Since then, other researchers, such as the
California Air Resources Board (CARB), have found that the
accuracy of the device for measuring HC emissions, while less
accurate than for CO, is within a practical range for roadside
monitoring. For example, CARB researchers recently reported to
the CARB I/M Review Committee17 that the device, under highly
controlled operating conditions, yielded results that compared to
calibrated on-board measurements as follows: The remote sensors
accurately measured CO within  5% and HC within + 15% of the
instrumented vehicle measurements, respectively. EPA, however,
knows of no current RSD methodology for detecting and measuring
NOx emissions, although developmental work is being done in this
area. EPA encourages the states to be innovative in fulfilling
the on-road testing requirement.
There have been and continue to be a number of efforts in the
area of RSD evaluation, including those at the University of
17 D. Lawson, J. Gunderson, "In-Use Emission Study and High Emitter Phase,"
Presentation to I/M Review Committee, Sacramento, California, January 29,
1992.
-Ill-

-------
Denver, where the first RSD testing strategies were developed.
The bibliography18 of research in this area continues to grow.
Currently, it is difficult for EPA to project a standard
"emission credit" for on-road testing for the purpose of
performance standard modeling. Hence, for the purpose of this
cost-benefit analysis, the impact of on-road testing is not
addressed. Nonetheless, emission reduction credits will be
assessed for on-road testing efforts once additional experience is
gained in the actual use of various on-road testing strategies,
including RSD technology. Under EPA's current proposal, on-road
testing programs required by the Act "shall provide information
about the emission performance of in-use vehicles, by measuring
on-road emissions through the use of remote sensing devices or
roadside pullovers including tailpipe emission testing. The
program shall collect, analyze and report on-road testing data" as
part of the state's annual report to EPA. EPA shall use this
data, in conjunction with data gathered as part of the Agency's
on-going investigation of these, testing strategies, to develop
testing protocols and guidance.
18 In addition to the sources referenced in Appendix J, the following works
have contribute to the body of information concerning RSD.
1.	D.R. Lawson, P.J. Groblicki, et . al. , "Emissions for In-use Motor
Vehicles in Los Angeles: A Pilot Study of Remote Sensing and the Inspection
and Maintenance Program," Journal of the Air Waste Management Association,
40 (8) : 1096 (1990)
2.	R.D. Stevens and S.H. Cadle, "Remote Sensing of Carbon Monoxide
Emissions," Journal of the Air Waste Management Association, 40<1):39
(1990)
3.	G.A. Bishop, D.H. Stedman, et. al., "IR Long-Path Photometry, A Remote
Sensing Tool for Automobile Emissions,n Analytical Chemistry, 61, 671A-677A
(1989)
4.	D.H. Stedman and G.A. Bishop, "Evaluation of a Remote Sensor for Mobile
Sources CO Emissions," Report to the Environmental Protection Agency, EPA-
600-S4-90-032.
5.	D.H. Stedman, G.A. Bishop, et. al., On-Road CO Remote Sensing in the Los
Anaeles Basin. Final Report on Contract No. A932-189, California Resources
Board, Research Division, Sacramento, 1991.
6.	D.H. Stedman and G.A. Bishop. An Analysis of On-Road Remote Sending as
a Tool for Automobile Emissions Control, ILENR/RE-AQ-90/05, Final Report to
Illinois Department of Energy and Natural Resources, Springfield, IL, 1990.
-112-

-------
9.0 ALTERNATIVE TESTS
9.1 Status of Alternative Exhaust Tests
In 1988, the State of California, Southwest Research
Institute, and Sierra Research, Inc. did developmental work on a
series of loaded steady-state test modes known .as Acceleration
Simulation Modes or ASMs. EPA was involved in reviewing the
results of the testing that California had undertaken at that
time. The testing, based on 18 vehicles, found that two ASM modes
- ASM5015 and ASM2525 (the first two digits refer to the load
factor while the second two refer to the speed of steady-state
operation) - had some potential for identifying vehicles with NOx
problems related to exhaust gas recirculation (EGR) valve
malfunctions (which had been induced in the vehicles tested). A
Society of Automotive Engineers (SAE) paper (#891120) was issued
and the authors found that the tests did poorly on the
identification of HC and CO failures. The SAE paper concluded
that retention of the idle and two-speed tests would be necessary
and that the primary benefit of the ASMs was for NOx testing.
In early 1992, five low mileage 1992 model year vehicles with
induced failures were tested by ARCO using the ASM5015 and the
ASM2535. ARCO reported that the ASM5015 test may identify excess
NOx emissions as well as effectively test for evaporative system
purge. ARCO suggested an equipment package consisting of a single
power absorption curve dynamometer with no inertia simulation
capability, a raw exhaust, concentration-type emission analyzer,
and a mass flow measuring device. ARCO did not specify a specific
flow measuring device and suggested that its testing indicates
that mass flow measurement may not be essential since an
approximation can be made on the basis of engine size and
dynamometer power absorption setting. This equipment may be
substantially less expensive than the transient test equipment,
which could in turn lead to a more cost-effective program, if the
emission reduction benefits of the test were found to be
comparable. However, ARCO suggested a more complete test program
would be necessary to assess the effectiveness of the procedure
and the equipment arrangement ARCO suggests.
CARB has also been testing the ASM5015 and the ASM2525 in a
laboratory setting. At the time of the proposal of this rule, EPA
expected that data from the CARB effort, along with data from the
FTP and other steady-state tests California was conducting in its
program, would provide better insight into the effectiveness of
the ASM tests. Unfortunately, the data developed by California
turned out to be defective in that it was produced using incorrect
dynamometer settings and the State has withdrawn the data from the
docket as a result.
-113-

-------
Environment Canada conducted lab ASM and FTP testing on 4 0
Canadian vehicles and forwarded the test results to EPA. Only 20
of the 40 vehicles are representative of the U.S. fleet (since
1981) because Canada has had lower standards in effect and
recruited vehicles from the older part of the fleet. The results
of this testing are discussed below.
Vancouver, British Columbia began pilot testing of the
ASM5015 and the ASM2525 along with idle and 2500 rpm modes in its
regular I/M lanes early this summer - the first time this has been
attempted in an I/M setting. Unfortunately, Vancouver's FTP lab
was not in operation in time to do tests on any of the vehicles
that were run through the trial program. Nevertheless, the
program has forwarded important information that contributes to
the discussion of the ASM procedures . British Columbia officials
found serious problems with the ASM5015 and the Provincedecided,
to drop the mode from its official test procedure. Thesefindings
leave serious questions about the viability and practicality of
the ASM5015 for actual I/M lane use and are discussed in the next
section.
Regardless of less-than-impressive preliminary findings, EPA
is pursuing the development of emission reduction credits for the
ASM tests and began performing ASM tests in Mesa, Arizona on
September 14, 1992 (although data from these tests were
unavailable for the analyses in this report). The test procedure
being used in Arizona was discussed and agreed to by
representatives of ARCO, the Society of Automotive Vehicle
Emission Reductions, Inc. (SAVER - represented by Allen
Testproducts, Inc.), Sierra Research, and the California BAR. The
procedure includes the ASM5015, the ASM2525, a 50-mph steady-state
mode, and an idle test. In light of the experience in Vancouver,
EPA believes it is likely that a preconditioning mode or immediate
opportunity for a second-chance test will be necessary to avoid
false failures on this test. EPA's testing program is designed to
address this possibility. This testing will also -help assess
whether the ASM5015 is a practical test mode for an I/M program
lane. The test program in Arizona is similar to that used for
evaluating the IM240, where vehicles coming to the station for a
regular I/M test are also given the test sequence under evaluation^
and an IM240. Vehicles will be recruited for FTP testing at a
contractor lab. EPA also plans to evaluate the performance of the
test in ensuring adequate repairs. At this point, sufficient data
are not available to determine the emission reduction benefits for
this four-mode test.
9. 2 Current Analysis of Available Data on ASM Tests
EPA. has completed an analysis of the available ASM data,
using a database of 31 vehicles. The data were gathered from
programs performed by three different organizations: Environment
-114-

-------
Canada19, Sierra Research20, and ARCO Products21. As stated above,
EPA started performing ASM tests in Mesa, Arizona on September 14,
1992, but these data were unavailable in time for this analysis.
Detailed discussions of this database and EPA's analysis follow in
the subsequent subsections of this report.
The small sample, the lack of representativeness, and the
fact that these are laboratory data would normally lead EPA to
hesitate making any comments until additional information is
available. There is intense interest, however, in the ASM tests;
so, limited, preliminary findings are included for the sake of
this report. As mentioned previously, EPA plans to have a more
complete analysis prepared by the end of the calendar year and
will be in a position at that time to say something more
definitive about the ASM tests. Not only will more EPA data be
available, but also data from Vancouver and California.
In brief, the two-mode ASM tests have been found to be
considerably less well correlated with the FTP than is the IM240
under controlled laboratory conditions, as evidenced by subjective
analyses of the scatter plots (see Appendix M) and objective
measurements using the standard error statistic. Testing at real-
world I/M lanes will add considerably more variability to both ASM
and IM240 tests because of conditions known to affect emissions
such as temperature, humidity, and vehicle operating conditions
prior to the test. Variability on the ASM or.IM240 test will cause
a reduction in the quality of the correlation with the FTP test.
For the IM240, lane-to-FTP data is available . and demonstrates good
correlation.	The uncontrolled lane variables may add
proportionally more variability to a steady-state test like the
ASM, but not enough data has been accumulated to confirm this
hypothesis. It is possible, however, that the loss in correlation
due to increased variability associated with actual I/M testing may
be somewhat offset for the ASM by adding two additional modes; a 50
mph steady-state mode at road-load horsepower, and an idle mode.
Of course, it is also possible that these additional modes may
19 Ballantyne, Vera F. Draft. Steady State Testing	Report	3 ft <3	Data,
Environment Canada, August 28, 1992.
20 Austin, Thomas C., Sherwood, Larry, Development of Improved Loaded-Mode
Te3t Proceduresfar	Inspection and Maintenance Programs. Sierra Research,
Inc. and California Bureau of Automotive Repair, SAE Paper No. 891120,
Government/Industry Meeting and Exposition, May 2-4, 1989.
Boekhaus Kenneth L., et al . Evaluation of Enhanced Inspection
Techniques	OH	State-of-the-Art Automobiles. ARCO Products
Company Report, May 8,1992.
-115-

-------
contribute error-of-commission problems of their own. This four-
mode ASM procedure is currently being performed by EPA as part of
the Mesa, Arizona I/M test program, and EPA looks forward to having
a better database in the near future. Once an adequate database is
available, emission reduction credits can be assigned and official
test procedures established.
Although not part of this analysis (due to a lack of FTP
testing capability at the time of the pilot program) the
experience of the Vancouver pilot program provides some very
telling information regarding the ASM tests . Vancouver, British
Columbia began official, mandatory testing in its I/M program on
September 1, 1992 after several months of pilot testing its four-
mode test in the actual I/M lanes. The Vancouver program was
designed to include the ASM5015 and the ASM2525 along with idle
and 2500 rpm modes. This pilot program represents the first time
ASM tests have been used in an actual I/M program setting.
Unfortunately, as previously mentioned, Vancouver's lab was not in
operation in time to do FTP tests on any of the vehicles that were
run through the trial program.
Problems with the ASM5015 reportedly became apparent during
the pilot phase of the Vancouver program and, ultimately, the test
was dropped as an official test procedure. Information from
Vancouver indicates that the inspection contractor's drivers were
having great difficulty maintaining the 15 mph cruise within the
1.5 mph required for the ASM5015 (intuitively, driving a steady
15 mph against substantial load on a dynamometer with low inertia
would be difficult) . It was reported that vehicles with small
engines produced excessive engine lugging and spark knock.
Drivers had difficulty selecting the smoothest-running gear on
vehicles with manual transmissions. Vancouver also experienced
problems with suspiciously high failure rates on the test. For
example, 1992 model year vehicles were failing at rates of 8%
according to data supplied by the Province using extremely loose
NOx emission standards. While no FTPs could be done to verify
that nothing was wrong with these vehicles, EPA's experience in
Hammond, Indiana showed no NOx failures among 1991 and/or 1992
model year vehicles. It is therefore likely that these were false
failures. Vancouver decided to drop the ASM5015 from the test
sequence and to add preconditioning for all vehicles.
British Columbia officials also reported that false failures
were a problem across the board with the test procedure, probably
because all vehicles were not being preconditioned. Vancouver
added preconditioning to control the false failure problem. At
this point Vancouver is running the ASM2525, along with the 2500
rpm and idle tests, and the FTP lab is now in operation. EPA
looks forward to additional information becoming available on this
three-mode test procedure. The ASM2525 is very much like the
steady-state loaded test that EPA approved for I/M use in 1980.
Like the idle and 2500 rpm tests, EPA "believes this test has not
been very effective in identifying high-emitters and insuring
-116-

-------
effective repair. The ASM2525 was also reported in SAE paper
#891120 to be less effective at identifying high NOx cars. So, it
may be that the ASM2525 alone (or in combination with the 2500 rpm
and idle) will not be sufficient.
9.3 Alternative Puree Tests
Of the potential alternatives to EPA's recommended tests, the
one which has garnered the most attention is the suggestion by
some that steady-state loaded testing using a simple non-inertial
dynamometer (or a dynamometer with some small fixed inertia) can
be used to perform the purge check. EPA pursued transient testing
instead of steady-state because our best engineering and technical
judgement suggested that steady-state testing as a mechanism for
conducting the purge check would lead to higher errors-of-
commission, and, ironically, higher overall costs per ton of
emission reductions produced because each error of commission
would lead to extra costs for attempted repairs, retests, and
special administrative handling. If false failures are too
frequent, emission reductions themselves would be imperiled by
adverse public reaction and a skeptical and negligent attitude by
inspectors, administrators, and technicians. As expressed in the
draft of this report, the rationale behind the assumption that
higher errors-of-commission rates would result is the fact that
purge strategies vary from vehicle to vehicle, and the possibility
of developing a few-mode steady-state test that successfully
addresses this variety by catching each car in one of its purging
conditions is small to none. New analysis of test data supports
this rationale.
Figure L-l in Appendix L depicts instantaneous purge data
during the IM240 from the vehicles described in Table 9-1. All
vehicles passed the purge test. By comparing the top trace in the
figure, which represent vehicle speed during the IM240, to the
instantaneous purge rates, it is clear that different vehicle
purge systems respond differently to the same operating mode.
Test vehicles 238 and 393 behave somewhat similarly in that the
purge is generally initiated during accelerations, and is
generally maintained during the reasonably steady-state portions
of the IM240 (i.e., between 60 seconds and witness line #3, and
between 140 seconds and witness line #5). Vehicles represented by
these tests would would be expected to pass a steady-state purge
test rather easily. However, it is clear that the calibration of
the design in test vehicle 238 uses almost double the purge flow
rate of the design in test vehicle 393.
In contrast to test vehicles 238 and 393, test vehicle 354
shows a greater degree of purge sensitivity to speed changes, and
turns off or reduces purge flow under some conditions to a greater
extent than test vehicles 238 or 393. Test vehicle 118 appears to
be extremely sensitive to acceleration, and seems to act almost in
an on-or-off mode. It is particularly important to note that
during steady-state operation from about 70 seconds to witness
-117-

-------
line #3, the purge flow in test vehicle 118 drops to very low
levels. Similar performance is also noted between 140 and 165
seconds for this test. Whereas a vehicle with a purge design
similar to test vehicle 354 would likely pass a steady-state test,
it would be more difficult to make such a judgement on vehicles
with a purge design similar to test vehicle 118 - particularly if
the calibration of the design operating like test vehicle 118 used
a lower flow rate during steady-state operation.
Table 9-1
Purge Vehicle Descriptions
Teat Veh. # Mod Yr Make	Model	Purge Vol	(1)
118
' 87
Nissan
Sentra
56.2
236
'88
Ford
Taurus
7 . 1
238
CD
Chev
Sprint
178. 0
354
' 91
Plym
Acclaim
43.7
393
' 87
Mits
Tredia
25 . 4
427
<33
00
Line
Cont11
00
rM
The most marked difference in purge design is apparent in
test vehicles 236 and 427. Neither test vehicle exhibits any
significant flow until well after 150 seconds. Prior to 150
seconds, test vehicle 236 exhibits a series of spikes with
extremely low flow typically at the end of an acceleration, and
the purge system appears to respond to the slight variations in
speed during the steady-state portions, but again with extremely
low flow. In the case of test vehicle 427, a purge delay or warm-
up timer might be assumed to be the cause for the delay of
significant purge flow. However, this car shows practically zero
purge flow in the steady-state section between 140 to 165 seconds
after some purge flow is evident earlier. Even more telling is
the fact that the engine size, engine family, and evaporative
family is the same between test vehicles 236 and 427. The only
difference is that the evaporative systems have different
calibrations.
The difference in these calibrations is highlighted in Figure
L-2 in Appendix L. Whereas Figure L-l represented instantaneous
purge flow, Figure L-2 shows the accumulation of the instantaneous
rates over time. For test vehicle 236 all of the little spikes
add up so that the vehicle exceeds the one liter outpoint by about
70 seconds, and the total flow accumulated is around 7 liters. On
the other hand, test vehicle 427 does not exceed the cutpoint
until around 140 seconds, and accumulates over 18 liters.
Recognizing that these cars were certified to a cycle similar to
the IM240, it is clear that the calibration engineer made
conscious trade-offs between timing of the flow and accumulated
volume over the cycle to meet the new certification standard.
Further, as an indication of different design philosophies, a
vehicle "with only a marginal increase in accumulated flow (test
-118-

-------
vehicle 393 in Figure L-2) over test vehicle 427 exceeds the purge
cutpoint in about 15 seconds on the IM240. Although vehicles that
require extended time to begin purging may represent a measurable
portion of the fleet (i.e., both samples were Ford Motor Company
vehicles), most of the vehicles purge fairly quickly (i.e., in the
first 30 seconds) on the first acceleration of the IM240, and
therefore, extended purge vehicles should not significantly affect
average IM240 test time when employing fast purge algorithms.
Clearly, purge strategies vary substantially among existing
vehicles. The degree of difference among existing designs is such
that no one steady-state test could avoid falsely failing some
vehicles. It might be possible to add an acceleration mode to a
steady-state test, but to insure proper test consistency, the base
inertia of all dynamometers used throughout the country would need
to be exactly the same, and a prescribed acceleration profile
would need to be maintained (probably with a video monitor) .
Adding these two quality control features would increase the cost
of the steady-state purge dynamometer, making it comparable to the
IM240 dynamometer. In addition, the acceleration test on the
steady-state dynamometer would lengthen the average test time. In
any event, no data is available on any specific steady-state
acceleration test that would allow an informed judgement to be
made.
Since EPA does not dictate design strategy, and because new
vehicles will be required to meet additional, evaporative
requirements for certification, EPA cannot predict the purge
strategies that might be used by vehicle manufacturers in the
future. The result of failing to address the full range of
current and future purge strategies in an I/M program is easy to
predict: Cars that should pass will fail, leading to unnecessary
expense and hardship for motorists, with no environmental benefit.
Clearly, using the IM240 - which is similar to the new car
certification test - is a prudent and conservative way to avoid
incorrectly failing cars that should pass. Given the lack of hard
test data on other possible approaches, EPA has no choice but to
proceed with the IM240 purge test as proposed for the purposes of
establishing the enhanced I/M performance standard.
Another purge test alternative has been proposed which calls
for a variation not on the test cycle, but on the test procedure
itself. In EPA's proposed purge test, a flow meter is inserted
into the evaporative purge line between the canister and the
engine. Some have proposed use of an alternative, tracer gas
technique. This alternative purge test strategy uses the
concentration of the tracer gas measured at some point down-
stream, and the known quantity supplied upstream to determine the
dilution of the injected gas. From the dilution of the known
quantity, the flow can be determined.
In this proposed alternative procedure, the known quantity of
tracer gas (helium) would be introduced into the gas tank through
-119-

-------
the gasoline filler neck. The down-stream measurement would take
place in the exhaust stream after it enters the CVS. Although
this technique is intriguing and elegant, there are several issues
that need to be considered. First, what is the detectable limit
of the tracer gas detector? Depending on the particular purge
system, after the tracer gas leaves the gas tank it has the
opportunity to be diluted to an unknown extent by the atmospheric
vent in the canister. During canister purging, the tracer gas is
again diluted by the engine intake air. If the car has a
secondary air system, the tracer gas gets diluted in the exhaust
system. And finally, the entire exhaust is diluted upon entering
the CVS. In each of these dilution steps the degree of dilution
will depend on the calibration of the entire emission control
system. As shown in Figure L-l, purge strategies can vary
significantly.
Given the multiple dilutions that occur, making a measurement
of purge volume comparable to the standard procedure (e.g., 1
liter 100%) would seem to be difficult. Among other things, the
accuracy of the amount of tracer gas injected would need to be
very precise. Some have suggested that any detection of the
tracer gas in the exhaust should be sufficient to indicate purge
flow. At this point, EPA has no data to support this contention.
In either case, however, the detectable limit would need to be set
sufficiently low to avoid falsely failing vehicles with low purge
flow designs, such as test vehicle 236 in Figure L-2. It should
also be pointed out that under a tracer gas scenario, multiple
dilutions could increase the amount of time necessary to determine
fast pass for purge.
On the vehicle side, consideration needs to be given to the
amount of inert tracer gas introduced into the gas tank.
Normally, there is a mixture of fuel and air in the gas tank, and
a fuel mixture or just air in the canister. The engine management
system is designed to handle both. However, if the inert tracer
gas displaces a significant quantity of mixture or air, the inert
tracer gas behaves as additional EGR, thus altering the engine
operation. As a result, tracer gas purge testing may have to be
performed separately from exhaust emission analysis for HC, CO,
and NOx, further lengthening the overall test time.
The final consideration is background levels of tracer gas in
the test facility. Normally, background levels of helium are very
low; But, with the multiple dilutions in the system, measurement
levels may approach background levels, particularly if the test
itself contributes to the background. This could occur after the
tracer gas is introduced into the system and the gas cap is re-
sealed, if during the driving cycle, the pressure in the fuel tank
increases (because of temperature increases), and the purge valve
shuts off (see Figure L-l). In this case, the fuel-air mixture in
the fuel tank would flow to the canister, where the fuel would be
retained, and the air, including the tracer gas, would e'xit the
atmospheric vent in the canister. Air flow from the cooling fan
-120-

-------
would likely carry the tracer gas under the vehicle, and into the
mixing funnel%for the CVS. In fact, there could be less dilution
from the canister vent to the CVS, than in the path through the
engine. In this scenario, the potential for passing a car with a
completely inoperative purge valve seems high.
Two similar alternatives have been suggested for the pressure
test. The pressure test as proposed involves locating the fuel
tank vent line at the canister, disconnecting it, and pressurizing
the fuel tank though the vent line. After pressurization, the
amount of leakage is determined by monitoring the pressure drop
over two minutes. If the pressure drop is less than allowed, the
system passes. Given the intrusive nature of the test procedure,
commenters have expressed concerns about the ability of an
inspector to find the canister, whether there is physical access
to the canister, and potential damage that could occur during
removal and re-attachment of the vent line.
Both alternatives to EPA's proposed pressure test involve
pressurizing the gas tank through the filler neck with a special
adapter. In one case, the helium used for an alternative purge
check would also be used for the pressure check, and a probe would
sniff for helium around and under the car. A concern with this
alternative is that the degree of leakage is not quantifiable.
Additionally, the helium molecule is much smaller than diatomic
nitrogen (N2) . Therefore, the size of the leak detected by the
helium would be significantly smaller than than that detected by
N2  The fact that this alternative would not provide a
quantifiable measure of the leak could lead to the improper
identification of inconsequential leaks (i.e., false failures).
Furthermore, this procedure appears to require an operator to
manually probe around the cars to detect leaks, thus reintroducing
the potential for human error in the test results and violating
the Clean Air Act's requirement that testing procedures be
computerized.
Another proposed alternative to EPA's pressure test procedure
also uses the filler neck as the avenue for pressurizing the
evaporative system. However, this alternative uses diatomic
nitrogen, and monitors the pressure drop over the specified time
interval. This system has some apparent advantages, but upon
closer inspection, they are illusory. The first apparent
advantage is that by pressurizing the system through the filler
neck, the inspector does not need to locate the canister. This is
not true. To be able to pressurize the system with this
alternative the canister must be located, and the vent line
plugged or pinched-off. If the line is plugged, the vent line had
to be removed, and so the system could just as easily be
pressurized from the vent line. If the line is to be pinched-off,
there are several considerations. Typically, vise-grip type
pliers would be used. If the canister- is difficult to get to in
the first place, there may also be a problem in having sufficient
clearance-room to actuate the handles of the pliers. Secondly, if
-121-

-------
the pliers do not completely close-off the vent line, this could
result in a false failure. In addition, some systems use plastic
lines with rubber nipples at the. ends (i.e., at the tank and at
the canister). Attempting to pinch a plastic line could easily
crack it, and because plastic lines are generally not easily
deformable, the seal would be questionable. Furthermore, the
outcome of the test is more subject to operator influence (i.e.,
how good is the seal) than is EPA's proposed test procedure.
Another issue to consider is that each lane will need to
maintain a series of filler neck adaptors to accommodate various
cars. Some have suggested that only 6 adaptors may be needed.
However, the inspector will still need, to make a judgement in
selecting the proper adaptor for each car.
Finally, there is the question of the interface between the
gas cap seal and the vehicle's filler neck. On older cars,
particularly in northern climates, the filler neck can become
corroded leaving a rough sealing surface. If the seal in the
mating gas cap is also weathered and non-compliant, a leak in the
system can occur (leaks around the gas cap are a common cause for
pressure test failures). Such a leak would not likely be detected
when testing the components separately with special adaptors. On
the filler neck side, the adaptor would generally have a new
compliant seal that could conform to the corrosion pits in the
filler neck. And on the gas cap side, the non-compliant seal
would more likely seal on a smooth adaptor surface.
Of all of the alternatives to the evaporative tests proposed
by EPA (i.e., the steady-state loaded-mode purge test, and the
tracer gas purge and pressure tests), the only one which appears
to warrant more study is the pressure test which uses diatomic
nitrogen introduced through the filler neck. Nevertheless, EPA is
open to demonstrations by states or their representatives that
proposed alternative testing strategies are equal or superior to
EPA's proposed tests in terms of identifying excess emissions and
keeping false failures to a minimum.
9.4 Alternative N0X Testing
Section 182(c) (3) of the Act requires that programs in
enhanced I/M areas achieve NOx reductions. EPA has found that N0X
emission testing (as opposed to visual inspection of emission
control devices) is essential for NOx emission reductions.
Some have suggested that a heavier loaded, steady-state test
(i.e., one using a heavier load than the EPA-approved steady-state
loaded test currently being used in Arizona) is an adequate
alternative to transient emission testing for N0X In particular,
ARCO and others have proposed that an ASM test be allowed in lieu
of the IM240 exhaust .test. As noted previously, the ASM concept
was first publicized in SAE paper #891120, by Austin and Sherwood,
and was intended primarily as a method to improve the
-122-

-------
effectiveness of no-load I/M procedures by providing a method for
measuring NOx. Also as previously noted, in 1992, the Province of
British Columbia began a pilot program utilizing the ASM test
prior to official implementation of an I/M program in Vancouver
for HC, CO, and NOx.
As it has currently evolved, the ASM concept involves
operating a car at lower vehicle speeds <15 or 25 mph) while
loading the vehicle at a fraction of the inertia load needed to
accelerate the vehicle at 3.3 mph/sec^ plus the windage load at
the test speed. The 3.3 mph/sec^ acceleration is the maximum
acceleration that occurs on the transient test used to certify new
cars. The ASM modes are designated by the fraction of the load
and by the test speed (i.e., ASM5015 represents 50% of the inertia
load for a 3.3 mph/sec^ acceleration at 15 mph). The SAE paper by
Austin and Sherwood concluded that the current 2 500 rpm/ldle test
was better than the ASM test in identifying HC and CO emitters,
and that the only benefit of the ASM test was for N0X. Subsequent
data and comments provided to the EPA support this earlier
conclusion.
An issue with the ASM proposal arises from the requirement
under Section 182(c)(3) of the Act that programs in enhanced I/M
areas must achieve NOx benefits. A question EPA must evaluate is
whether the ASM adequately identifies high NOx emitters to the
extent that NOx benefits can be quantified, and whether the ASM
falsely fails low NOx emitters.
It is claimed that the ASM more heavily loads the vehicle
than other steady-state tests, and that this heavier loading
results in the ability to test for NOx. The load for the ASM test
is determined by dividing the inertia weight of the vehicle by a
constant. A separate constant is used for each of the two ASM
modes proposed (i.e., the ASM5015 and the ASM2525) . Figure L-3
and Figure L-4 show the relationship of load versus speed for the
ASM, the EPA steady-state loaded test, and the IM240 for a 2,200
pound vehicle and a 3,000 pound vehicle. For a 2,200 pound
vehicle, which would likely have a 3 or 4 cylinder engine, the
ASM5015 clearly would load the vehicle more than the EPA steady-
state loaded test, and would require the vehicle to meet the load
at a lower speed. The ASM2525 would also load the vehicle
somewhat higher, but at the same speed. It is also clearly
evident that the IM240 loads the vehicle much greater than either
the ASM or EPA's steady-state loaded test.
For a 3,000 pound vehicle (Figure L-4) which will likely have
a 6 to 8 cylinder engine, the ASM5015 load is only marginally
higher than the upper limit for the EPA steady-state loaded test,
and the ASM2525 is effectively the same as the Arizona load. As
with the 2,200 pound car, the IM240 loads the vehicle
significantly greater than either the ASM or the EPA steady-state
loaded test.
-123-

-------
The load imposed on a vehicle is not the only factor in its
N0X production; also important are the rates at which the load and
speed change, and the NOx control strategy used for the vehicle.
The instantaneous second-by-second NOx emission (gpm) data in
Figure L-5 helps identify which operations in the IM240 cycle
produce N0X. Clearly, all of the vehicles described in Table 9-2
produce NOx during acceleration. However, it is equally clear
that under steady-state conditions similar to those encountered in
the ASM (i.e., segments 1 and 2), NOx is particularly low. In
nearly all cases the average N0X over these steady-state portions
is below a cutpoint of 2 gpm.
It should be noted that the time interval for segments is
around 10 to 15 seconds (which might be a typical measurement
window for an ASM test) after the emissions from vehicle have
stabilized at the specified test speed.
Table 9-2
NOx Vehicle Description*
Test #
Model Yr.
Make
Model
HC (aom)
CO (aom)
NOx (apm)
238
' 86
Chev
Sprint
1. 07
32 . 90
0.87
343
'86
Ford
Escort
0 .13
0.50
4 .55
393
' 87
Hits
Tredia
0 . 37
1. 90
2.93
435
' 86
Honda
Accord
0.76
9. 00
2. 93
461
88
Pont
Grd Am
0.16
4. 10
1. 64
*A11 gpm numbers are IM240 measurements
In reviewing the NOx performance of the vehicle represented
by test. 393 in segment 1 and 2, it is difficult to distinguish
test 393 from tests 238 or 461. In fact, the accumulated NOx over
segment 1 for tests 238 and 461 clearly exceeds that in test 393.
It is less clear when making this comparison in segment 2 .
However, the important point is that while test 393 produced
nearly 3 gpm of NOx over the IM240 cycle, both tests 238 and 461
were well below the 2 gpm cutpoint for the IM240 (the NOx measured
for test 238 was 0.87 gpm; for test 461, 1.64 gpm). In
reconciling the differences between test 393 and the two passing
tests in IM240 NOx emissions, it is obvious that heavy
accelerations were the major cause for the differences.
Another interesting comparison is that while the complete
IM240 produced over 4.5 gpm NOx in test 343, the NOx performance
in segments 1 and 2 would suggest that a steady-state test would
result in only around 2 gpm, or less than half the NOx produced
during a full transient test. Here again, the heavy accelerations
contributed the most NOx in this test.
The fact that accelerations contribute the most to NOx
production should not come as a surprise. Granted, accelerations
-124-

-------
require that the engine put out more power than a steady cruise.
However, the response of the feedback control system and its
sensors can also have a significant effect. For instance, a slow
oxygen sensor that has recognized a deceleration with a resulting
deceleration lean-out, might not immediately recognize a following
acceleration, resulting in a lean condition at the start of the
acceleration and higher-than-normal N0X emissions. Such a
condition would not be identified by a steady-state test like the
ASM, because the vehicle would be operated at one speed long
enough for the sensor to catch-up, which would not be the case in
real-world driving. In other cases, the duration of the ASM
steady-state test would generally be sufficiently long for the
feedback feature in the emission control module (ECM) to "learn"
how to be clean. On the mechanical side, a partially plugged EGR
passage could allow sufficient flow at the lower speeds of the ASM
to pass the ASM cutpoint, but restrict the EGR flow necessary for
the heavy acceleration at about second 160 in the IM240.
Based on the evidence, it appears that it would be unlikely
that a steady-state test can fully characterize the NOx
performance of in-use vehicles. Therefore, it would be difficult
for EPA to quantify the NO>: reductions from such tests without
additional data.
Another issue that needs investigation is the possibility of
errors of commission resulting from use of the ASM test. The
standards for the ASM5015 proposed in the SAE paper by Austin and
Sherwood were concentration-based standards, and were based on a
2% error-of-commission rate relative to the FTP.	The
concentration value (in ppm) of the standard was determined by
dividing the inertia weight of the vehicle into the constant,
753x(10)3. In developing this equation, data from fifteen 1982
and later closed loop cars, along with 3 mid- to late-1970s open
loop vehicles were used. Using this equation would result in a
concentration standard of 228 ppm for a 3,000 pound vehicle (3,000
pounds curb weight plus 300 pounds).
When the Vancouver ASM study program began, constants of
3100x(10)3 and 2650x(10)3 were used for the ASM5015 and ASM2525,
respectively. These new equations resulted in concentration
standards of 939 ppm and 803 ppm. These concentrations represent
more than a four-fold increase over the original proposal.
However, after testing more than 7,000 vehicles, the program
office determined that even these standards, when combined with
other failure modes (e.g., HC and CO) would result in an
unacceptable overall failure rate - particularly since the program
office did not have its FTP lab operational, and could not confirm
the NOx failures.. Therefore, prior to implementing the official
I/M program on September 1, 1992, the program office revised the
NOx standards, effectively setting a NOx cutpoint of 1000 ppm as
the minimum standard for three-way catalyst equipped feed-back
cars. Using the 1000 ppm NOx standard, and the equation presented
-125-

-------
in the SAE paper for estimating ASM concentration in gpm would
result in a NOx level of 9.55 gpm for a 3,000 pound vehicle. Even
considering that the SAE conversion equation possibly
overestimates NOx mass emissions, the Vancouver NOx benefits are
likely to be small with a 1000 ppm cutpoint.
In addition to revising the standards, the program office in
Vancouver dropped the ASM5015, and only retained the ASM2525.
There were a variety of reasons for dropping the ASM5015.
Analyses of the Vancouver vehicles indicated little difference in
NOx failure rates between the ASM5015 and the ASM2525. This
observation is contrary to the observation by Austin and Sherwood
that the ASM5015 was clearly superior to other ASM modes in
finding high NOx emitters. A 1992 report by Boekhaus, Sullivan,
and Gang of ARCO also reached a similar conclusion. Quite
possibly the high concentration standards used in the Vancouver
program account for the difference. Austin and Sherwood used NO>;
concentration standards in the 200 ppm range, while ACRO used a
NOx standard of 0.7 gpm.
However, even at the high standards used during the study
period (800 to 900 ppm), the .Vancouver program office reported
that 9 of 112 (or 8%) of 1992 model year cars tested during the
study failed for NOx  Anecdotal information on calls from the
public and new car dealers to the program office commenting that
nothing was apparently wrong with a relatively new vehicle which
failed NO;i, suggests that some of the late model NOx failures
could be false failures. Even one of the cars tested by ARCO, a
1992 Chevrolet (on an ASM2535 mode) , would have been a false
failure, and would have failed the 1000 ppm Vancouver standard,
even though that car registered only 1.75 gpm on the IM240 (1.5
gpm on the FTP) . The program office suggested that a possible
cause for the late model failures could have been due to extended
idling or engine shut-down in the test lanes. However, while a
similar situation existed in the IM240 lane (and at a 2 gpm IM240
standard) no recorded NOx failures for 1991 or 1992 model year
cars have been observed at this point in time.
In addition to the recorded late model NOx failures in the
Vancouver study, the program office indicated that it was
sometimes difficult to stay within the 1.5 mph window at 15 mph
with the ASM5015 load, although this is the same tolerance that is
used in the Arizona I/M program at higher vehicle speeds. Also,
in some cases, vehicles with small engines produced excessive
engine lugging and spark knock which could disturb the public -
particularly if the inspector selected an incorrect gear for
testing with manual transmissions. Because of the suspected
vehicle cool down in the lane, the potential lugging and pinging
problem, and anecdotal evidence that simply replicating the first
ASM mode (which happened to be the ASM5015) improved the chances
of passing, the Vancouver program office elected to substitute a
preconditioning mode for the ASM5015 test mode-.
-126-

-------
The second-by-second N0X traces in Figure L-5 clearly show
that the majority of the NOx is produced during acceleration, and
that N0X levels can be fairly low under steady-state conditions
for even dirty cars. Under such conditions it appears that it
would be difficult to discriminate between dirty N0X cars and
clean ones. This perceived difficulty in discrimination is likely
at the heart of the problems encountered in the Vancouver program.
The evidence of false ASM N0X failures in the ARCO data (when
realistic cutpoints are applied) simply serves to confirm this
hypothesis.
Based on this evidence, it appears unlikely that a steady-
state test can fully characterize the N0X performance of in-use
vehicles, and it would be inappropriate for EPA to consider
substituting the ASM for the IM240 at this time. Nevertheless, as
indicated earlier, EPA is open to demonstrations by states or
theirrepresentatives that proposed alternative testing strategies
are equal or superior to EPA's proposed tests in terms of
identifying excess emissions and keeping false failures to a
minimum.
9.5 Repair Grade IM240 Testing
The argument has been made that high-tech testing will have
limited success due to the fact that I/M programs will still need
to ensure successful repairs to net the emission reduction
benefits of the program. One complaint is that by separating
testing and repair, and introducing a costly test procedure, EPA
is making it impossible for repair facilities to confirm the
effectiveness of their repairs, and, in effect, is requiring the
repair industry to perform repairs in the dark. One rationale for
trying to develop cheaper alternative tests is, in fact, to fill
this diagnostic and confirmatory testing niche.
In response to this clear need, EPA is developing an
inexpensive repair-grade IM240 emission measurement system. This
repair-grade system is primarily designed to aid the service and
repair industry in verifying repair of vehicles which have failed
an official IM240 emission test. This equipment is designed to
provide an approximate measurement of IM240 mass emissions levels.
By measuring the vehicle's emissions before and after vehicle
repairs, the mechanic can determine the direction and approximate
magnitude of any changes in mass emission levels.
The current, direction of the repair-grade system is based on
a chassis dynamometer with inertia weights, an exhaust dilution
system, a BAR90 analyzer, and an appropriate computer and
software. The dynamometer will have a fixed inertia weight of
2,500 pounds with additional dynamic inertia provided by the power
absorber (if available, a function of speed and absorber type).
Because of installation concerns, only electric power absorbers
will be evaluated. Two exhaust dilution systems are being
-127-

-------
evaluated as part of this diagnostic system. The first emission
dilution system uses a 100 standard cubic feet per minute (SCFM)
critical flow venturi for a flow controller and the configuration
of the system is similar to a laboratory type unit . The second
emission dilution system uses a squirrel cage type blower and low
velocity air flow in order to reduce power requirements. The
trade-off of the second system is that while the air flow would
not be strictly constant, it would be assumed to be constant for
calculation purposes, with some error resulting. The estimated
costs for the individual components of this equipment system are
listed in Table 9-3.
Table 9-3
Estimated Costs for Repair-Grade IM24Q Emission System
ITEM	Mew Equipment	Retrofit
Dynamometer	$14,000.00	$14,000.00
386-based BAR90 w/ extras	$15,000.00	$3,000.00
CVS Venturi	$1,800.00	$1,800.00
CVS Blower and Motor	$2,000.00	$2,000.00
Squirrel Cage Type Blower	$500.00	$500.00
Tubing for dilution system	$600.00	$600.00
Total with CVS:	$33,400.00	$21,400.00
Total w/ Squirrel Cage $30,100.00	$18,100.00
Type Blower:
Emission analysis of the diluted sample is performed by
either a BAR84 or BAR90 emission analyzer. The emission analyzer,
which operates with either of the above dilution systems, samples
and analyzes the diluted flow and transmits the information to the
computer. For the CVS system, the computer calculates the
instantaneous and average emission values, using the flow
conditions, which are then stored in a file for later use. At a
minimum, an 80386-based IBM-compatible computer is required to
perform the computational and control functions for the equipment
system. The squirrel cage system would not require instantaneous
flow measurements to be calculated, but would still require
emission measurement computations during the test cycle.
The dollar figures in Table 9-3 are based on start-up
numbers; mass production of these items is expected- to
significantly lower costs. For example, the individual cost of a
dynamometer in a very large order, or for a market known to be
very - large, might be below $10, 000. The BAR90 estimates are
slightly higher than current street prices, but the high estimate
is expected to cover the additional cost of special programs and
driver cards for integrating the sample, computing the CVS flow,
interfacing with the dynamometer, and providing a drivers' aid.
The cost to retrofit a BAR90 unit, if a service facility already
has one in use, should be only about $3, 000.00 (a savings of
approximately $12,000.00 on the estimated new equipment price)
Other costs (for example, the cost quoted for a squirrel cage
-128-

-------
blower) are based on our purchase costs or prices obtained from
supply catalogs such as W.W. Granger. Because BAR grade analyzers
currently measure only HC and CO, a N0X channel will need to be
added. Currently, a fuel-cell type NOx analyzer would add between
$2K and $6K to the system ($2200 for ESP, and $5900 for Allen),
although this cost range is not reflected in the above estimates.
Because of the interest in the ASM test, a comparison of
repair-grade equipment costs based on the two tests has been
developed. In determining the price difference between equipment
for IM240 repairs and ASM repairs, it is assumed that the ASM
equipment will include the same analyzer as in the IM240 set-up
(i.e., a BAR90 with NOx capabilities) and a dynamometer, but would
not include a CVS unit. The lack of a CVS unit would save between
$1,100 and $4,400. The dynamometer would be somewhat simpler than
the IM240 which would have a base inertia of 2,000 pounds.
Compensating for the lower base inertia in the ASM dynamometer
might save $1,000. Additionally, the ASM equipment would not
require as extensive a software upgrade in the BAR90 as the IM240
equipment, but would still require significant upgrades. The
software savings may only be around $1,000.
Compiling the numbers (using the values in Table 9-3), the
estimated price to upgrade an existing BAR90 for repair-grade
analysis with NOx is between $20.3K and $27.3K. Subtracting the
savings in the previous paragraph for ASM would result in a range
of estimates for the ASM repair equipment between $17.2K and
$20.9K. Thus, the reduction in price to upgrade BAR90 repair
equipment for ASM as opposed to IM240 could be as low as $3,100 or
a high as $6, 400 .
If a BAR 90 analyzer were not available for upgrading, adding a
BAR90 unit would increase the price about $12K, but the increased
price would apply equally to IM240 and ASM repair-grade equipment.
Adding a BAR90 analyzer to the ASM upgrade price estimate would
result in a price range of $29, 200 to $32, 900. This range
compares favorably with the price of around $30,000 for a BAR90
(w/o NOx) with a dynamometer (somewhat comparable to that which
would be expected in ASM repair-grade equipment) that is currently
marketed in limited quantity in Florida.
-129-

-------
Appendix	A
Appendix	B
Appendix	C
Appendix	D
Appendix	E
Appendix F
Appendix G
Appendix H
Appendix I
Appendix J
Appendix K
Appendix L
Evaporative Emissions and Running Loss Emission Factor
Derivation
Purge and Pressure Test Effectiveness Figures and
Spreadsheet
Exhaust Short Test Accuracy: IM240 vs. Second-chance
2500 rpm/ldle Test
MOBILE4.1 Technology Distribution and Emission Group
Rates and Emission Levels
Regression Analyses and Scatter Plots for Fuel
Injected 1983 and Later Vehicles
IM240 Cutpoint Tabel Analysis
Evaporative System Purge and Pressure Diagrams
Evaporative System Failures and Repairs
MOBILE4.1 Performance Standard Analyses, By Option
Identifying Excess Emitters with a Remote Sensing
Device: A Preliminary Analysis
Model Year Failure Rates by Test Type
Comparative Purge Flow Data
-130-

-------
APPENDIX A
EVAPORATIVE EMISSIONS AND RUNNING LOSS EMISSION FACTOR
DERIVATION

-------
EVAPORATIVE HC EMISSIONS
i.0 Int roduct ion
Th'ere have been more vehicles tested by EPA's Emission Factors
Program (EFP) since the issuance of M0BILE4. From the EFP test
facility located at Motor Vehicle Emissions Laboratory (MVEL) in
Ann Arbor, Michigan (EFP-MVEL), EPA continues recruiting in-use
vehicles and testing them with the certification test procedure --
a diurnal test, followed by exhaust emissions test on a Federal
Testing Procedure (FTP) cycle, and then a hot soak test. A
certification test fuel with fuel volatility level of 9.0 psi Reid
Vapor Pressure (RVP) is used. Two additional fuels with 10.4 and
11.7 psi RVPs were also used during FY84 through FY89 to
characterize the evaporative emissions effect due to different fuel
volatility levels that were commercially available Cor in-use
vehicles. Since FY90, the 9.0 psi RVP fuel has been the only
gasoline fuel used in the EFP-MVEL testing.
EPA added "Hammond Program" (EFP-Hammond) to the so-called
MVEfL "traditional" EFP during FY90. A brief description of this
new program is given in section 1.1. Due to the addition of this
new test program, a different modeling approach is used for
MOBILE4,l's evaporative diurnal and hot soak emissions.
For example, in the MOBILE4 evaporative diurnal and hot soak
emissions model, there were seven elements of evaporative
emissions: standard level, emissions due to insufficient canister
purge, emissions due to excess .fuel volatility, problem-free
emission levels, emissions due to malmaintenance and/or defect,
non-tampered emission levels, and emissions from tampering. The
problem-free emission rates were the sum of the first three
elements. The non-tampered emissions were the emissions from
problem-free vehicles plus the emissions from malmaintenance and/or
defect. Finally, the in-use evaporative diurnal and hot soak
emissions were estimated by accounting for emissions from both the
non-tampered and tampered fractions of the vehicle fleet.
However, in the MOBILE4.1 evaporative emissions model, there
are three different types of evaporative HC emissions: emissions
from pass vehicles, emissions from vehicles that failed pressure
test, and emissions from vehicles that failed purge test. These
three types of emissions are defined for both diurnal and hot soak
emissions in MOBILE4.1.
-1-

-------
L . 1 Hammond Program
EPA's Hammond program was initiated in FY90. This program
utilizes the testing facilities for the State of Indiana's
Inspection and Maintenance (I/M) programs located at Hammond,
Indiana. While waiting for the required I/M tests, in-use vehicle
owners were asked to participate additional tests defined by EPA's
new EFP. One of the tests those I/M vehicles were asked to
participate was a transient cycle called "IM240," which is a
shortened version of the certification FTP cycle. All
participating vehicles were also tested for their evaporative
emissions control system to see if there exists either pressure
and/or purge failure.
The pressure test is performed by using a gauge to measure the
pressure between a vehicle's fuel tank rollover valve and
evaporative canister, by introducing nitrogen at 14" of water
pressure. Two minutes after the initiation of this test, if the
evaporative system holds less than 8" pressure, the vehicle is
considered to have pressure problem in its evaporative emissions
control system, thus failing the pressure test.
The purge test is performed by using a flow meter measuring
the purge air between a vehicle's canister and engine. The
cumulative purge ait over the entire IM240 cycle is measured. If
the flow rates are less than 1.0 cubic liter, the vehicle's
evaporative emissions control system is considered to have
insufficient purge, thus failing the purge test.
As of January 30, 1991, 2,497 I/M vehicles were tested for
potential pressure and purge failures under the Hammond program.
The model years of these in-use vehicles ranged from 1976 to 1991.
Overall failure rates were: 12.7 percent for the pressure test,
11.5 percent for the purge test, and 21.1 percent for either.
These failure rates were analyzed and then characterized in terms
of vehicle's age, and are summarized in Table 1.
Some of the failed vehicles were also tested for their diurnal
and hot soak emission levels. The results seem to suggest that
emissions from vehicles that failed pressure test are similar to
the emissions from vehicles with no evaporative emissions control
system, and are somewhat higher than the emissions from vehicles
that failed purge test.
1.2 Modeling Approach
As mentioned in section 1.0, the MOBILE4.1 evaporative
emissions model include: emissions from pass vehicles, emissions
from vehicles that failed pressure test, and emissions from
vehicles that failed purge test.
-2-

-------
Test results obtained from EFP-MVEL were used to:
a)	characterize the fuel volatility effect on emissions,
b)	derive emission levels for pass vehicles, and,
c)	describe the emissions effects due to other parameters,
such as fuel delivery system, vehicle type, etc.
Results obtained from EFP-Hammond program were used to:
a)	define pass versus pressure and/or purge failure vehicles,
b)	estimate the emission levels due to pressure and/or purge
failures, and,
c)	see if the pressure/purge test results are applicable to
and/or are consistent with mechanics' visual diagnosis
check in EPA's EFP at MVEL.
In addition to test data, a theoretical vapor generating model
(based on Wade equation) was also used for the diurnal emissions
model to:
a)	determine the emissions effect from a combination of fuel
volatility, temperature rise, fuel tank capacity, and
tank fill level,
b)	describe the emissions from vehicles that failed either
pressure or purge test, and,
c)	define the emission levels from vehicles that have two or
more consecutive no-driving days.
The "pass" vehicles are defined as vehicles that are not
tampered, nor failing pressure/purge tests. Therefore, in terms of
the MOBILE4 evaporative emissions categories, the pass vehicles
include vehicles that have no visible nor measurable problems in
their evaporative emissions control system components, or problems
that do not appear to dijectly. relate to pressure leaks or purge
deficiency.
Table 2 provides a list of the criteria used in defining
pressure/purge failures as well as tampered vehicles from the
EFP-MVEL data base. As can be seen, "tampered" in the MOBILE4.1
model is now a subsample of the failure vehicles by definition. In
general, pressure failures indicate the potential existence of a
leakage on vehicle's evaporative emissions control system
components, which include: gas cap, filler neck, sending unit,
rollover valve, and vent hoses. Purge failures are usually caused
by an in-operative canister purge solenoid or valve, or
disconnected, missing, or damaged purge hoses.
-3-

-------
2.0 Diurnal Emissions
As in M0BILE4, the diurnal emissions {in unit of grams per one
hour test) are described as a function of Uncontrolled Diurnal
Index (
-------
DIpa si = a +
b * UDI + c *
UDIz

(2.1)
where:





Vehicle Type

Regression Coefficient

/Fuel System

a
b
c

LDGV/CARB

-4.3304
5 .9904
0.0

LDGV/PFI

-0.0879
0 . 0
1.4179

LDGV/TBI

-0.8233
0 . 0
2.0433

LDGT/CARB

-7.6803
9.7703
0 . 0

LDGT/FINJ

-3.8537
4.9437
0 . 0

The coefficients
f rom
equat ion
(2.1) were der i
ved from
EFP-
data, where the
three
UDI values corresponding
to 9.0,
10.4,
-MVEL
and
11.7 psi RVP fuel volatility levels were 1.0, 1.4567, and 2.0677,
respectively. The constant terms of equation (2.1) (denoted as "a"
above) were adjusted so that when UDI is 1.0 the predicted diurnal
emissions are equal to the average diurnal emissions at 9.0 psi RVP
from the EFP-MVEL sample.
For carbureted vehicles, the equation (2.1) is to be
extrapolated when the UDI values are greater than 2.0677. However,
when the, UDI values are less than 1.0, new sets of coefficients
listed below are used. These coefficients were derived from a
two-step analysis: first, all carbureted vehicle data were used to
fit an exponential function, then, the predicted emissions for UDIs
at 1.0 and lower were used to fit a linear regression line. Again,
the constant terms were adjusted so that when UDI is 1.0 the
predicted diurnal emissions are equal to the average diurnal
emissions at 9.0 psi RVP from the EFP-MVEL sample. In addition, a
lower bound of the diurnal emissions is set to equal to the resting
loss emissions.
Vehicle Type	Regression Coefficient	
/Fuel System	a	b	c
LDGV/CARB	-0.2888	1.9488	0.0
LDGT/CARB	0.1412	1.9488	0.0
For fuel-injected vehicles, equation (2.1) is to be
extrapolated when the UDI values are beyond the existing data
(i.e., UDI values are either less than 1.0 or greater than
2.0677). At low UDI values the calculated diurnal emissions may
become negative. A lower bound of the diurnal emissions is set to
equal to the resting loss emissions.
Graphic presentations of 1981+ LDGV/LDGT diurnal emissions as
a function of UDI for pass vehicles are shown in Figures 1 and 2.
-5-

-------
2.1.2 Failed Vehicles
To estimate the diurnal emission levels for vehicles that
failed either pressure or purge test, two steps of analyses were
used: 1) the theoretical vapor generating model, and, 2) data from
EFP-Hammond.
The theoretical vapor generating model was used to describe
the relationships between UDI and the emission levels from failed
vehicles (see Attachment). It is assumed that fuel delivery system
is not a significant parameter on failed diurnal emission levels,
although the emissions simulation was based on a late model-year
fuel-injection system. Uncontrolled diurnal emissions (in grams)
were calculated at various fuel volatility levels (from 7.5 to
11.0 psi RVP) for the temperature rise of 72 to 96F. These
calculated uncontrolled diurnal emissions were used for the
pressure failure levels. The purge failure levels were the
calculated uncontrolled emissions subtracted by a backpurge
effect. The g/test diurnal emissions for failed LDGVs are
expressed as linear functions of LTD I:
DIFa,,ea pressure = 2.2002 + 7.5626  UDI
DI F a i led Purge = -4.2897 + 7.4563 * UDI
At UDI = 1.7448 (72 to 96F temperature rise, 9.0 psi RVP fuel, and
40% tank fill), the diurnal emissions estimated from the above two
equations are 15.40 and 8.72 grams for vehicles that failed
pressure and purge test, respectively.
From EFP-Hammond data base, 30 LDGVs (as listed in Table 4)
were also tested for diurnal emissions under the conditions of 72
to 96F temperature rise, 9 psi RVP fuel, and 40% tank fill (i.e.,
also with UDI => 1.7448). These vehicles were found to have failed
either pressure or purge test at the Hammond I/M Lane. Their
average diurnal emissions were:
G/Test
Catego ry	N	Diurnal
Failed Pressure 12	27.73
Failed Purge	18	15.58
The average diurnal emission levels from EFP-Hammond were
compared with the estimated emissions from the theoretical vapor
generating model. The data versus model ratios of 1.801185
( = 27.73/15.40 for the pressure failure) and 1.786687
( = 15.58/8.72 for the purge failure) were the multiplicative
factors used to derive new diurnal emission equations for LDGVs
that failed either pressure or purge test:
-6-

-------
DI F a I led Pressure = 1.801185 * ( 2.2002 + 7.5626 * UDI)
= 3.9630 + 13.6216  UDI	(2.2)
DI|Talle. Purge = 1.786687 * (-4.2897 + 7.4563 * UDI)
= -7.6643 + 13.3221 * UDI	(2.3)
Note that equations (2.2) and (2.3) were derived based on a
sales-weighted LDGV fuel tank capacity of 16.2 gallons. The fuel
tank capacities of LDGTs are, in general, larger than the LDGVs.
Many of the LDGTs are also equipped with an auxiliary fuel tank of
a similiar capacity as the primary tank. Based on EPA's
Certification data base, the sales-weighted primary fuel tank
capacity for model year 1988 LDGTls was 19.0 gallons. When
including the auxiliary fuel tank, the sales-weighted capacity for
LDGTls became 20.8 gallons. Therefore, an average fuel tank
capacity ratio of 1.17 (19.0 gallon/16.2 gallon) is used to
estimate the diurnal emissions of failed LDGTs:
Dlfaiied LDGTs = 1.17 * Dlpjilea LDOVs	(2.4)
This approach implies that the diurnal emissions from failed
LDGTs are 117 percent higher than those from failed LDGVs. This
assumption is well-supported from EFP-MVEL test data. As can be
seen from Table 3, the average diurnal emissions.from failed LDGTs
are 125 percent higher at 11.7 psi RVP fuel, and 138 percent higher
at 9.0 psi RVP fuel, than the average emissions from failed LDGVs.
Under FTP conditions, the diurnal emissions for vehicles that
failed either pressure or purge test are calculated to be:
Diurnal (G/Test)
Vehicle	Failed	Failed
Type	Pressure	Purge
LDGVs	17.58	5.66
LDGTs	20.57	6.62
In general, diurnal emission levels for vehicles that failed
either pressure or purge test at any other UDI values for 1981 +
LDGVs and LDGTs can be calculated from equations (2.2) through
(2.4). Note that the pressure failure equation can be extrapolated
down to UDI = 0.0 without getting negative emission values. When
UDI is at 0.58 or lower, however, the purge failure equation is
predicting zero or negative diurnal emissions. A lower bound of
the failed purge diurnal emissions is also set to equal to the
resting loss emissions.
Graphic presentations of 1981+ LDGV/LDGT diurnal emissions as
a function of UDI for vehicles failed either pressure or purge test
are shown in Figures 3 and 4.
-7-

-------
2.1.3 Pre-1981 LDGV Diurnal Emission Rates
Pre-1981 model year LDGVs are predominately carbureted
vehicles. In this section, the diurnal emissions from pass
vehicles and from vehicles that failed either pressure or purge
test far pre-1981 LDGVs are determined. The following is a list of
methodologies used for deriving these emissions:
1) For pre-1971 no evaporative emission standard vehicles,
all three emissions are set to be the same as the M0BILE4
uncontrolled emission rates. However, if the MOBILE4 uncontrolled
emission rates are lower than the estimated emission levels from
vehicles that failed pressure test, the new failed pressure
emission levels are used.
2)	Emissions are calculated from equations derived for 1981 +
LDGV group (e.g., for diurnal emissions, use equations (2.1)
through (2.4) defined in sections 2.1.1 and 2.1.2).
3)	As the MOBILE4 emission rates came from the average values
of test data, it is assumed that data from various model years were
results from vehicles tested at about the same age by EPA' s EFP.
The average mileages of the current 1981+ sample were:
VTYP	CARB	PFI	TBI	FINJ
LDGV 55,057 42,115	48,890
LDGT 52,794	44,759
These average mileages fell into 4-5 year old category. The
fractions of the 4 year old LDGV fleet that failed either pressure
or purge test are (from Table 1) 0.053 and 0.060, respectively,
with the emission levels from vehicles that failed either pressure
or purge test being already defined, the following relationships
are established:
E = 0.887 * Epa3s + [0.053 -* Era I I ed Pressure]
[0.06 * Emailed Purge]
Therefore, the emission levels for pass vehicles can be back
calculated from the above equation.
4)	Assume the same MOBILE4 uncontrolled emission rates.
5)	Use the emission ratio from:
(MOBILE4 h y ri / MOBILE4 h y r 2)
= (MOBILE4 . 1 y, , / MOBILE 4 . Uyrj)
6)	Use the emission ratio from:
(Failed Pressurenyri / Failed PressureMyrz)
= (Failed PurgenYRi / Failed PurgeMYR2)
-8-

-------
The MOBILE4 FTP diurnal emissions in g/test for pre-1981 LDGVs
ac two different RVP fuels were:
MYR Group
Pre-1971
19 71
1972-77
1978-80
M4 Diurnal (G/Test)
9.0 RVP 11.5 RVP
26 . 08
16.28
8 .98
5. 16
47.99
38 . 58
23 .53
14 .47
The derived MOBILE4.1 LDGV FTP diurnal emissions Eor the two RVP
fuels are:
	FTP Diurnal Emissions (G/Test)	


at 9.0 RVP


at 11.5
RVP


Fa iled
Failed

Fa iled
Fa iled
MYR Group
Pass
Pu rqe
Pressure
Pass
Pu rqe
Pressure
Pre-1971
26 . 081
26 . 081
26.081
47.99 1
47.99'
47.991
1971
15 . 0 3 J
26.08 1
26.08 1
37.38 J
47.99'
47.99'
1972-77
8.27s
9 . 33 J
20.39s
22.80s
23 .83J
35.45"
1978-80
4 . 39'
5 . 66 2
17.58'
13. 243
18 .42'
30.64'
1981+ Carb
1.66'
5 . 6 6 2
17.58'
7.40'
18.42'
30.64'
Note that the 1981-1- carbureted diurnal emissions were calculated
from equations (2.1) through (2.3). They were listed above for
comparison purposes only, since all pre-1981 LDGVs are of
carbureted fuel delivery system. As in MOBILE4, the pre-1981 FTP
LDGV diurnal emissions in g/test for any other RVP fuels are
calculated by interpolating between these two given emission
levels.
Using the assumption that in-use vehicles were tested at the
age of four, the MOBILE4.1 FTP diurnal emissions in g/test for
LDGVs at two different RVP fuels are:
MYR Group
Pre-1971
1971
1972-77
1978-80
1981+ Carb
M4 .1 Diurnal (G/Test)
9.0 RVP 11.5 RVP
26.08
16.28
8.98
5 . 16
2 .74
47.99
38.58
23 . 53
14 .47
9 .29
-9-

-------
2-1.4 Pre-1981 LDGT1 Diurnal Emission Rates
In MOBILE4, as well as its earlier versions, pre-1981 LDGV
emission rates were used directly for pre-1981 LDGTls, since there
were insufficient test data from LDGTls to derive their own
emission rates and LDGTls have the same evaporative emission
standards as the LDGVs. The MOBILE4 FTP diurnal emissions in
g/test for pre-1981 LDGTls at two different RVP fuels were:
MYR Group
M4 Diurnal (G/Test)
9.0 RVP	11.5 RVP
Pre-1971
1971
1972-77
1978-80
26.08
16.28
8 .98
5 . 16
47.99
38.58
23 . 53
14 .47
However, as discussed in section 2.1.2, LDGTls in general are
equipped with larger capacity of fuel tank than the LDGVs. From
model years 1981+ EF data, where the majority of truck data were
available, the average LDGT1 diurnal emissions are found to be
significantly higher than their LDGV counterpart (see Table 3).
Thus, it is reasonable for LDGT1's to have higher diurnal emissions
than the LDGVs, whether for pass vehicles, or for vehicles that
failed either pressure or purge test.
The MOBILE4.1 LDGT1 diurnal emission rates were estimated by
multiplying a factor of 1.17 (the ratio of the average LDGT1 and
LDGV fuel tank capacities, as discussed in section 2.1.2).
FTP Diurnal Emissions (G/Test)


at 9.0 RVP


at 11.5
RVP


Failed
Fa iled

Failed
Fa iled
MYR Group
Pass
Purqe
Pressure
Pass
Purqe
Pressure
Pre-1971
30.51
30.51
30.51
56. 15
56 . 15
56 . 15
1971
17.59
30.51
30.51
43.73
56 . 15
56 . 15
1972-77
9 . 68
10 .92
23.86
26. 68
27 . 88
41.48
1978-80
5 . 14
6 . 62
20 .57
15.49
21.55
35.85
1981+ Carb
2 . 09 2
6 . 62
20 .57
11.452
21.55
35.85
Note that the 1981+ carbureted diurnal emissions were calculated
from equations (2.1) through (2.4). They were listed above for
comparison purpose only, since all pre-1981 LDGTls are of
carbureted fuel delivery system. As in MOBILE4, the pre-1981 FTP
LDGT1 diurnal emissions in g/test for any other RVP fuels are
calculated by interpolating between these two given emission levels.
-10-

-------
Using the assumption that in-use vehicles were tested at the
age of four, the MOBILE4.1 FTP diurnal emissions in g/test nor
LDGTls at two different RVP fuels are:
M4.1 Diurnal (G/Test)
MYR Group 9.0 RVP	LI.5 RVP
Pre-1971	30.51	56.15
1971	19.05	45.13
1972-77	10.51	27.54
1978-80	6.05	16.93
1981+ Carb	3.34	13.35
As can be seen, the new LDGT1 diurnal emissions for MOBILE4.1 are
slightly higher than the diurnal rates assumed for MOBILE4.
2.1.5 Pre-1981 LDGT2 Diurnal Emission Rates
In MOBILE4, the same post-1979 LDGT1 diurnal emissions were
assumed for LDGT2s, based on their similar evaporative emissions
standards and their similar fuel tank capacities. The diurnal
emissions for pre-1979 LDGT2s were assumed to be the same as the
pre-controlled HDGV diurnal emission levels, as there were no
separate vehicle category of LDGT2 before the model year of 1979.
The MOBILE4 FTP diurnal emissions in g/test for LDGT2s at two
different RVP fuels were:
M4 Diurnal (G/Test)
MYR Group 9.0 RVP	11.5 RVP
Pre-1979	42.33	77.89
1979-80	5.16	14.47
Using the same methodology as in	LDGTls (section 2.1.4), the
MOBILE4.1 FTP g/test diurnal emissions	for LDGT2s at two different
RVP fuels are:
	FTP Diurnal Emissions (q/test)	
	at 9.0 RVP			at 11.5 RVP
Failed Failed	Failed Failed
MYR Group Pass Purge Pressure	Pass Purge Pressure
Pre-1979
1979-80
1981+ Carb
42.33 42.33
5.14	6.62
2 . 09 2 6.62
42.33 77.89
20.57 15.49
20.57 11.45'
77.89 77.89
21.55 35.85
21.55 35.85
-11-

-------
Note that the 1981 + carbureted diurnal emissions were calculated
from equations (2.1) through (2.4). They were listed above tor
comparison purpose only, since all 1979-80 LDGT2s are of carbureted
fuel delivery system. As in MOBILE4, the pre-1981 FTP LDGT2
diurnal emissions in g/test for any other RVP fuels are calculated
by interpolating between these two given emission levels.
Using the assumption that in-use vehicles were tested at the
age of four, the MOBILE4.1 FTP diurnal emissions in g/test for
LDGT2s at two different RVP fuels are:
M4.1 Diurnal (G/Test)
MYR Group 9.0 RVP 11.5 RVP
Pre-1979	42.33	77.89
1979-80	6.05	16.93
1981+ Carb 3.34	13.35
2.1.6 HDGV Diurnal Emission Rates
The MOBILE4 FTP diurnal emissions in g/test for HDGVs at two
different RVP fuels were:
M4 Diurnal (G/Test)
MYR Group 9.0 RVP 11.5 RVP
Pre-1985	42.33	77.89
1985+	4.68	17.94
The same algorithm as MOBILE4 HDGV evaporative emissions were
used to estimate model year 1985+ HDGV diurnal emission levels for
pass vehicles, as well as HDGVs that failed either pressure or
purge test. The bases used in the calculations were the 1981 +
LDGT1 diurnal emission rates (i.e., 2.09/11.45 g/test foe pass
vehicles, 20.57/35.85 g/test for vehicles that failed pressure
test, and 6.62/21.55 g/test for-vehicles that failed purge test, at
9.0 and 11.5 psi RVP fuels, respectively). In general, the
following equation is used to estimate HDGV diurnal emission rates:
DIhdgv - [DIlboti * 1.5 " 0.875]
+ [DIldgti * 2.0 * 0.125]
where, the constant values of 1.5 and 2.0 were the ratios of the
evaporative emission standards between HDGVs and LDGTls. The
constant values of 0.875 and 0.125 were the estimated market shares
of HDGVs under each evaporative emissions standards.
-12-

-------
The M0BILE4.1 FTP g/test diurnal emissions for HDGVs at two
different RVP fuels were:
MYR Group
Pre-1985
1985 +
FTP Diurnal Emissions (q/test)
at 9.0 RVP
at 11.5 RVP
Pass
42.33
3 .27
Failed Failed	Failed
Purge Pressure Pass Purge
42.33
10 .34
42.33
32.14
77'. 89
17 . 89
77 . 89
33 . 67
Fa iled
Pressure
77 . 89
56 . 02
As in MOBILE4, the FTP HDGV diurnal emissions in g/test for
any other RVP fuels are calculated by interpolating between these
two given emission levels. Assuming that in-use vehicles were
tested at the age of four, the MOBILE4.1 FTP diurnal emissions in
g/test for HDGVs at two different RVP fuels are:
MYR Group
Pre-1985
1985 +
M4.1 Diurnal (G/Test)
9.0 RVP	11.5 RVP
42.33
5 .22
77 .89
20.86
2.1.7 Motorcycle Diurnal Emission Rates
There were no new data available for the motorcylce diurnal
emissions since the release of MOBILE4 model. Therefore, the
MOBILE4.1 diurnal emissions for motorcycles are the same as
those used for MOBILE4.
-13-

-------
2 . 2
Other Types of Diurnal Emissions
Three other types of diurnal emissions used in the
MOBILE4/MOBILE4.1 models are: in-use full diurnal emissions (FDI)
partial diurnal emissions (PDI), and multiple diurnal emissions
(MDI). The g/test diurnal emissions discussed in the previous
sections (2.1, 2.1.1 through 2.1.7) were all derived from one hour
heat build test results.
Under in-use conditions, however, vehicles may experience
diurnal emissions from a much slower temperature rise for a much
longer time frame, for example, for the entire period of six hours
when the ambient temperatures are increasing from the minimum to
the maximum. Sometimes one full diurnal may be interrupted (or cut
short) by the start of a trip during the ambient temperature rising
period so that vehicles may only experience a partial diurnal.
Vehicles may have multiple diurnal emissions when they have not
being driven for two or more consecutive days. Some vehicles may
have short but frequent trips or one long trip that they have no
diurnal emissions at all during one particular day. The fractions
of occurrences of these various types of diurnal emissions are
discussed in section 5.0. The levels of these various types of
diurnal emissions are discussed in the following sections.
2.2.1 In-Use G/Day Full Diurnal (FDI) Emissions
In MOBILE4, the in-use (6-hour) g/day full diurnal emissions
are the FTP one-hour emission rates (g/test) adjusted for 121 and
124 percent at 9.0 and 11.5 psi RVP fuels, respectively.
There has been very limited data available in this area since
the release of MOBILE4 model. The adjustment factors of 1.21 and
1.24 were derived from testing of one late model year carbureted
vehicle with its canister disconnected. It is unclear whether the
differences in emissions between a slower (six hours) temperature
rise and faster (one hour) heat build should be classified as
diurnal or resting loss emissions, as the levels seem to be more
similar to the latter. For these reasons, it is assumed that the
21 (or 24) percent emission increases due to the slower temperature
rise are the resting loss emissions and the in-use six-hour full
diurnal emission rates are the same as the one-hour FTP diurnal
emissions. Consequently, the MOBILE4 adjustment factors of 1.21
and 1.24 used for in-use full diurnal emission calculations are no
longer applicable in MOBILE4.1.
-14-

-------
2.2.2 Partial Diurnal (PDI) Emissions
Many of the MOBILE4 assumptions used for partial diurnals
remained unchanged for the MOBILE4.1 model, with one exception.
By definition, partial diurnals occur only when the vehicle's
engine has been turned-off for at least three hours, but less than
six hours, between the daily minimum and maximum temperatures when
the ambient temperatures are increasing (i.e., between 7 AW and
3 PM) . The two temperatures used to calculate the partial diurnal
emissions are:
1)	the ambient temperature at two hours (rather than the
one hour assumed in MOBILE4) after the end of a trip,
and,
2)	the ambient temperature at the time when a new trip
starts.
The three types of partial diurnals and their characteristics are
described below:
Description
8 AM - 11 AM
10 AM - 3 PM
8 AM - 2 PM
Temperature
Minimum
T l 0 A M
T l z ? M
T 1 o A M
Maximum
T | 1 A M
T 3 p M
Ti p*
Percent of
Occu rrence
32.36
7 .09
3 . 69
The temperatures at 10AM, 11AM, and 2PM are calculated by the
following equations:
Tio AM = Tmin + 0.53 * (Tmax - Tmin)	(2.5)
Ti, AM = Tmin + 0.71 * (Tmax - Tmin)	(2.6)
T2 pm = Tmin + 0.94 * (Tma'x - Tmin)	(2.7)
where:
Tmin - the minimum temperature of	the'day, in F
Tmax - the maximum temperature of	the day, in F.
The ambient temperature at 3PM is just the maximum ambient
temperature of the day.
-15-

-------
2.2.3 Multiple Diurnal (MDI) Emissions
In M0BILE4, due to a lack of data, the adjusted in-use diurnal
emissions from tampered vehicles were used for multiple diurnal
emission rates. It has been suggested that this assumption may not
be technically correct. with vehicles parked for two or more
consecutive days, their diurnal emission levels are dependent upon
many factors: ambient temperature rise, backpurge during the
cooling of the ambient temperature, canister capacity and
efficiency, fuel weathering, and tank fill level. When all these
parameters are being considered, the multiple diurnal emissions may
be lower (on a grams-per-day basis) than the diurnal emissions
under uncontrolled or tampered conditions.
Similar to the FTP one-hour (or in-use full) diurnal
emissions, the MOBILE4.1 multiple diurnal emissions model also
includes three elements: emissions from pass vehicles, emissions
from vehicles that failed pressure test, and emissions from
vehicles that failed purge test. Multiple diurnal emissions for
vehicles that failed either pressure or purge test are assumed to
be the same as their FTP one-hour diurnal counterparts. These
emissions were based on both the theoretical vapor generating model
and test data from EFP-Hammond (see discussions in section 2.1.2).
Multiple diurnal emission rates for pass vehicles were based
on the theoretical vapor generating model. Overall grams per day
emissions modeled from late model year fuel-injected technology
vehicles were simulated at various fuel volatility levels for one
to seven consecutive no-driving days. The temperature rise of 72
to 96 F and 40% tank fill level were assumed. Also, it is assumed
that backpurge occures in 50% of the vehicles experiencing multiple
diurnals.
To summarize results, Table 2 from Attachment has been copied
and presented at the top portion of Table 5. Ratios of emissions
from any day (day 2 through day 7) to emissions at day 1 for each
fuel volatility level were calculated and listed at the lower
portion of Table 5. These ratios, then, were weighted according to
the re-normalized fractions of occurrence for days 2 through 7.
Finally, a nonlinear regression line was fitted through the
weighted ratios. The equational form for the ratio as a function
of fuel volatility is:
CFhoi , pass = EXP (4.8491 - 0.33424  RVP)	(2.8)
Both the weighted and predicted ratios at each fuel volatility
level are given in Table 5. The multiple diurnal emissions for
pass vehicles are calculated by:
MDIPais = Dlpai! * CFmoi.P.ss	(2.9)
-16-

-------
Using equations <2.8) and (2.9), the multiple diurnal
emissions for pass vehicles are sometimes calculated to be higher
than the emissions from vehicles that failed purge, especially for
pre-2.0 SHED evaporative emissions standard vehicles (e.g.,
pre-1981 LDGVs). One of the reasons is that the emissions
calculated to derive equation (2.8) were based on late model year
fue1-injected vehicle performance, while all pre-2.0 SHED
evaporative emissions standard vehicles are of carbureted
technology. Also, for the pre-2.0 SHED evaporative emissions
standard era, the emission levels for pass vehicles were similar to
the emission levels for vehicles that failed purge. Therefore, it
is reasonable to set the maximum levels of multiple diurnal
emissions for pass vehicles equal to the emissions from vehicles
that failed purge, i.e.,
MDIp a j 5 = MDIFaliea Purge	(2 . 10)
To summarize, multiple diurnal emissions for pass vehicles are
calculated by the following two methods:
1)	use equations (2.8) and (2.9), or,
2)	use equation (2.10), if the calculated emission levels
from equations (2.8) and (2.9) are greater than the
multiple diurnal emissions from vehicles that failed
purge.
These two methods are used for all LDGVs and LDGTs, pre-1981 model
years as well as 1981+ model years. As there is no multiple
diurnals assumed for HDGVs and MCs, these two methods are not
applicable to these two vehicle types.
-17-

-------
3.0 Hot Soak Emissions
As in MOBILE4, the hot soak emissions (in unit of grams per
one hour test) are described as two step functions of fuel
volatility, in psi RVP and ambient temperature in F. The hot soak
emissions are measured at FTP test conditions, i.e., with 9.0 psi
RVP fuel, 40% tank fill, and at the ambient temperature of 82F.
In the model, hot soak emissions are first adjusted for fuel
volatility effect, then corrected for the temperature effect at
ambient temperatures 4lF and higher.
3.1 1981* LDGVS and LDGTS
As in the MOBILE4.1 diurnal emissions model, the g/test hot
soak emissions are derived for pass vehicles, for vehicles that
failed pressure test, and for vehicles that failed purge test.
Table 3 also summarizes the average hot soak emissions from the
current EFP-MVEL data base.
3.1.1 Pass Vehicles
Hot soak
EFP-MVEL data. For
trucks, the derived
are:
emissions for pass vehicles were derived from EPA' s
model years 1981+ pass light-duty vehicles and
equational forms for g/test hot soak emissions
HS
Pass
= a
RVP
(3.1)
or,
HS p a s i
where:
Vehicle Type
/Fuel System
= c + d * RVP + e * RVP
Regression Coefficient
(3.2)
LDGV/CARB
0 . 14593
0. 13823
-1.436657
0.0
0
LDGV/PFI
-0.46673
0. 10297
18.437
-4.2538
0
LDGV/TBI
0 .198327
0.041297
-4.669710
0 . 58219
0
LDGT/CARB
-0 .16407
0. 13823
-5.196600
0.69740
0
LDGT/FINJ
0 .078327
0.041297
-4.789710
0.58219
0
034897
25072
0
-18-

-------
Equation (3.1) is used when fuel volatility is Less than
9.0 psi RVP, while equation (3.2) is used when fuel volatility is
equal to or high than 9.0 psi RVP. The coefficients from equation
(3.1)	were regression results from data where the fuel rvp is
9.0 psi and from limited emissions data at lower fuel volatility
levels of 6.5 and 8.0 psi RVP. The coefficients from equation
(3.2)	were derived from EFP-MVEL data, where the three fuel
volatility levels were 9.0, 10.5, and 11.7 psi RVP.
The constant terms of equations (3.1) and (3.2) (denoted as
"a" and "c" above) were adjusted so that at 9.0 psi RVP the
predicted hot soak emissions are equal to the average values at
9.0 psi RVP of the EFP-MVEL sample. Note that the lower bound of
fuel volatility foe equation (3.1) is 5.0 psi RVP, and the Lower
bound of hot soak emissions is set to equal to the resting loss
emissions.
Graphic presentations of 1981+ LDGV/LDGT hot soak emissions
for pass vehicles as a function of fuel volatility are shown in
Figures 5 and 6. There has been no new data available to develop
new hot soak temperature correction factors. Therefore, the
equations used to correct for temperature effect from MOBILE4 will
be used again in MOBILE4.1 for pass vehicles.
3.1.2 Failed Vehicles
During FY91, five vehicles were tested for hot soak emissions
with canister connected and disconnected at two levels of fuel
volatilities (9.0 and 10.5 psi RVP) and at three levels of ambient
temperatures (70, 82, and 95F). These five vehicles were:
Model Year	Make	Fuel System
1979
1983
1983
1990
1990
Cut lass
Cut lass
Skylack
Taurus
Lumina
Carbureted
Carbureted
TBI
PFI
PFI
Overall average hot soak emissions with canister disconnected
(or uncontrolled hot soak emissions) at various ambient
temperatures and fuel volatilities are summarized below:
Temperature
in
70 . 0
82 . 0
95 . 0
Hot Soak (G/Test)
9 . 0 RVP
5 . 62
7.93
20.08
10 . 5 RVP
10.71
16.11
39.27
-19-

-------
Regression equation has been derived to describe the
relationships of the uncontrolled hot soak emissions, versus the
fuel volatility, and ambient temperature:
HSu ncontroi lea = EXP [2.2307 + 0.4443 * (RVP - 9.0)
+ 0.05114 * (TEMP - 82.0)]
where:
HS = uncontrolled hot soak emissions in g/test,
EXP = exponential function,
RVP = fuel volatility in psi, and
TEMP = ambient temperatures in F.
Due to limited available data, it is assumed that the estimated
uncontrolled hot soak emissions are the emission levels for
vehicles that failed pressure test.
From EFP-Hammond data base, 18 of the 30 LDGVs (as listed in
Table 4) were also tested for hot soak emissions at 92F ambient
temperature, and with 9.0 psi RVP fuel. These vehicles were found
to have failed purge test at the Hammond I/M Lane. The average hot
soak emissions were 10.56 g/test. Therefore, for vehicles that
failed purge test, the hot soak emissions can be estimated from the
above failed pressure hot soak emission equation but normalized at
the average emissions from Hammond data.
The hot soak emissions equations for failed vehicles are:
HS F a i led Pressure = EXP [2.2307 + 0.4443 * (RVP-9.0)
+ 0.05114  (TEMP-82.0)]	(3.3)
HS f a i led Purge = EXP [ 1.84567 + 0.4443 * (RVP-9.0)
+ 0.05114 * {TEMP-8 2.0)]	(3.4)
Using these two equations (3.3 and 3.4), the predicted hot soak
emissions for LDGVs that failed either pressure or purge test at
various ambient temperatures and fuel volatilities are shown in the
table	below:
Predicted Hot Soak (G/Test)

9 . 0
RVP
10 . 5
RVP
11.7
RVP
Temperature
Fa iled
Failed
Failed
Failed
Fa iled
Failed
in F
Pres
Purqe
Pres
Purqe
Pres
Purqe
70 .0
5 . 04
3.43
9 .81
6 . 68
16 . 72
11. 38
82 .0
9.31
6 .33
18 . 12
12.33
30.89
21.02
92.0
15 .52
10 .56
30.22
20.56
40.00
35 . 05
95 . 0
18 . 09
12 . 31
35.23
23.97
40.00
40 . 00
-20-

-------
Note that these equations can be extrapolated beyond the
ambient temperature ranges of 70.0 and 95.0F and fuel volatility-
levels of 9.0 and 10.5 psi RVP. To be consistent with the hot soak
emissions calculations for pass vehicles, lower limits are also set
at 5.0 psi RVP fuel volatility and 41F ambient temperature. There
are also a lower bound ot hot soak emissions being equal to the
resting loss emissions and an upper bound of 40.00 g/test of hot
soak emissions for LDGVs.
Equations (3.3) and (3.4) are also used to calculate hot soak
emissions for all other vehicle types (such, as LDGTls, LDGT2s, and
HDGVs) that failed pressure or purge test.
Graphic presentations of 1981+ LDGV/LDGT hot soak emissions as
a function of fuel volatility at 82F for vehicles that failed
either pressure or purge test are shown in Figure 7.
3.1.3 Pre-1981 LDGV Hot Soak Emission Rates
The same. list of methodologies used to derive diurnal
emissions, as described in section 2.1.3, is also used for hot soak
emissions calculations. The MOBILE4 FTP hot soak emissions in
g/test for pre-1981 LDGVs at two different RVP fuels were:
The derived MOBILE4.1 LDGV FTP hot soak emissions for the two RVP
fuels are:
M4 Hot Soak (G/Test)
MYR Group 9.0 RVP	11.5 RVP
Pre-1971
1971
1972-77
1978-80
14 . 67
10 . 91
8 . 27
2.46
22 .45
16. 15
12 .32
4 .30
FTP Hot Soak Emissions (q/test)
at 9.0 RVP
at 11.5 RVP
MYR Group
Failed Failed
Pass Purge Pressure Pass
Failed Failed
Purge Pressure
Pre-1971
1971
1981+ Carb
1972-77
1978-80
14.67'	14.67'	14.67'	28.26'	28.26'	28.26
10.431	14.67'	14.67'	14.61J	28.261	28.26
1.111	9.98*	14.67'	11.10s	19.23'	28.26
1.79 J	6 . 33 2	9 . 312	3.89 s	19.232	28.26
1.3 9 2	6 . 33 2	9 . 3 1 2	3.18 2	19.23 2	28.26
-21-

-------
Note that the 1981+ carbureted hot soak rates were calculated from
equations (3.2) through (3.4). They were listed above tor
comparison purpose only, as all pre-1981 LDGVs are of carbureted
fuel delivery system. As in M0BILE4, the pre-1981 FTP LDGV hot
soak emissions in g/test for any other RVP fuels are calculated by
interpa 1 ating 'between these two given emission levels.
Using the assumption that in-use vehicles were tested at the
age of four, the M0BILE4.1 FTP hot soak emissions in g/test for
LDGVs at two different RVP fuels are:
M4 .1 Hot Soak (G/Test)
MYR Group 9.Q RVP	11.5 RVP
Pre-1971	14.67	28.26
1971	10.91	16.15
1972-77	8.27	12.50
1978-80	2.45	6.10
1981+ Carb	1.91	5.47
3.1.4 Pre-1981 LDGT1 Hot Soak Emission Rates
As in MOBILE4, the pre-1981 LDGV hot soak emission rates are
used also for pre-1981 LDGTls because of their similar evaporative
emission standards. The MOBILE4 FTP hot soak emissions in g/test
for pre-1981 LDGTls at two different RVP fuels were:
MYR Group
M4 Hot Soak (G/Test)
9.0 RVP 11.5 RVP
Pre-1971
1971
1972-77
1978-80
14 .67
10.91
8.27
2.46
22 . 45
16 . 15
12 .32
4 .30
The derived MOBILE4.1 LDGT1 FTP hot soak emissions for the two RVP
fuels are:
FTP Hot Soak Emissions (g/test)


at 9.0 RVP
at
11.5 RVP



Failed
Failed

Failed
Failed
MYR Group
Pass
Purqe
Pressure
Pass
Purqe
Pressure
Pre-1971
14 . 67
14 . 67
14. 67
2 8.26
28.26
28.26
1971
10. 43
14 . 67
14.67
14. 61
28.26
28.26
1972-77
7 . 77
9. 98
14 . 67
11. 10
19.23
28.26
1978-80
1 . 79
6.33
9.31
3 . 89
19 . 23
28.26
1981+ Carb
1 . 08
6.33
9.31
2 .82
19.23
28 .26
-22-

-------
Note that the 1981+ carbureted hot soak rates were calculated from
equations (3.2) through (3.4). They were listed above for
comparison purpose only, as all pre-1981 LDGTls are of carbureted
fuel delivery system. As in M0BILE4, the pre-1981 FTP LDGT1- hot
soak emissions in g/test for any other RVP fuels are calculated by
interpolating between these two given emission levels.
Assuming that in-use vehicles were tested at the age of four,
-he MOBILE4.1 FTP hot soak emissions in g/test for LDGTls at two
different RVP fuels are:
M4.1 Hot Soak (G/Test)
MYR Group 9.0 RVP	11.5 RVP
Pre-1971	14.67	28.26
1971	10.91	16.15'
1972-77	8.27	12.50
1978-80	2.46	6.10
1981+ Carb	1.83	,5.15
3.1.5 Pre-1981 LDGT2 Hot Soak Emission Rates
As in MOBILE4, the same post-1979 LDGT1 hot soak emission
rates are assumed for LDGT2s, based on their similar evaporative
emission standards. The pre-1979 LDGT2 hot soak emission rates
were the same as the pre-controlled HDGV hot soak rates, as there
were no separate vehicle category of LDGT2s before the model year
of 1979. The MOBILE4 FTP hot soak emissions in g/test for pre-1981
LDGT2s at two different RVP fuels were:
M4 Hot Soak (G/Test)
MYR Group 9.0 RVP	11.5 RVP
Pre-1979	18.08	27.66
1979-80	2.46	4.30
The derived MOBILE4.1 LDGT2 FTP hot soak emissions for the two RVP
fuels are:
FTP Hot Soak Emissions (q/test)	
at 9.0 RVP	at 11.5 RVP
Failed Failed	Failed Failed
MYR Group	Pass Purge Pressure Pass Purge Pressure
Pre-1979
1979-80
1981+ Carb
18.08 18.08
1.79 6.33
1.08 6.33
18.08 44.16
9.31	3.89
9.31	2.82
44.16 44.16
19.23 28.26
19.23 28.26
-23-

-------
As in M0BILE4, the pre-1981 FTP LDGT2 hot soak emissions in g/test
for any other RVP fuels are calculated by interpolating between
these two given emission levels.
Assuming that in-use vehicles were tested at the age of four,
the M0BILE4.1 FTP hot soak emissions in g/test for LDGT2s at two
different RVP fuels are:
M4.1 Hot Soak (g/test)
MYR Group 9.0 RVP	11-5 RVP
Pre-1979	18.08	44.16
1979-80	2.46	6.10
1981+ Carb 1.83	5.15
3.1.6 HDGV Hot Soak Emission Rates
The MOBILE4 FTP hot soak emissions in g/test for HDGVs at two
different RVP fuels were:
M4 Hot Soak (G/Test)
MYR Group 9.0 RVP	11.5 RVP
Pre-1985	18.08	27.66
1985+	2.12	4.77
As in the diurnal emissions calculations, the same algorithm
from MOBILE4 HDGV evaporative emissions are used to estimate 1985+
hot soak emission rates for pass vehicles, as well as HDGVs that
failed either pressure or purge test. The bases used in the
calculations were the 1981+ LtiGTl hot soak emission rates
(1.08/2.82 g/test for pass vehicles, 9.31/28.26 g/test for vehicles
that failed pressure test, and 6.33/19.23 g/test for vehicles that
failed purge test, at 9.0 and 11.5 psi RVP fuels, respectively).
The new MOBILE4.1 FTP g/test hot soak emissions for HDGVs at
two different RVP fuels are:
	FTP Hot Soak Emissions (q/test)		
at 9.0 RVP		at 11.5 RVP	
Failed	Failed	Failed	Failed
MYR Group Pass Purge	Pressure	Pass	Purge	Pressure
Pre-1985 18.08 18.08	18.08	44.16	44.16	44.16
1985+ 1.69 9.89	14.55	4.41	30.05	44.16
-24-

-------
As in M0BILE4, the FTP HDGV hot soak emissions in a/test for
any other RVP fuels are calculated by interpolating between these
two given emission levels.
Assuming that in-use vehicles were tested at the age of four,
the MOBILE4.1 FTP hot soak emissions in g/test for HDGVs at two
different RVP fuels are:
Hot Soak (G/Test)
MYR Group -9.0 RVP 11.5 RVP
Pre-1985	18.08	44 .15
1985+	2.86	8.06
3.1.7 Motorcycle Hot Soak Emission Rates
There were no new data available for motorcylces hot soak
emission rates since the release of MOBILE4 model. Therefore, the
MOBILE4.1 hot soak emissions for motorcycles are the same as those
used for MOSILE4.
-25-

-------
40 High Altitude
In M0BILE4 model, a set of altitude adjustment factors was
used to calculate the evaporative diurnal and hot soak emissions
for vehicles operating in high altitude region. These altitude
adjustment factors were based on either actual test data or from
their emissions standard ratio. For example, the M0BILE4 FTP
diurnal and hot soak emissions in g/test for high altitude LDGVs at
two different RVP fuels were:
Diurnal	(G/Test)	Hot Soak	(G/Test)
MYR Group	9.0 RVP	11.5 RVP	9.0 RVP	11.5 RVP
Pre-1971	33.90	62.39	19.07	29.18
1971	21.16	50.15	14.13	20.99
1972-76	17.15	44.93	17.15	20.96
1977	8.98	23.53	8.27	L2.32
1978-80	13.36	36.85	6.37	LI.15
Note that the emission rates for model years 1972-76 were actual
test results from EPA's EFP test facility in Denver, Colorado. The
emissions for model year 1977 were the same as the low altitude
rates (because of their same evaporative emissions standard). The
emission rates for model years 1978-80 were estimated by
multiplying a factor of 2.59 to the low altitude rates. Finally,
the emission rates for all other model year groups were calculated
by multiplying an adjustment factor of 1.30.
The same set of altitude adjustment factors is used again in
MOBILE4 . 1 and are summarized in Table 6. The only exception is
that, for model years 1978-80, the MOBILE4 adjustment factor of
2.59 has been changed to 1.30. ' These high altitude adjustment
factors are used for both diurnal and hot soak emissions of pass
vehicles, as well as vehicles that failed either pressure or purge
test. For example, the derived MOBILE4.1 high altitude LDGV
diurnal emissions at two RVP fuel Levels are:
LDGV Diurnal Emissions (G/Test)

at
9.0 RVP


at 11.5
RVP


Failed
Failed

Fai led
Fa iled
MYR Group
Pass
Purqe
Pressure
Pass
Purqe
Pressure
Pre-19 71
33 .90
33.90
3 3.90
62.39
62.39
62 . 39
1971
19 . 54
33 . 90
33 . 90
48.59
62.39
62 . 39
1972-76
10 . 16
13 . 82
27.46
33.13
45.06
56 . 15
1977
8 .27
9 . 33
20 . 39
22.80
23 .83
35.45
1978-80
5 . 71
7.36
22 .85
17.21
23.95
39 . 83
1981-83 Carb
2.16
7.36
22 .35
9 . 62
23.95
39 .83
1984+ Carb
1.66
5 . 66
17 . 58
7.40
13.42
30 . 64
-26-

-------
Similarly, the derived M0BILE4.1 high altitude LDGV hot soak
emissions at two RVP fuel levels are:
LDGV Hot Soak Emissions (G/Test)
at 9  0 RVP		at 11.5 RVP


Fa iled
Fai led

Failed
Fa iled
iMYR Group
Pass
Purqe
Pressure
Pass
Pu rge
Pressure
Pre-1971
19 . 07
19 . 07
19 . 07
36.74
36.74
36.74
1971
13.56
19 . 07
19 .07
18 .99
36.74
36.74
1972-76
10 .03
14 . 52
19 . 07
18 .87
27.30
36. 74
1977
7 . 77
9 .98
14 . 67
11. 10
19 .23
28.26
1978-80
2 . 33
8 .23
12 . 10
5 .06
25.00
36 . 74
1981-83 Carb
1 . 81
8 .23
12 . 10
4 . 13
25 .00
36.74
1984+ Carb
1 . 39
6 .33
9.31
3. 18
19 . 23
28.25
Assuming that in-use vehicles were tested at the age of four,
the MOBILE4 . 1 FTP diurnal and hot soak emissions in g/test for high
altitude LDGVs at two different RVP fuels are:
Diurnal (G/Test) Hot Soak (G/Test)
MYR Group	9.0 RVP 11.5 RVP 9.Q RVP 11.5 RVP
Pre-1971	33.90	62.39	L9.07	36.74
1971	21.16	50.15	14.18	20.99
1972-76	11.30	35.07	10.91	20.32
1977	8.98	23.53	8.27	12.50
1978-80	6.72	18.81	3.20	7.94
1981-83 Carb	3.57	12.08	2.74	7.11
1984+ Carb	2.74	9.29	2.11	5.47
-2.7-

-------
5.0 Trip Related Estimates
Most of the M0BILE4 trip-related estimates, such as trips per
day (TPD), miles per day (MPD), tractions of occurrences for FDI,
MDI, and PDI are also used in MOBILE4.1, with some modifications.
There is no new data available to derive new estimates. Therefore,
the data base where these estimates were derived from is the same
1979 GM-NPD survey data used in MOBILE4.
5.1 Trips Per Day and Miles Per Day
The trips per day and miles per day equations used for LDGVs
and LDGTs have the following forms:
where: TPD = estimated number of trips per day,
AGE = vehicle age in years,
MPD = estimated miles driven per day,
JAMAR = January 1 annual mileage accumulation rate, which is
also a function of vehicle age.
These equations were used in MOBILE4 to calculate the estimated
trips per day and miles per day for vehicles at ages 1 through 20.
In MOBILE4.1, equations (5.1) and (5.2) are used to obtain TPD and
MPD estimates for LDGVs and LDGTs for ages 1 through 25.
The TPD and MPD estimates for HDGVs and MCs remain unchanged
in MOBILE4.1:
TPD = 4.7187 - 0.058508.* AGE
MPD = JAMAR / 3 6 5.0
(5.1)
(5.2)
HDGVs MCs
TPD
MPD
6.88 1.35
33.97 10.02
-28-

-------
5.2 Other Trip-Related Estimates	-- LDGVS and LDGTS
Percents of other trip-related estimates for LDGVs and LDGTs
used in M0BILE4 were:
Tot a 1
Descript ion	Percent	Pe rcent
Driving Days	76.15	76.15
Single Full Diurnal Days	22.37
Partial Diurnal Days	43.14
i) 8 AM to 11 AM	3 2.3 6%
ii) 10 AM to 3 PM	7.09%
iii) 8 AM to 2 PM	3.69%
3+ multiple Diurnal Days	3.89
No Diurnal Days	6.75
No Driving Days	23.85	23.85
Single Full Diurnal Days	11.59
2+ Multiple Diurnal Days	12.26
In MOBILE4.1, all the above estimates are expressed in terms
of vehicle's age. For example, the proportions of vehicles having
at least one trip on any given day for age 1 vehicles are higher
than those for age 2 vehicles, etc. The equational forms for
various types of driving days (in percent) are:
Driving Days = 87.07 - 1.72 * AGE	{5.3)
Single Full Diurnal = 25.58 - 0.5 * AGE	(5.4)
Partial Diurnal Days:
8AM-11AM Partial Diurnal = 37.0 - 0.73 * AGE	(5.5)
10AM-3PM Partial Diurnal = 8.11 - 0.16 * AGE	(5.6)
8AM-2PM Partial Diurnal = 4.22 - 0.08 * AGE	(5.7)
3+ Multiple Diurnal = 4.45 - 0.09 * AGE	(5.8)
No Diurnal Days = 7.72 - 0.15 * AGE
The equational forms for various types of driving days (in percent)
are:
No Driving Days = 12.93 + 1.72 * AGE
Single Full Diurnal Days = 6.31 + 0.84 * AGE	(5.9)
2+ Multiple Diurnal Days = 6.62 + 0.88 * AGE	(5.10)
Table 7 summarizes the results.
-29-

-------
5 . 5 Other Trip-Re la bed Estimates for HDGVs
Percents
MOBILE4 were:
of all trip-related estimates for HDGVs used
1 n
Description
Driving Days
No Driving Days
Single Full Diurnal Days
2 Multiple Diurnal Days
3+ Multiple Diurnal Days
Percent
27 . 10
72 . 90
49.80
13.70
9.40
Tota I
Percent
27.10
72.90
Due to a lack of data, no equations as functions of vehicle's
age were derived. These estimates are used in MOBILE4.1 again.
-30-

-------
Table 1
Pressure/Purge Failure Rates
by Vehicle Age
Source: EPA's EFP-Harrunond Data
Age of		LDGV Failure Rates
Vehicle
Pu rqe
Pressure
Ei ther
01
0 .043
0 .043
0 .080
02
0 . 043
0.043
0.080
03
0 .043
0.043
0 .080
04
0 . 060
0.053
0 .096
05
0 .060
0 .053
0 . 103
06
0 .069
0 .053
0 . 120
07
0 . 094
0 . 053
0.150.
08
0 . 09 4
0 . 106
0 . 188
09
0 . 142
0 . 152
0.2 58
10
0 . 214
0 . 171
0 . 323
11
0.224
0 . 236
0.389
12
0 .229
0 . 336
0 .442
13 +
0 . 229
0 . 336
0 .451
-31-

-------
Table 2
Criteria Used to Define Pressure and Purge Failures
Emission Factors Programs
Evaporative System Component		Diagnosis	
Purge Failures
Canister Purge Solenoid/Valve	Missing
Disconnected or Bypassed
Leaks Vacuum
Sticking
Inoperative
Vacuum or Vent Lines	Disconnected or Missing
Plugged or Damaged
Mis routed
Purge Hose	disconnected or Missing
Split or Not sealed
Canister Purge TVS	Stuck
ECM signal to purge solenoid	None detected
Pressure Failures
Gas Cap	Non OEM or Missing
Leaking
Sending Unit Gasket	Leaking
Fuel Tank Rollover	Valve leaking or Stuck
Fuel Tank Filler Neck	Leaking
Both Pressure and Purge Failures
Canister	Missing
Carburetor Bowl Vent Line	Disconnected or Leaks
Fuel Line to Canister	Disconnected
EFE Control Switch	Missing
Bowl Vent Solenoid	Disconnected
-32-

-------
Table 3
Date: April 2 6, 1991
Statistics on Evaporative Emissions
(Average G/Test Hot Soak and Diurnal)
Source: EPA's EFP-MVEL
Fue 1
9.0 RVP
10.4 RVP
11.7 RVP
System
N
HS
DI
N
HS
DI
N
HS
DI

Light
-Duty
Gasoline-Powered
Vehicles



All









Carbureted
386
1.99
2.65
158
3 .43
5 . 52
380
4 . 17
9 . 59
PF I
347
0 . 70
1. 90
69
2 .09
4 .80
254
3 .84
6 .85
TBI
3 15
0 .82
1 . 75
95
1 .42
3 .90
274
2 .99
8.91
Fa iled Pu rge









Carbureted
31
5 .32
5.11
17
7 . 14
8 .76
30
10.25
15 . 19
PF I
13
4 . 27
6 .93
2
16.21
7 . 70
9
21.07
12 . 63
TBI
13
3.26
7 . 19
2
3 . 27
13 . 52
10
13 .69
21.71
Failed Pressure








Ca rbu reted
37
3 . 15
7.44
17
6 . 14
9 . 76
37
5 . 10
15.93
PF I
12
3 .22
10 .84
3
9 . 16
9.36
7
10 .96
25.27
TBI
9
2 . 77
9 .21
2
1 . 25
11.77
7
13 .92
24.42
(FPurge,FPressure)








All
115
3 . 85
7 .22
43
6 . 86
9.51
100
9 .97
17 . 24
Tampered
Ca rbureted
PFI
TBI
7 7.77 10.28
1	0.29 13.79
2	11.49 10.52
2 16 . 00 22 . 70
0
7 9.0 5 19.3 6
1 0.20 21.44
1 36.43 28.28
Pass
Carbureted
258
1.21
1.38
95
1 . 89
3 . 74
253
3 .25
7.25
PF I
299
0.41
1.03
61
1.36
3 . 66
219
2 .97
5 . 64
TBI
233
0.47
0 . 93
79
0.85
3 .32
206
1.77
7 -. 68
Others1
Carbureted
53
2.27
3.02
27
3 . 88
5 . 80
53
3 . 79
11.85
PFI
22
1. 19
5.42
3
0 .51
21 . 57
18
3.25
10 . 78
TBI
58
1 .01
2.38
12
4.91
4 . 77
50
3.71
8 .86
(Pass,Others)
Carbureted
PF I
TBI
311 1.39 1.66 122 2.33 4.20
321 0.46 1.33 64 1.32 4.50
291 0.57 1.22 91 1.39 3.51
306 3.34 8.05
237 2.99 6.03
256 2.15 7.91
*A11 vehicles that had some non-zero ECOMP codes but were not
defined as either purge, failure, pressure failure, or tampered.
-33-

-------
Table 3 (Continued)
Date: April 26, 1991
Statistics on Evaporative Emissions
(Average G/Test Hot Soak and Diurnal)
Source: EPA's EFP-MVEL
Fue 1
Delivery		 9.0 RVP
System	N
10.4 RVP
11.7 RVP
HS
DI
N
HS
DI
HS
DI
All
Carbureted 170
Fuel-Injected 30
Failed Purge
Carbureted	7
Fuel-Injected 1
Failed Pressure
Carbureted	10
Fue1-Injected 2
(FPurge,FPres)
All
Tampered
Carbureted
Fuel-Injected
Pass
Carbureted 129
Fuel-Injected 69
Others *
Carbureted	20
Fuel-Injected 7
(Pass,Others)
Carbureted 149
Fuel-Injected 76
Light-Duty Gasoline-Powered Trucks
6 5.35 10.01
1.53
0.58
5. 06
0. 19
2.93
1.9 0
9 .44
0. 62
2.89 8.71
0.40 22.42
20 3.27 9.93
4 8.94 8.41
1 11.10 23.32
1. 00
0.43
1.59
0. 62
1.08
0.45
1. 69
0.95
4 . 68
2.50
2.09
1.09
1 21.68 25.47
0 -
0
0
199
42
4.13 14.02
1.32 8.54
12 10.20 17.36
0 -
11 7.23 23.65
1 6.61 54.35
1 21.68 25.47 24 8.70 21.79
0
0
4
0
1
0
5
0
2.31 3.63
1.17 20.07
2.08 6.92
6 19.47 32.06
1 26.94 53.62
128 2.19 10.54
37 0.36 6.75
42 5.31 18.58
3 2.83 1.70
170 2.96 12.52
40 0.55 6.37
"All vehicles that had some non-zero ECOMP codes but were not
defined as either purge, failure, pressure failure, or tampered.
-34-

-------
Table 4
Failed Pressure/Purge Test Results
9 2  F Hot Soak, 9 psi RVP Fuel
7 2-96 F Diurnal Heat Build, 40% Tank Fill
Source: EPA's EFP-Haminond
Vehicle # Model Year
576
640
660
665
724
443
736
1541
1635
1640
1530
1552
722
1542
740
1524
673
1526
1612
1532
743
1537
1544
1529
659
1533
1548
1555
730
1525
1981
1981
1981
1981
1982
1983
1983
1983
1984
1984
1984
1984
1985
1985
1985
1985
1985
1985
1985
1986
1986
1986
1986
1987
1988
1988
1988
1988
1989
1989
Fue 1
System
Carb
PFI
Carb
Carb
Carb
TBI
Ca rb
TBI
PFI
PFI
TBI
TBI
PFI
PFI
Carb
Carb
TBI
TBI
TBI
PFI
TBI
TBI
TBI
PFI
PFI
PFI
PFI
TBI
PFI
PFI
N:
Average:
Failed Purge
Hot Soak- Diurnal
1 .95
15.96
6 . 17
12 .05
1 . 62
3 . 70
26 .08
1.73
9 .'26
5 . 54
33.30
25.37
0 .96
2 . 77
0 . 52
20.97
7.36
14 . 71
TaT
10.557
13 .32
11.93
2 . 68
13 . 36
20	. 08
21	.92
25 .55
19 .30
22 .85
13 . 01
22.22
14.40
2. ao
21.34
0 . 65
34 . 61
15 .44
4 .90
18
15.576
Failed Pressure
Diurnal
76 . 22
77.92
24 . 75
17.55
10.85
17.01
16.38
11.38
14.47
5.41
33 .45
27 . 37
12
27.727
-35-

-------
Table 5
Multiple Diurnal Emissions
Pass Vehicles
Source: Table 2 of Attachment
Fractions
of
Day Occurrence
7.5
RVP
8 . 0
RVP
8 . 5
RVP
8 . 7
RVP
9 . 0
RVP
Calculated Emissions (in Grams)
Calculated Emission Ratios
(Relative to Day L Emissions)
10 . 0
RVP
11. 0
RVP
1
0.341
0.2
0 . 4
0.7
0 . 8
1. 1
2 . 5
5 . 3
2
0 . 078
1. 6
2 . 4
3 . 4
3 . 8
4 . 7
8 . 7
15 . 5
3
0.037
2 . 6
3 . 6
5 . 0
5 . 7
6 . 8
11 . 9
20 . 0
4
0.021
3 . 3
4 . 5
6 . 2
6 . 9
8 . 2
13 . 8
22 . 3
5
0.014
3 . 8
5 . 3
7 . 0
7.9
9 . 2
15 . 0
23 . 6
6
0 . 008
4 . 3
5 . 8
7 . 7
8 . 6
10 . 0
15 . 8
24.3
7
0 . 003
4 . 7
6 . 3
8 . 2
9 . 0
10 . 5
16 . 4
24 . 7
2
0 .484
8 . 0
6 . 0
4 . 9
4 . 8
4 . 3
3 . 5
2 . 9
3
0.230
13 . 0
9 . 0
7 . 1
7 . 1
6 . 2
4 . 8
3 . 8
4
0 . 13 0
16.5
11.3
8 . 9
8 . 6
7.5
5 . 5
4 . 2
5
0.087
19 . 0
13 .3
10 . 0
9.9
8 . 4
6 . 0
4 . 5
6
0 .050
21.5
14 . 5
11 .0
10 . 8
9. 1
6 . 3
4 . 6
7
0 .019
23 . 5
15 . 8
11.7
11.3
9 . 5
6 . 6
4 . 7


Occurrence
Weighted
Ratio





12. 2
8 . 6
6.8
6.7
5 . 8
4 . 5
3 . 5



Predicted Ratio*





10.4
8. 8
7 . 4
7 . 0
6.3
4 . 5
3 . 2
Predicted ratios are calculated from equation
Ratio = EXP (4.8491 - 0.33424 * RVP)
-36-

-------
Table 6
High. A11 i tude Adjustment Factors
Evaporative Emissions
Vehicle	Model Year	Adjustment
Type	Group	Factor
LDGVs	Pre-1972	1.30
1972-76	*
1977	1.00
1978-81	1.30
1982-83	1.30
1984+	1.00
LDGTls	Pre-1972	1.30
1972-75
1977	1.00
1978-81	1.30
1982+	1.30
LDGT2S	All	1.30
HDGVs	All	1.3 0
MCs	All	1.3 0
* Based on actual test data.
-37-

-------
TABLE 7
Trip-Related Estimates (in Percent)
LDGVs and LDGTs
Source: 1979 GM-NPD Survey Data


Trip Days



No
Trip Days



S ingle
Partial



S i ng le



Fu 11
Diurna1 Days
3 +
No
No '
Full
2 +
Veh
Trip
DT
8AM-
10AM 8AM
MDI
DI
Trip
DI
MDI
Age
Days
Days
11AM
-3PM -2PM
Days
Days
Days
Days
Days
01
85.35
25 . 08
36.27
7.95 4.14
4.36
7 . 57
14 . 65
7 .15
7 ; 50
02
83 . 63
24 . 58
35 . 54
7.79 4.06
4.27
7 .42
16.37
7 .99
8 .38
03
81.91
24 .08
34 .81
7.63 3.98
4 . 18
7 .27
18 . 09
8 .83
9 . 2'6
04
80 . 19
23 .58
34 . 08
7.47 3.90
4 .09
7 . 12
19 .81
9 .67
10 . 14
05
78 . 47
23.08
33 . 35
7.31 3.82
4 . 00
6 .97
21.53
10.51
11.02
06	76.75	22.58	32.62	7.15	3.74	3.91	6.82	23.25	11.35	11.90
07	75.03	22.08	31.89	6.99	3.66	3.82	6.67	24.97 12.19	12.78
08	73.31	21.58	31.16	6.83	3.58	3.73	6.52	26.69	13.03	13.66
09	71.59	21.08	30.43	6.67	3.50	3.64	6.37	28.41 13.87	14.54
10	69.87	20.58	29.70	6.51	3.42	3.55	6.22	30.13	14.71	15.42
11	69.87	20.08 28.97	6.35	3.34	3.46	6.07	31.85	15.55	16.30
12	68.15	19.58	28.24	6.19	3.26	3.37	5.92	33.57 16.39	17.18
13	66.43	19.08	27.51	6.03	3.18	3.28	5.77	35.29	17.23	18.06
14	64.71	18.58	26.78	5.87	3.10	3.19	5.62	37.01	18.07	18.94
15	62.99	18.08 26.05	5.71	3.02	3.10	5.47	38.73	18.91	19.82
16	61.27	17.58	25.32	5.55	2.94	3.01	5.32	40.45	19.75	20.70
17	59.55	17.08	24.59	5.39	2.86'2.92	5.17	42.17	20.59	21.58
18	57.83	16.58	23.86	5.23	2.78	2.83	5.02	43.89	21.43	22.46
19	56^11	16.08	23.13	5.07	2.70	2.74	4.87	45.61	22.27	23.34
20	54.39	15.58	22.40	4.91	2.62	2.65	4.72	47.33	23.11	24.22
21	50.95
22	49.23
23	47.51
24	45.79
25+ 44.07
15 .08
14 .58
14 . 08
13 .58
13 .08
21. 67
20.94
20 .21
19 .48
18.75
75 2
59 2
43 2
27 2
11 2
-54 3
46 3
38 3
30 3
22 3
46 6.07
37 5.92
28 5.77
19	5.62
10 5.47
31.85
33 . 57
35 .29
37.01
38 . 73
15.55
16.39
17.23
18.07
18.91
16 .30
17. 18
18 .06
18	.94
19	.82
-38-

-------
ATTACHMENT
-39-

-------
Figure 1
1981+ Passing Vehicles LDGV Diurnal Emissions (g/tesi)
16 T
TBI
\
PFI
H	1	1-
0.3 0.4 0.5 0.6 07 0.8 0.9 1
H	1	1	1	1	1	\	b	1-
1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9
UDI
H	\	1	1	1	^	1-
2 2.1 2.2 2.3 2.4 2.5 2.6
10:34 AM 7/25/91

-------
Figure 2
1981+ Passing Vehicles LDGT Diurnal Emissions (g/iest)
CARB
10:30 AM 7/25/9 I

-------
Figure 3
1981+ Failed Vehicles LDGV Diurnal Emissions (g/test)
Failed
Pressure
failed
Purge
10:31 AM 7/25/91

-------
Figure 4
1961+ Failed Vehicles LDGT Diurnal Emissions (g/iest)
Failed
Pressure
Failed
Purge
8:38 AM 6/25/91

-------
Figure 5
1981+ Passing Vehicles at 82F LDGV Hot Soak Emissions (g/lest)
3.5
2.5
CARB
0.5
TBI
	
5
5.5
6
6.5
7
7.5
8
8.5
9
10
10.5
Fuel Volatility (in psi RVP)
9:11 AM 6/25/91

-------
Figure 6
1981+ Passing Vehicles at 82F LDGT Hot Soak Emissions (g/lest)
Fuel Volatility (in psi RVP)
10:10 AM 6/25/91

-------
Figure 7
1981+ Failed Vehicles at 82F LDGV & LDGT Hot Soak Emissions (g/lest)
ji.
o\

32

30
H
28
0
26
t


24
S
22
o

a
20
k
18

16
E
m
14
i
12
s

s
10
i
8
0

n
6

s
4

2

0
Failed
Purge
Failed
Pressure
5.5 6 6.5 7 7.5 8 8.5 9 9 5 10 10.5 11 115 12
Fuel Volalilily (in psi RVP)
9:14 AM 7/3/91

-------
RUNNING LOSS EMISSIONS
1.0 Introduction
Running loss emissions test program has been performed and
data collected by Automotive Testing Laboratories, Inc. (ATL)
located at South Bend, Indiana under EPA's Emission Factors Program
contract.
The test program was designed to test in-use vehicles with
three different driving cycles:
1)	a low speed cycle (known as the New York City Cycle
[NYCC]) with an average speed of 7.1 mph,
2)	a Federal Testing Procedure (FTP) LA-4 driving cycle
with an average speed of 19.6 mph, and,
3)	a high speed cycle (or Highway Fuel Economy Test [HFET])
with an average speed of 47.9 mph.
The duration of the running loss test is approximately one hour for
each driving cycle. Therefore, the NYC driving cycle is repeated
six times (6 bags), the two portions of the LA-4 cycle, are repeated
three times (6 bags), and the HFET driving cycle is repeated five
times (5 bags). The distances, average speeds, and durations of
these three driving cycles are summarized in the following:


Distance
Speed
Duration
Dr iving

in
i n
in
Cycle
Baqs
Miles
mh
Minutes
NYCC
1 to 6
1. 18
7. 1
10.0
LA-4
1,3,5
3 . 59
25.6
8 . 4

2,4,6
3.89
16.2
14 . 4
HFET
1 to 5
10.20
47.9
12.8
Under this test program, many vehicles were tested only
cycles. Fewer vehicles were tested on HFET cycle.
on LA-4
The running loss emissions test program	was designed to
collect data at four levels of fuel volatility	(7.0, 9.0, 10.4,
11.7 psi in Reid Vapor Pressure [RVP]) and three	levels of ambient
temperature (80, 95, and 105F).
-47-

-------
Not all vehicles were tested for all combinations of fuel RVPs
and ambient temperatures, however. There is usually no testing at
extreme conditions, such as the combinations of high RVP fuel and
high ambient (11.7 psi/105F) , and low RVP fuel and low ambient
(7.0 psi/80F), because of their less likely occurrences in the
real world. Also, if from a test vehicle the running loss emission
results are low (less than 0.5 grams) at certain fuel and
temperature combination (for example, 9.0 psi/95F), it is assumed
that at the combinations of lower fuel volatilities and/or lower
ambient temperatures (i.e., 7.0 psi/95F, 9.0 psi/80F, and,
7.0 psi/80F), this vehicle would have emissions at a similarly low
level. To save resources, the vehicle is not tested for the
combinations of lower fuel volatilities and lower ambient
temperatures. Further, there has been no testings on 11.7 psi RVP
fuel shortly after the issuance of MOBILE4 in 1989.
-48-

-------
2.0 True Vapor Pressure
In MOBILE4 model, when the test data is. not available at
certain combinations of fuel volatility and ambient temperature,
the g/riii running loss emissions were estimated from a variable
called "True Vapor Pressure (TVP)." In MOBILE4.1 model, this TVP
is used to correlate with the running loss emissions from failed
vehicles. These TVPs by bag are expressed as functions of fuel
volatility and fuel tank temperature, and are calculated by the
following five steps of equations:
U7 = 1.0223 *RVP + [0.0357 *RVP / ( 1.0 - 0.0368*RVP)]
V7 = 66.401 - 12.718 *U7 + 1.3067*U7**2 - 0.077934*U7**3
+ 0 . 0018407*U7*M
T0 i = Tmin + DTB i	for i=l,2,...,6
CD7Bi = 262.0 / [(V 7/6.0 + 560.0) - 0.0133] * (100.0 - Ta ; )
+ V7
TVPs = 14.697 - 0.53 OS *CD7 a i + 0.0077215*CD7B, **2
- 0 .000055631*CD7B; **3 + 1.769* 10" '*CD73, * *4
for i=l,2,...,6
where:
RVP = fuel volatility in psi,
TBi = cumulative tank temperature
Tmin = initial tank temperature in
DTai =cumulative tank temperature
and,
TVPBi = calculated bag i TVP value.
The TVP values were calculated for all combinations of fuel
volatilities (7.0, 9.0, 10.4, and 11.7 psi RVP) and tank
temperature profiles (with the initial tank temperatures at 80, 87,
95, and 105F).
for bag i in F,
F,
rise for bag i in F,
The tank temperature rises for each bag were estimated from a
44-car tank temperature data collected at Laredo, Texas during the
time frames of September 11 through October 23, 1989, by ATL under
EPA' s contract. In- this program, vehicles' tank temperatures were
measured and recorded while being driven under the LA-4 driving
cycle repeated for three times (thus, a total of six bags). The
ini.tial tank temperatures were adjusted to be within + '2F of the
ambient temperatures at the start of the test. An ambient
temperature of 95F was used. The fuel tanks were filled at 40%
level with 9.0 psi RVP fuel.
-49-

-------
The initial and final tank temperatures for each bag from
these 44 vehicles were examined and an average tank temperature
rise for each bag was calculated. The resulting average tank
temperature rises for the six bags are: 5, 9, 4, 5, 2, and 3F,
respectively. The cumulative tank temperature rises (5, 14, 18,
23, 25,' and 28F), then, were used in the above five equations to
calculate the corresponding TVP values. Therefore, the initial and
final tank temperatures for the four temperature profiles were:
Initial Tank
Temp in "F
BO. 0
87. 0
95.0
105. 0
	Final Tank Temperature in F	
Bag 1 Bag 2 Bag 3 Bag 4 Bag 5 Bag 6
85 . 0
92 . 0
100 . 0
110 . 0
94 . 0
101. 0
109 . 0
119 .0
98 . 0
105.0
113 . 0
123 . 0
103 . 0
110.0
118.0
128 . 0
105.0
112 . 0
120.0
130 . 0
108 . 0
115 . 0
123 . 0
133 . 0
Table 1 summarizes the calculated TVP values for each bag under
each fuel volatility
temperature.
level and each initial ambient (tank)
-50-

-------
3.Q Running Loss Emissions Data
Similar to EPA's Hammond program, all vehicles under the
running loss emissions test program since FY90 were also checked
for pressure and/or purge failures of their evaporative emissions
control system. Descriptions of these pressure/purge tests were
given in the Evaporative HC Emissions Chapter. Failure rates from
either pressure and/or purge test were described as a function of
vehicle's age. These failure rates are the same as those used for
the diurnal/hot soak emissions model.
Therefore, the M0BILE4.1	running loss emissions model include
three portions: emissions	for pass vehicles, emissions for
vehicles that failed pressure	test, and emissions for vehicles that
failed purge test.
For vehicles tested prior to the Hammond program, diagnoses
and comments from mechanics and their running loss emissions test
results at 9.0 psi/95F were examined to see which category the
vehicles should be grouped into. For example, if diagnoses and
comments from mechanics indicated that there exists a leakage on
one of the evaporative emissions control system components {such as
gas cap, filler neck, sending unit, rollover valve, or vent hoses),
the vehicle is categorized as "failed pressure test". If there
exists an in-operative canister purge solenoid or valve, or
disconnected, missing, or' damaged purge hoses, the vehicle is
grouped as "failed purge test." If there were no problems
detected, the vehicle is categorized as "pass vehicle."
Among those pre-Hammond vehicles that had problems detected in
their evaporative emissions control system components  by the
mechanics' diagnoses and comments, some of them had relatively low
running loss emissions. Among those pre-Hammond vehicles that had
no visible problems detected in their evaporative emissions control
system components, some of them had very high running loss
emissions. Due to this lack of consistency, it was decided that
all LDGVs tested prior to Hammond program be excluded from
MOBILE4.1 model. For LDGTs, however, the sample sizes were very
limited, their running loss emissions test results were relatively
consistent, with no visible problems detected. For these reasons,
all LDGT data were used as pass vehicles for the LDGT fleet.
-51-

-------
3.1 Exception Vehicles
Among the data used for M0BILE4.1 analysis, there were two
vehicles with "unusual" test results. One of them is a 1986
Chevrolet Camaro, identified as vehicle #1532, which is equipped
with a' 5. OL engine, ported fuel injection fuel delivery system, by
engine family G1G5.0V8NTA8, and with evaporative emissions control
system 6B0-1A. This vehicle, when tested at Hammond I/M lane,
passed pressure check but failed purge flow^ check. The other
vehicle is a 1986 Pontiac Sunbird, identified as vehicle #1578,
which is equipped with a 1.6L engine, throttle body fuel injection
fuel delivery system, by engine family G2G1.8V5TDG2, and with
evaporative emissions control system 6A0-2B. This vehicle was
found, to have a leaking gas cap. It failed pressure check but
passed purge flow check.
Two things were observed on these two vehicles: their tank
temperature rises were extremely high, and their emission levels
were equally high under as-received as well as after repair
conditions. As discussed above (in section 2.0), the average tank
temperature rise from an initial ambient of 95F is 28F. The tank
temperature rises for these two vehicles, however, were more than
4 0  F. The cumulative temperature rises from the six bags were:
12, 22 , 33 , 40, 44, 47F for the Camaro, and 9, 26, 32 , 37, 41,
43F for the Sunbird. The following table shows their running loss
emissions in total grams (the sum of 6 bags) tested at 9.0 psi/95F
from LA-4 driving cycle:
Vehicle	Total .Emissions
Number		Status		(G rams)
1532	as-received (failed purge)	315.5
after reconnect purge line	330.8
after replace gas cap	319.9
1578	as-received (failed pressure)	263.2
after repai r	325.5
Note that these repair/maintenance steps of reconnecting purge
line and replacing gas cap usually will reduce a major portion of
the evaporative running loss emissions for most' vehicles. The
above table has illustrated that for these two vehicles, however,
the high levels of emissions may not be caused by either pressure
or purge problems in their evaporative emissions control system.
The repair work failed to decrease their emissions, even increased
the total grams slightly. Thus, their extremely high tank
temperature rises could be the main reason why these two vehicles
had such high running loss emissions. For these reasons, it was
decided that these two vehicles be treated as exceptions. Their
average emissions from as-received and after repair are to be
included in both the pass and failed vehicle categories.
-52-

-------
Market shares of these two engine families for model years
1983 through 1990 were obtained from EPA's Certification files.
The "Camaros" included all models of Caniaro and Firebird with
either 5.0L or 5.7L engines. Their sales ranged from a low 0.46 to
a high 2.02 percent, with . an average of 1.06 percent. The
"Sunbicds" were models under the names of Sunbird, Skyhawk, and
Firenza with either 1.8L or 2.0L engines. Their sales ranged from
0.48 to 2.01 percent, with an average of 1.28 percent. An
alternative was to derive an estimate based on the sample. As
there were a total of 60 LDGVs in the MOBILE4.1 running loss
emissions test program sample, weighting factors of 0.01667 each
are used to represent the Camaros and the Sunbirds in the fleet
(i.e., a weighting factor of 0.96666 is used for the "regular"
vehicles).
Table 2 summarizes the calculated TVP values for each bag
under each fuel volatility level and each initial ambient
temperature for these two exception vehicles. Table 3 shows the
running loss emissions (in cumulative grams) on these two vehicles
to be used for failed vehicles. Table 6 shows the running loss
emissions (in grams per nile) of these two vehicles to be used for
pass vehicles.
-53-

-------
3.2 Pass Vehicles
Among the "regular" vehicles, those that passed both pressure
and purge checks are categorized as the "pass" vehicles. For
example, there are 32 model year 1981+ pass LDGVs tested at
9.0 psi'/95F for the LA-4 driving cycle.
The running loss emissions for pass vehicles are expressed in
unit of grams per mile (g/mi) for each bag (bags 1 through 6) under
each driving cycle (NYCC, LA-4, and HFET). Note that there are
only five bag testing results from the HFET driving cycle. As the
emission levels from five bags under the combinations of 95F and
fuel volatilities of 9.0 and 11.7 psi RVP are relatively stable,
the emissions from the sixth bag of the HFET driving cycle are
assumed to be the same as the emissions from the fifth bag.
Therefore, there are a total of 18 (6 bags * 3 speeds) emission
values for the combination of four fuel volatility levels (7.0,
9.0, 10.4, 11.7 psi RVP) and four ambient temperature profiles (80,
87, 95, 105 F).
The g/mi running loss emissions are derived by dividing the
total grams by the cumulative trip distances in miles. For
example, for bags 1 and 2:
= RLgi,j / Dsi.mi
RL 0 ! , g/m i
= (RLb i , j + RLb j , g) / (Dai,mi .+ Dez.mi)
where:
D = trip distance in miles.
The average g/mi running loss emissions for each bag under each
driving cycle for 1981+ "regular" pass LDGVs and 1981+ pass LDGTs
are given in Tables 7 and 9. Note that the sample sizes for the
combinations of fuel volatility and ambient temperature vary, due
to the limitation of the test data discussed previously in sections
1.0 and 3.0. When there is no actual test data available, the
sample size is denoted as zero ("0"), and the emission values are
estimated by linear interpolation/extrapolation. Table 8
summarizes the g/mi emissions used for MOBILE4.1 model 1981+ pass
LDGVs, including the "regular" 'pass vehicles and two exception
vehicles.
-54-

-------
3.3 Failed Vehicles
Due to the limited sample sizes, there is no differentiation
made between vehicles that failed pressure test and vehicles that
failed purge test. The criteria used to define vehicles that
failed -either pressure or purge test are the same as those used for
hot soak/diurnal emissions model.
The running loss emissions' for failed vehicles in the
M0BILE4.1 model are expressed in terms of cumulative grams by bag
for the combinations of fuel volatilities and ambient
temperatures. Based on their similar emission levels in cumulative
grams, results from the NYC and LA-4 cycles were combined. The
emissions are also used for HFET cycle. The emissions in
cumulative grams are to be converted onto grams per mile unit (and
adjusted for speed) in the M0BILE4.1 model, as to be discussed
later in section 4.0. Emissions for failed LDGVs are used also for
LDGTs, due to a lack of LDGT data. Further, failed emission levels
are also used for vehicles made during the pre-control era.
Table 4 summarizes the running loss emissions used for
"regular" failed vehicles. The actual test data by each bag were
used to fit equations as a function of True Vapor Pressure (as
discussed in section 2.0). The predicted emissions in cumulative
grams are used in MOBILE4.1 model for the failed emission levels.
Table 5 shows failed emissions used for LDGVs and LDGTs, including
the "regular" failed vehicles and two exception vehicles.
-55-

-------
4.0 Driving Characteristics
In M0BILE4, a VMT weighted urban type driving characteristics
was built into the running loss emissions model so that the
emission rates were independent of the vehicle speed. This set of
urban VMT weighting factors was derived from tabulated frequencies
by trip distance in miles and average speed in mph matching the
three driving cycle characteristics used in the running loss
emissions test program. The data sources were the 1979 GM-NPD
Survey Data, and an Operational Characteristics Study sponsored by
EPA in Columbus, Ohio.
In MOBILE4.1, however, a VMT weighted driving characteristics
in terms of trip duration alone is used in the model so that the
emission rates can be adjusted by any given vehicle speed. The
1979 GM-NPD Survey Data again are used. The frequency table was
based on trip durations in 10-minute intervals, and the deriving
VMT weighting factors are:
Trip Duration	Corresponding Percent
in minutes	Bag #	VMT
10
and less
1
33.227
11
- 20
2
32.883
21
- 30
3
14.871
31
- 40
4
7 . 886
41
- 50
5
3 . 645
52
and up
6
7.488


Total:
100 . 0
This new set of percent VMT weighting factors is used to represent
each bag of the three driving cycles.
4 .1 Pass Vehicles
For pass vehicles in the M0BILE4.1 model, the running loss
emissions are given for all six bags of the three driving cycles in
grams per mile unit, for each of the four fuel volatilities and
four ambient temperatures. The speed adjustment algorithm for pass
vehicles are described in the following.
At any given average vehicle speed between 7.1 and 47.9 mph,
the emissions are calculated by linear interpolation between	the
emission rates at the two adjacent default vehicle speeds.	For
example, emissions at 10.0 mph are to be calculated based on
emissions at 7.1 mph (NYCC) and emissions at 19.5 mph (LA-4),	and
emissions at 30.0 mph are based on emissions at 19.6 mph (LA-4)	and
emissions at 47.9 mph (HFET):
-56-

-------
ADJ = (SPD - SPD1) / (SPD2 - SPD1)	(4.1)
where:
ADJ = calculated speed adjustment factor,
SPD = user input speed in mph, which is between 7.0 and
47.9 mph,
SPD1 = average speed in mph for the default driving
cycle 1 (either NYCC or LA-4),
SPD2 = average speed in mph for the default driving
cycle 2 (either LA-4 or HFET).
Then, with the calculated speed adjustment factor, the g/mi running
Loss emissions are estimated by subtracting the resting loss
emissions, and weighted by the VMT weighting factors:
RESTL &/m; = RESTLg~h r / (6.0 * SPDmi/hr)	(4.2)
RL b j ,s p d
= [ RL B i , S P D I + ADJ * ( RL 9 i , S P D 2 - RL B i , S P D 1 )
- RESTL g/mi ] * VMTaI tor i = l,2	6	(4.3)
whe re:
resting loss emissions,
calculated g/mi bag i running loss emissions at
speed SPD,
g/mi bag i running loss emissions for the default
driving cycle 1 (either NYCC or LA-4),
g/mi bag i running loss emissions for the default
driving cycle 2 (either LA-4 or HFET),
bag i VMT weighting factor.
For any average speed, below 7.1 mph but greater than 2.5 mph,
the emissions are calculated from the emissions at 7.1 mph and an
adjustment factor calculated from the ratio of the two speeds (the
input speed divided by 7.1 mph). This same algorithm is applicable
to where, the speed is above 47.9 mph but less than 65.0 mph.
Therefore, equations (4.1) and (4.3) above are now reformulated as:
ADJ = SPD / SPD1
RL g i ,s p d
= [RLbi.spdi + ADJ * RLb i , s pd i  RESTL g / m , ]
* VMTbi	for i = 1,2, . . . , 6
whe re:
RESTL =
RL B i , S P D =
RL B i , S P D I =
RL B i , S P D 2 =
VMTa i =
SPD = user input speed in mph, which is either between 2.5
and 7.1 mph, or between 47.9 and 65.0 mph,

-------
SPD1 = average speed of 7.1 mph from NYC driving cycle if SPD
is less than 7.0 mph, or average speed of 47.9 mph from
HFET driving cycle if SPD is greater than 47.9 mph,
RLbl.spdi = g/mi bag i running loss emissions from NYC
driving cycle if SPD is below 7.1 mph, or those
from HFET driving cycle if SPD is above 47.9 mph.
Finally, the running loss emissions are the total of the g/mi
emissions over the six bags:
RL = ELb I , S PD + Rla 2 , S PD + RL B 3 , S P D + RL B 4 , S PD
+ RLbs,spd + RLiB6,spd	(4.4)
4.2 Failed Vehicles
The running loss emissions for failed vehicles in the
MOBILE4.1 model are expressed in cumulative grams. For each
combination of fuel volatility and ambient temperature, there are
only six emission values, each representing the cumulative bag
emissions for all three driving cycles. The speed adjustment
algorithm for failed vehicles are based on the following two
assumptions:
1)	the durations for each bag of each driving cycle are
approximately the same, (i.e., about 10 minutes),
2)	the cumulative emissions in grams by bag under each
driving cycle from failed vehicles are the same.
For failed vehicles, the speed and VMT adjustment algorithm
includes converting the emissions from cumulative grams onto grams
per mile unit by the given speed, as well as subtracting off the
resting loss emissions and multiplying VMT weighting factor:
RESTL g / I 0 min= RESTLg/rir / 6 . 0 | o m i n rt r
RLb 1 , I = [ ( RL a , , g' - RESTL) * VMT B i ] / DSi,mi
= (RLb i . ^ - RESTL) * VMTs, * 6 / SPD
RLB2.g/m, = [ (RLei , s - RESTL) * VMT a z ]
/ (Db 1 ,mi + Da ! ,mi ) 
= (RLS2,K - RESTL) * VMTBi / (2 * DB, ,mi )
= (RLbz.k - RESTL) * VMTsi * 6 / (2 * SPD)
t hen,
RL b i , g/m i = (RLm.g - RESTL) * VMT B i * 6 / (i * SPD)
for i = 1,2,. ..,6
-58-

-------
and,
RLs P D = RLb 1 , g /m i + RL BJ , g/ni + RLb 3 , J /m i
+ RLb 4 , g / tn I	RL B 5 t g / rn i +	0 6 , g / m i
where:
RESTL = resting loss emissions,
RLbi ,g/mi = calculated g/mi bag i running loss emissions at
speed SPD,
RLbi,g = bag i cumulative grams running loss emissions,
VMTbi = bag i VMT weighting factor,
DBi,mi = trip distance in miles for bag i, and,
SPD = user input speed in mph.
-59-

-------
5.0 Other Model Years and Other Vehicle Types
The running loss emissions data described in Tables 5, 7, and
9 are used in MOBILE4.1 or other model year groups and other
vehicle 'types according to the following mapping:
Data
Pass/
Fail
LDGV
Vehicle Type
LDGT1
LDGT2
HDGV
1981+ Pass LDGV
Pass
Pass
Pass
1972-77
1978-80
1981 +
1985 +
1981+ Pass LDGT1
Pass
Pass
Pass
1972-77
1978-80
1981 +
1979-80
1981 +
1981+ Failed
LDGV All
Failed
Pre-19 72
1972 +
Pre-1972
1972 +
Pre-19 79
1979 +
Pre-1985
1985 +
-60-

-------
Table 1
True Vapor Pressure
Regular Vehicles
Fuel Initial
RVP Ambient 	True Vapor Pressure
in psi
7.0
9". 0
10 . 4
11.7
Temp  F
Baq 1
Baq 2
Bad 3
Baq 4
Baq 5
Baq 6
80.a
5 . 58
6 . 67
7.21
7. 94
8.24
8 . 72
87. a
6.42
7 . 64
8 . 24
9 . 04
9 . 38
9 .91
95.0
7. 50
a .88
9.56
10.46
10 . 84
11. 44
105 . 0
9 . 04
10 . 65
11.44
12 .48
12 . 92
13 . 60
80.0
7 .28
8 . 63
9 .30
10 . 19
10.56
11. 14
87.0
8.32
9 . 8 2
10 . 56
11 . 54
11.95
12 . 60
95 . 0
9 . 64
11.34
12 . 17
13 . 27
13 .73
14 . 45
105 . 0
11.54
13 .50
14 .45
15.71
16 . 24
17 . 06
80 . 0
8 . 63
10 . 19
10.95
11.96
12.39
13 .05
87. 0
9 . 82
11.5"5
12.39
13 . 50
13 . 97
14 . 70
95 . 0
11. 34
13 . 27
14.21
15 .46
15 .98
16 . 80
105 . 0
13 . 50
15 . 72
15 . 80
18 .22
18.82
19 . 74
80.0
9 .76
11.47
12.31
13 .42
13 .89
14 .61
87.0
11. 07
12.96
13 .89
15 . 11
15 . 63
16 . 43
95 . 0
12 . 74
14 .86
15 . 89
17.26
17.83
18 .71
105 . 0
15.11
17 .54
18.71
2 P . 2 7
20 .92
21. 93
-61-

-------
Table 2
True Vapor	Pressure
Exception	Vehicles
Fuel Initial
RVP Ambient 			True Vapor Pressure	
in psi Temp F Bag 1 Bag 2	Bag 3	Bag 4	Bag 5	Bag 6
Camaro
7 Q 80.0 6.42 7.79	9.56	10.84	11.64	12.27
37.0 7.35 8.88	10.84	12.27	13.14	13.84
95.0 8.55 10.28	12.48	14.07	15.05	15.82
105-0 10.28 12.27	14.80	16.63	17.75	18.63
9 o 80.0 8.32 10.00	12.17	13.73	14.69	15.45
87.0 9-47 11.34	13.73	15.45	16.51	17.34
95.0 10.94 13.04	15.71	17.62	18.80	19.72
105.0 13.04 15.45	18.50	20.68	22.02	23.07
10	4 80.0 9.82 11.75	14.21	15.98	17.07	17.93
87 0 11.14 13.27	15.98	17.93	19.12	20.06
95.0 12.82 15.21	18.22	20.38	21.70	22.74
105.0 15.21 17.93	21.37	23.81	25.31	26.48
11	7 80 0 11.07 13.19	15.89	17.83	19.02	19.95
87.0 12.52 14.86	17.83	19.95	21.25	22.27
95.0 14.37 16.97	20.27	22.62	24.06	25.18
105.0 16.97 19.95	23.69	26.35	27.97	29.24
Sunbird
7 . 0
80 . 0
6 . 05
8 .40
9.38
10 . 28
11.04
11.44

87 . 0
6 . 94
9 .56
10 . 65
11.64
12.48
12 .92

95 . 0
8 . 09
11 . 04
12 . 27
13 . 37
14.31
14 .80

105 .0
9 . 73
13 . 14
14 . 56
15 . 32
16 .90
17 .46
9 . 0
80 . 0
7 . 86
10 .75
11.95
13 .04
13 .96
14 . 45

87.0
8 . 96
12 . 17
13 . 50
14 .69
15.71
16 . 24

95 . 0
10 . 37
13 . 96
15.45
16 . 78
17.91
18 . 50

105.0
12 . 38
16 .51
18 . 21
19 .72
21.01
21.68
10	4 80 0	9.30	12.60	13.97	15.21	16.25	16.80
87.0	10.56	14.21	15.72	17.07	18.22	18.82
95.0	12.17	16.25	17.93	19.43	20.71	21.37
105.0	14.46	19.12	21.03	22.74	24.18	24.93
11	7 80.0	10.49	14.13	15.63	16.97	18.12	18.71
87.0	11.88	15.89	17.54	19.02	20.27	20.92
95.0	13.65	18.12	19.95	21.59	22.97	23.69
105.0	16.16	21.25	23.33	25.18	26.75	27.56
-62-

-------
Table 3
Cumulative Grams Running Loss Emissions
Exception Vehicles
Fuel
Ambient
in psi
Temp F
Bag 1
Bag 2
Bag 3
Bag 4



Carnaro

7.0
80 . 0
1.83
2 . 27
3 . 83
8 .06

87.0
1.96
3 .99
7 . 98
11.20

95. 0
2 .15
7. 00
16 . 62
32 .01

105 . 0
2.47
12 . 07
30 . 78
82 .27
9 . 0
80 . 0
2.11
4.12
5 . 07
33 . 30

87. 0
2.31
8 . 62
19 .69
45.99

95 . 0
2 .61
15 . 16
40.76
130.51

105 . 0
3 . 11
25 .98
75 .27
378.61
10 . 4
80 . 0
2 .38
10 . 12
24 .59
59 .71

87 . 0
2 . 65
16.13
43.92
148 .49

95 . 0
3 . 05
24 . 80
71 . 58
347.42

105 . 0
3 .71
39.04
116.55
789.42 .
11.7
80 . 0
2 . 64
15 . 78
42.82
142.28

87 . 0
2 .98
23 . 17
66.43
305.45

95 . 0
3.47
33.78
100 .05
614.02

105 . 0
4 . 28
51.11
154 .42
818.00*



Sunbi rd

7 . 0
80 . 0
0 .71
2 . 00
4 . 24
6 . 68

87 . 0
1.40
4.66
5 .43
17 . 12

95.0
2.42
8 . 48
12 . 92
43 . 89

105 . 0
4 . 10
22 . 72
34 . 93
87 . 78
9 . 0
80 . 0
1 . 52
6 . 73
17 . 02
18 . 73

87 . 0
2 . 97
15 . 79
21.95
48 . 02

95 . 0
5 . 10
28 .93
51.91
123.67

105 . 0
8 .66
50 .59
139.24
24 7.17
10 . 4
80 . 0
3 .45
18 . 83
26 . 83
65 . 67

87 . 0
5.41
30 . 89
58 . 18
135.05

95 . 0
8.26
48 . 24
128.15
234.04
Bag 5 Bag 6
12.55	36.78
16.73	49.04
47.79	122.60
120.61	240.39
50.25	100.62
67.15	134.16
192.70	336.79
565.10	664.77
87.34	181.88
219.75	367.54
518.46	630.51
818.00*	818.00*
210.54	357.42
455.53	582.20
818.00*	818.00*
818.00*	818.00*
10.05	16.20
25.76	4 1.53
66.05	94.39
132.10	178.09
28.42	41.96
72.88	107.60
185.66	241.92
369.14	460.06
99.29	139.12
202.69	262.23
349.77	437.11
105.0	13.01 76.64 283.75 394.43 587.29	719.06
11.7	80.0	5.29 30.19 55.95 131.16 196.95	255.44
87.0	7.72 44.99 113.31 215.68 322.60	404.86
95.0	11.24 66.18 221.83 335.71 500.51	616.16
105.Q	17.07 100.64 441.95 529.16 719.06*	719.06*
*An upper limit based on actual test results.
-63-

-------
Table 4
Cumulative Grams Running Loss Emissions
Failed Vehicles
Fuel
RVP Ambient
in psi Temp F
7.0
9 .0
80.0
87.0
95.0
105.0
80.0
87 . 0
95.0
105.0
Running Loss Emissions (Cumulative Grams),
Bag 1 Bag 2 Bag 3 Bag 4 Bag 5 Bag 6
0	. 50
1. 02
1. 52
3 . 25
1	.23
2	. 52
3	.75
4	. 79
1.44
2 .48
7 . 27
14 . 13
6 .32
10 .92
16 . 79
25 . 13
3.26
6 . 80
11.34
26.24
9 .28
19 . 29
32 .04
50 . 13
4.33
10 . 07
17 . 37
44 . 20
13 . 69
31.71
54 . 63
87 . 09
4.91
12 . 27
21. 52
58.84
16.41
41.49
73 . 37
118.47
5 . 83
15134
27.40
77 . 40
20 . 58
54 . 20
96.87
157.15
10 . 4
11.7
80 . 0
87.0
95.0
105 . 0
80 . 0
87 .0
95.0
105 .0
85
90
72
90
12.34
17 . 59
24 .28
33.74
22.37
33.79
4 8.28
68 . 78
37.28
57.80
83 . 80
120 . 49
49.25
77.78
113.91
164.85
64 . 60
102 .78
151.10
219.11
3.85	17.30	33.15	56.67 76.21	100.70
4.62	23.08	45.69	79.17 107.49	142.53
4.97 30.42	61.58	107.64	147.02	195.33
6.66 40.78	83.99	147.69	202.59	269.48
-64-

-------
Table 5
Cumulative Grams Running Loss Emissions
1981+ Failed LDGVs and LDGTs
Fuefl
RVP Ambient	Running Loss	Emissions (Cumulative	Grams)
in psi Temp F Bag 1	Bag 2	Bag 3	Bag 4	Bag 5	Bag 6
7.0 80.0	0.53 1.46	3.29	4.43	5.12	6.52
37.0	1.04 2.54	6.80	10.21	12.57	16.34
95.0	1.55 7.29	11.45	18.06	22.70	30.10
105.0	3.25	14.24	26.46	45.56	61.09	81.80
9.0	80.0
87.0
95 . 0
105 . 0
10.4	80.0
87 . 0
95.0
105 .0
11.7	80.0
87 . 0
95 . 0
105 .0
1.25	6.29
2.52	10.96
3.76	16.97
4.83	25.57
2.85	12.41
3.90	17.79
4.75	24.69
5.02	34.54
3.85	17.49
4.64	23.45
5.05	31.07
6.79	4 1.95
9.34	14.10
19.34	32.22
32.52	57.05
52.04	94.62
22.48 38.13
34.37 60.60
50.00 90.70
73.16	136.21
33.69	59.34
47.16	85.22
64.89	119.88
91.13	165.22
17.17	22.27
42.44	56.42
77.23	103.29
130.10	170.66
50.72	67.80
82.23	109.85
124.59	163.86
182.78	237.42
80.46	107.56
116.88	154.23
164.10	212.73
222.58	287.77
-65-

-------
Table 6
Fuel
RVP
Grams per Mile Running Loss Emissions
Exception Vehicle: Camaro
Ambient 	Running Loss Emissions (Grams/Mile)
in psi Temp F	Bag 1	Bag 2	Bag 3	Bag 4	Bag 5	Bag 6
New	York City Cycle
7.0 80.0	1.55	0.96	1.08	1.71	2.13	5.19
87.0	1.66	1.69	2.25	2.37	2.84	6.93
95.0	1.82	2.97	4.69	6.78	8.10	17.32
105.0	2.09	5.11	8.69	17.43	20.44	33.95
9 0 80.0	1.79	1.75	1.43	7.06	8.52	14.21
87.0	1.96	3.65	5.56	9.74	11.38	18.95
95.0	2.21	6.42	11.51	27.65	32.66	47.57
105.0	2.64	11.01	21.26	80.21	95.78	93.89
10 4 80.0	2.02	4.29	6.95	12.65	14.80	25.69
87.0	2.25	6.83	12.41	31.46	37.25	51.91
95.0	2.58	10.51	20.22	73.61	87.87	89.06
105.0	3.14	16.54	32.92	167.25	138.64*	115.54"
11.7 80.0	2.24	6.69	12.10	30.14	35.68	50.48
87.0	2.53	9.82	.18.77	64.71	77.21	82.23
95.0	2.94	14.31	28.26	130.09	138.64*	115.54*
105.0	3.63	21.66	43.62	173.31	138.64*	115.54*
LA-4 Driving Cycle
7 . 0
80 .0
0.51
0.30
0 . 35
0 . 54
0 . 68
1.63

87.0
0.55
0 . 53
0 . 72
0 . 75
0.90
2 . 18

95.0
0 . 60
0 . 93
1 . 50
2 . 13
2 . 57
5.45

105 .0
0 . 69
1. 61
2 . 78
5 . 48
6.49
10 . 68
9 . 0
80 . 0
0.59
0.55
0.46
2 . 22
2 . 70
4.47

87.0
0 . 64
1. 15
1 . 78
3 . 07
3 .61
5.96

95.0
0 . 73
2 . 02
3 . 68
8 . 70
10 .37
14 .97

105 . 0
0 . 87
3 .46
6 . 79
25.24
30.40
29 .55
10 . 4
80 .0
0 .66
1.35
2 . 22
3 .98
4 . 70
8 . 08

87 . 0
0 . 74
2 . 15
3 .96
9.90
11.82
16.34

95.0
0 .85
3.31
6 . 45
23.16
27.89
28 .02

105 . 0
1 . 03
5.21
10 . 51
52 . 63
44 . 00*
36.36
11.7 80.0	0.74	2.10	3.86	9.49	11.33	15.89
87.0	0.83	3.09	5.99	20.63	24.50	25.88
95.0	0.97	4.50	9.02	40.93	44.00*	36.36*
105.0	1.19	6.81	13.92	54.53	44.00*	36.36*
-66-

-------
Table 6 (Continued)
Grams per Mile Running Loss Emissions
Exception Vehicle: Sunbird
Fuel
in psi
Temp F
Bag 1
Bag 2
Bag 3
Bag 4
Bag 5
Bag 6


New
York City Cycle



7 . 0
80 . 0
0.60.
0 . 85
1.20
1.42
1. 70
2 .29

87.0
1. 19
1.97
1. 53
3 . 63
4.37
5.87

95 . 0
2 . 05
3 . 59
3.65
9.30
11.19
13 .33

105 . 0
3.47
9 .63
9 . 87
18 .60
22.39
25.15
9 . 0
80 . 0
1 .29
2 .85
4.81
3 .97
4 .82
5.93

87 . 0
2. 52
6 . 69
6.20
10 . 17
12.35
15.20

95 . 0
4 .32
12 .26
14 .66
26.20
31.47
34 . 17

105 . 0
7 .34
21.44
39 . 33
52 .37
62 .57
64 .98
10 . 4
80 . 0
2.92
7.98
7 . 58
13.91
16 .83
19.65

87. 0
4 . 58
13 .09
16.44
28.61
34.35
37.04

95.0
7 . 00
20.44
36.20
49 . 58
59 .28
61.74

105 . 0
11.03
32.47
80.16
83 . 57
99.54
101.56*
11.7
80 . 0
4 .48
12 . 79
15.81
27 . 79
33.38
36.08

87. 0
6 . 54
19 .06
32.01
45.69
54 . 68
57. 18

95.0
9 . 53
28 . 04
62.66
71.13
84 . 83
87. 03

105 . 0
14 . 47
42.64
124 .84
112.11
133.28
101.56*
LA-4 Driving Cycle
7 . 0
80 . 0
0 . 20
0 .27
0.38
0.45
0 . 54
0 . 72

87 . 0
0.39
0 . 62
0 .49
1 . 14
1 . 39
1.85

95.0
0 .67
1. 13
1. 17
2 . 93
3 .55
4 .20

105 . 0
1 . 14
3 . 03
3 . 15
5 . 85
7.11
7.92
9.0
80 . 0
0 .42
0.-90
1. 53
1 .25
1.53
1.86

87.0
0. 83
2 . 11
1.98
3 . 20
3 . 92
4 . 78

95 . 0
1.42
3 .86
4 . 68
8 . 24
9 .99
10. 75

105 . 0
2.41
6.75
12.56
16.48
19 .86
20.45
10 . 4
80. 0
0 .96
2 .51
2.42
4.38
5 .34
6 . 18

87.0
1.51
4 . 12
5.25
9 .00
10.90
11.65

95 . 0
2.30
6 . 43
11.56
15.60
18 .81
19 .43

105 . 0
3 .62
10 . 22
. 25.59
26.30
31.59
31.96*
11.7
80.0
1.47
4 . 03
5 .05
8 . 74
10.59
11.35

87 . 0
2 . 15
6 . 00
10 . 22
14 .38
17 .35
17 .99

95. 0
3 . 13
8 . 82
20.00
22 .38
26.92
27.38

105 . 0
4 .75
13 . 42
39 . 85
35 . 28
42.30
31.96*
-67-

-------
Table 7
Grams per Mile Running Loss Emissions
Regular Pass Vehicles
Fuel
RVP Ambient
ps i Temp F
7.0
9 .0
10 . 4
ao. o
87.0
95.0
105 . 0
80.0
87 . 0
95.0
105 . 0
80 . 0
87.0
95 . 0
105 . 0
Sample
Size
0
0
12
12
0
0
8
14
0
0
0
0
	Running Loss Emissions	(Grams/Mile)
Bag 1 Bag 2 Bag 3	Bag 4	Bag 5	Bag 6
New York City Cycle
0.28 0.30 0.31	0.31	0.30 0.30
0.28 0.30 0.31	0.31	0.30 0.30
0.28 0.30 0.31	0.31	0.30 0.30
0.29 0.31 0.32	0.33	0.32 0.32
0 . 28
0 .29
0 .33
0 . 67
0 . 29
0.36
0.63
1.37
0.30
0.31
0 .35
0.91
0.31
0 . 40
0 . 85
2 .25
0.31
0 .32
0	. 37
1	.72
0 .32
0	. 50
1	. 58
6 .09
0.31
0.32
0 .38
2 . 80
0.32
0.61
2 . 55
13.71
0 .30
0.31
0 . 53
3 . 90
0.31
0 . 85
3 . 55
22 . 93
0.30
0.31
0 . 63
4 .87
0.31
1.04
4.45
29.25
11.7
80 . 0
87 . 0
95.0
105 . 0
0
0
0
0
0 .35
0	.59
1	. 10
1 . 95
0.38
0 . 77
1.73
3 .35
0.45
1.	39
4.40
9 . 65
LA-4 Driving Cycle
0.53
2 . 20
9 . 53
22 . 60
0. 74
3 . 08
15. 63
38 . 43
0 .73
3 .85
19 .88
49.17
7 . 0
80 . 0
0
0 .07.
0.09
0 .09
0 . 09
0 . 09
0 . 09

87 . 0
0
0 . 07
0 . 09
0 .09
0 .09
0 . 09
0 . 09

95.0
12
0 .07
0 .09
0 . 09
0, 09
0 . 09
0. 09

105 .0
12
0 .09
0 .11
0 . 18
0 . 11
0 .11
0.11
9 . 0
80 . 0
0
0.07
0 .09
0 . 09
0 .09
0 . 09
0.09

87.0
0
0 . 08
0.10
0 . 10
0 . 10
0 . 10
0. 10

95.0
32
0 . 12
0 . 17
0 . 18
0.21
0 . 23
0 .30

105 . 0
14
0 . 12
0 .22
0 . 29
0 . 34
0 . 65
1.00
10.4
80.0
4
0 .08
0 . 10
0 .10
0 . 10
0 . 10
0 . 10

87.0
0
0 . 12
0 . 17
0 . 19
0.22
0 .27
0.36

95.0
0
0 . 12
0 .22
0 .28
0.33
0 . 61
0 .92

105 . 0
0
0.25
0.42
1 .32
4 . 23
6 . 28
7 .58
11.7
80 . 0
0
0 . 12
0 . 17
0 . 19
0 .22
0 .26
0 .34

87 . 0
0
0 . 12
0.21
0 .27
0.31
0 . 55
0 . 82

95 . 0
0
0 . 20
0 . 34
0 .92
2 . 74
4 . 12
5 .05

105 . 0
0
0.36
0 . 58
2 .16
7.40
10 . 86
12.96
-68-

-------
Table 7 (Continued)
Grams per Mile Running Loss Emissions
Regular Pass Vehicles
Fue 1
RVP -Ambient
psi Temp 0F
7.0
9 . 0
80.0
87 . 0
95 .0
105 . 0
80 .0
87 . 0
95 . 0
105 . 0
Sample	Running Loss Emissions (Grams/Mile)
Size	Bag I Bag 2 Bag 3 Bag 4 Bag 5 Bag 6
0
0
0
0
0
0
7
0
Highway Fuel Economy Test
0.02 0.02 0.02
0 . 02
0 . 02
0 . 06
0 . 02
0 . 02
0 . 02
0 . 09
0 .02
0 . 02
0 . 02
0 . 02
0 . 02
0 . 02
0 . 02
0 . 02
0 . 02
0 . 02
0 . 02
0 . 02
0 . 02
0 . 02
0 . 02
0 . 02
0 . 02
0 . 02
0 . 02
0. 02
0 . 02
0 . 02
0 . 02
0 . 02
0 . 02
0 . 02
0 . 02
0 . 02
0 . 02
0 . 02
0 . 02
0 . 02
0 .02
0 .02
0 .02
0 .02
0.02
0.02
10.4 80.0	0
87 . 0	0
95.0	0
105.0	0
11.7 80.0	0
87.0	0
95.0	7
105.0	0
0.02	0.02	0.02
0.02	0.02	0.02
0.08	0.02	0.02
0.09	0.03	0.03
0.02	0.02	0.02
0.05	0.03	0.03
0.09	0.05	0.04
0.09	0.05	0.04
0.02	0.02	0.02
0.02	0.02	0.02
0.02	0.02	0.02
0.03	0.03	0.03
0.02	0.02	0.02
0.03	0.03	0.03
0.03	0.03	0.03
0.03	0.03	0.03
-69-

-------
Table 8
Fue 1
Grams per Mile Running Loss Emissions
1981+ Pass LDGVs
Running Loss Emissions (Grams/Mile)
in psi
Temp F
Baq 1
Bag 2
Bag 3
Baq 4
Baq 5
Baq 6


New
York City
Cycle



7. Q
80 . 0
0.31
0 .32
0.34
Q .35
0.35
0.41

87 . 0
0.32
0.35
0 .36
0.40
0.41
0.50

95.0
0 . 34
0 .40
0.44
0.57
0.61
0.80

105.0
0.37
0 . 55
0 . 62
0.92
1. 02
1.29
9 . 0
80.0
0.32
0 .37
0.40
0 .48
0.51
0 . 63

87.0
0 . 35
0.47
0 . 50
0 . 64
0 . 70
0 . 87

95.0
0 . 43
0.65
0.79
1 .27
1.58
1.97

105 . 0
0 .81
1.42
2.67
4 .92
6.41
7.36
10.4
80 . 0
0 . 36
0 .50
0.55
0 .76
0 .83
1.06

87 . 0
0.46
0.72
0.96
1 . 59
2 . 02
2 .49

95.0
0 . 77
1 .34
2.47
4 . 52
5 .89
6 . 82

105 . 0
1. 56
2 .99
7.78
17 .42
26 . 14
31.89
11.7 80.0	0.45	0.70	0.90	1.48	1.87 2.15
87.0	0.72	1.23	2.19	3.97	5.18	6.05
95.0	1.27	2.38	5.77	12.57	18.83	22.59
105.0	2.18	4.31	12.14	26.60	41.68	55.39
LA-4	Driving	Cycle
7.0 80.0	0.08	0.10	0.10	0.10	0.11	0.13
87.0	0.08	0.11	0.11	0.12	0.13	0.15
95.0	0.09	0.12	0.13	0.17	0.19	0.25
105.0	0.12	0.18	0.21	0.30	0.3 3	0.42
9;o 80.0	0.08	0.11	0.12	0.14	0.16	0.19
87.0 ' 0.10	0.15	0.16-	0.20	0.22	0.28
95.0	0.15	0.26	0.31	0.49	0.56	0.72
105.0	0.17	0.38	0.60	1.02	1.47	1.80
10.4 80.0	0.11	0.17	0.18	0.24	0.27	0.34
87.0	0.15	0.27	0.34	0.53	0.64	0.82
95.0	0.17	0.37	0.57	0.96	1.37	1.68
105.0	0.32	0.66	1.88	5.40	7.33	8.47
11.7 80.0	0.15	0.27	0.33	0.51	0.61	0.79
87.0	0.17	0.35	0.53	0.88	1.23	1.53
95.0	0.26	0.55	1.37	3.70	5.17	5.94
105.0	0.44	0.90	2.98	8.65	11.94	13.74
-70-

-------
Table 8 (Continued)
Grams per Mile Running Loss Emissions
1981+ Pass LDGVs
Fuel
RVP Ambient
in psi Temp F
7 .0
80 - 0
87 . 0
95.0
105 . 0
	Running Loss Emissions (Grams/Mile)	
Bag 1 Bag 2 Bag 3 Bag 4 Bag 5 Bag 6
Highway Fuel Economy Test
0 . 02
0 . 02
0 .02
0.06
0 . 02
0 . 02
0 . 02
0 . 02
0. 02
0 . 02
0.02
0. 02
0 . 02
0 . 02
0 .02
0.02
0 . 02
0 . 02
0 . 02
0 . 02
0 . 02
0 . 02
0 .02
0 . 02
9.0	80.0
87 . 0
95.0
105 . 0
10.4	80.0
87 . 0
95.0
105 . 0
11.7	80.0
87.0
95.0
105 . 0
0.02	0.02
0.02	0.02
0.02	0.02
0.09	0.02
0.02	0.02
0.02	0.02
0.08	0.02
0.09	0.03
0.02	0.02
0.05	0.03
0.09	0.05
0.09	0.05
0.02	0.02
0.02	0.02
0.02	0.02
0.02	0.02
0.02	0.02
0.02	0.0 2
0.02	0.02
0.03	0.03
0.02	0.02
0.03	0.03
0.04	0.03
0.04	0.03
0.02	0.02
0.02	0.02
0.02	0.02
0.02	0.02
0.02	0.02
0.02	0.02
0.02	0.02
0.03	0.03
0.02	0.02
0.03	0.03
0.03	0.03
0.03	0.03
-71-

-------
Table 9
Grams per Mile Running Loss Emissions
1981+ Pass	LDGTs
Fuel
RVP Ambient		Running Loss Emissions (Grams/Mile)	
in psi Temp "F Bag 1 Bag 2	Bag 3	Bag 4	Bag 5	Bag 6
New York City Cycle
7.0 80.0 0.45 0.41	0.34	0.29	0.29	0.32
87.0 0.49 0.45	0.41	0.37	0.39	0.44
9 5*0 0.55 0.50	0.48	0.45	0.43	0.42
105.0 0.37 0.55	0.62	0.92	1.02	1.29
9.0 80.0 0.61 0.61	0.77	0.35	0.99	1.21
87.0 0.60 0.54	0.62	0.67	0.78	0.93
95.0 0.56 0.46	0-53	0.61	0.71	0.78
105.0 0.81 1.42	2.67	4.92	6.41	7.36
10.4 80.0 0.62 0.56	0.66	0.74	0.88	1.05
87.0 0.71 0.63	0.86	1.07	1.34	1.66
95 0 0.80 0.71	1.14	1.57	2.08	2.67
105.0 1.56 2.99	7.78	17.42	26.14	31.89
11.7 80.0 0.66 0.57	0.72	0.82	1.11	1.66
87.0 0.77 0.74	1.23	1-96	2.78	3.80
95.0 0.89 0.90	1.74	3.09	4.40	5.93
105.0 2.18 4.31	12.14	26.60	41.68	55.39
0.41
0 .34
0.29
0.29
0.45
0.41
0.37
0.39
0 .50
0.48
0 .45
0.43
0.55
0 . 62
0 .92
1.02
0.61
0 . 77
0 .85
0 . 99
0. 54
0.62
0 . 67
0 .78
0 .46
0. 53
0 .61
0.71
1.42
2. 67
4 . 92
6.41
0 .56
0 . 66
0 . 74
0.88
0 . 63
0.86
1 .07
1.34
0.71
1. 14
1.57
2 . 08
2.99
7 .78
17 .42
26 . 14
0.57
0.72
0 . 82
1. 11
0. 74
1 . 23
1 .96
2 . 78
0.90
1.74
3 . 09
4.40
4.31
12 . 14
26.60
CO
ID
i1
sr
Driving
Cyc le


0. 14
0.11
0 . 07
0.06
0 . 14
0.11
0 .08
0.07
0 . 12
0 . 13
0 . 17
0 . 19
0.25
0.22
0 . 23
0.26
0.12
0 . 11
0 . 12
0.11
0. 10
0.09
0 . 09
0.09
0. 10
0 , 09
0 . 09
0.09
0 . 18
0 . 15
0.28
0.35
0. 15
0 . 13
0 . 14
0. 14
0 . 15
0 . 14
0 . 18
0 .20
0 . 16
0 . 16
0.24 .
0 . 28
0.16
0 . 16
0.26
0.31
0. 12
0 . 10
0.09
0.08
0 . 16
0 . 19
0.41
0 . 57
0 . 16
0 . 23
0 .53
0 .74
0 .16
0.21
0 . 58
0.91
7.0	80.0	G.12	0.14	0.11	0.07	0.06	0.05
87.0	0.13	0.14	0.11 0.08	0.07	0.07
95.0	0.09	0.12	0.13	0.17	0.19	0.25
105.0	0.23	0.25	0.22	0.23	0.26	0.32
9.0 aO.O	0.12	0.12	0.11	0.12	0.11	0.11
87.0	0.10	0.10	0.09	0.09	0.09	0.L1
95.0	0.10	0.10	0.09	0.09	0.09	0.11
105.0	0.25	0.18	0.15	0.28	0.35	0.51
10.4	80.0	0.15	0.15	0.13	0.14	0.14	0.16
87.0	0.16	0.15	0.14	0.18	0.20	0.24
95.0	0.18	0.16	0.16	0.24 . 0.28	0.39
105.0	0.18	0.16	0.16	0.26	0.31	0.43
11.7	80.0	0.17	0.12	0.10	0.09	0.08	0.11
87.0	0.16	0.16	0.19	0.41	0.57	0.95
95.0	0.16	0.16	0.23	0.53	0.74	1.07
105.0	0.23	0.16	0.21	0.58	0.91	1.72
-72-

-------
Table 9 (Continued)
Grams per Mile Running Loss Emissions
1981+ Pass LDGTs
Fuel
RVP- Ambient
in psi Temp F
7 . 0
80 . 0
87.0
95 . 0
105.0
Running Loss Emissions (Grams/Mile)
Bag 1 Bag 2 Bag 3
Bag 4
Highway Fuel Economy Test
0 . 02
0 . 02
0 . 02
0 . 06
0 . 02
0 . 02
0 . 02
0 . 02
0.02
0 .02
0 .02
0.02
0 . 02
0 . 0.2
0 . 02
0 . 02
Bag 5 Bag 6
0.02
0 .02
0 .02
0 .02
0 . 02
0.02
0 . 02
0 . 02
9 . 0
80 . 0
87.0
95 . 0
105 . 0
0 . 02
0 . 02
0 .02
0 . 09
0 . 02
0 . 02
0. 02
0.02
0.02
0 .02
0.02
0.02
0 . 02
0 .02
0 - 02
0.02.
0.02
0 . 02
0 .02
0.02
0 . 02
0 . 02
0 . 02
0 . 02
10 . 4
80 . 0
87 . 0
95 . 0
105 . 0
0 . 02
0 . 02
0 . 08
0 . 09
0 . 02
0 . 02
0 . 02
0 . 03
0 .02
0.02
0.02
0 .03
0 .02
0 .02
0 . 02
0 .03
0 . 02
0 .02
0 . 02
0 . 03
0 . 02
0 .02
0 .02
0 . 03
11.7
80.0
87.0
95 . 0
105 .0
0 . 02
0. 05
0 . 09
0 .09
0 . 02
0 . 03
0 . 05
0 .05
0 . 02
0.03
0 . 04
Q .04
0 .02
0 . 03
0 . 03
0 .03
0 .02
0 . 03
0 . 03
0 .03
0 .02
0 . 03
0 . 03
0 .03
-73-

-------
M0BILE4.1
Technology Distribution for Gasoline Light-Duty Trucks

CUP
CUP
CUP
Oplp
Oplp
Oplp
CUP
CUP
CLLP
Oplp
Oplp
Oplp
Oplp
Oplp
Modal
3wy
3wiy
Sway
3w*y
3way
3wsy
Ox3way
Ortway
OxSway
Ox] way
Ox3way
OrttMy


XtK
*fe
MPR
IBI
Cab
MPR
IBJ
Gath
MPR
IBI
Cub
HEB
IBI
Ox Id
Nona
1M1
0
0
0
13589
0
0
26593
0
0
0
0
0
1429886
46007
1982
57339
0
0
0
0
0
906
0
0
0
0
0
1429253
0
1983
74447
3438
0
0
0
0
183194
0
0
0
0
0
1658256
51277
1984
109748
65296
0
20757
0
0
631304
0
0
51116
0
0
2061626
0
1985
255529
206437
144973
16454
0
0
546263
0
368
66637
0
0
1868467
0
1986
167151
661000
326016
17760
0
0
308664
0
163878
0
0
0
1776740
0
1987
260447
716266
780032
48068
0
0
188663
429429
202385
64044
0
0
814666
0
1988
260231
1099548
1455516
15660
0
38829
101129
534243
283427
69315
0
0
72566
0
1989
174645
1544412
1435056
10998
0
0
185103
974850
288728
0
0
0
53200
0
1990
7965
1948880
1533017
9998
0
0
87597
684251
200673
0
0
0
40500
4969
1991
21900
2177487
1815964
6000
0
0
53055
712782
223736
0
0
0
0
0

CLLP
CUP
CUP
Oplp
Oplp
Oplp
CLLP
CLLP
CLLP
Oplp
Oplp
Oplp
Oplp
Oplp
IfcxM
Sway
Iwty
3wy
3way
3way
^Miy
Ox3ray
Ortway
Ox3nray
Ox3way
Ox3way
Ox3way


It*
Cafe
MPR
IBI
Cafe
MPR
IBI
Smtt
MEB
IBI
Cafe
MPR
IBI
QiM
ttfiDf
1981
0.0%
0.0%
0.0%
0.9%
OjO%
0.0%
ia%
0j0%
0.0%
0.0%
0.0%
0.0%
943%
3.0%
1982
3.9%
00%
op%
0.0%
0.0%
0.0%
01%
0.0%
0.0%
0.0%
0.0%
0.0%
96.t%
0.0%
1983
3.8%
02%
0.0%
0.0%
00%
0.0%
63%
0.0%
0.0%
00%
00%
00%
84 1%
2 6%
1984
3.7%
22%
0.0%
0.7%
00%
0.0%
213%
0.0%
0.0%
1.7%
00%
0.0%
70 3%
0.0%
1985
82%
6.6%
4 7%
05%
0.0%
9.0%
17.6%
0 0%
0.0%
22%
0 0%
0.0%
602%
0.0%
1996
4.6%
238%
9 1%
05%
00%
00%
85%
0.0%
45%
0.0%
0.0%
0.0%
49.0%
0.0%
1987
7.4%
20 4%
223%
14%
00%
0.0%
5 4%
123%
5.8%
18%
0.0%
00%
232%
0.0%
1988
6.6%
28.0%
37 0%
0.4%
0.0%
1.0%
2 6%
13.6%
72%
18%
00%
0.0%
1.8%
0.0%
1989
3.7%
33.1%
30.7%
02%
00%
00%
4 0%
20.9%
62%
0.0%
0 0%
0.0%
1.1%
0.0%
1990
02%
41 3%
32.5%
02%
0.0%
00%
1.9%
18.7%
43%
0 0%
0 0%
0 0%
0.9%
0 1%
1991
0.4%
435%
362%
0.1%
0.0%
0.0%
1.1%
142%
45%
00%
0.0%
0 0%
0 0%
0.0%
Ifcxtt
lam
CUP
CUP
CUP
Oplp

MckM
Any
Any
Any

Uodal
Any
Any
Any
Cm*
mph
IBI
AC*

Iter
Cede
MPR
IBI

*M(
iwy
0x?my
Ox Id
1981
1.8%
0.0%
0.0%
982%

1981
100.0%
0.0%
0.0%

1981
0.9%
1.8%
943%
1982
3.9%
0.0%
0.0%
96.1%

1982
100.0%
0.0%
0.0%

1982
3.9%
0.1%
96 1%
1983
13.1%
02%
0.0%
86.8%

1983
99.8%
02%
0.0%

1983
4.0%
9 3%
84 1%
1984
25 0%
22%
0.0%
72.8%

1984
97 8%
22%
0.0%

1984
6.6%
23.1%
70 3%
1985
2&a%
6.6%
4.7%
62 6%

1985
88 7%
6 6%
4 7%

1985
20.0%
19 8%
602%
1986
13.1%
23.8%
13.6%
49.5%

1986
62.7%
23.8%
13.6%

1986
37 9%
13 0%
49 0%
1987
12.8%
32.7%
28 0%
26 4%

1987
39.3%
32 7%
28.0%

1987
515%
252%
232%
1988
92%
41 6%
442%
5.0%

1988
132%
41.6%
452%

1988
73.0%
25.1%
1.8%
1989
77%
54.0%
36.9%
1.4%

1989
9.1%
54.0%
36.9%

1989
67.8%
31.0%
11%
1990
2.0%
60.1%
36.7%
12%

1990
32%
60.1%
36.7%

1990
742%
24.9%
0.9%
1991
15%
57.7% 
40.7%
0.1%

1991
i 6%
57 7%
40 7%

1991
803%
19 7%
0.0%
A# vehicle counts taken trom CAFE estimates, except 1989 and newer model years, which ate (torn General Label predictions
7/2AJ1

-------
APPENDIX B
PURGE AND PRESSURE TEST EFFECTIVENESS FIGURES AND SPREADSHEET

-------
Purge-Pressure Detection Rales
Annual
* W/Program
D No-Program
D '
-U i I
r>
LI
P

wU 4 t i
10
Age Index
15

-------
Purge-Pressure Detection Rates
Biennial
M
0.500
0.450
0.400
0.350
0.300
QC
 0.250
0.200
0.150
0.100
0.050
0.000
W/Program
u No-Program
.-o
~ -j-
Lw
.LI
LI
P
P
P'
 LI I j i.
f.r
/
/
[ i [j Lj I i
10
i : i
15
Age Index
20

-------
Revised Purge/Pressure Failure Rates for M0BILE4.1
ANNUAL





Ir.-lana
:!evf



Age
No-programi Detect
Annual
Annual
Annuai
Index
fail rate
Rate
After
averaoe
benefit
1
0 .060
0.0800
0.0000
0.0800
0 . 000
2
0 . 060
0.0800
0.0600
0.0600
0.238
2
o.oeo
0.0 600
0.0000
0.0080
0 . 853
4
0 .096
0.0160
0.0000
0.0035
0. 910
5
0 .103
0.0070
oooo
0.0005
0. 83 6
6
0 .120
3 .0170
0.0000
0.0150
0.789
7
0.15.0
0.0300
0.0000
0.0190
0.7 65
8
0 . 138
0.0380 -
0.0000
0.0350
0.732
o
0 . 253
0.0700
i.0000
0.0325
0 . 798
10
0 . 323
0.0 650
0.0000
0.0330
0. 837
11
0. 339
0.0 660
0 .0000
0.0265
0.903-
12
0.442
0.0530
0.0000
0.0125
0. 967
13
0.451
0.0250
0.0000
0.0035
0 . 992
14
9 . 451
0.0070
0.00 0 0
0.0085
0 . 981
15
0.451
0.0170
0.0000
0.0150
0. 967
15
0 . 451
0.0300
0.0000
0.0190
0. 956
17
0 .451
0.03S0
0.0000
0.0350
0 . 922
18'
0. 451
0.0700
. .1.0000
0.0325
0 . 92 a
L9
0 . 451
0.0650
>.0000
0.0330
"). 927
:o
). 451
: .0660
?.0000
0.0265
) . 941
21
0 . 451
? . 0530
0.0000
0.012 5
0 . 97 2
2 2
0.451
0.0250
0.0000
0.0035
0. 992
23
0 . 451
0.0070
0.0000
0.0085
0 , 981
24
0. 451
0.017Q
0.0000
0.0150
0 . 967
25
0 . 451
0.0300
0.0000
0.0190
0 . 958


0.0380





0.0412

Averaoe
0 . 908
BIENNIAL
New Detect




Ma-program
1



Ago
Ln-Lane
0
Bien.
Bien.
Bien.
Index
fail rate
Rate
After
averaae
benefit
1
0 . 080
0.0800
0.0800
0.0600
0 .000

0 . 080
 0.0800
0.0600
0.0600
0.23 8
3
0.080
0.0600
0.0000
0.0080
0 . 853
4
0 . 096
0.0160
Q.QOOO
0.0035
0 :910
5
0. 103
0.0070
0.0070
0.0155
0.779
6
0.120
0.0240
0.0000
0.0150
0 . 789
7
0 .150
0.0300
0.0300
0.0490
0.612
8
0.108
0.0680
0.0000
0.0350
0.732
9
0.258
0.0700
0.0700
0.1025
0.582
10
0. 323
0.1350
0.0000
0.0330
0 . 837
11
0.389
0.0660
0.0660
0.0995
0.733
12
0.442
0.1330
0.0000
0.0125
0. 967
13
0. 451
0.0250
0.0250
0.0250
0 . 94 5
14
0. 451
0.0250
0.0000
0.0120
0. 97 3
15
0. 451
0.0240
0.0240
0.0240
0 . 94 7
16
0. 451
0.0240
0.0000
0.0340
0. 925
17
0.451
0.0680
0.0680
0.0680
0 .849
10
0.451
0.0630
0.0000
0.0675
0 . 850
19
0 .451
0.1350
0.1350
0.1350
0.701
20
0.451
0.1350
0.0000
0.0665
0. 853
21
0 . 451
0.1330
0.1330
0.1330
0. 705
22
0.451
0.1330
0.0000
0.0125
0 . 97 2
23
0. 451
0.0250
0.0250
0.0250
0. 945
24
0.451
0.0250
0.0000
0.0120
0 . 973
25
0.451
0.0240
0.0240
0.0240
0. 947


0.0240





0.0630

Averaae
0. 843
3

-------
r:i :;iu <_ fjiiuic I
AIIIJUAL
1 ri  ) dni 
Ui'M





A Jw
No-prc-ji jii.
IKjI^ -I
J.... i

! ..i.o.i i


Indu;.
fail rate
fr.at fe	
a; ti
dVt.l a J 
\ L-vinv : i c


I
0.08
-BL

-|W5Vi.| .
- (5 L 5 - a 11  i i
i/ i: i  i

-llA5
0.G8
-E5
- 0 . 7 j  V o
- (Wf.4iOj  :
- (sl6-xci / u:
1 1. Cl n
.
-l IA6
o.oe
-N6

-(wvvfa) :
-ci.7v:7i m:
I 1 1 ll.vl

-] *A7
0. 09l
-BB-bV
ij
- (Wo p v*) :
- Ci.b-x&j / {ir

.
-WAS
0.103
-BV-B6
u
-twvviij y.
-1si.y-v.v-1ir
l.vcu-

-HAS
o. i:
-BK'-BIj
0
- fWU-V)l ) u
- :i 11 -1
fsl]it:
1 1
-ltXJ]
o.iaa
-B12-B11
0
- (HI :1VI31i 2
-(SL12-X12)-(
151. i r

-l A12
0.250
-B13-B12
0
-(Ml 3 * VI 4 ) -2
-|5L13-X13)  (
HUM'.
4 i
-ltJU3
0.323
-B14-B13
0
- (WH + VI 5) /2
- (5L14->;1 4 1 / (
(5li j~:
i 1
-11/U4
0.389
 B15-B14 t (bo-ii! i
0
-(H15+V]6),2
- CSL15->;i 5)  1
151.1 5C
o 1
-liAl5
0. 44?
-B 1 6-B15 t (&T-B6]
0
-|W]6(VI7) -"2
-(5L16-XJ6) (
(s 1.1 f. ~:
7 I
- WM 6
0. 451
-Bll'-BU* (B 0 -E7 )
0
- (W1 7 ( VI o t > 2
- f 5 L1 7 - X ] 7 )  |
:i6  i
<$Lib<:
t-i
hMB
0.451
-vi..
;
- |W1:i9)  I
r

-l +A19
0.451
-VI1
u
-|W20tV:i1/2
- (5L20-Xl'Oj  |
i r i.:  i r
1
- J * A2G
0.451
-VI2
0
-(h:itv::i-2
-(5L21-X21)  1
i5i.ii i r

-nx21
0.451
-VI 3
c.
- (:: < y2 2i 2
-(IL22-X2;)  (
isi.:i . c

-1 i AJir
0.451
V) 4
c
"2
- :23| - (
1 CM *
2 ;
-HA23
0.451
-VI 5
r.
-IKUiCii 2
- i  i
icli-jic
I
-]+ A24
0.451
-VIC
t
-tH25iV2fc) . 2
- in.25-7.2t) * 1
151.21.. i;
i.i
-HA25
0.451
-VI 7
o
-(H:61v;7 J -2
- (C1.26->;26| |
(51.1 ^ i C
^ j
-MA26
0.451
-via
0
-fW?7 *VLdj /?
- isi.:7->:: h . t
151.27  :
>.)
-i*M7
0. 451
-VI 9
0
- |W?di V2Sj '2
- {:t.26->:2& -1
i ci.ro  r

-hub
0.151
-V20
-VTl
-AVER ACE |V5 : Yjii.
0
-  / ( (5I-C7

-HA67
0.323
V67 (B66-[n'.7j i i t* o ~- - -1: L. v
1 -H66MVC57-(V6fa-W6t| )
-(W68iVC9
!
- {S1.6 8 - X 66 > / | | $L6b
51  *i)
-11A66
0. 389
-tB^V-B66) i l&CG-b5^>
-V9
-(WC9 V70
, ->
- (51.6t - X691 / ( (5l-6i
51 7f<>
-1+A69
0. 442
- V61 (B70-fc6 9 ) t (H61-LC-)
-W68 t (VJ 5 7 * r/7t'-HC6| I
-(W70
- (L74 -X74 ) ' ( [$1.74
Cl Hi
-1tA74
0.451
-W7 4 ^ |V66-Ht6j
-V7 5
-
, 
- ($L75~X75> t ( ($1.75
51 if.)
-14A75
0.451
-W7S (V67-H67)
-H74 * (VJ57-(V76-W74)j
- {W7.a'J7
>
- CL76-X76) , f ($1.76
S 1-7 V |
-14A76
0.451
-Wlti(V68-W6Bi
-V77
- (W 7 7  V 7 6
t -)
- (5L77-X77I / ( (51-77
51 7(m
-]+A77
0.451
-Ulli (V6^-W6t)
-H7 6t (V557  (V7&-W76))
-(H7&IV7 9
12
- (JL76-X70) , (f$L7fa
5L7>)
-1+A78
0.451
-W78*(V70-H70)
-V7 9
-(W7 4 4 V&0
n
- (SL79->:79t /( (SL79
51-601
-WA79
0.451
-W7 9~ IV71-H711
-H781 (V551' (Vflf.-W7B| )
-fwaotvei
} 2
- (5L0C--XaO| i ( (Sl-00
51.81 i
-14ABO
0.451
-W60 *(V7 2-H721
-V81
-|W014V8?
/2
- (5L01-X61)/((5L81
5 LB2)
-1 i AB1
.451
-H81(V73-W73)
-WBOt (V557*lva2-H0O) )
-(W82 t VB 2
/ 2
-(5L02-X82)/(($LS:
51.6 j i
-HA82
0.451
-H82~
-------
APPENDIX C
EXHAUST SHORT TEST ACCURACY: IM240 VS. SECOND-CHANCE 2500
RPM/IDLE TEST

-------
Figure 1
Comparison of IM240 to Second Chance 2500/Idle foi
1983 & Newer Port Fuel Injected Vehicles from Hammond Indiana
IM240 Cutpoints = 0.8 & 15 g/mi Composite and 0.5 &12 g/mi Bag 2
Second Chance 2500/Idle Ctpts = 220 ppm & 1.2%
90% -r
80% 
70% -
60% 
50% 
40%
30%
20% 
10%
0%
81%
83%
70%
62%
13%
10%
2.5%
0.0% 0.0%
HCIDR
COEDR
I/M Fail
Rate
I/M Fail
Rate for
FTP
Passing
I/M Fail
Rate for
Normal
Emitters
~ 2-Spd (220ppm & 1.2%)
IM24Q (0.8 & 15 g/mi composite; 0.5 & 12 g/mi Bag 2)
C-2

-------
Figure 2
Comparison of IM240 to Second Chance 2500/Idle for
1983 & Newer PFI Vehicles from Hammond Indiana
IM240 Cutpoints = 0.8 & 15 g/mi Composite; 0.5 & 12 g/mi Bag 2
Second Chance 2500/Idle Qpts = 220 ppm & 1.2% @2500; 100 ppm & 1.0% @Idle
90%
80%
70% 
60% -
50%
40%
30%
20% -1
10% -
0%
81%
83%
73%
78%
29%
22%
13%
12%
0.0%
2.5%
HC EDR
COIDR
I/M Fail
Rate
I/M Fail
Rate for
FTP
Passing
I/M Fail
Rate for
Normal
Emitters
~ 2-Spd (220/1 OOppm & 1.2/1.0%)
IM240 (0.8 & 15 g/mi composite; 0.5 & 12 g/mi Bag 2)
C-3

-------
Figure 3
Comparison of IM240 to Second Chance 2500/IdIe for
1983 & Newer Port Fuel Injected Vehicles from Hammond Indiana
IM240 Cutpoints = 0.8 & 15 g/mi Composite; 0.5 &12 g/mi Bag 2
Second Chance 2500/Idle Ctpts = 100 ppm & 0.5% both modes
90%
80%  78%
70% 
60% -
50%
40% 
30% -
20% --
10% 
0% -
84%
81%
83%
33%
27%
15%
13%
0.0%
HC IDR
CO IDR
I/M Fail
Rate
I/M Fail
Rate for
FTP
Passing
I/M Fail
Rate for
Normal
Emitters
~ 2-Spd (lOOppm & 0.5%)
IM240 (0.8 & 15 g/mi composite; 0.5 & 12 g/mi Bag 2)
C-4

-------
figure 4
Comparison of IM240 to Second Chance 2500/Idle for
1983 & Newer Throttle Body Injected Vehicles from Hammond Indiana
IM240 Cutpoints = 0.8 & 15 g/mi Composite; 0.5 &12 g/mi Bag 2
Second Chance 2500/Idie Ctpts = 220 ppm & 1.2%
100% -
90% -
80% -
70% -
60%
50%
40%
30%
20%
10% -1-
0%
92%
82%
75%
71%
20%
9.1%
4.4%
5.9%
HCIDR
COIDR
VM Fail
Rate
l/M Fail
Rate for
FTP
Passing
I/M Fail
Rate for
Normal
Emitters
~ 2-Spd (220ppm & 1.2%)
IM240 (0.8 & 15 g/mi composite; 0.5 & 12 g/mi Bag 2)
C-5

-------
Figure 5
Comparison of IM240 to Second Chance 2500/Idle for
1983 & Newer TBI Vehicles from Hammond Indiana
IM240 Cutpoints = 0.8 & 15 g/mi Composite; 0.5 & 12 g/mi Bag 2
Second Chance 2500/Idle Ctpts = 220 ppm & 1.2% @2500; 100 ppm & 1.0% @Idle
100% -T-
90% 
80%
70%
60%
50% 
40% -
30% -
20%
10% 
0 %
91% 92%
83% 82%
33%
23%
20%
4.5%
HCIDR
COIDR
I/M Fail
Rate
I/M Fail
Rate for
FTP
Passing
I/M Fail
Rate for
Normal
Emitters
~ 2-Spd (220/100ppm & 1.2/1.0%)
IM240 (0.8 & 15 g/mi composite; 0.5 & 12 g/mi Bag 2)
C-6

-------
Figure 6
Comparison of IM240 to Second Chance 2500/Idle for
1983 & Newer Throttle Body Injected Vehicles from Hammond Indiana
IM240 Cutpoints = 0.8 & 15 g/mi Composite; 0.5 & 12 g/mi Bag 2
Second Chance 2500/ldle Ctpts = 100 ppm & 0.5% both modes
100%
90% 
80%
70%
60%
50%
40% 
30%
20% 
10%
0%
91% 92%
84%
82%
37%
HC IDR
CO EDR
I/M Fail
Rate
I/M Fail
Rate for
FTP
Passing
I/M Fail
Rate for
Normal
Emitters
~ 2-Spd (lOOppm & 0.5%)
IM240 (0.8 & 15 g/mi composite; 0.5 & 12 g/mi Bag 2)
C-7

-------
Figure 7
Comparison of EM240 to Second Chance 2500/IdIe for
1983 & Newer Carbureted Vehicles from Hammond Indiana
IM240 Cutpoints = 0.8 & 15 g/mi Composite; 0.5 &12 g/mi Bag 2
Second Chance 2500/Idle Ctpts = 220 ppm & 1.2%
80% -r
70%
60% 
50% 
40%
30%
20%
10% 
0%
78%
77%
57%
49%
5.6%
0.0% 0.0%
1.2%
0.3%
HC IDR
CO IDR
I/M Fail
Rate
I/M Fail
Rate for
FTP
Passing
I/M Fail
Rate for
Normal
Emitters
~ 2-Spd (220ppm & 1.2%)
IM240 (0.8 & 15 g/mi composite; 0.5 & 12 g/mi Bag 2)
C-8

-------
Figure 8
Comparison of IM240 to Second Chance 2500/IdIe for
1983 & Newer Carbureted Vehicles from Hammond Indiana
IM240 Cutpoints = 0.8 & 15 g/mi Composite; 0.5 & 12 g/mi Bag 2
Second Chance 2500/Idle Ctpts = 220 ppm & 1.2% @2500; 100 ppm & 1.0% @Idle
80% -r
70%
60%
50%
40% -
30% -
20% 
10%
0%
78%
77%
68%
60%
6.9%
5.6%
2.3%
0.0% 0.0%
0.3%
HC IDR
CO IDR
I/M Fail
Rate
I/M Fail
Rate for
FTP
Passing
I/M Fail
Rate for
Normal
Emitters
~ 2-Spd (220/1 OOppm & 1.2/1.0%)
IM240 (0.8 & 15 g/mi composite; 0.5 & 12 g/mi Bag 2)
C-9

-------
Figure 9
Comparison of IM240 to Second Chance 2500/Idle for
1983 & Newer Carbureted Vehicles from Hammond Indiana
IM240 Cutpoints = 0.8 & 15 g/mi Composite; 0.5 & 12 g/mi Bag 2
Second Chance 2500/Idle Ctpts = 100 ppm & 0.5% both modes
80%
70% --
60%
50% -
40% 
30% 
20% --
10% 
0%
78% 78%
78%
77%
30%
' Ov ' '' &
27%
0.0% 0.0%
HC IDR
CO IDR
I/M Fail
Rate
I/M Fail
Rate for
FTP
Passing
I/M Fail
Rate for
Normal
Emitters
~ 2-Spd (lOOppm & 0.5%)
Hi CvI24G (0.S & i5 g/mi composite; 0.5 & 12 g/mi Bag 2)
C-10

-------
APPENDIX D
M0BILE4.1 TECHNOLOGY DISTRIBUTION AND EMISSION GROUP RATES
AND EMISSION LEVELS

-------
M0BILE4 1
Technology Distribution lor Gasoline Passenger Cars

CLLP
CLLP
CLLP
Oplp
Oplp
Oplp
CLLP
CLLP
CLLP
Oplp
Oplp
Uodal
3w*y
3way
3w*y
Sway
3w*y
3way
OxSiwy
OvSway
Oilwy
Ortway
Ortway

CMb
ypR
161

MPR
nu
Ccb
MEB
mi
Cafe
MEG
1981
1571360
413972
0
48696
0
0
2731528
1434
201346
932653
284
1982
1040443
403436
435887
54160
0
0
2265172
0
2S8479
1181258
0
1883
151414
591266
794934
0
0
0
3211332
15517
471542
870343
409
1984
761159
938060
1414498
25877
0
0
4140952
40669
1091249
488429
0
1085
666873
2182617
432479
25478
0
0
2085078
116557
1125292
547055
0
1886
594368
2607047
1058186
50
0
150047
1936299
504156
1041046
40209
0
1987
415441
2239841
1397261
16821
64
41918
1188253
156340
942058
48809
650
1988
277445
2998983
2206819
0
65
0 .
404916
322283
542380
0
0
1989
613700
4652439
2034135
0
90
0
449631
431553
308598
25000
0
1990
10800
5177046
1395418
0
0
0
122142
311612
186117
7000
0
1991
9000
6419242
1570776
0
0
0
15690
176552
103342
0
0

CLLP
CUP
CLLP
Oplp
Oplp
Oplp
CLLP
CLLP
CLLP
Oplp
Oplp
Mod*!
Sway
3w*jr
Sway
3ny
Sway
3way
Ox3wy
Ox3way
Ox3vway
Ox3wy
Ox3way
Xcar
*fr
MPR
ibi
Sab
MEB
3B1
Cfb
MPR
III
Cafe
MPR
1981
23.0%
6.0%
0.0%
0.7%
0.0%
0.0%
39.9%
0.0%
2.9%
13.6%
0.0%
1B82
15.9%
62%
6.7%
0.8%
0.0%
0.0%
34.7%
0.0%
4.0%
18.1%
0.0%
1983
22%
85%
115%
00%
0.0%
0.0%
46.4%
02%
6.8%
12.6%
0.0%
1984
8.6%
105%
15.9%
03%
0.0%
0.0%
465%
05%
123%
55%
00%
1985
12.9%
292%
5.8%
0.3%
0.0%
0.0%
27-9%
1.6%
150%
73%
0.0%
1986
75%
32.9%
13,3%
0.0%
0.0%
1.9%
24.4%
6 4%
131%
05%
0.0%
1987
6.4%
34.7%
21.7%
0.3%
0.0%
0.7%
18.4%
2.4%
14.6%
08%
0 0%
1988
4.1%
44.4%
32.7%
0.0%
00%
00%
6 0%
4.8%
8.0%
0.0%
0.0%
1989
7 2%
54.6%
23.9%
00%
0.0%
0.0%
5.3%
5.1%
3 6%
03%
0.0%
1990
0.1%
71.8%
19.4%
0.0%
0.0%
0.0%
1.7%
43%
2.6%
01%
0.0%
1991
0.1%
77.4%
18.9%
0.0%
0.0%
0.0%
02%
2.1%
12%
0.0%
0.0%
MocM
CLLP
CLLP
CLLP
Oplp

Modal
Any
Any
Any

Modal
iMt
CMfe
MEB
nu
An*

XtK
Ctcfe
MPR
IB1

1m
1961
62.9%
6.1%
2.9%
28.1%

1981
91.0%
61%
2.9%

1981
1982
50.6%
6.2%
10.6%
325%

1982
832%
62%
10j6%

1982
1983
48.6%
8 8%
18.3%
244%

1983
729%
6.8%
183%

1983
1984
55.1%
11.0%
28 2%
5.8%

1984
60.9%
110%
282%

1984
I98S
40.8%
30.7%
20 8%
77%

1985
48 4%
30.7%
20 8%

1985
1986
31.9%
392%
265%
2 4%

1986
32.4%
392%
284%

1986
1987
24.9%
372%
36.3%
1.7%

1987
25.9%
372%
36.9%

1987
1988
10.1%
492%
40.7%
0.0%

1988
10.1%
492%
40.7%

1988
1989
125%
59 7%
27.5%
03%

1989
12.8%
59 7%
275%

. 1989
1990
1.8%
76.1%
21 9%
0.1%

1990
1.9%
76 1%
21.9%

1990
1991
0.3%
79 5%
202%
0.0%

1991
0.3%
795%
202%

1991
J) vehicle counis taken Irom CAFE estimates. except 1989 and newer model years, which are from General Label predictions.
7/2/51
Oplp
Oplp
Oplp



Ox3wny





m
Oxld
Hor
All
CM PI
%GMF
0
943390
38
6644901
133369
1.9%
0
889494
10
6528339
636022
9.7%
0
817824
0
6924581
83H42
12.0%
0
0
0
8900893
1444036
162%
0
0
0
7481429
1644484
22.0%
0
0
O
7931408
2087185
26.3%
0
0
0
6447456
1673716
26.0%
0
0
0
6752891
2420729
35.8%
0
0
0
8515146
3018191
35.4%
0
0
0
7210135
2522963
35.0%
0
0
0
8294602
3639046
43.9%
Oplp
Oplp
Oplp



Ox3wray





IB)
fiAjd
tteos



0.0%
13.8%
0.0%



00%
13.6%
0 0%



0.0%
118%
00%



0 0%
0.0%
0.0%



0.0%
0.0%
0 0%



0 0%
0.0%
0.0%



0.0%
0.0%
0.0%



0.0%
0.0%
0.0%



00%
0.0%
0.0%



0.0%
0.0%
00%



0.0%
0.0%
0.0%



Any
Any
Any



3way
0x?m>y
Ox Id



29.7%
565%
13.8%



29.6%
56.8%
13 6%



222%
66.0%
11 8%



353%
64.7%
0.0%



482%
51.8%
0 0%



55 6%
44.4%.
0.0%



63.8%
362%
0 0%



812%
18.8%
0.0%



85.7 %
143%
0.0%



91.3%
8.7%.
00%



96 4%
3.6%
0 0%




-------
MOBILE4.1
Emitter Group Emission Levels'
Uodel Emission
HC	HC
Emissions Emission*
CO
Emissions
CO
Eml salons
NOx
NOx
Emissions Emissions
Yes*
Lewsl
Technology
HC
HC
at 50,000
al 100,000
CO
CO
al 50,000
at 100,000
NOx
NOx
al 50,000
al 100.0C
OfQMQ
Grow
Orotic
ZML
DEI
tutu
Mllei
ZML
eo
Mile*
Mllet
ZML
BET
Miles
Miles
1061-82
Norms)
UPR
0 186
0 0450
0.113
0.638
1.548
0.5807
4 452
7.355
0.360
0.1280
1 020
1.660
1961-81
Normal
TBI
0.278
0.0161
0.360
0 440
3.422
0.2520
4.682
5.642
0 545
0.1414
1.252
1.959
1081-82
Normal
Cart)
0.268
0.02S3
0.415
0.541
3.067
0.3281
4708
6.348
0.678
0.0696
1.026
1.374
1081-82
Normal
Oplp
0.306
0.0230
0 421
0.536
3.368
0.2677
4.807
6.245
0.649
0.0440
0 869
1.089
1083-85
Normal
UPR
0.266
0.0115
0.327
0.384
2.598
0.1554
3.375
4 152
0.689
0.0185
0 782
0.874
108)45
Normal
TBI
0.242
0.0134
0.306
0.376
2.653
0 1572
3.736
4.525
0.585
0.0470
0.820
1 055
1063-85
Normal
Cart)
0.222
0.0166
0.322
0.421
2.209
02282
3.350
4.461
0.692
0.0558
0.971
1.250
1063-85
Normal
Oplp
0.334
0.0248
0.458
0.582
4.063
0.2063
5.140
6.186
0.524
0.0540
0.794
1.064
1086+
Normal
UPR
0.266
0.0115
0.327
0 384
2.598
0.1554
3.375
4 152
0.539
0.01 B5
0.632
0.724
1086+
Normal
TBI
0.242
0.0134
0.306
0.376
2.653
0.1572
3.736
4.525
0.316
0.0470
0.551
0.786
1086+
Normal
Cart)
0222
0.0160
0.322
0421
2.206
0.2282
3.350
4.461
0.478
0.0558
0.757
1 036
1986+
Normal
Oplp
0.334
0.0240
0.458
0.582
4.093
02063
5.140
6.186
0 524
0 0540
0.794
1.064
1081-82
High
UPR
0.374
0.0450
0.599
0.824
3878
0.5807
6.7B3
9.686
1081-82
High
TBI
0.758
0.0161
0.836
0.616
6.843
0.2520
8.103
6363
1081-82
High
Cut
0.781
0.0253
0.900
1.034
8.496
03281
10.137
11.777
1681-82
High
Oplp
0.757
0.0230
0.B72
0.687
8.142
02877
6.581
11.016
1063+
High
UPR
0.667
0.0115
1.055
1.112
7.602
0 1554
8.379
6.156
1981+
High
TBI
0.762
0.0134
0.829
0.866
8.820
01572
6.606
10.362
1081+
High
Cait>
0.703
0.0166
0.803
0.602
8.577
0.2282
6718
10.856
1081+
High
Oplp
0.840
0.0248
0.664
1.088
8.386
02063
6 433
10.476
1981-82
Vary High
UPR
0.606
0.0450
0.831
1.056
18.046
0.5807
20.953
23.856
1081-82
Vary High
TBI
1.163
0.0161
1.244
1.324
22.069
02520
23.316
24.576
1081-82
Vary High
Carto
1.650
0.0253
2.077
2.203
33.396
0.3281
35.037
36677
108142
Vary High
Oplp
1.569
0.0230
1704
1.816
27.650
0.2877
29.086
30.527
1983+
Vary High
UPR
2.016
0.0115
2.077
2.134
22.301
0.1554
23.078
23.655
1B8J+
Vary High
TBI
2.242
0.0134
2.306
2.376
44416
0.1572
45.202
45.688
1081+
Vary High
Cart>
2.002
0.0166
2.102
2.201
36.130
02282
37.271
38.412
1083+
Vary High
Oplp
1.352
0.0248
1.476
1.600
34.021
02063
35.068
36.114
1981-82
Super
UPR
15.256
N/A
15.259
15.256
173.840
N/A
173.840
173.840
1061-82
Super
TBI
15.256
N/A
15256
15.256
173.840
N/A
173.840
173.840
1081-82
Super
Cart)
15.256
N/A
15.256
15.256
173.840
N/A
173.840
173B40
1061-82
Super
Oplp
0 000
N/A
0.000
0.000
0.000
N/A
0.000
0.000
1083+
Super
UPR
11.123
N/A
11.123
11.123
189.000
U/A
189.000
186.000
1083+
Super
TBI
18.083
WA
18.083
18.083
183.590
N/A
183.560
183.560
1963+
Super
Caib
12.870
N/A
12.870
12.670
246.850
N/A
246.850
246.050
1063+
Super
Oplp
0.000
N/A
0.000
0.000
0.000
N/A
0.000
0.000
' All emission levels in grams per mile. DET In grams per mile per 10.000 miles.
BLOCK.XLS

-------
MOBILE4.1
Emitter Group Rates



Growth
Rate
Percent
Percent
Model
Emission

Rate
Increase
at
at
Year
Level
Technology
per 10,000
Beyond
50,000
100,000
Group
Group
Group
MHea
50.000 Miles
MIIm
Miles
1981-82
Normal
MPF1
N/A
N/A
85.8%
29.9%
1981-82
Normal
TBI
N/A
N/A
89.5%
37.2%
1981-82
Normal
Carb
N/A
N/A
66.4%
33.3%
1981-82
Normal
Oplp
N/A
N/A
68.8%
37.7%
1983+
Normal
MPF1
N/A
N/A
90.3%
65.0%
1983+
Normal
TBI
N/A
N/A
87.0%
45.5%
1983+
Normal
Carb
N/A
N/A
84.6%
60.0%
1983+
Normal
Oplp
N/A
N/A
91.3%
82.7%
1981-82
High
MPF1
0.0065
13.8353
3.3%
48.2%
1981-82
High
TBI
0.0065
13.8353
3.3%
48.2%
1981-82
High
Carb
0.0270
0.9643
13.5%
26.5%
1981-82
High
Oplp
0.0270
1.0000
13.5%
27.0%
1983+
High
MPF1
0.0090
4.4709
4.5%
24.6%
1983+
High
TBI
0.0040
15.2076
2.0%
32.4%
1983+
High
Carb
0.0140
2.3200
7.0%
23.2%
1983+
High
Oplp
0.0130
1.0000
6.5%
13.0%
1981-82
Very High
MPFI
0.01933
N/A
9.7%
19.3%
1981-82
Very High
TBI
0.01198
N/A
6.0%
12.0%
1981-82
Very High
Carb
0.03762
N/A
18.8%
37.6%
1981-82
Very High
Oplp
0.03530
N/A
17.7%
35.3%
1983+
Very High
MPFI
0.00840
N/A
4.2%
8.4%
1983+
Very High
TBI
0.02012
N/A
10.1%
20.1%
1983+
Very High
Carb
0.01487
N/A
7.4%
14.9%
1983+
Very High
Oplp
0.00433
N/A
2.2%
4.3%
1981-82
Super
MPFI
0.00257
N/A
1.3%
2.6%
1981-82
Super
TBI
0.00257
N/A
1.3%
2.6%
1981-82
Super
Carb
0.00257
N/A
1.3%
2.6%
1981-82
Super
Oplp
0.00000
N/A
0.0%
0.0%
1983+
Super
MPFI
0.00194
N/A
1.0%
1.9%
1983+
Super
TBI
0.00194
N/A
1.0%
1.9%
1983+
Super
Carb
0.00194
N/A
1.0%
1.9%
1983+
Super
oplp
0.00000
N/A
0.0%
0.0%
BLOCK.XLS
3
11/8/91

-------
APPENDIX E
REGRESSION ANALYSES AND SCATTER PLOTS FOR FUEL INJECTED 1983
AND LATER VEHICLES

-------
Table of Contents
Appendi:-: E
E-l	IM240 Vehicles Tested to Date	11/14/91
PFI Vehicles
E-2	Selecting One Score From the 2500/Idle Test
E-3	Regression Analyses of FTP HC versus HC Emissions Over
IM240, Idle Test, and 2500/Idle Test for PFI Vehicles
E-4	HC Emissions - FTP vs Lane IM240
E-6	HC Emissions - FTP vs Lane 2500/Idle
E-8	HC Emissions - FTP vs Lane Idle
E-10	Regression Analyses of FTP CO versus CO Emissions
Over: IM240, Idle Test, and 2500/Idle Test for PFI
Vehicles
E--11	CO Emissions - FTP vs Lane IM240
E-13	CO Emissions - FTP vs Lane 2500/Idle
E-15	CO Emissions - FTP vs Lane Idle
E-17	Regression Analyses of FTP NOx versus IM240 NOx
Emissions
E-18	NOx Emissions - FTP vs Lane IM240
E-20	Vehicles Tested by ATL at Southbend, Indiana
E-21	HC Emissions - FTP vs Indolene-IM240
E-23	CO Emissions - FTP vs Indolene-IM240
E-25	NOx Emissions - FTP vs Indolene-IM240
E-27	EF Vehicles Tested at EPA's MVEL
E-28	HC Emissions - FTP vs Indolene-IM240
E-30	CO Emissions - FTP vs Indolene-IM240
E-32	NOx Emissions - FTP vs Indolene-IM240
E-34	Description of Repair Regressions Analysis
i

-------
E-35 Statistical Summaries for Variables Used in PFI Before
and After Regression Analyses
E-36 Regression Analyses for the Change in Emission
Constituents Before ana After Repairs of the PFI
Vehicles for the FTP vs IM240-A
E-37 HC Emissions - AFTP vs AIM240-A
E-39 CO Emissions - AFTP vs Alndolene-IM240
E 41 NOx Emissions - AFTP vs AIM240-A
E-43 Fuel Economy - AFTP vs AIM240-A
TBI Vehicle*
E-45 Selecting One Score From the 2500/Idle Tests
E-46 Regression Analyses of FTP HC versus HC Emission Over:
IM240, Idle Test, and 2500/Idle Test for TBI Vehicles
E-47 HC Emissions - FTP vs Lane IM240
E-49 HC Emissions - FTP vs Lane 2500/Idle
E-51 HC Emissions - FTP vs Lane Idle
E-53 Regression Analyses of FTP CO versus CO Emissions
Over: IM240, Idle Test, a'nd 2500/Idle Test for TBI
Vehicles
E-54 CO Emissions - FTP vs Lane IM240
E-56 CO Emissions - FTP vs lane 2500/Idle
E-58 CO Emissions - FTP vs Lane Idle
E-60 Regression Analyses of FTP NOx versus IM240 NOx
Emissions
E-61 NOx Emissions - FTP vs Lane IM240
E-63 Vehicles Tested by ATL at Southbend, Indiana
E-64 HC Emissions - FTP vs Indolene-IM240
E-66 CO Emissions - FTP vs Indolene-IM240
E-68 NOx Emissions - FTP vs Indolene-IM240
E-70. EF Vehicles Tested at EPA's MVEL
ii

-------
E-73
E-80
E-32
E-84
E - 6 6
description of Repair Regression Analysis
Statistical Summaries :cr Variables Used in 751
Before ana After Regression Analyses
Regression Analyses f;r the Change in Emission
Constituents Before ana After Repairs of the T2I
Vehicles for the FTP vs IM240-A
HC Emissions - AFT? vs IM240-A
CO Emissions - AFTP - AIM2 4 0-A
MOx Emissions - AFT? vs AIM240-A
Fuel Economv - AFT? vs AIM2 4 0-A
iii

-------
IM240 VEHICLES TESTED TO DATE
1 2/13/9 1
Test
As Received
In Micro
Indiana Ann Arbor
RM#1
In Micro
Indiana Ann Arbor
RM #2
In Micro
Indiana Ann Arbor
RM 03
In Micro
Indiana Ann Arbor
FTP
453
314
213
75
35
Lane IM240
IM240 Lane
Qiange of Owner
MISC Repeat
7313
b
14
NA
NA
NA
73
0
0
NA
NA
NA
1(1
t)
(t
NA
NA
NA
NA
NA
NA
PJ
I
Tank Fuel 1M240
Diagnostic IM240
TanL Fuel IM240
After Exhusi Repair
435
15
NA
(I
0
NA
0
0
14
NonCITP IM240
IM 240 Performed in Lab. Independent
of at! other tests.
2H
2<>7
70
Indoluic CD'P-A
IM240-A (A)
CDH226-A (B)
Steady Slalc-A
378
369
0
30
31
0
ina
180
o
15
15
It
2')
29
0
Indolent CITP-1J
1K1244H3 (Dj
CDH226-U (C|
Steady Slale-B
365
366
0
31
31
(i
180
179
0
15
15
(I
2>)
2<>
0
. I)
0
0

-------
Regression Analyses - Port Fuel-Injected Vehicle Sample:
As part of EPA's regression analysis. FTP results for HC and CO emissions were
each regressed against the corresponding emission results on the Lane 1M240, the second
chance 2500/Idle Test, and the Idle Test. Regressions were also performed comparing
NOx results for the FTP and Lane IM240 (NOx emissions are not measured as part of
either Idle Test and are therefore not included in this comparison). Lastly, FTP results for
HC, CO, and NOx were each regressed against corresponding emissions from Lab IM240s
run on Indolene fuel at the ATL lab in South Bend and at MVEL in Ann Arbor. This
section provides the results of these regressions performed on a sample of port fuel injected
(PFI) vehicles recruited as pan of the Hammond study.
EPA removed thirteen vehicles from its sample of PFI vehicles. Three were
removed because they received repairs for exhaust leaks during the interim between
receiving an I/M test at the Hammond lane and being FTP tested at the lab. The ten
remaining PFI vehicles were excluded due to inconsistent dynamometer settings for the
Lane IM240 and the FTP. Vehicles were only removed if the horsepower or inertia weight
settings were at least 15% different between the lab and the lane.
These differences in dynamometer settings arose from the use of different
dynamometer look-up tables at the lane and lab. Traditional certification test car lists used
for emission factor testing are too detailed for efficient use in the I/M lane. After becoming
aware of the differences in the two tables, EPA opted to use the look-up table that had been
simplified for lane use at both testing sites to avoid future inconsistencies. The resulting
sample consisted of 74 '83 and newer PFI vehicles.
Selecting One Score From the Second Chance 2500/Idle Test:
To illustrate the correlation of FTP HC and CO results to the corresponding results
on the second chance 2500/Idle Test, we had to select an HC score and CO score from the
first chance or second chance test, and from the 2500 rpm mode or the idle mode.
Concerning the first chance versus second chance results, we chose the better results
relative to the HC/CO cutpoint being evaluated.
For the 2500 rpm versus idle modes, we chose the larger of the corresponding emissions
measured. This approach is based on the assumption that the same cutpoints would be
used on each of the modes. Hence, a vehicle would pass the two mode test if and only if
the higher of its HC scores (whether Idle or 2500 rpm) met the HC cutpoint (with the same
also following for the CO results).
Graphing the Data:
On each of the graphs comparing HC emissions, the three horizontal dotted lines
are positioned at the boundaries separating:
the normal emitters from the high emitters <0 X2 g/mi).
the high emitters from the very high emitters i I .M g/mi). and
 the very high emitters from the super emitters i 10.00 g/mi).
E-2A

-------
Similarly, on each of the graphs comparing CO emissions, the three horizontal
dotted lines are positioned at die boundaries separating:
the normal emitters from the high emitters (10.2 g/mi),
the high emitters from the very high emitters (13.6 g/mi), and
the very high emitters from the super emitters (150.0 g/mi).
E-2B

-------
Regression Analyses of FTP HC versus HC Emissions Over:
IM240, Idle Test, and 2500/Idle Test
For PFI Vehicles
Dependent variable is:
HCFTP


RA2 = 80.3%




s = 1.016 with
74 - 2 = 72 degrees of freedom


Source
Sum of Squares
df
Mean Square
F-ratlo
Regression
302.637
1
303
293
Residual
74.3928
72
1.03323

Variable
Coefficient
s.e. of Coeff
t-ratlo

Constant
0.198655
0.1372
1.45

HCIM24
1 .1 3037
0.066
17.1

Dependent
variable is:
HCFTP


RA2 = 39.3%





s = 1.784
with
74 - 2 = 72 degrees of freedom


Source

Sum of Squares
df
Mean Square
F-ratlo
Regression

148.004
1
148
46.5
Residual

229.026
72
3.18092

Variable

Coefficient
s.e. of Coeff
t-ratlo

Constant

0.692609
0.2313
2.99

Higher(ldle
HC..
0.003966
0.0006
6.82

Dependent variable is:
HCFTP


RA2 = 44.3%




s = 1.709 with
74 - 2 = 72 degrees of freedom


Source
Sum of Squares
df
Mean Square
F-ratlo
Regression
166.842
1
1 67
57.2
Residual
210.188
72
2.91927

Variable
Coefficient
s.e. of Coeff
t-ratlo

Constant
0.706016
0.2184
3.23

Idle.HC
0.004541
0.0006
7.56

E-3

-------
FTP HC
(fl/ml)
12 -
HC Emissions -- FTP vs Lane IM240
(74 1983 & Newer PFI Vehicles)
10 --
8 -1-
-v
6 -

Regression Line
RA2 = 80.3%
' 


6	8
IM240 HC Emissions (g/ml)
1 0
1 2
E - 4

-------
HC Emissions - FTP vs Lane IM240
(74 1983 & Newer PFI Vehicles)
Enlarged for Better Detail near Origin
Regression Line
= 60.3%
I	{
1	t.5	2	2.
IM240 HC Emissions (g/ml)

-------
10 
6 T
2 --
HC Emissions - FTP vs Lane 2500/ldle
(74 1983 & Newer PFI Vehicles)
\
\
Regression Line
RA2 = 39.3%

200	400	600	800	1000 1200 1400 1600 1800 2000
2500/ldle HC Emissions (ppm)
E-6

-------
FTP HC
(g/ml)
3 -
2 T
1.5 T
HC Emissions  FTP vs Lane 2500/idle
(74 1983 & Newer PFI Vehicles)
Enlarged for Better Detail near Origin
Regression Line
RA2 = 39.3%
f:   

o -!	
50	100	150	200	250
2500/ldle HC Emissions (ppm)
E - 7

-------
FTP HC
(fl/ml)
12 -
HC Emissions -- FTP vs Lane Idle
(74 1983 & Newer PFI Vehicles)
10 J- ;
8 ~
6 -
4 ~

Regression Line
RA2 = 44.3%
200	400	600	800	1000 1200 1400 1600 1800 2000
Idle HC Emissions (ppm)
E-8

-------
HC Emissions -- FTP vs Lane Idle
(74 1983 & Newer PFI Vehicles)
Enlarged for Better Detail near Origin
, > >'
 
# -
Regression Line
RA2 = 44.3%
/
/
/
I '
t
200

-------
Regression Analyses of FTP CO versus CO Emissions Over:
IM240, Idle Test, and 2500/Idle Test
For PFI Vehicles
Dependent variable is:
COFTP


RA2 = 90.4%




s = 17.10 with
74 - 2 = 72 degrees of freedom


Source
Sum of Squares
df
Mean Square
F-ratlo
Regression
199166
1
200000
681
Residual
21052.3
72
292.392

Variable
Coefficient
3.e. of Coeff
t-ratlo

Constant
0.446807
2.21 7
0.202

COIM24
1 .05072
0.0403
26.1

Dependent variable is:
COFTP


RA2 = 76.3%





s = 26.91
with
74 - 2 = 72 degrees of freedom


Source

Sum of Squares
df
Mean Square
F-ratlo
Regression

168093
1
200000
232
Residual

52125.7
72
723.968

Variable

Coefficient
s.e. of Coeff
t-ratlo

Constant

4.01842
3.446
1.17

Hfgher(ldle
CO..
23.5992
1.555
15.2

Dependent variable is:
COFTP


RA2 = 61.8%




S = 34.19 with
74 - 2 = 72 degrees of freedom


Source
Sum of Squares
df
Mean Square
F-ratfo
Regression
136071
1
100000
116
Residual
84147.5
72
1 168.71

Variable
Coefficient
s.e. of Coeff
t-ratio

Constant
8.59165
4.291
2

Idle.CO
27.9077
2.586
10.8

E-10

-------
FTP CO
(g/ml)
250 -
200 -
150 -T---
100
50 -
CO Emissions -- FTP vs Lane IM240
(74 1983 & Newer PFI Vehicles)

Regression Line
RA2 = 90.4%
50	100	150	200	250
IM240 CO Emissions (g/ml)
E-1 1

-------
CO Emissions - FTP vs Lane IM240
(74 1983 & Newer PFI Vehicles)
Enlarged for Better Detail near Origin
(g/ml)
50 	Regression Line
"	R*2 = 90.4%
40 "
30
20 +
10 t"
0 -i:	i	r
1 0	1 5	20	25	30	35	40	45	50
1M240 CO Emissions (g/ml)
E -1 2

-------
FTP CO
(g/ml)
250 -
200
150
100 -
50

o  * 
CO Emissions -- FTP vs Lane 2500/ldie
(74 1983 & Newer PFI Vehicles)
Regression Line
RA2 = 76.3%
.-n
t
0	1	2	3	4	5	6
2500/ldle CO Emissions {percent)
E -1 3

-------
FTP CO
(g/ml)
50 -
40 x
30 -
20 -
CO Emissions  FTP vs Lane 2500/ldle
(74 1983 & Newer PFI Vehicles)
Enlarged for Better Detail near Origin
Regression Line
R*2 = 76.3%
10 --	
I
0 T

0.2	0.4	0.6	0.8	1	1.2	1.4	1.6	1.8
2500/ldle CO Emissions (percent)
E -1 4

-------
CO Emissions - FTP vs Lane Idle
(74 1983 & Newer PFI Vehicles)
FTP CO

-------
FTP CO
(g/ml)
50 -
CO Emissions  FTP vs Lane Idle
(74 1983 & Newer PFI Vehicles)
Enlarged for Better Detail near Origin
40
X
30 --
20
Regression Line
RA2 = 61.8%
10
'	
Lll "

0
0.25
0.5
0.75	1	1.25
Idle CO Emissions (percent)
1.5
E-16

-------
Regression Analyses of FTP NOx versus IM240 NOx Emissions
(Note: NOx is not Measured for the Idle Test)
Dependent variable is:
NOFTP


RA2 = 60.3%




S = 0.3277
with 74 - 2 =
72 degrees of freedom

Source Sum of Squares
df
Mean Square
F-ratlo
Regressior
1 1.758
1
11.76
109
Residual
7.73394
72
0.107416

Variable
Coefficient
s.e. of Coeff
t-ratlo

Constant
0.259283
0.0587
4.41

NOIM24
0.427093
0.0408
10.5

E-17

-------
FTP NOx
(g/ml)
3 -
NOx Emissions - FTP vs Lane IM240
(74 1983 & Newer PFI Vehicles)
2.5 -
2 --
1.5 t
1 -r
0.5 x

m -"
Regression Line
RA2 = 60.3%
0.5
1.5	2	2.5	3
IM240 NOx Emissions (g/ml)
3.5
4.5
E-18

-------
FTP NOx
(fl/ml)
1.5 -
1.25 --
1 -
NOx Emissions - FTP vs Lane IM240
(74 1983 & Newer PFI Vehicles)
Enlarged for Getter Detail near Origin
"M

0.5 -	,
I	-" 
0.25
'..I- 
Regression Line
RA2 = 60.3%
0 -i	r
0	0.25	0.5	0.75	1	1.25	1.5	1.75
IM240 NOx Emissions (g/ml)
E -1 9

-------
Vehicles Tested by ATL at Southbend, Indiana
Regressions of FTP vs lndolene-IM240
for 63 1983 & Newer PFI Vehicles
Dependent variable is
HCFTP



R*2 = 89.2%




s = 0.8037 with 63
- 2 = 61 degrees of freedom


Source
Sum of Squares
df
Mean Square
F-ratlo
Regression
325.428
1
325
504
Residual
39.4065
61
0.646008

Variable
Coefficient
s.e. of Coeff
t-ratlo

Constant
0.43699
0.1 132
3.86

HC.IM24.lndo...
1.1596
0.051 7
22.4

Dependent variable is
COFTP



RA2 = 90.3%




s = 18.57 with 63 -
2 = 61 degrees of freedom


Source
Sum of Squares
df
Mean Square
F-ratlo
Regression
195708
1
200000
568
Residual
21030.6
61
344.764

Variable
Coefficient
3.e. of Coeff
t-ratlo

Constant
5.6588
2.571
2.2

CO.IM24.lndo...
0.991509
0.0416
23.8


Dependent variable is
NOx.FTP



RA2 = 89.1%




s = 0.1780 with 63
- 2 = 61 degrees of freedom


Source
Sum of Squares
df
Mean Square
F-ratlo
Regression
15.8477
1
15.85
500
Residual
1.93337
61
0.031695

Variable
Coefficient
s.e. of Coeff
t*ratlo

Constant
0.145473
0.0356
4.09

NOx.IM24.lnd...
0.757394
0.0339
22.4

E-20

-------
FTP HC
(g/ml)
15 -
HC Emissions - FTP vs lndolene-IM240
(63 1983 & Newer PFI Vehicles)
12.5
10 --
7.5
5 -
2.5
Regression Line
R*2 = 89.2%

2.5
7.5
IM240 HC Emissions (g/ml)
E-21

-------
HC Emissions - FTP vs lndolene-IM240
(63 1983 & Newer PFI Vehicles)
Enlarged lor Better Detail near Origin
Regression Line
R"2 = 89.2%
1	1.5
IM240 HC Emissions (g/ml)
E-22

-------
FTP CO
(g/ml)
250 -
150 -
100
50
0
*
CO Emissions - FTP vs lndolene-IM240
(63 1983 & Newer PFI Vehicles)
Regression Line
200 -f	RA2 = 90.3%
0	50	100	150	200	250
IM240 CO EmUslonB (g/ml)
E - 23

-------
CO Emissions - FTP vs lndolene-IM240
(63 1983 & Newer PFI Vehicles)
Enlarged for Better Detail near Origin
FTP CO
(g/ml)
50	r
45
40	--
35	--
30
Regression Line
2 5	~ R*2 = 90.3%
20	-
1 5
'0	T /"
5 T
0

0	5	10	15	20	25	30	35	40	45
IM240 CO Emissions (g/ml)
E-24

-------
FTP NO*
(g/ml)
4.00 -
3.00
2.00
1.00 -
NOx Emissions - FTP vs lndolene-IM240
(63 1983 & Newer PFI Vehicles)
Regression Line
RA2 = 89.1%
\
 \
\
\
n
 -m

	" a -'
m
	m m
0.00
0.00	1.00	2.00	3.00	4.00
IU240 NOx Emissions (g/ml)
E - 25

-------
1.00
0.75
0.50 -1-
0.25 j
NOx Emissions - FTP vs lndolene-IM240
(63 1983 & Newer PFI Vehicles)
Enlarged for Better Detail near Origin
Regression Line
RA2 = 89.1%
FTP NOx
(g/ml)
2.00 -
1.75 --
1.50 --
1.25 -	\	
- 11
	 "v " * 
	.J*1 
tv 
0.00 -K-
0.00	0.25	0.50	0.75	1.00	1.25	1.50	1.75	2.00
IM240 NOx Emissions (g/ml)
E-26

-------
EF Vehicles Tested at EPA's MVEL
Regressions of FTP vs Indolene IM240
for 143 1985-90 PFI Vehicles
Dependent variable is:	HC.FTP
RA2 = 95.7%
s = 0.5416 with 143 - 2 = 141 degrees of freedom
Source
Sum ol Squares
df
Mean Square
F-ratlo
Regression
927.841
1
928
3163
Residual
41.3557
141
0.293303

Variable
Coefficient
3.e. of Coeff
t -ratio

Constant
0.088877
0.0472
1.88

HCIM24
1.91326
0.034
56.2


Dependent variable is:
CO-FTP


RA2 = 95.4%




s = 5.335 with
143 - 2 = 141 degrees of freedom


Source
Sum of Squares
df
Mean Square
F-ratlo
Regression
83635
1
83635
2939
Residual
4012.44
141
28.457

Variable
Coefficient
s.e. of Coeff
t-ratlo

Constant
1.68729
0.4594
3.67

COIM24
1.14873
0.0212
54.2

Dependent variable is:
NOx.FTP


RA2 = 90.5%




s = 0.2012
with 143 - 2 = 141 degrees of freedom

Source
Sum of Squares
df
Mean Square
F-ratlo
Regression
54.6898
 1
54.69
1351
Residual
5.70772
141
0.04048

Variable
Coefficient
s.e. of Coeff
t-ratlo

Constant
0.10242
0.0237
4.33

NOIM24
0.81782
0.0222
36.8

E-27

-------
HC Emissions -- FTP vs lndolene-IM240
(143 1985-90 PFI Vehicles)
(Tested at EPA's Ann Arbor Lab)
FTP HC
(g/mi)
30.00 -
Regression Line
RA2 = 95.7%
10.00 x

0.00
0.00	5.00	10.00
IM240 HC Emissions (g/mi)
E-28

-------
HC Emissions - FTP vs lndolene-IM240
(143 1985-90 PFI Vehicles)
(Tested at EPA's Ann Arbor Lab)
Enlarged for Better Detail near Origin
FTP HC

3.00 -
X
1.00 - 
L '-
0.00
0.00
0.50
Regression Line
R*2 = 95.7%
1.00	1.50	2.00
IM240 HC Emissions (g/ml)
2.50
3.00
E- 29

-------
FTP CO
(g/ml)
300.00
250.00
CO Emissions - FTP vs lndolene-IM240
(143 1985-90 PFI Vehicles)
(Tested at EPA's Ann Arbor Lab)
Regression Line
BA2 = 95.4%
200.00 -
150.00
v
100.00
50.00 -
0.00-

0.00
50.00
100.00	150.00
IM240 CO Emissions (g/ml)
200.00
250.00
E-30

-------
FTP CO
rml)
.00 -
40.00 -
30.00
i
1
20.00 -
J
J
CO Emissions - FTP vs lndolene-IM240
(143 1985-90 PFI Vehicles)
(Tested at EPA's Ann Arbor Lab)
Enlarged for Better Detail near Origin
Regression Line
RA2 = 95.4%
10.00 T " a J "
I 
IT 
0.00 	r
0.00	10.00	20.00	30.00	40.00	50.00
IM240 CO Emissions (g'ml)
E - 31

-------
FTP NOx
(g/ml)
2.00
NOx Emissions - FTP vs lndolene-IM240
(143 1985-90 PFI Vehicles)
(Tested at EPA's Ann Arbor Lab)
4.00	Regression Line
R*2 = 90.5%
3.00 -

0.00	1.00	2.00	3.00
IM240 NOx Emissions (g/ml)
E-32

-------
FTP NOx
(g/ml)
2.00 -
1.50 -
1.00 ~
NOx Emissions - FTP vs lndolene-IM240
(143 1985-90 PFI Vehicles)
(Tested at EPA's Ann Arbor Lab)
Enlarged for Better Detail near Origin
Regression Line
R*2 = 90.5%
\
	m 
: & . 
0.50 -
- i , * ,
  v*i.
- SV.- 

0.00
0.00	0.50	1.00	1.50	2.00
IM240 NOx Emissions (g/ml)
E - 33

-------
Description of Repair Regression Analysis
Regression Analyses
The difference in emission levels before and after repairs for the FTP's and Lab
EM240's run on Indolene fuel was calculated such that a positive difference would
correspond to a decrease in emissions or an increase in fuel economy. The AFTP values
(AHC, ACO, ANOx and AMPG) were regressed against the corresponding Indolene
AIM240 values. The results and graphs of these regressions are included.
Data Description
The data used in this analysis consisted of FTP and Indolene IM240 scores for tests
performed before and after repairs. These results consisted of HC, CO, NOx, and fuel
economy values. The database was restricted to 1983 and newer PH and TBI vehicles
which received repairs and whose as received FTP scores exceeded twice the FTP standard
for either HC or CO. Also excluded from the analysis were vehicles which received repairs
before the as received FTP and vehicles which had differences in dynamometer settings of
greater than 15%. The resulting data set contained 41 TBI vehicles and 23 PF1 vehicles.
Summary Statistics for the variables used in the PFI analyses follow.
E-34

-------
Statistical Summaries for Variables Used in
PFX Before and After Regression Analyses
Summary statistics for: AHC.IM240
NumNumeric = 23
Mean = 1.7848
Median = 0.59000
Midrange = 3.7950
Standard Deviation = 2.3719
Range = 8.2500
Minimum = -0.33000
Maximum = 7.9200
Summary statistics for: AHC.FTT
NumNumeric = 23
Mean = 2.4678
Median = 1.9700
Midrange = 4.2600
Standard Deviation = 2.9115
Range = 9.0800
Minimum = -0.28000
Maximum = 8.8000
Summary statistics for: ^CO.EM240
NumNumeric = 23
Mean = 53.061
Median = 9.6200
Midrange = 119.73
Standard Deviation = 75.023
Range = 248.85
Minimum = -4.6900
Maximum = 244.16
Summary statistics for: ACO.FTP
NumNumeric = 23
Mean = 60.184
Median = 14.580
Midrange = 115.89
Standard Deviation = 74.656
Range = 239.46
Minimum = -3.8400
Maximum = 235.62
Summary statistics for: ANOX.IM240
NumNumeric = 23
Mean = -0.36174
Median = -0.23000
Midrange = -0.90000
Standard Deviation = 0.52710
Range = 2.3000
Minimum = -2.0500
Maximum = 0.25000
Summary statistics for: ANOX.FTP
NumNumeric = 23
Mean = -0.29522
Median =-0.15000
Midrange = -0.78000
Standard Deviation = 0.48197
Range = 2.2800
Minimum = -1.9200
Maximum = 0.36000
Summary statistics for: AMPG.IM240
NumNumeric = 23
Mean = 2.5904
.Median = 1.5500
Midrange = 3.2800
Standard Deviation = 2.9593
Range = 10.780
Minimum = -2.1100
Maximum = 8.6700
Summary statistics for AMPG.FTP
NumNumeric = 23
Mean = 2.5183
Median = 1.7100
Midrange = 5.3200
Standard Deviation = 3.5426
Range = 17.720
Minimum = -3.5400
Maximum = 14.180
E-35

-------
Regression Analyses for the Change in Emission Constituents
Before and After Repairs of the PR Vehicles for the FTP vs IM240-A
Dependent variable is:
AHC.FTP


R*2 = 90.8%




s = ' 0.9031
with 23 - 2 = 21
degrees of freedom

Source
Sum of Squares df
Mean Square
F-ratlo
Regression
169.361
1
169
208
Residual
17.1275
21
0.815595

Variable
Coefficient
s.e. of Coeff
t-ratlo

Constant
0.380093
0.2376
1.6

AHC.IM240
1.16974
0.0812
14.4

Dependent variable is:
ACO.FTP


RA2 = 88.4%




s = 26.01
with 23 - 2 = 21
degrees of freedom

Source
Sum of Squares df
Mean Square
F-ratlo
Regression
108409
1
100000
1 60
Residual
14207.5
21
676.547

Variable
Coefficient
s.e. of Coeff
t-ratlo

Constant
10.5355
6.693
1.57

ACO.IM240
0.93568
0.0739
12.7

Dependent variable is:
ANOX.FTP

RA2 = 71.2%



s = 0.2649
with 23 - 2 = 21
degrees of freedom
Source
Sum of Squares
df
Mean Square F-ratlo
Regression
363679
1
3.6368 51.8
Residual
1.47378
21
0.07018
Variable
Coefficient
s.e. of Coeff
t-ratlo
Constant
-0.016187
0.0675
-0.24
ANOX.IM240
0.771357
0.1072
7.2
Dependent variable is:
AMPG.FTP


RA2 = 36.7%




s = 2.885 with 23 - 2 = 21
degrees of freedom

Source
Sum of Squares df
Mean Square
F-ratlo
Regression
101.306
1
101
12.2
Residual
174.8
21
8.32381

Variable
Coefficient
s.e. of Coeff
t-ratlo

Constant
0.639853
0.8074
0.793

AMPG.IM240
0.725132
0.2079
3.49

E-36

-------
HC Emissions  AFTP vs Alndolene-IM240
(23 1983 & Newer PFI Vehicles with Repairs and FTP > 2*(FTP.Stnd))
A FTP
HC (g/ml)
10 T
6 -
\
4 -r
\
2 T
it
2 -
5
Regression Line
RA2 = 90.8%
A IM240 HC (g/ml)
E - 37

-------
HC Emissions - AFTP vs Alndolene-IM240
(23 1983 & Newer PFI Vehicles with Repairs and FTP > 2*(FTP.Stnd))
A FTP
HC (g/ml)
4 -
3.5
2.5
\
2 -
1.5 -
1 ^
0.5 +.
0.5
-
-0.5 -
0.5
1.5
A IU240 HC (g/ml)
Regression Line
R A2 = 90.B%
2.5
Enlarged for Belter Detail near Origin
-|
3
E-38

-------
CO Emissions - AFTP vs Alndolene-IM240
(23 1983 & Newer PFI Vehicles with Repairs and FTP > 2'FTP.Stnd)
a FTP
CO (g/mi)
250 -
200
150 -
100 -
Regression Line
RA2 = 88.4%
-50
50 -
I
0
50
100
1 50
200
2 50
-50 -
A IM240 CO (g/ml)
E-39

-------
CO Emissions - AFTP vs Alndolene-IM240
(23 1983 & Newer PFI Vehicles with Repairs and FTP > 2*FTP.Stnd)
A FTP 100 -
CO (g/mi)
80 -
60 -
40
Regression Line
R*2 = 88.4%
-20
20
i
20
40
60
80
100
-20 -
Enlarged lor Better Detail near Origin
A 111240 CO (g/mi)
E -40

-------
NOx Emissions - A FTP vs Alndolene-IM240
(23 1983 & Newer PFI Vehicles with Repairs and FTP > 2*FTP.Stnd)
A IM240
NOx (g/ml)
A FTP
NOx (g/ml)
0.5
-2.5
- 2
-1.5
- 1
_    ^ T
0.5 I V Q
0.5
Regression Line
R*2 = 71.2%
-0.5
- 1
-1.5
2 -
E-41

-------
NOx Emissions - AFTP vs Alndoiene-IM240
(23 1983 & Newer PFJ Vehicles with Repairs and FTP > 2*FTP.Stnd)
A IM240
NOx (g/ml) ,
-0.5
-0.4
-0.3i
-0.2
-0,1-
0.4
0.2
-G-
-0.2
& FTP
NOx (g/ml)
I
!
0.1
0.2
0.3
-0.4
Regression Line
R*2 = 71.2%
-0.6 -
0.8
- 1 -
Enlarged for Belter Detail near Origin
E -42

-------
Fuel Economy -- AFTP vs Alndolene-IM240
(23 1983 & Newer PFI Vehicles with Repairs and FTP > 2'FTP.Stnd)
- 4
A FTP UPG
(mi/gal)
1 5
10
2
Regression Line
RA2 = 36.7%
\
\
Y

1 0
5 -
A IM240 MPG (ml/gal)
E 43

-------
Fuel Economy -- AFTP vs Alndolene-IM240
1983 & Newer PFI Vehicles with Repairs and FTP > 2*FTP.Stnd)
A FTP MPG
(ml/gal)
5 -
Regression Line
RA2 = 36.7%
- i
0	
- 1
0
-1 -
- 2 -
- 3
- 4 -
A IM240 MPG (ml/gal)
5 -
Enlarged lor Beller Detail near Origin
E-44

-------
Regression Analyses - Throttle-Body Injected Vehicle Sample:
As pait of EPA's regression analysis, FTP results for HC and CO emissions were
each regressed against the corresponding emission results on the Lane IM240, the second
chance 2500/Idle Test, and the second chance Idle Test. Regressions were also performed
comparing NQx results for the FTP and Lane IM240 (NOx emissions are not measured as
pan of either idle test and are therefore not included in this comparison). Lastly, FTP
results for HC, CO, and NOx were each regressed against corresponding emissions from
Lab IM240s run on Indolene fuel at the ATL lab in South Bend and at MVEL in Ann
Arbor. This section provides the results of these regressions performed on a sample of
throttle-body injected (TBI) vehicles recruited as part of the Hammond study.
EPA removed a total of twenty-two vehicles from its sample of TBI vehicles. Six
were removed because of repairs performed after their I/M test but before their FTPs, all
but two of which were exhaust leak repairs. Of the remaining two pre-H P repaired
vehicles, one received a new fuel pump and the other received an alternator and battery
replacement. In addition to these six pre-FTP repair exclusions, sixteen TBI vehicles were
removed from the analyses due to differences of 15% or greater between lane and lab
dynamometer settings. The resulting database consisted of 108 TBI vehicles.
Selecting One Score From the Second Chance 2500/Idle Test:
To illustrate the correlation of FTP HC and CO results to the corresponding results
on the second chance 2500/Idle Test, we had to select an HC score and CO score from the
first chance or second chance test, and from the 2500 rpm mode or the idle mode.
Concerning the first chance versus second chance results, we chose the better results
relative to the HC/CO cutpoint being evaluated.
For the 2500 rpm versus idle modes, we chose the larger of the corresponding emissions
measured. This approach is based on the assumption that the same cutpoints would be
used on each of the modes. Hence, a vehicle would pass the two mode test if and only if
the higher of its HC scores (whether Idle or 2500 rpm) met the HC cutpoint (with the same
also following for the CO results).
Graphing the Data:
On each of the graphs comparing HC emissions, the three horizontal dotted lines
are positioned at the boundaries separating:
	the normal emitters from the high emitters (0.82 g/mi),
the high emitters from the very high emitters (1.64 g/mi), and
the very high emitters from the super emitters (10.00 g/mi).
Similarly, on each of the graphs comparing CO emissions, the three horizonial dotted lines
are positioned at the boundaries separating:
	the normal emitters from the high emitters (10.2 g/mi).
the high emitters from the very high emitters (13.6 g/mi), and
	the very high emitters from the super emitters (150.0 g/mi).
E-45

-------
Regression Analyses of FTP HC versus HC Emissions Over:
IM240, Idle Test, and 2500/idle Test
For TBI Vehicles
Dependent variable is:
HCFTP

R*2 = 64.4%



s = 2.397
with 108 - 2 = 106
degrees of freedom
Source
Sum of Squares
dt
Mean Square F-ratlo
Regression
1 102.31
1
1102' 192
Residual
609.049
106
5.74575
Variable
Coefficient
3.e. of Coeff
t-ratlo
Constant
-0.149912
0.261
-0.574
HCIM24
1.45609
0.1051
13.9
Dependent variable is:
HCFTP


R*2 = 8.4%




s = 3.845
with 108 - 2 = 106
degrees of freedom

Source
Sum of Squares
df
Mean Square
F-ratlo
Regression
143.863
1
144
9.73
Residual
1567.5
1 06
14.7877

Variable
Coefficient
s.e. of Coeff
t-ratlo

Constant
0.89518
0.4242
2.1 1

Hlgher(ldle
0.003197
0.001
3.12

Dependent variable is:
HCFTP


RA2 = 14.0%




s = 3.727
with 108 - 2 = 106
degrees of freedom

Source
Sum of Squares
df
Mean Square
F-ratlo
Regression
238.906
1
239
17.2
Residual
1472.46
106
13.8911

Variable
Coefficient
s.e. of Coeff
t-ratlo

Constant
0.651022
0.418
1.56

Idle.HC
0.005218
0.0013
4.15

E-46

-------
FTP HC
(g/ml)
HC Emissions -- FTP vs Lane IM240
(108 1983 & Newer TBI Vehicles)
35 -
30 x
15 -r
Regresslon Line
RA2 = 64.4%

10 -
5 x
0
6	8
IM240 HC Emissions (g/ml)
1 0
1 2
1 4
E - 4 7

-------
HC Emissions -- FTP vs Lane IM240
(108 1983 & Newer TBI Vehicles)
FTP HP
(g/ml)	Enlarged for Better Detail near Origin
5 T
4.5
4
3.5 JL
3
2.5
2
1.5
1 +


RegrcLMC'i. Line
R',2 ; M.4%
i 
4&:-
^ " - -

0.5	1	1.5	2	2.5	3	3.5
IM240 HC Emissions (g/mi)
E - 48

-------
FTP HC
(g/ml)
HC Emissions -- FTP vs Lane 2500/ldle
(108 1983 & Newer TBI Vehicles)
35 -
I
30 -
25 -
I
Regression Line
2000
2500/ldle HC Emissions (ppm)
E-49

-------
HC Emissions - FTP vs Lane 2500/ldle
(108 1983 & Newer TBI Vehicles)
ftp hc	Enlarged for Better Detail near Origin
(g/ml)
5 -	"
4.5 --
4 --
3.5
3 -
2.5 -
2 -
1.5 --
1 "J----
0.5 -r"*
Regression I
RA2 = 8.2';,
0
1 %;. / ' 1
* ' * * "
0	50	100	150	200	250	300	350	400	450
2500/ldle HC Emissions (ppm)
E-50

-------
FTP HC
(g/ml)
35 v
HC Emissions -- FTP vs Lane Idle
(108 1983 & Newer TBI Vehicles)
30
25
20
Regression Line
RA2 = 14.0%
\
15 
10 -f
500
1000
idle HC Emissions (ppm)
1 500
2000
E 51

-------
FTP HC
(g/ml)
5 -
4.5
4 -
3.5 -
3
2.5
1.5
1 -
0.5
HC Emissions - FTP vs Lane Idle
(108 1963 & Newer TBI Vehicles)
Enlarged for Better Detail near Origin
Regression Line
R*2 = 14.0%
 -v
50
1 00
150	200	250
Idle HC Emissions (ppm)
300
350
400
450
E-52

-------
Regression Analyses of FTP CO versus CO Emissions Over:
IM240. Idle Test, and 2500/Idle Test
For TBI Vehicles
Dependent
variable is:
COFTP


RA2 = 79.9%




s = 20.97
with 108 - 2 = 106
degrees of freedom

Source
Sum of Squares
df
Mean Square
F-ratlo
Regression
185395
1
200000
421
Residual
46625.6
106
439.864

Variable
Coefficient
s.e. of Coeff
t -ratio

Constant
-3.46603
2.331
-1 .49

COIM24
1 .41279
0.0688
20.5


Dependent
vaiiable is:
COFTP
4


RA2 = 46.9%




s = 34.10
with 108 - 2 = 106
degrees of freedom

Source
Sum of Squares
df
Mean Square
F-ratlo
Regression
108761
1
100000
93.5
Residual
123260
106
1162.83

Variable
Coefficient
s.e. of Coeff
t-ratlo

Constant
4.43526
3.677
1.21

Hlqher(ldle
15.0538
1.557
9.67

Dependent
variable is:
COFTP


RA2 = 51.1%




s = 32.73
with 108 - 2 = 106
degrees of freedom

Source
Sum of Squares
df
Mean Square
F-ratlo
Regression
1 18502
1
100000
11 1
Residual
1 13519
106
1070.93

Variable
Coefficient
s.e. of Coeff
t-ratlo

Constant
5.15188
3.47
1.48

Idle.CO
1 7.2888
1.644
10.5

E-53

-------
FTP CO
(g/ml)
CO Emissions - FTP vs Lane IM240
(108 1983 & Newer TBI Vehicles)
300
250
200 -1-
150	
100 -
50 --
Regression Line
R*2 = 79.9%
  -
25
50
75	100	125
IM24Q CO Emissions (g/ml)
1 50
175
200
E-54

-------
CO Emissions - FTP vs Lane IM240
(108 1983 & Newer TBI Vehicles)
Enlarged, lor Better Detail near Origin
50 -
45 --
40
35 --
30 -
25 --
20
15 +
10
5 -
Regression Line
RA2 = 79.9%
r.

:.*>%< ; 
10	1 5	20	25	30	35	40
IM240 CO Emissions (g/ml)
E- 55

-------
CO Emissions - FTP vs Lane 2500/idle
(108 1983 & Newer TBI Vehicles)
Regression Line
RA2 = 46.9%
\
\
\
\
5
4	5	6
2500/ldle CO Emissions (percent)
E - 56

-------
FTP CO
(g/ml)
50 r
4 5 -
40 --
35
30 --
25 --
20 --
15
CO Emissions - FTP vs Lane 2500/Jdle
(108 1983 & Newer TBI Vehicles)
Enlarged (or Better Detail near Origin
Regression Line
RA2 = 46.9%
10 i "b 

 
5 iti* * \ "
0.2
0.4
0.6	0.8	1	1.2
2500/ldle CO Emissions (percent)
1.4
1.6
1.8
E-57

-------
FTP CO
(B'ml)
300 T
CO Emissions ~ FTP vs Lane Idle
(108 1983 & Newer TBI Vehicles)
250
200 -f
Regresslon Line
RA2 = 51.1%
150
\
100 +
50
 -r
~T
2
4	5	6
Idle CO Emissions (percent)
1 0
E -58

-------
CO Emissions -- FTP vs Lane Idle
(108 1983 & Newer TBI Vehicles)
Enlarged for Better Detail near Origin
FTP CO
(g'ml)
50


Regression Line
45


RA2 = 51.1%
\
40


\
\
35


\
\
30


 
\

25



20
-

 "M .
	
1 5

.. < a

10
5
0
1 
		 "
1 " 
my-*

a '
n
 .
m
0	0.2	0.4	0.6	0.8	1	1.2	1.4	1.6	1.8	2
idle CO Emissions (percent)
E -59

-------
Regression Analyses of FTP NOx versus IM240 NOx Emissions
(Note: NOx is not Measured for the Idle Test)
Dependent variable is:
NOFTP.TBI


RA2 = 80.7%




s = 0.4677
with 108 - 2 = 106 degrees of freedom

Source
Sum of Squares
df
Mean Square
F-ratlo
Regression
96.941
1
96.94
443
Residual
23.1857
1 06
0.218733

Variable
Coefficient
s.e. of Coeff
t-ratlo

Constant
-0.069973
0.071
-0.985

NOIM24.TBI
0.734578
0.0349
21.1

E-60

-------
NOx Emissions - FTP vs Lane IM240
ftpnox	(108 1983 & Newer TBI Vehicles)
(g/mi)
7 -
Regression Line
RA2 = 80.7%
4 --
3 --
2 --
0
i ***:
-	V
%  
-	
0	1	2	3	4	5
IM240 NOx Emissions (g/ml)
E-61

-------
NOx Emissions - FTP vs Lane IM240
(108 1983 & Newer TBI Vehicles)
FTPN0*	Enlarged for Belter Detail near Origin
(g/ml)
2 :
1.75 -
1.5
1.25 -
0.75
0.5
0.25 -
0
Regression Line
R*2 = 80.7%
\
 " .
 mm'  1
'.V--  "
 IV. 
  ,  *:
? v
0	0.25	0.5	0.75	1	1.25	1.5
IM240 NOx Emissions {g/ml)
E -62

-------
Vehicles Tested by ATL at Southbend, Indiana
Regressions of FTP vs lndolene-IM240
for 91 1983 & Newer TBI Vehicles
Dependent variable i
RA2 = 90.1%
s = 1.354 with 91
Source
Regression
Residual
HCFTP
2 = 89 degrees of freedom
Sum of Squares	df
1491.43	1
163.188	89
Mean Square F-ratto
1491	313
1.83357
Variable	Coefficient s.e. of Co eft t-ratlo
Constant	0.002929	0.1537	0.019
IHC.IM24.lndO...	1.66576	0.0584	28.5
Dependent variable
is COFTP



R*2 = 90.3%




s = 15.12 with 91
- 2 = 89 degrees of freedom


Source
Sum of Squares
df
Mean Square
F-ratlo
Regression
190004
1
200000
831
Residual
20343.9
89
228.583

Variable
Coefficient
s.e. of Coeff
t-ratlo

Constant
-0.174774
1.758
-0.099

CO.IM24.lndo...
1.34497
0.0467
28.8

Dependent variable if
NOx.FTP



RA2 = 92.6%




S = 0.2989 with 91
- 2 = 89 degrees of freedom


Source
Sum of Squares
df
Mean Square
F-ratlo
Regression
99.1838
r
99.18
1110
Residual
7.9495
89
0.08932

Variable
Coefficient
s.e. of Coeff
t-ratlo

Constant
0.073869
0.0443
1.67

NOx.IM24.lnd...
0.845099
0.0254
33.3

E-63

-------
FTP HC
(g/ml)
30 -
10 -
5 -
rAtf
HC Emissions - FTP vs lndolerie-IM240
(91 1983 & Newer TBI Vehicles)
Regression Line
= 90.1%
\
\

X
10	15
IM240 HC Emissions (g/ml)
E-64

-------
FTP HC
(g/ml)
3 -
2.5 -
2 -
1.5 -
HC Emissions - FTP vs lndolene-)M240
(91 1983 & Newer TBI Vehicles)
Enlarged for Better Detail near Origin

I 
0.5 -i

# 
Regression Line
RA2 = 90.1%
0.5	1	1.5	2	2.5
IM240 HC Emissions (g/ml)
E-65

-------
FTP CO
(g/ml)
300 -
150 -
100 --
CO Emissions - FTP vs lndolene-IM240
(91 1983 & Newer TBI Vehicles)

Regression Line
RA2 = 90.3%
\
x
20	40	60	60	100	120	140
IM240 CO Emissions (g/ml)
E-66

-------
FTP CO
(g/ml)
50 -
CO Emissions - FTP vs lndolene-IM240
(91 1983 & Newer TBI Vehicles)
Enlarged lor Better Detail near Origin
40 -
20 -

4 til 
3',. 
1 0
20	30
IM240 CO Emissions 
-------
FTP NOx
(g/ml)
6.00 -
5.00 --
2.00 -r
NOx Emissions - FTP vs lndolene-IM240
(91 1983 & Newer TBI Vehicles)
Regression Line
RA2 = 92.6%
X.

0.00
*
1.00	2.00
3.00	4.00	5.00
IM240 NOx Eml8*lons (g/ml)
6.00	7.00
8.00
E -68

-------
0.75 -
0.50 -r
0.25 -
0.00 -
NOx Emissions -- FTP vs lndolene-IM240
(91 1983 & Newer TBI Vehicles)
Enlarged for Better Detail near Origin
FTP NOx
(g/ml)
2.00 -
Regression Line
1.75 -1-	R*2 = 92.6%
1.50

%
 n
n
m
v-  i 
-'' 
i  
0.00	0.25	0.50	0.75	1.00	1.25	1.50	1.75	2.00
IM240 NOx Emissions (g/ml)
E-69

-------
EF Vehicles Tested at EPA's MVEL
Regressions of FTP vs Indolene IM240
for 62 1985-88 TBI Vehicles
Dependent variable is: HC.FTP


RA2 = 87.7%



S = 1.703 with
62 - 2 = 60 degrees of freedom


Source
Sum of Squares df
Mean Square
F-ratlo
Regression
1241.84 1
1242
428
Residual
173.984 60
2.89973

Variable
Coefficient s.e. of Coeff
t-ratlo

Constant
0.146231 0.2279
0.642

HCIM24
1.71372 0.0828
20.7

Dependent variable is:
CO. FTP


RA2 = 92.5%




s = 11.59 with
62 - 2 = 60 degrees of freedom


Source
Sum of Squares
df
Mean Square
F-ratlo
Regression
99522.9
1
99523
741
Residual
8056.07
SO
134.268

Variable
Coefficient
s.e. of Coeff
t-ratlo

Constant
2.56957
1.584
1.62

COIM24
1.10181
0.0405
27.2

Dependent variable is: NOX.FTP


RA2 = 90.9%



S = 0.1594
with 62 - 2 = 60 degrees of freedom


Source
Sum of Squares df
Mean Square
F-ratlo
Regression
15.2107 1
15.21
599
Residual
1.52398 60
0.0254

Variable
Coefficient s.e. of Coeff
t-ratlo

Constant
0.067854 0.0323
2.1

NOIM24
0.780974 0.0319
24.5

E-70

-------
HC Emissions - FTP vs lndolene-IM240
(62 1985-88 TBI Vehicles)
(Tested at EPA's Ann Arbor Lab)
FTP HC
(g/ml)
Regression Line
30.00 --	RA2 = 87.7%
\
20.00 --
10.00 -1-
Jl1-
o.oo 	
0.00
5.00	10.00	15.00
IM240 HC Emissions (g'ml)
E 7 1

-------
FTP HC
(a/ml)
3.00 -
2.00 -
HC Emissions -- FTP vs lndolene-IM240
(62 1985-88 TBI Vehicles)
(Tested at EPA's Ann Arbor Lab)
Enlarged for Better Detail near Origin
Regression Line
R2 = 87.7%
 
1.00 +	.	" 
 r

o.oo
0.00	0.50	1.00	1.50	2.00
IM240 HC Emissions (g/ml)
E-72

-------
FTP CO
(g/ml)
200.00 --
150.00 -
100.00 -
50.00 --
CO Emissions -- FTP vs lndolene-IM240
(62 1985-88 TBI Vehicles)
(Tested at EPA's Ann Arbor Lab)
Regression Line
RA2 = 92.5%

S
0.00
0.00
50.00
100.00
IM240 CO Emissions (fl/ml)
150.00
E- 73

-------
FTP CO
(g/ml)
50.00 -
CO Emissions - FTP vs lndolene-IM240
(62 1985-88 TBI Vehicles)
(Tested at EPA's Ann Arbor Lab)
Enlarged for Better Detail near Origin
40.00
30.00
20.00
Regression Line
R*2 = 92.5%
10.00 T
 
0.00
0.00
10.00
20.00	30.00
IM240 CO Emissions (g/ml)
40.00
50.00
E- 74

-------
FTP NOx
2.50
2.00 -
1.50
1.00
0.50 -
J
i .
0.00 
NOx Emissions -- FTP vs lndolene-IM240
(62 1985-88 TBI Vehicles)
(Tested at EPA's Ann Arbor Lab)
3.00 -j-	Regression Line
R"2 = 90.9%
\
\
\
m
u
- - :
-"
0.00	0.50	1.00	1.50	2.00	2.50	3.00
IM240 NOx Emissions (g/mi)
E-75

-------
FTP NOx
(a/ml)
2.00
1.50
1.00
0.50
NOx Emissions - FTP vs lndofene-IM240
(62 1985-88 TBI Vehicles)
(Tested at EPA's Ann Arbor Lab)
Enlarged for Better Detail near Origin
Regression Line
RA2 = 90.9%
0.00 	
0.00

N. c
ill*.

0.50
1.00
IM240 NOx Emissions (g'ml)
1.50
2.00
E-76

-------
Description of Repair Regression Analysis
Regression Analyses
The difference in emission levels before and after repairs for the FTP's and Lab
IM240's run on Indolene fuel was calculated such that a positive difference would
correspond to a decrease in emissions or an increase in fuel economy. The AFTP values
(AHC, ACO, ANQx and AMPG) were regressed against the corresponding Indolene
AIM240 values. The results and graphs of these regressions are included.
Data Description
The data used in this analysis consisted of FTP and Indolene EM240 scores for tests
performed before and after repairs. These results consisted of HC, CO, NOx, and fuel
economy values. The database was restricted to 1983 and newer PFI and TBI vehicles
which received repairs and whose as received FTP scores exceeded twice the FTP standard
for either HC or CO. Also excluded from the analysis were vehicles which received repairs
before the as received FTP and vehicles which had differences in dynamometer settings of
greater than 15%. The resulting data set contained 41 TBI vehicles and 23 PFI vehicles.
Summary Statistics for the variables used in the TBI analyses follow.
E-77

-------
Statistical Summaries for Variables Used in
TBI Before and After Regression Analyses
Summary statistics for AHC.IM24
NumNumeric = 41
Mean = 1.3563
Median = 0.2200
Midrange = 7.8500
Standard Deviation = 3.4020
Range = 17.660
Minimum = -0.98000
Maximum = 16.680
Summary statistics for AHC.FTP
NumNumeric = 41
Mean = 2.4634
Median = 0.3700
Midrange = 14.595
Standard Deviation = 6.0860
Range = 30.590
Minimum = -0.70000
Maximum = 29.890
Summary statistics for ACO.IM24
NumNumeric = 41
Mean = 24.092
Median = 4.8800
Midrange = 84.310
Standard Deviation = 47.490
Range = 186.72
Minimum = -9.0500
Maximum = 177.67
Summary statistics for: ANOx.IM24
NumNumeric = 41
Mean = 0.04683
Median = -0.07000
Midrange = 2.2850
Standard Deviation = 1.2167
Range = 8.3700
Minimum = -1.9000
Maximum = 6.4700
Summary statistics for: AMPG.IM24
NumNumeric = 41
Mean = 2.4132
Median = 1.4300
Midrange = 5.2600
Standard Deviation = 3.4721
Range = 15.420
Minimum = -2.4500
Maximum= 12.970
Summary statistics for ACO.FTP
NumNumeric = 41
Mean = 38.661
Median = 5.4500
Midrange = 128.11
Standard Deviation = 69.2007
Range = 265.65
Minimum = -4.7200
Maximum = 260.93
Summary statistics for ANOx.FTP
NumNumeric = 41
Mean = -0.06439
Median = -0.06000
Midrange = 1.5800
Standard Deviation = 1.0174
Range = 6.8800
Minimum = -1.8600
Maximum = 5.0200
Summary statistics for: AMPG.FTP
NumNumeric = 41
Mean = 2.0820
Median = 0.97000
Midrange = 3.6850
Standard Deviation = 3.7642
Range = 21.810
Minimum = -7.2200
Maximum = 14.590
E-78

-------
Regression Analyses for the Change in Emission Constituents
Before and After Repairs of the TBI Vehicles for the FTP vs IM240-A
Dependent variable is:
AHC.FTP


RA2 = 87.2%




s = 2.209
with 41 - 2 = 39 degrees of freedom


Source
Sum of Squares
df Mean Square
F-ratlo
Regression
1 291.3
1
1 291
265
Residual
190.271
39
4.87875

Variable
Coefficient
s.e. of Coeff
t-ratlo

Constant
0.198146
0.372
0.533

AHC.IM24
1.67013
0.1027
16.3

Dependent variable is:
ACO.FTP


RA2= 82.1%




s = 29.57
with 41 - 2 = 39 degrees o< freedom

Source
Sum of Squares
df
Mean Square
F-ratlo
Regression
156368
1
200000
1 79
Residual
34110.7
39
874.633

Variable
Coefficient
s.e. of Coeff
t-ratlo

Constant
6.94316
5.192
1.34

ACO.IM240
1.31656
0.0985
13.4

Dependent variable is:
ANOX.FTP


RA2 = 84.9%




S = 0.4008
wi(h 41 - 2 = 39
degrees of freedom

Source
Sum of Squares df
Mean Square
F-ratlo
Regression
35.1435
1
35.14
219
Residual
6.26452
39
0.160629

Variable
Coefficient
9.e. of Coeff
t-ratlo

Constant
-0.100466
0.0626
-1.6

ANOX.IM240
0.770364
0.0521
14.8

Dependent variable is:
ampg.ftp


RA2 = 58.3%




s = 2.489 with
41 - 2 = 39 degrees of freedom


Source
Sum of Squares
df Mean Square
F-ratlo
Regression
325.213
1
325
52.5
Residual
241.558
39
6.19379

Variable
Coefficient
s.e. of Coeff
t-ratlo

Constant
0.100224
0.4753
0.158

AMPG.IM240
0.821213
0.1133
7.25

E-79

-------
A FTP
HC (g/ml)
HC Emissions  A FTP vs Alndolene-IM240
(41 1983 & Newer TBI Vehicles with Repairs and FTP > 2'FTP.Stnd)
35 T
30
25 t
20 -
1 5
10

-2
- 5 -
Regression Line
R*2 = 87.2%
1 0
1 2
t 4
16
18
A 111240 HC (g/ml)
E -80

-------
(41 1983
HC Emissions -- AFTP vs Alndolene-IM240
& Newer TBI Vehicles with Repairs and FTP > 2*FTP.Stnd)
A FTP HC
(g/ml)
2.5
1.5
0.5


-0.5 ~
/
/
i 
0.5
-0.5 -
- 1 ~ Expanded lor Belter Delail near Origin
\
Regression Line
R*2 = 87.2%
1.5
2.5
A IM240 HC (g/ml)
E 81

-------
A FTP
CO (g/ml)
300
CO Emissions - AFTP vs Alndolene-IM240
(41 1983 & Newer TBI Vehicles with Repairs and FTP  2*FTP.Stnd)
250
200
150 -
Regression Line
RA2 = 82.1%
100
50 -L
 *

2t)
20
40
60
80
100	120	140	160
1 80
-50 -
A 111240 CO (g/ml)
E -82

-------
A FTP
CO (g/mi)
CO Emissions - A FTP vs Alndolene-IM240
(41 1983 & Newer TBI Vehicles with Repairs and FTP > 2'FTP.Stnd)
50 -
40
30 -
20
Regression Line
R*2 = 82.1%
hi
10 -
-0-"-

10^
t 0
20
30
-10 -J-
Expanded for Better Detail near Origin
A IU240 CO (gymi)
40
-I
50
E -83

-------
NOx Emissions  AFTP vs Alndolene-IM240
(41 1983 & Newer TBI Vehicles with Repairs and FTP > FTP.stnd)
A FTP
NOx (g/ml)
6 -
5 -
4 -
3 -
Regression Line
RA2 = 84.9%
- 2
#-1
#1^
" 1
2 -
A IM240 NOx (g/mi)
E-84

-------
NOx Emissions - A FTP vs Alndolene-IM240
(41 1983 & Newer TBI Vehicles with Repairs and FTP > FTP.stnd)
A IM240
NOx fa/mll
1.5
6. FTP NOx
(g/ml)
2 -
1.5 -
1 -
0.5 -
Regression Line
R*2 = 64.9%
-0.5
m fr\
 * o 
0.5
1.5
-0.5
1 x
1.5
- 2 -
Expanddd for Better Detail near Origin
E -85

-------
Fuel Economy - AFTP vs Alndolene-1M240
(41 1983 & Newer TBI Vehicles with Repairs and FTP > 2*FTP.Stnd)
15 -
A FTP
MPG (ml/gal)
10
Regression Line
R*2 = 57.4%
5 -
 

f
10
12	14
-5 -
-1 0 -
A IM240 UPG (ml/gal)
E -86

-------
Fuel Economy - AFTP vs Alndolene-IM240
(41 1983 & Newer TBI Vehicles with Repairs and FTP > 2*FTP.Slnd)
A FTP MPG
(ml/gal) 5
- 3
v1-
2 --
* 1

 1
-	2
-	3 -f
- 5
Regression Line
R*2 = 57.4%

 " 
A IM240 MPG
(ml/gal)
Expanded for Better Detail near Origin
E -87

-------
APPENDIX F
IM2 4 0 CUTPOINT TABEL ANALYSIS

-------
APPENDIX F:
IM240 Cutpoint Table Analysis
The objective of this analysis was to investigate the performance of the
IM240 as an I/M test using various outpoints and cutpoint combinations
and pick outpoints for the enhanced I/M model program. The main
characteristics investigated in this analysis were failure rate (% of
vehicles exceeding an IM240 cutpoint combination), error of commission
rate (% of FTP passing vehicles exceeding a cutpoint), and excess
emission'identification rate <% of emissions above the FTP standards
identified by a cutpoint). Tables at the end of this appendix
illustrate the results of this analysis.
Since FTP results were needed to calculate the error-of-commission rate
(Ec rate) and the excess emission identification rate (excess IDR), the
data used were restricted to vehicles recruited for FTP testing from
the Hammond I/M lane to ATL's New Carlisle laboratory. Other factors
described below further restricted the available data.
	Since the IM240 is targeted toward identifying newer technology cars
only 1983 and newer cars were used.
	In order to maintain consistency with similar analyses performed for
the MOBILE emission factor model, only vehicles having as-received
Idle/2500 data, lane IM240 data, lab tank fuel IM240 data and as-
received FTP data were used.
	Vehicles receiving repairs prior to the initial FTP test but
following the lane IM240 test were excluded from the database. Such
repairs are expected to cause emission changes that prohibit using
the data for these analyses.
	Similarly, vehicles tested at substantially different dynamometer
settings (>15% different) were also removed from the database. These
differences in dynamometer settings arose from the use of different
dynamometer look-up tables at the lane and lab, and has since been
corrected. Traditional look-up "tables used for emission factor
laboratory testing are too detailed for efficient use in the I/M lane
so simpler tables were developed for use at the lane. Because the
simplified tables developed for I/M lane are also derived from a
broader sample of option packages within a vehicle model, the
simplified tables are also more representative than the traditional
tables which are restricted to vehicles tested for certification.
After becoming aware of the dynamometer setting differences in the
two tables, EPA opted to also use the simplified I/M lane table for
the laboratory tests.
The resulting database consisted of 274 vehicles; 90 were port fuel
injected (PFI), 129 were throttle body injected (TBI), and 55 were
carbureted (Carb). Because NOx was not originally a laboratory
recruitment criterion, it was felt that the database did not adequately
represent typical NOx failing vehicles. Therefore, 16 vehicles
recruited specifically for NOx repairs based on their IM240 NOx
emissions were added. These vehicles were subject to the previously
F-l

-------
described restrictions and consisted of 6 TBI,  PFI and 4 carbureted
cars.
The recruiting criteria for these laboratory vehicles were based on
IM240- emission levels and targeted toward higher emitting vehicles.
Because a higher fraction of dirty cars were recruited to the
laboratory, these cars were not representative of the sample of vehicles
encountered at the Hammond lane. In order to approximate the
distribution of vehicles at the Hammond lane and offset the sample bias
caused by the IM240-based recruitment criteria, the database was
weighted using the IM240 recruitment criteria. For example, the ratio
of vehicles exceeding the recruitment criteria in the database to the
vehicles exceeding the criteria at the lane was used to weight the
"high" emitting vehicles. The "low" emitting vehicles were similarly
weighted. The result was a weighted database with a similar
distribution as the Hammond lane sample about the recruitment criteria.
This weighted database was used to produce real-world estimates of Ec's
and Excess IDR's. The following section describes the tables which
display the results of this analysis.
The tables that follow were calculated using the two-ways-to-pass
criteria. For this method, a vehicle fails only if its composite
emissions exceed the HC or CO composite cutpoints, and its Bag 2
emissions exceed the HC or CO Bag 2 cutpoints. In other words, a
vehicle can pass by having low emissions in Bag 2 even if its Bag 1
emissions were high. NOx cutpoint criteria are unaffected by the two-
ways-to-pass algorithm. EPA has found that this method reduces the
chances of inappropriately failing clean vehicles without sacrificing
the indentification of dirty cars, and therefore, is proposing this
approach to IM240 cutpoints.
Separate tables were calculated for each fuel metering group. The first
four columns of these tables refer specifically to HC/CO-only cutpoints.
These columns include the cutpoint criteria, estimated HC excess
emission IDR's, estimated CO excess emission IDR's and estimated error
of commission rates associated with the HC/CO cutpoint criteria. The
next four columns specifically address NOx-only cutpoints and include
the estimates for NOx excess emission IDR's and error of commission
rates associated with the NOx cutpoints. The final three columns
illustrate actual Hammond Lane failure rates associated with the HC/CO-
only-, NOx-only and combined HC/CO/NOx cutpoints. Excess IDR's and Ec's
can be estimated simply by combining the IDR's and Ec's of the HC/CO-
only and NOx-only outpoints, although this will underestimate the IDRs,
somewhat, making them conservative. This approach was used because the
MOBILE model uses IDRs for NOx cutpoints separate from HC/CO cutpoints,
so using the same approach for this document trades off conservative IDR
estimates for the sake of consistency.
F-2

-------
1983 and newer Port Fuel Injected Vehicles
HC/CO OUTPOINTS ONLY
NOX CUTPOINTS ONLY
Actual
Actual
Actual
HC/CO
HC Excess
CO Excess
HC/CO

NOx Excess

Hammond
Hammond
Hammond
CUTPOINT
Emissions
Emissions
only
NOx
Emissions
Nox Only
HC/CO
NOx
Combined
Composite-Bag 2
Identified
Identified
Ec Rate
Cutpoint
Identified
Ec Rate
Failure Rate
Failure Rate
Failure Rate
0.4/20-0.25/16
93%
89%
1.5%
1.75
88%
2.2%
17.9%
7.1%
21.5%
0.6/20-0.375/16
88%
85%
0.0%
1.75
88%
2.2%
9.2%
7.1%
14.0%
0.8/20-0.5/16
81%
82%
0.0%
1.75
88%
2.2%
6.2%
7.1%
14.4%
1.0/20-0.625/16
56%
73%
0.0%
1.75
88%
2.2%
4.9%
7.1%
13.6%
1.2/20-0.75/16
55%
72%
0.0%
1.75
88%
2.2%
4.1%
7.1%
13.1%
0.4/18-0.25/14.4
93%
89%
1.5%
1.75
88%
2.2%
17.9%
7.1%
23.6%
0.6/18-0.375/14.4
88%
85%
0.0%
1.75
88%
2.2%
9.4%
7.1%
16.6%
0.8/18-0.5/14.4
81%
82%
0.0%
1.75
88%
2.2%
6.5%
7.1%
14.5%
1.0/18-0.625/14.4
56%
73%
0.0%
1.75
88%
2.2%
5.3%
7.1%
13.7%
1.2/18-0.75/14.4
55%
72%
0.0%
1.75
88%
2.2%
4.6%
7.1%
13.3%
0.4/15-0.25/12
93%
89%
1.5%
1.75
88%
2.2%
18.2%
7.1%
23.8%
0.6/15-0.375/12
88%
85%
0.0%
1.75
88%
2.2%
9.9%
7.1%
17.1%
0.8/15-0.5/12
81%
83%
0.0%
1.75
88%
2.2%
7.1%
7.1%
15.0%
1.0/15-0.625/12
56%
74%
0.0%
1.75
88%
2.2%
6.0%
7.1%
14.3%
1.2/15-0.75/12
55%
73%
0.0%
1.75
88%
2.2%
5.3%
7.1%
13.9%
0.4/12-0.25/9.6
94%
90%
1.5%
1.75
88%
2.2%
18.9%
7.1%
24.5%
0.6/12-0.375/9.6
89%
86%
0.0%
1.75
88%
2.2%
11.6%
7.1%
18.4%
0.8/12-0.5/9.6
82%
84%
o:o%
1.75
88%
2.2%
9.2%
7.1%
16.7%
1.0/12-0.625/9.6
66%
80%
0.0%
1.75
88%
2.2%
8.2%
7.1%
16.0%
1.2/12-0.75/9.6
66%
80%
0.0%
1.75
88%
2.2%
7.6%
7.1%
15.7%
0.4/10-0.25/8
94%
90%
1.5%
1.75
88%
2.2%
20.1%
7.1%
25.6%
0.6/10-0.375/8
89%
86%
0.0%
1.75
88%
2.2%
13.6%
7.1%
20.2%
0.8/10-0.5/8
82%
84%
0.0%
1.75
88%
2.2%
11.6%
7.1%
18.8%
1.0/10-0.625/8
66%
80%
0.0%
1.75
88%
2.2%
10.7%
7.1%
18.3%
1.2/10-0.75/8
66%
80%
0.0%
1.75
88%
2.2%
10.4%
7.1%
18.1%
0.4/20-0.25/16
93%
89%
1.5%
2
74%
0.0%
17.9%
7.1%
21.5%
0.6/20-0.375/16
88%
85%
0.0%
2
74%
0.0%
9.2%
7.1%
14.0%
0.8/20-0.5/16
81%
82%
0.0%
2
74%
0.0%
6.2%
7.1%
11.8%
1.0/20-0.625/16
56%
73%
0.0%
2
74%
0.0%
4.9%
7.1%
10.8%
1.2/20-0.75/16
55%
72%
0.0%
2
74%
0.0%
4.1%
7.1%
10.3%
0.4/18-0.25/14.4
93%
89%
1.5%
2
74%
0.0%
17.9%
7.1%
21.5%
0.6/18-0.375/14.4
88%
85%
0.0%
2
74%
0.0%
9.4%
7.1%
14.1%
0.8/18-0.5/14.4
81%
82%
0.0%
2
74%
0.0%
6.5%
7.1%
11.9%
1.0/18-0.625/14.4
56%
73%
0.0%
2
74%
0.0%
5.3%
7.1%
11.0%
1.2/18-0.75/14.4
55%
72%
0.0%
2
74%
0.0%
4.6%
7.1%
10.5%
0.4/15-0.25/12
93%
89%
1.5%
2
74%
0.0%
18.2%
7.1%
21.8%
0.6/15-0.375/12
88%
85%
0.0%
2
74%
0.0%
9.9%
7.1%
14.5%
0.8/15-0.5/12
81%
83%
0.0%
2
74%
0.0%
7.1%
7.1%
12.4%
1.0/15-0.625/12
56%
74%
0.0%
2
74%
0.0%
6.0%
7.1%
11.5%
1.2/15-0.75/12
55%
73%
0.0%
2
74%
0.0%
5.3%
7.1%
11.1%
F-3

-------
1983 and newer Port Fuel Injected Vehicles
HC/CO OUTPOINTS ONLY
NOX CUTPOINTS ONLY
Actual
Actual
Actual
HC/CO
HC Excess
CO Excess
HC/CO

NOx Excess

Hammond
Hammond
Hammond
CUTPOINT
Emissions
Emissions
only
NOx
Emissions
Nox Only
HC/CO
NOx
Combined
Composite-Bag 2
Identified
Identified
Ec Rate
Outpoint
Identified
Ec Rate
Failure Rate
Failure Rate
Failure Rate
0.4/12-0.25/9.6
94%
90%
1.5%
2
74%
0.0%
18.9%
7.1%
22.4%
0.6/12-0.375/9.6
89%
86%
0.0%
2
74%
0.0%
11.6%
7.1%
15.9%
0.8/12-0.5/9.6
82%
84%
0.0%
2
74%
0.0%
9.2%
7.1%
14.2%
1.0/12-0.625/9.6
66%
80%
0.0%
2
74%
0.0%
8.2%
7.1%
13.4%
1.2/12-0.75/9.6
66%
80%
0.0%
2
74%
0.0%
7.6%
7.1%
13.1%
0.4/10-0.25/8
94%
90%
1.5%
2
74%
0.0%
20.1%
7.1%
23.6%
0.6/10-0.375/8
89%
86%
0.0%
2
74%
0.0%
13.6%
7.1%
17.8%
0.8/10-0.5/8
82%
84%
0.0%
2
74%
0.0%
11.6%
7.1%
16.4%
1.0/10-0.625/8
66%
80%
0.0%
2
74%
0.0%
10.7%
7.1%
15.8%
1.2/10-0.75/8
66%
80%
0.0%
2
74%
0.0%
10.4%
7.1%
15.6%
0.4/20-0.25/16
93%
89%
1.5%
2.25
65%
0.0%
17.9%
5.0%
20.2%
0.6/20-0.375/16
88%
85%
0.0%
2.25
65%
0.0%
9.2%
5.0%
12.2%
0.8/20-0.5/16
81%
82%
0.0%
2.25
65%
0.0%
6.2%
5.0%
9.8%
1.0/20-0.625/16
56%
73%
0.0%
2.25
65%
0.0%
4.9%
5.0%
8.9%
1.2/20-0.75/16
55%
72%
0.0%
2.25
65%
0.0%
4.1%
5.0%
8.3%
0.4/18-0.25/14.4
93%
89%
1.5%
2.25
65%
0.0%
17.9%
5.0%
20.2%
0.6/18-0.375/14.4
88%
85%
0.0%
2.25
65%
0.0%
9.4%
5.0%
12.4%
0.8/18-0.5/14.4
81%
82%
0.0%
2.25
65%
0.0%
6.5%
5.0%
10.1%
1.0/18-0.625/14.4
56%
73%
0.0%
2.25
65%
0.0%
5.3%
5.0%
9.1%
1.2/18-0.75/14.4
55%
72%
0.0%
2.25
65%.
0.0%
4.6%
5.0%
8.6%
0.4/15-0.25/12
93%
89%
1.5%
2.25
65%
0.0%
18.2%
5.0%
20.5%
0.6/15-0.375/12
88%
85%
0.0%
2.25
65%
0.0%
9.9%
5.0%
12.8%
0.8/15-0.5/12
81%
83%
0.0%
2.25
65%
0.0%
7.1%
5.0%
10.6%
1.0/15-0.625/12
56%
74%
0.0%
2.25
65%
0.0%
6.0%
5.0%
9.7%
1.2/15-0.75/12
55%
73%
0.0%
2.25
65%
0.0%
5.3%
5.0%
9.2%
0.4/12-0.25/9.6
94%
90%
1.5%
2.25
65%
0.0%
18.9%
5.0%
21.1%
0.6/12-0.375/9.6
89%
86%
0.0%
2.25
65%
0.0%
11.6%
5.0%
14.3%
0.8/12-0.5/9.6
82%
84%
0.0%
2.25
65%
0.0%
9.2%
5.0%
12.5%
1.0/12-0.625/9.6
66%
80%
0.0%
2.25
65%
0.0%
8.2%
5.0%
11.7%
1.2/12-0.75/9.6
66%
80%
0.0%
2.25
65%
0.0%
7.6%
5.0%
11.4%
0.4/10-0.25/8
94%
90%
1.5%
2.25
65%
0.0%
20.1%
5.0%
22.3%
0.6/10-0.375/8
89%
86%
0.0%
2.25
65%
0.0%
13.6%
5.0%
16.2%
0.8/10-0.5/8
82%
84%
0.0%
2.25
65%
0.0%
11.6%
5.0%
14.7%
1.0/10-0.625/8
66%
80%
0.0%
2.25
65%
0.0%
10.7%
5.0%
14.1%
1.2/10-0.75/8
66%
80%
0.0%
2.25
65%
0.0%
10.4%
5.0%
13.8%
F-4

-------
1983 and newer Throttle Body Injected Vehicles
HC/CO OUTPOINTS ONLY
NOX OUTPOINTS ONLY
Actual
Actual
Actual
HC/CO
HC Excess
CO Excess
HC/CO

NOx Excess

Hammond
Hammond
Hammond
OUTPOINT
Emissions
Emissions
only
NOx
Emissions
Nox Only
HC/CO
NOx
Combined
Composite-Bag 2
Identified
Identified
EcRate
Outpoint
Identified
Ec Rate
Failure Rate
Failure Rate
Failure Rate
0.4/20-0.25/16
97%
89%
4.0%
1.75
95%
0.0%
32.0%
17.7%
38.1%
0.6/20-0.375/16
94%
84%
1.2%
1.75
95%
0.0%
19.7%
17.7%
28.8%
0.8/20-0.5/16
89%
76%
1.2%
1.75
95%
0.0%
13.2%
17.7%
28.4%
1.0/20-0.625/16
77%
70%
0.0%
1.75
95%
0.0%
9.4%
17.7%
26.9%
1.2/20-0.75/16
70%
66%
0.0%
1.75
95%
0.0%
7.2%
17.7%
26.3%
0^4/18-0.25/14.4
97%
89%
4.0%
1.75
95%
0.0%
32.0%
17.7%
40.8%
0.6/18-0.375/14.4
94%
83%
1.2%
1.75
95%
0.0%
19.9%
17.7%
32.3%
0.8/18-0.5/14.4
89%
76%
1.2%
1.75
95%
0.0%
13.5%
17.7%
28.7%
1.0/18-0.625/14.4
72%
62%
0.0%
1.75
95%
0.0%
9.8%
17.7%
27.2%
1.2/18-0.75/14.4
64%
54%
0.0%
1.75
95%
0.0%
7.7%
17.7%
26.7%
0.4/15-0.25/12
97%
89%
4.0%
1.75
95%
0.0%
32.2%
17.7%
41.0%
0.6/15-0.375/12
94%
84%
1.2%
1.75
95%
0.0%
20.5%
17.7%
32.8%
0.8/15-0.5/12
92%
82%
1.2%
1.75
95%
0.0%
14.6%
17.7%
29.4%
1.0/15-0.625/12
80%
76%
0.0%
1.75
95%
0.0%
11.3%
17.7%
28.1%
1.2/15-0.75/12
74%
73%
0.0%
1.75
95%
0.0%
9.5%
17.7%
27.8%
0.4/12-0.25/9.6
97%
89%
4.2%
1.75
95%
0.0%
32.4%
17.7%
41.2%
0.6/12-0.375/9.6
95%
86%
1.6%
1.75
95%
0.0%
21.8%
17.7%
33.8%
0.8/12-0.5/9.6
93%
84%
1.6%
1.75
95%
0.0%
16.6%
17.7%
30.9%
1.0/12-0.625/9.6
82%
79%
0.3%
1.75
95%
0.0%
14.1%
17.7%
30.0%
1.2/12-0.75/9.6
77%
77%
0.3%
1.75
95%
0.0%
12.7%
17.7%
29.7%
0.4/10-0.25/8
97%
90%
4.2%
1.75
95%
0.0%
33.5%
17.7%
41.9%
0.6/10-0.375/8
95%
87%
1.6%
1.75
,95%
0.0%
24.1%
17.7%
35.4%
0.8/10-0.5/8
93%
86%
1.6%
1.75
95%
0.0%
19.5%
17.7%
32.9%
1.0/10-0.625/8
86%
83%
0.3%
1.75
95%
0.0%
17.5%
17.7%
32.2%
1.2/10-0.75/8
82%
81%
0.3%
1.75
95%
0.0%
16.8%
17.7%
32.0%
0.4/20-0.25/16
97%
89%
4.0%
2
89%
0.0%
32.0%
17.7%
38.1%
0.6/20-0.375/16
94%
84%
1.2%
2
89%
0.0%
19.7%
17.7%
28.8%
0.8/20-0.5/16
89%
76%
1.2%
2
89%
0.0%
13.2%
17.7%
24.6%
1.0/20-0.625/16
77%
70%
0.0%
2
89%
0.0%
9.4%
17.7%
22.8%
1.2/20-0.75/16
70%
66%
0.0%
2
89%
0.0%
7.2%
17.7%
21.9%
0.4/18-0.25/14.4
97%
89%
4.0%
2
89%
0.0%
32.0%
17.7%
38.1%
0.6/18-0.375/14.4
94%
83%
1.2%
'2
89%
0.0% .
19.9%
17.7%
29.0%
0.8/18-0.5/14.4
89%
76%
1.2%
2
89%
0.0%
13.5%
17.7%
24.9%
1.0/18-0.625/14.4
72%
62%
0.0%
2
89%
0.0%
9.8%
17.7%
23.1%
1.2/18-0.75/14.4
64%
54%
0.0%
2
89%
0.0%
7.7%
17.7%
22.2%
0.4/15-0.25/12
97%
89%
4.0%
2
89%
0.0%
32.2%
17.7%
38.3%
0.6/15-0.375/12
94%
84%
1.2%
2
89%
0.0%
20.5%
17.7%
29.5%
0.8/15-0.5/12
92%
82%
1.2%
2
89%
0.0%
14.6%
17.7%
25.5%
1.0/15-0.625/12
80%
76%
0.0%
2
89%
0.0%
11.3%
17.7%
24.0%
1.2/15-0.75/12
74%
73%
0.0%
2
89%
0.0%
9.5%
17.7%
23.3% |
F-5

-------
1983 apd newer Throttle Body Injected Vehicles
HC/CO OUTPOINTS ONLY
NOX CUTPOINTS ONLY
Actual
Actual
Actual
HC/CO
HC Excess
CO Excess
HC/CO

NOx Excess

Hammond
Hammond
Hammond
CUTPOINT
Emissions
Emissions
only
NOx
Emissions
Nox Only
HC/CO
NOx
Combined
Composite-Bag 2
Identified
Identified
Ec Rate
Cutpoint
Identified
Ec Rate
Failure Rate
Failure Rate
Failure Rate
0.4/12-0.25/9.6
97%
89%
4.2%
2
89%
0.0%
32.4%
17.7%
38.5%
0.6/12-0.375/9.6
95%
86%
1.6%
2
89%
0.0%
21.8%
17.7%
30.5%
0.8/12-0.5/9.6
93%
84%
1.6%
2
89%
0.0%
16.6%
17.7%
27.1%
1.0/12-0.625/9.6
82%
79%
0.3%
2
89%
0.0%
14.1%
17.7%
26.0%
1.2/12-0.75/9.6
11%
77%
0.3%
2
89%
0.0%
12.7%
17.7%
25.4%
0.4/10-0.25/8
97%
90%
4.2%
2
89%
0.0%
33.5%
17.7%
39.3%
0.6/10-0.375/8
95%
87%
1.6%
2
89%
0.0%
24.1%
17.7%
32.2%
0.8/10-0.5/8
93%
86%
1.6%
2
89%
0.0%
19.5%
17.7%
29.3%
1.0/10-0.625/8
86%
83%
0.3%
2
89%
0.0%
17.5%
17.7%
28.5%
1.2/10-0.75/8
82%
81%
0.3%
2
89%
0.0%
16.8%
17.7%
28.2%
0.4/20-0.25/16
91%
89%
4.0%
2.25
84%
0.0%
32.0%
14.5%
36.7%
0.6/20-0.375/16
94%
84%
1.2%
2.25
84%
0.0%
19.7%
14.5%
26.9%
0.8/20-0.5/16
89%
76%
1.2%
2.25
84%
0.0%
13.2%
14.5%
22.4%
1.0/20-0.625/16
11%
70%
0.0%
2.25
84%
0.0%
9.4%
14.5%
20.1%
1.2/20-0.75/16
70%
66%
0.0%
2.25
84%
0.0%
7.2%
14.5%
19.1%
0.4/18-0.25/14.4
97%
89%
4.0%
2.25
84%
0.0%
32.0%
14.5%
36.7%
0.6/18-0.375/14.4
94%
83%
1.2%
2.25
84%
0.0%
19.9%
14.5%
27.1%
0.8/18-0.5/14.4
89%
76%
1.2%
2.25
84%
0.0%
13.5%
14.5%
22.6%
1.0/18-0.625/14.4
72%
62%
0.0%
2.25
84%
0.0%
9.8%
14.5%
20.4%
1.2/18-0.75/14.4
64%
54%
0.0%
2.25
84%
0.0%
7.7%
14.5%
19.5%
0.4/15-0.25/12
97%
89%
4.0%
2.25
.84%
0.0%
32.2%
14.5%
36.9%
0:6/15-0.375/12
94%
84%
.1.2%
2.25
84%
0.0%
20.5%
14.5%
27.6%
0.8/15-0.5/12
92%
82%
1.2%
2.25
84%
0.0%
14.6%
14.5%
23.4%
1.0/15-0.625/12
80%
76%
0.0%
2.25
84%
0.0%
11.3%
14.5%
21.5%
1.2/15-0.75/12
74%
73%
0.0%
2.25
84%
0.0%
9.5%
14.5%
20.7%
0.4/12-0.25/9.6
97%
89%
4.2%
2.25
84%
0.0%
32.4%
14.5%
37.1%
0.6/12-0.375/9.6
95%
86%
1.6%
2.25
84%
0.0%
21.8%
14.5%
28.6%
0.8/12-0.5/9.6
93%
84%
1.6%
2.25
84%
0.0%
16.6%
14.5%
25.1%
1.0/12-0.625/9.6
82%
79%
0.3%
2.25
84%
0.0%
14.1%
14.5%
23.7%
1.2/12-0.75/9.6
77%
77%
0.3%
2.25
84%
0.0%
12.7%
14.5%
23.0%
0.4/10-0.25/8
97%
90%
4.2%
2.25
84%
0.0%
33.5%
14.5%
38.0%
0.6/10-0.375/8
95%
87%
1.6%
2.25
84%
0.0%
24.1%
14.5%
30.4%
0.8/10-0.5/8
93%
86%
1.6%
2.25
84%
0.0%
19.5%
14.5%
27.4%
1.0/10-0.625/8
86%
83%
0.3%
2.25
84%
0.0%
17.5%
14.5%
26.2%
1.2/10-0.75/8
82%
81%
0.3%
2.25
84%
0.0%
16.8%
14.5%
25.9%
F-6

-------
1983 and newer Carbureted Vehicles
HC/CO CUTPOINTS ONLY
NOX CUTPOINTS ONLY
Actual
Actual
Actual
HC/CO
HC Excess
CO Excess
HC/CO

NOx Excess

Hammond
Hammond
Hammond
CUTPOINT
Emissions
Emissions
only
NOx
Emissions
Nox Only
HC/CO
NOx
Combined
Composite-Bag 2
Identified
Identified
Ec Rate
Cutpoint
Identified
Ec Rate
Failure Rate
Failure Rate
Failure Rate
0.4/20-0.25/16
94%
87%
3.4%
1.75
86%
0.0%
51.7%
32.5%
63.1%
0.6/20-0.375/16
91%
85%
3.4%
1.75
86%
0.0%
37.5%
32.5%
54.1%
0.8/20-0.5/16
75%
75%
0.0%
1.75
86%
0.0%
29.7%
32.5%
54.2%
1.0/20-0.625/16
73%
73%
0.0%
1.75
86%
0.0%
26.0%
32.5%
52.0%
1.2/20-0.75/16
73%
73%
0.0%
1.75
86%
0.0%
23.0%
32.5%
51.1%
0.4/18-0.25/14.4
94%
87%
3.4%
1.75
86%
0.0%
51.9%
32.5%
66.1%
0.6/18-0.375/14.4
91%
85%
3.4%
1.75
86%
0.0%
38.0%
32.5%
58.4%
0.8/18-0.5/14.4
72%
69%
0.0%
1.75
86%
0.0%
30.7%
32.5%
54.9%
1.0/18-0.625/14.4
66%
57%
0.0%
1.75
86%
0.0%
27.4%
32.5%
53.1%
1.2/18-0.75/14.4
66%
57%
0.0%
1.75
86%
0.0%
25.0%
32.5%
52.3%
0.4/15-0.25/12
94%
87%
3.4%
1.75
86%
0.0%
52.2%
32.5%
66.3%
0.6/15-0.375/12
91%
85%
3.4%
1.75
86%
0.0%
39.2%
32.5%
59.2%
0.8/15-0.5/12
77%
78%
0.0%
1.75
86%
0.0%
32.7%
32.5%
56.2%
1.0/15-0.625/12
77%
78%
0.0%
1.75
86%
0.0%
29.9%
32.5%
54.8%
1.2/15-0.75/12
77%
78%.
0.0%
1.75
86%
0.0%
28.0%
32.5%
. 54.2%
0.4/12-0.25/9.6
97%
98%
3.4%
1.75
86%
0.0%
53.1%
32.5%
67.0%
0.6/12-0.375/9.6
94%
96%
3.4%
1.75
86%
0.0%
41.1%
32.5%
60.6%
0.8/12-0.5/9.6
82%
92%
0.0%
1.75
86%
0.0%
35.8%
32.5%
58.5%
1.0/12-0.625/9.6
82%
92%
0.0%
1.75
86%
0.0%
34.0%
32.5%
57.9%
1.2/12-0.75/9.6
82%
92%
0.0%
1.75
86%
0.0%
32.7%
32.5%
57.5%
0.4/10-0.25/8
97%
98%
3.4%
1.75
86%
0.0%
54.0%
32.5%
67.7% 
0.6/10-0.375/8
95%
96%
3.4%
1.75
86%
0.0%
43.3%
32.5%
62.0%
0.8/10-0.5/8
83%
92%
0.0%
1.75
'86%
0.0%
38.8%
32.5%
60.1%
1.0/10-0.625/8
83%
92%
0.0%
1.75
86%
0.0%
37.5%
32.5%
59.8%
1.2/10-0.75/8
83%
92%
0.0%
1.75
86%
0.0%
36.7%
32.5%
59.6%
0.4/20-0.25/16
94%
87%
3.4%
2
85%
0.0%
51.7%
32.5%
63.1%
0.6/20-0.375/16
91%
85%
3.4%
2
85%
0.0%
37.5%
32.5%
54.1%
0.8/20-0.5/16
75%
75%
0.0%
2
85%
0.0%
29.7%
32.5%
49.6%
1.0/20-0.625/16
73%
73%
0.0%
2
85%
0.0%
26.0%
32.5%
47.4%
1.2/20-0.75/16
73%
73%
0.0%
2
85%
0.0%
23.0%
32.5%
46.3%
0.4/18-0.25/14.4
94%
87%
3.4%
2
85%
0.0%
51.9%
32.5%
63.3%
0.6/18-0.375/14.4
91%
85%
3.4%
2
85%
0.0%
38.0%
32.5%
54.5%
0.8/18-0.5/14.4
72%
69%
0.0%
2
' 85%
0.0%
30.7%
32.5%
50.4%
1.0/18-0.625/14.4
66%
57%
0.0%
2
85%
0.0%
27.4%
32.5%
48.5%
1.2/18-0.75/14.4
66%
57%
0.0%
2
85%
0.0%
25.0%
32.5%
. 47.6%
0.4/15-0.25/12
94%
87%
3.4%
2
85%
0.0%
52.2%
32.5%
63.6%
0.6/15-0.375/12
91%
85%
3.4%
2
85%
0.0%
39.2%
32.5%
55.4%
0.8/15-0.5/12
77%
78%
0.0%
2
85%
0.0%
32.7%
32.5%
51.9%
1.0/15-0.625/12
77%
78%
0.0%
2
85%
0.0%
29.9%
32.5%
50.3%
1.2/15-0.75/12
77%
78%
0.0%
2
85%
0.0%
28.0%
32.5%
49.6%
F-7

-------
1983 and newer Carbureted Vehicles
HC/CO CUTPOINTS ONLY
NOX CUTPOINTS ONLY
Actual
Actual
Actual
HC/CO
HC Excess
CO Excess
HC/CO

NOx Excess

Hammond
Hammond
Hammond
CUTPOINT
Emissions
Emissions
only
NOx
Emissions
Nox Only
HC/CO
NOx
Combined
Composite-Bag 2
Identified
Identified
Ec Rate
Cutpoint
Identified
Ec Rate
Failure Rate
Failure Rate
Failure Rate
0.4/12-0.25/9.6
97%
98%
3.4%
2
85%
. 0.0%
53.1%
32.5%
64.3%
0.6/12-0.375/9.6
94%
96%
3.4%
2
85%
0.0%
41.1%
32.5%
56.9%
0.8/12-0.5/9.6
82%
92%
0.0%
2
85%
0.0%
35.8%
32.5%
54.3%
1.0/12-0.625/9.6
82%
92%
0.0%
2
85%
0.0%
34.0%
32.5%
53.5%
1.2/12-0.75/9.6
82%
92%
0.0%
2
85%
0.0%
32.7%
32.5%
53.0%
0.4/10-0.25/8
97%
98%
3.4%
2
85%
0.0%
54.0%
32.5%
65.0%
0.6/10-0.375/8
95%
96%
3.4%
2
85%
0.0%
43.3%
32.5%
58.4%
0.8/10-0.5/8
83%
92%
0.0%
2
85%
0.0%
38.8%
32.5%
56.2%
1.0/10-0.625/8
83%
92%
0.0%
2
85%
0.0%
37.5%
32.5%
55.8%
1.2/10-0.75/8
83%
92%
0.0%
2
85%
0.0%
36.7%
32.5%
55.5%
0.4/20-0.25/16
94%
87%
3.4%
2.25
75%
0.0%
51.7%
25.4%
59.5%
0.6/20-0.375/16
91%
85%
3.4%
2.25
75%
0.0%
37.5%
25.4%
49.4%
0.8/20-0.5/16
75%
75%
0.0%
2.25
75%
0.0%
29.7%
25.4%
44.3%
1.0/20-0.625/16
73%
73%
0.0%
2.25
75%
0.0%
26.0%
25.4%
42.0%
1.2/20-0.75/16
73%
73%
0.0%
2.25
75%
0.0%
23.0%
25.4%
40.7%
0.4/18-0.25/14.4
94%
87%
3.4%
2.25
75%
0.0%
51.9%
25.4%
59.7%
0.6/18-0.375/J4.4
91%
85%
3.4%
2.25
75%
0.0%
38.0%
25.4%
49.8%
0.8/18-0.5/14.4
72%
69%
0.0%
2.25
75%
0.0%
30.7%
25.4%
45.1%
1.0/18-0.625/14.4
66%
57%
0.0%
2.25
75%
0.0%
27.4%
25.4%
43.1%
1.2/18-0.75/14.4
66%
57%
0.0%
2.25
75%
0.0%
25.0%
25.4%
42.0%
0.4/15-0.25/12
94%
87%
3.4%
2.25
75%
0.0%
52.2%
25.4%
60.0%
0.6/15-0.375/12
91%
85%
3.4%
2.25
75%
0.0%
39.2%
25.4%
50.8%
0.8/15-0.5/12
77%
78%
0.0%
2.25
75%
0.0%
32.7%
25.4%
46.7%
1.0/15-0.625/12
77%
78%
0.0%
2.25
75%
0.0%
29.9%
25.4%
45.1%
1.2/15-0:75/12
77%
78%
0.0%
2.25
75%
0.0%
28.0%
25.4%
44.2%
0.4/12-0.25/9.6
97%
98%
3.4%
2.25
75%
0.0%
53.1%
25.4%
60.9%
0.6/12-0.375/9.6
94%
96%
3.4%
2.25
75%
0.0%
41.1%
25.4%
52.4%
0.8/12-0.5/9.6
82%
92%
0.0%
2.25
75%
0.0%
35.8%
25.4%
49.3%
1.0/12-0.625/9.6
82%
92%
0.0%
2.25
75%
0.0%
34.0%
25.4%
48.5%
1.2/12-0.75/9.6
82%
92%
0.0%
2.25
75%
0.0%
32.7%
25.4%
48.0%
0.4/10-0.25/8
97%
98%
3.4%
2.25
75%
0.0%
54.0%
25.4%
61.6%
0.6/10-0.375/8
95%
96%
3.4%
2.25
' 75%
0.0%
43.3%
25.4%
54.0%
0.8/10-0.5/8
83%
. 92%
0.0%
2.25
75%
0.0%
38.8%
25.4%
51.5%
1.0/10-0.625/8
83%
92%
0.0%
2.25
75%
0.0%
37.5%
25.4%
51.0%
1.2/10-0.75/8
83%
92%
0.0%
2.25
75%
0.0%
36.7%
25.4%
50.7%
F-8

-------
APPENDIX G
EVAPORATIVE SYSTEM PURGE AND PRESSURE DIAGRAMS

-------
Pressure		Prorgrinre
The evaporative pressure test is used to determine the integrity
a vehicle13 evaporative system, and fuel tank. It is conducted :v
troducing nitrogen pressure into the fuel tank through the zar.fc-tc-
n.ster vapor vent line near the canister. The Nitrogen is introduced
;; the system until the pressure in the fuel tank stabilizes at about
inches of	(0.5 ?SI). ~uei tank pressuri zation is done co-
ntinually module; inc the Nitrogen flow into trie fuel system 'cy
coessive opening and closing cf the control valve by the operator,
dulating the Nitrogen flow into the system allows a higher pressure
trogen flow to be safely used to pressurize the system. Without
dulating the flew, the Nitrogen pressure wouid have to be low, thus
nsiderably iengthir.g the test. If too high a pressure is used, a .ear.
por hose might bulge and rupture.
After the vehicle's evaporative system is pressurioed, it is
lowed to stand for up to two minutes to determine if it oar. oor.tinue
hold pressure. A vehicle is recorded as a failure if :he fuel system
assure drcos to less than 8 inches cf water within the 2 minute time
a me.
P rt3 aura Teat Equipment
Figure 2 shows a schematic of the Pressure test set-up which is
ed. The required equipment includes an air or nitrogen gas bottle, a
andard regulator, and a magnehelic to provide finer control while
essurizing a vehicle's evaporative system. Other pieces of equipment
elude clamps to close off vapor lines and other assorted fittings.
PRESSURE TEST

ENGINE
CVAPOHATTVC
NITROGEN ctunoer
1

-------
Pury Teat	ProfoHhto
The evaporative purge test is conducted during an IM240 transient
dy r.aomete r test to detect vehicles whose evaporative canister puree
system are inoperative. The test procedure includes disconnecting the
test vehicle's vapor purge line running from the canister to the engine,
end installing a gas flowmeter in the line. .Cn most vehicles this is a
relatively simple and quick procedure. However, in scrr.e cases tne
canister and its purge lines are difficult to find in the a.letted t-rr.e,
and as a resuit, the vehicles cannot re tested.
After installing the flow meter in the evaporative purge system,
the vehicle is ope rated over the IM240 transient cycle, and the
cumulative vapor purge flow in units of liters are recorded. The
vehicle is recorded as a failure if its cumulative vapor purge is less
than 1.0 liter.
Piirgo Tpat	Ecru inmsnt
Figure 1 shows a schematic of the Purge test set-up which is used.
The required equipment includes a transient dynamometer (r.oc 'Shown in
Tigure 1) on which to conduct the IM240, and a gas flowmeter which
measures the instantaneous and cumulative vapor purge flow between the
evaporative canister and the engine during the IM240 test. The
iyr.amometer which is used is a standard Clayton ECE-50 twin roll with
125 pound inertia weights and Road Load power control. The flowmeter
which is used is a Sierra Total flow meter series 730 capable of
measuring flows from 0 to 50 liters per minute.
PURGETEST
FILLEF	FILLER NECK
ROLLOVER VALVE
EVAPORATIVE
CANISTER
FUEL TAN*
FLOW METEF
2

-------
APPENDIX H
EVAPORATIVE SYSTEM FAILURES AND REPAIRS

-------
APPENDIX I
MOBILE4.1 PERFORMANCE STANDARD ANALYSES, BY OPTION

-------
1 1 1





I I


Running Loss Emission Levels at 95 F and 9.0 RVP Fuel





Number of Failures by Type

















AVE RL
AVE RL
AVE RL


Purge
Pressure

SOLENOID /
PURGE
VENT

TANK LEAK/
MFG
Purge Failure
Press Failure
After Repair

ALL
Failures
Failures
GAS CAP
ELECTRICAL
UNE
UNE
OTHER
SENDING UNIT

g/mi
g/mi
g/ml

#
#
#
#
#
#
#
#
n














GM
7.65
7.28
1.15

45
21
24
14
9
9
3
3
7
FORD
6.08
6.30
0.91

22
13
9
1
7
5
7
0
2
CHRY
4.81
4.20
0.70

6
5
1
0
1
4
0
0
1
OTHER
3.94
2.37
0.23

7
2
5
4
1
0
0
1
1














ALL
6.74
6.76
1.01

80
41
39
19
18
18
10
4
11




























Notes



























A total of 39 cars were repaired











Vehicle #1578 (Pontiac Sunbird) was not included in the after repair running loss averaqe because ft did not pass the purge test after repair.



Vehicle #1532 (Chevrolet Camero 228) was not included in the after repair runninq loss averaqe because it did not pass the pressure test after repair.


I I I III, II



-------



 -
i
i
. __L_ 		
A* of 08 /
 	
-- - -
	
	 	
Veh
MYR
MAKE
MODEL
EN a. F AM IPuro
Ptm
TE*.
< OTAL RL
RL Redud
Hot Soak
Hum*!
COMMENTS







tf ml
a'rrt



"660
81
ford"
GRAN


PASS
~RECV
0.83

_ . 	


2. SAX
FAIL

Purpe Solenoid Inoperative





1460
85
FORD
CROWN
S OL TBI
FAIL
PASS
RECV
7.15



Electronic Puroe Solenoid Inoperative
1460
85
FORD
CROWN
5.0L TBI
PASS
PASS
RM1
0.25
6.90



~1462~
1462
bT^
88







	



FORD
MUST
2.3LPFI
PASS
FAI_
RECV
9.17
	

Plasbc Vent Line Irom Canister Missing
FORD
MUST
2.3L PFI
PASS
PASS
RM1
2.39
6.78















1525
89
FORD
MUST
5.0L PFI
FAIL
PASS
RECV
11.63



Connector at Can. does not make connectloi
1525
89
FORD
MUST
5.0LPFI
PASS
PASS
RM1
0.82
10.81
















1563
85
FORD
TEMP
FFM2.3V5HCF4
FAIL
PASS
RECV
5.26



Pum Solenoid Inoperative
1563
65
FORD
TEMP
FFM2.3V5HCF4
PASS
PASS
RM1
0.24
5.03
















1574
87
FORD
ESCO
HFM1.8V5FFF1
PASS
FAL
RECV
3.56



Sendna Unll Gasket does not sea)
1574
87
FORD
ESCO
HFM1.9VSFFF1
PASS
PASS
RM1
0.11
3 48
















1575
89
FORD
TAUR
KFM2.5V5HCF1
FAIL
PASS
RECV
10.41



Puroe Vac Lino burned throuah
1575
89
FORD
TAUR
KFM2JV5HCF1
PASS
PASS
RM1
0.08
10.33
















1647
86
FORD
MUST
GFM5.0V5HBF9
PASS
FAL
RECV
e.60



Fuel Tank Vent Ine disconnected at Canlsle
1647
86
FORD
MUST
GFM5.0V5HBF9
PASS
PASS
RM1
0.36
8.24
















1676
64
FORD
CROWN
EFM5.0V5HBF7
FAIL
PASS
RECV
4.43



Puroe Vac Una Msroutod
1676
84
FORD
CROWN
EFMS.0V5HBF7
PASS
PASS
RM1
1.81
2.62
















2026
87
FORD
TAUR
HFM3.0V5FEG6
PASS
FAL
RECV
11.44



Vent Line Disconnected at Tank
2026
87
FORD
TAUR
HFM3.0V5FEG6
PASS
PASS
RM1
6.77
4.67
















2536
89
FORD
TEMP
KFM2.3V5HEF4
PASS
FAL
RECV
5.28



Vapor Leak at Sendnn Unit
2536
89
FORD
TEMP
KFM2.3V5HEF4
PASS
PASS
RM1
Vi4
4.14


























4 .


730
89
MERCURY
TOPA
KFM2.3V5HEF4
PASS
FAL
RECV
3.37



Tank Vent Ine to Canister Disconnected













736
83
MERCURY
LYNX
DFM1.6V2GDK6
PASS
FAL
RECV
1.86



Gas Ceo Leaks













1450
88
MERCURY
TRAC
JFM1.6V5FZK0
FAIL
PASS
RECV
122



Broken Puroe Line Fittings
1450
88
MERCURY
TRAC
JFM1.6V5FZK0
PASS
PASS
RM1
0.17
1.06
















1537
86
MERCURY
COUG
3.6LTBI
PASS
FAL
RECV
6.76



Vaoor Line Disconnected at Rollover Valvs
1537
86
MERCURY
COUG
3.8LTBI
PASS
PASS
RM1
0.05
6.71
















1561
85
MERCURY
TOPA
FFM2.3V5HCF4
FAIL
PASS
RECV
3.88



Puroe Solenoid Inoperative
1561
85
MERCURY
TOPA
FFM2.3V5HCF4
PASS
PASS
RM1
0.10
3.78
















1689
85
MERCURY
MAFO
5.0 L TBI
FAIL
PASS
RECV
3.49

0.68
0.74
TVS I* stuck In the Closed Position
168B
85
MERCURY
MARQ
5.0 L TBI
PASS
PASS
RM1
0.25
3.24
















1802
83
MERCURY
MARQ
5.0LTBI
FAIL
FAL
RECV
6.28



Recv  Tank Vent Ine pLooed. Battery Add
1602
83
MERCURY
MARQ
5.0C TBI
FAIL
FAL
RM1
6.34



destroyed Canister.
1802
83
MERCURY
MARQ
5.0L TBI
PASS
PASS
RM2
0.12
6.16


RM1 - Raoalmd Puroe Urns
2015
2015"
89
89











MERCURY
COUG
KFM3.8V5FAF6
PASS
FAL
RECV
6.64



Tank Vent Line Disconnected
MERCURY
COUG
KFU3.8V5FAF6
PASS
PASS
RM1
0.25
6.39



1548
" 1548

LINCOLN










88
88
MARK
JFM5.0V5HBF3
FAIL
PASS
RECV
12.26



Electrical Connection at Puroe Solenoid Bad
LINCOLN
MARK
JFM5.0V5HBF3
PASS
PASS
RM1
1.40
10.86
















1639
81
LINCOLN
TOWN
5.0CCF
FAIL
PASS
RECV
4.09



Electrical Grcundina Problem near ECM
w
" ~86~~











LINCOLN
TOWN
5.0L PFI
FAIL
PASS
RECV
8.06



Puroe Hose Disconnected
1713
86
LINCOLN
TOWN
5.0LPF1
PASS
PASS
RM1
0.09




1713
86
LINCOLN
TOWN
5.0L PR I PASS
PASS
REP
0.12
7.94




-------



1
A* of OS/ 10/02




VBi
MYR
MAKE
MODEL
ENG.FAM
Pwfl
Pm
TEST
TOTAL RL
RL FUKtud
Hot Soak
Diurnal
COtMENTS








tfml
aim
















1461
87
PONTIAC
GUN AM
2.SL TBI
PASS
FAL
RECV
11.53



Gas Cwi Leaks. Filler Neck lusted ol Cap
1461
87
PONT LAC
QRN AM
2.5L TBI
PASS
PASS
RM1
0.10
11.44
















1530
84
POMTIAC
6000
2.51 TBI
PASS
FAJL
RECV
7.73



Gas Cap Doas not Seal
1530
84
PONTIAC
6000
2.51 TBI
PASS
PASS
RM1
0.11
7.63
















1570
86
POMTIAC
SUNB
G2G1.8VSTDG2
PASS
FAL
RECV
11.70



Gas Cap Leaks
1578
86
PONTIAC
SUNB
G2G1.8V5TDG2
PASS
PASS
RM1
14.45
-2.75
















2011
87
PONT
BONN
H2G3.8VeXffi7
PASS
FAI
RECV
4.04



Sendng Unit Gaskel Leaks
2011
87
POMT
BONN
H2G3.8V6XEB7
PASS
PASS
RM1
1.45
2.59
















2513
84
POMT
sure
E1G2.0V5XAJ5
FAIL
PASS
RECV
339



Puree Una Disconnected
2513
84
POMT
sure
E1G2.0V5XAJ5
PASS
PASS
RM1
0.20
3.39


Reconnected Puree Line













2516
B5
POMT
SUNB
F2G1.8V5TDG1
PASS
FAL
RECV
2.52



Seam ol Gas Tank Leaks, Steel Une SpRI
2516
65
POMT
suw
F2G1.8V6TDG1
PASS
PASS
RM1
o.oe
2.44
















2537
83
POMT,
FIRE
D2G2.5V5TPG6
FAIL
FAL
RECV
4.56



Tank Vent Une worn by Speedometer Cable
2537
83
PONT
FIRE
D2Q2.5V6TPG6
PASS
PASS
RM1
0.06
4.47


Installed new Tank Vend Une













2644
85
PONT
FIRE
F1G5.7V8NEA9
PASS
FAL
RECV
22.28



Leak at Sendlna Urtt
2644
85
PONT
FIRE
F1G5.7V6NEA9
PASS
PASS
RM1
035
21.93
















724
82
BUIC
REGA
C4G3.8V2TMA5
PASS
FAL
RECV
3.51



Lock)no Gas Cap













740
85
BUJC
CENT
F4G3.8V2NSY6
PASS
FAL
RECV
1.00

1

Gas Cap Leaks













1457
as
BUCK
CENT
2.5L TBI
PASS
FAL
RECV
4.65



Crack In Seam ol Gas Tank and Fillet Neck
1457
es
BUICK
CENT
2.5L TBI
PASS
PASS
RM1
0.11
4.54
















1524
85
BUICK
LESA
5.0L
FAIL
PASS
RECV
3.31



TVS Broken
1524
B5
BUICK
LESA
5.0L
PASS
PASS
RM1
0.11
3-20
















1533
86
BUICK
REGAL
2.8L PFI
FAIL
PASS
RECV
0.07



l)i+novm Problems













1542
85
BUICK
PARK
F4G3.BV8XS3
PASS
FAL
RECV
2.86



Gas Cap Does not Seal
1542
85
BUICK
PARK
F4G3.BV8XEB3
PASS
PASS
RM1
0.44
2.42
















1544
86
BUICK
CENT
G2G2.5V5TPG9
FAIL
PASS
RECV
10.43



Vac Une trom Manifold to Pure Valve Broker
1S44
66
BUICK
CENT
G2G2.5V5TPG9
PASS
PASS
RM1
1.05
9.38
















1564
' 66
BUICK
CENT
G2G2.5V5TPG9
FAIL
PASS
RECV
7M



Purge Une Weathered. Collasoas when hot
1564
86
BUICK
CENT
G2G2.5VSTPG9
PASS
PASS
RM1
0.14
7.40
















1672
86
BUICK
LESA
G4G3.8V8XEB4
FAIL
PASS
RECV
7.80



Puree Solenoid Wlflna Disconnected
1672
86
BUICK
LESA
G4G3.8V8Xffi4
PASS
PASS
RM1
0A7
6.93
















2012
86
BUIC
SOME
G2G2.5V5TPG9
PASS
FAL
RECV
7.46



Gas Cap Leaks
2012
86
BUIC
SOME
G2G25V5TPG9
PASS
PASS
RM1
0.10
7.36
















2507
86
BUIC
SKYH
G2G1.8V5TDG2
PASS
FAL
RECV
14.79



Gas Cap Damaqed
2507
66
BUIC
SXYH
G2G1.fi V5TDG2
PASS
PASS
AMI
0.32
14.47

















-------
VEH





1 1








Ao<08/10/92



UYR
MAKE
OLDS
MODEL
DELT
B4G.FAM
H2G3.8VBXEB7
PuB
FAIL
Pm
pass;
JTEST
RECV
TOTAL RL
ff'ml
5.53
RLRmJucI
Hcrt Soak
Dtumal
COMMENTS	
Punje Line Disconnected
1585 ' 87
a/trt


1585 ' 97
OLDS
DELT
H2G3.8V8XS7
PASS
PASS
RM1
2.14
3.39



1











1661
1661
85
-
OLDS
OLDS
cim.
CUTL
F2G2.5V5TPG8
F2G2.5V5TPG8
PASS
PASS
FAL
RECV
B0



Qas Cap Leaks
PASS
RM1
022
8.58
	








1720
86
OLDS
CUTL
2.8LPFI
PASS
FAL
RECV
920

0.76
1.76
Gu Cod has a holed Drilled In H













2013
2013
86
66
OLDS
OLDS
DELT
DB.T
G4G3.8V8XEB4
[PASS
PASS
FAL
RECV
15.91



Sand no Unit Qaikel LaaKs
G4G3.8V8XEB4
PASS
RM1
0.07
15.84










2018
85
OLDS
CUTL
F2G2.5V5TPG8
PASS
FAL
RECV
9B4




2018
65
OLOS
CUTL
F2G2.5V5TPGB
PASS
PASS
RM1
0.16
9.68
1

2506
2506
"85
85
OLDS "
OLDS
CMj*
CALA

PASS
PASS







F2G2.SV5TPG8
F2G2.5V5TPG8
FAL
PASS
RECV
1.04



Leak In Gas Tank Seam
RM!
0.09
0.95
















252B
BO
OLDS
TORO
L2G3.8VBXEB2
PASS
FAL
RECV
12.27

1
Tank Van) Una Dticonnected
2626
SO
<*-DS
TORO
L2Q3.8V8XEB2
PASS
PASS
RM1
0.18
. 	
12.09
1



1

1526
85
CADLLAC
SEVI
4.1LTBI
FAIL
PASS
RECV ; 5.48


Canister Puree Solenoid Inoperative
1552"
1552
"84~
84
CADILLAC
cadlLac
sevT
SEVI
E6G4.1 V5NKA7
E6G4.1 V5NKA7





	
	
Canister Purpa Solenoid Bad 	
FAIL
PASS
PASS
PASS
RECV
RM1
7 11
0.20
" 6.91













1560
86
CADLLAC
B.DO
G6G4.1 V5NKA9
FAIL
PASS
RECV
6.88

i

Puroe Vanor Line on Iod of Can. Disconnect.
1560
2014
86
"83"
CADLLAC
B-DO
G6G4.1V5NKA9
PASS
PASS
RM1
0.16
6.71



CADLLAC







	


ELDO
D6G4.1 V5AGA4
FAIL
PASS
RECV
11.57


Puttie Solenoid Inoperative
2014
83
CADLLAC
BJX)
D6G4.1 V5AGA4
PASS
PASS
RM1
0.16
11.41
















2508
2508
84
84
CADLLAC
SEVI
E6G4.1 V5NKA7
FAIL
PASS
PASS
PASS
RECV
5.89

	

Purge Hose Spll at Canister
CADLLAC
SEVI
E6G4.1 V5NKA7
RM1
0.08
5.81

Replaced Broken Purge Una








2609
83
CADLLAC
COUP
D6G4.1V5AGA4
PASS
FAL
RECV
5.30



Qu Tank Leaklnu Irom Rust (tear In Tankl
2609
83
CADLLAC
COUP
D6G4.1V5AGA4
PASS
PASS
RM1
0.20
5.10


Patched Qas Tank
2624
2624
84
84
cadlLac
CADLLAC
FLEE
FLEE









E6G4.1V5NKA7
FAIL
PASS
PASS
RECV
7.12



Purna TVS Is disconnected
E6G4.1V5NKA7
PASS
RM1
0.14
6.98


Reconnect Pume TVS
1












-------



|







1 '














j
Ae of 08 /10 / 92




VEH
MYR
WAKE
MODEL
ENQ.FAM' Pura
Pre*
TEST
TOTAL RL
RL Rxfcjcl
Hot So**
Diurnal
COMMENTS







am
Cfml







|







659
88
CHEV
BERE
J1G2.8V8XRZ8 FAIL
PASS
RECV
2.10



No Canister Punw




j







678
81
CHEV
MONT
11D2AC FAIL
PASS
RECV
0.95



Puroe Una Disconnected












743
66
CHEV
CAVA
G1G2.0V5XAG2 IPASS
FAI.
RECV
1.14



Filter Neck Leaks












1455
68
CHEVY
CORS
2.0L TBI PASS
FAL
RECV
5.68



Gas Cap Leaks
1455
68
CHEVY
CORS
2.0L TBI PASS
PASS
RM1
3.06
2.81







|







H56
90
CHEVY
LUMI
L1G3.1 VBXGZ5 FAIL
PASS
RECV
11.66



Bad Purge Solenoid
1456
90
CHEVY
LUMI
L1G3.1 V8XGZ5 PASS
PASS
RM1
138
9.98















1459
90
CHEVY
LUMI
3.1LPFI FAIL
PASS
RECV
27.00



Puroo Solenoid Inoperative
1459
90
CHEVY
LUMI
3.1LPFI PASS
PASS
RM1
13.85
13.15















1532
86
CHEVY
Z28
G1G5.0V6NTA8 FAIL
PASS
RECV
14.02



Puree Una (rom Canister Disconnected
1532
86
CHEVY
228
G1G5.0V8NTA8 PASS
FAL
RM1
14.70



Qu Cap Leaklna
1532
66
CHEVY
Z28
G1G5.0V8NTA8 PASS
PASS
RM2
14.22
-0.20


Leak Possibly al Sendlna Unit Gasket












1560
90
CHEVY
CAVA
L1G2.2V5JFG2 PASS
FAI
RECV
5.63



Gas Cap Leaks
1580
90
CHEVY
CAVA
L1G2.2V5JFG2 PASS
PASS
RM1
0.34
5.30















1712
87
CHEVY
CAVA
2.0L TBI PASS
FAI.
RECV
8.48

0.45
0.43
Van) Una Leaks al connactlon w'lh Steel Urt
1712
67
CHEVY
CAVA
2.0LTBI PASS
PASS
RM1
7.78
0.69







|







2023
91
CHEV
CAPR
M1G5.7V5XEA1 FAIL
PASS
RECV
3.07



Puroe Solenoid Broken
2023
91
CHEV
CAPR
M1G5.7V5XEA1 PASS
PASS
RM1
0.11
2.96















2536
68
CHEV
BERE
J1G2.0V5)CAG7 PASS
FAL
RECV
0.05



Aooarenttr Falls Because of Design












2540
90
CHEV
LUMI
L1G2.5VETPG6 FAIL
PASS
RECV
859



Puroe Flow Une Klnted
2540
90
CHEV
LUMI
L1G2.5V5TPG6 |PASS
PASS
RM1
3.08
5.51







i







-------



i
A* of 06/10/ 02




VEH | MYR
MAKE
MODEL
ENG.FAM
Pura
Pree
TEST
TOTAL RL
RL Rftduel
Hoi Soak
DiumaJ
COMMENTS
673" T 85
CHRYSLER





qfiri
g'lri



Lase

FAIL"


3^42

	


FCR2.2V5FAA9
PASS
RECV
Purge Una Moiled













1555
88
CHRYSLER
LEBA
JCR2.5V5FBE6
FAIL
PASS
RECV
6.52



Tm llttlna In ourae Ine broken at Throttle Bo
1555
88
CHRYSLER
LBA
JCR2.5V5FBE6
PASS
PASS
RM1
0.05
547















1458
87
DODGE
LANC
2.2L TBI
PASS
FAJL
RECV
4.20



Vehicle his a Broken Rollover Valve
1458
87
DODGE
LANC
2.2L TBI
PASS
PASS
RM1
0.06
4.12



1587"
~86~

DAYT


PASS






DODGE
GCR2.2V5FAAX
FAIL
RECV
5.86



Purde Una Broken mar Enrtne
1587 I 66
DODGE
DAYT
GCR2.2V5FAAX
PASS
PASS
FM1
o.oe
5.78
















1635,
84
DODGE
DAYT
ECR22V5FAA8
FAIL
PASS
RECV
5.43



Canister Purge Line Is Burned In Hall











1559 1 89
PLYMOUTt-
ACCL
KCR2.5V5FBD6
FAIL
PASS
RECV
3.81



Punne Vatve Stuck













		
_	









		
	..

		
	


At of 06 /10 / 02


VEH
MYR
MAKE
MODEL
ENG.FAM
Pura
Ptm
TEST TOTAL RL
RLRkduct
HotSoek
DIumal
COMMENTS








o>ml
a'rrt



154f
"1541
~83








'


RENAULT
ALLIANCE
DAM1.4V5FFD3
PASS
FAi.
RECV
2.57



Fuel Sending Unrt Gasket
83
RENAULT
ALLIANCE
DAM1.4V5FFD3
PASS
PASS
RM1
0.06
2.51
















1667
85
RENAULT
ALLIANCE
FAM1.4V5FFA2
PASS
fail
RECV
S.10



Gas Cap Leaks
1667
B5
RENAULT
ALLIANCE
FAM1.4V5FFA2
PASS
PASS
RM1
0.07
5.12














1568
86
MAZDA
RX-7
GTK1.3V5HFDB
FAIL
PASS
RECV
4.70


IPuroes thru Ol Filter. Oil cap does not seal













665
B1
TOYO
COROLLA
BTY1.8V2HF3
PASS
FAI.
RECV i 0.37



Gas Cao Leaks













1640
1S40
84
ii-
TOYOTA
CELICA
ETY2.8V5FBB5
FAIL
PASS
PASS
RECV
3.06



TVS It Broken
TOYOTA
CBJCA.
ETY2.8V5FBB5
PASS
RM1
057
2.51













640
722 "
81
AUDI
4000
BAD1.7V6FF04004F
PASS
FAB.
RECV
2.08



Gas Cap Leaks
85".











AUDI
5000
FAD2.2V6FCF1
PASS
FAL
RECV
1.63



Locking Ga> Cap












VO

-------
ullBlll
Evaporative System Purge and Pressure Test
Failure Rate versus Model Year
60
50
40
Failure
Rate 30
IO/
20
10
Overall Failure Rate
7.9
6.0
2.2
16.1
9,426
Purge Only
Pressure Only
Both
Either
Sample Size
76 77 78 79 80 81 82 83 84 85
Model Year
86 87 88 89 90
91
Pressure Only
Purge Only
BOTH
EITHER
1.0 Liter Purge
Test Outpoint
All Model Years
Pressure Test
Conditions:
Init: 14 inches Water
Pinal: 8 inches Water

-------
Reduction in Running Loss Emissions from
Repair of Purge and Pressure Failures
10.0
9.0 --
8.0 --
7.0 --
6.0 --
LA4
Running e A
Losses 50 +
(g/mile)
4.0 4-
3.0 --
2.0 --
1.0 
0.0
^llll
llliilf
00
Bef Repair
Aft Repair
Purge
Pressure

-------
"IZQ wafrunqi
*>&ItJt4.1 CC| not rxJ4i rjs I'.1
?quLr*nwnt** CnlJiicn r*ct-ofa t
:.'H pr~ar*m al*ec>di
.'tirt y 1January 11;
".-s-1901 MYB ennoncv rsrx
"'.tat roa 1 yac cov*to;	'.?68
,4*C rrero*i- V*E vciq!	^"9
J*lv*c s pt-l?8tls
r*t (1301 ca nawn;	--*
*rnrc
-------
Mobile output trora CEM 4.1 1/24/92
;un *2 : 2a sic I . M
'V2BH.E4. 1 ( 4ov91>
:l 20 Wj ming :
MOBILE4.1 d?es ncc mc:
alver race <1961 and newecJ:
.*npilanc Racet
nsp^ction type:
nspecci.cn frequency
ehlele types covered:
1991 & Uc*r MYR tiC type:
Ldest model yeac covtrtd by:
1M240 Transient TesC
Purge System Check
ruei System Pressure Check
:?93
:
1?68
:02a
"2 . *
0.\
100.*
."entrallied
Annual
LCGV
LDCT1
LDCT2 " He
HDCV - No
Tdle
2020
:o?o
2020
HC
Period i P.VP: 11.5
Minimum Terrp:
r-rlcd - Rvp:
Maximum Ten: 2 . (F)
"clod 2 Start Tfr: 1992
' yz HC emission factors include evaporative HC emission iactcrs.
11. .'ear: :000
I *n
Prcaram:

\jro lent
ferro: i *. ?
- -. 
?~.0 in P.ealon: Low



Antl-c am.
Prcaram:
NO
ceracino
Mode: ;0.f 
:". J *
JO.6 Altitude: SoO.
ft.

7*h. Type:
LDGV
1TCT1
:-CT2
J2C7
2CV
~DV
_20T i-DOV

Ml Vh
'."-h. Speeds:
19.6
i? .t>
i?.6

".? . b
'.9.6
.9.6 1?.6
1?. 6

vmt Hi*:
0. se<
0.199
0.000

:.J35
1.C02
0.001 *7.093
1.007

":rroolce Emlss
Ion Faccori
(Cm/Kile)







VCC HCi
:. se
2.25
:. 59
C .35
; .54
"1.65
0.84 :.15
J.96
1.971
x&ausc HC:
0. 52
: .20
i. J7
1.25
:.07
J. 6 5
0.84 2.15
1.99
*3.9 37
viporie HCr
0. J6

0 . 46
0. 38
1.44


2 . 5 2
0.391
P.Cul L HC:
0.19
0.25
0.25
0.25
0.40



0.192
tuning L KC:
0.42
0.36
0.40
0.37
0.51



0.364
ftstlng L HCi
0.11
0.10
0.10
n.io
0.12


0. 54
0. 093
xhaut CO;
.86
11.96
14.40
12.66
36.27
1.60
1.74 11.21
24.79
10.021
Exhauat NOXi
0.73
L .15
1.2 5
1.18
4.51
L. 37
1.S0 9.30
0.77
1. 763
Cal Year: 2000
* I/M
ProQcamt
Mo
p
e
9
*4
I
Tettpt 07.5
/
87.5 /
87.5 m
Regiont Low



Antl-tam.
Program*
NO
Operating
Hodei 2 0.6
/
27.3 /
20.6 Altitude: 500.
Ft.

''eh. Type i
LDGV
LDGT1
LDGT2
IDGT
KDGV

L0DV
LDDT.
HDDV
MC
All Vah
\>h. Speedst
:?. ,
1?. 6
17.6

19.6

i?. 6
19. 4
:?.6
19.6

'>rr Mix:
0. 56 4
0. 199
1.000

0.035

-1. 002
0.001
n. 093
0.007

r.*n*>alte Emission Factors
(Cm/Klll









.'"C HC:
1 . 77
 .25
2.S9
:.35
1.54

:. 65
KS4
: .is
1. 96
:. o
-------
'2M4.1 Fun *2 : Sasic I/M
\:?al Parameter Selected:
:: : * / 9 2 ; 2: u: 5 9
20.6
ID DV
"oton: L;*
Alclcuae: C-Jio. Ft.
'-itnlmum Terr:
:'-a jcIjtbahv ferro:
3ase RVP: 11.5
!n-use HVPr 9.7
ttilcle Type:
'.'oh. JpeaOJ:
"jrbi fit Tert*>:
rperatlna Mcde: 2J.c
Invise scare '{r: 1??2
-rcTi ltgt:
 .5
j
hdcv
.. t	1 ? . 6
:1 Prcoram Selected:
I >1 start year ( January i: :
Tairoeclnq dc*rrenc jcarc:
?ldesc rrrdel year covtrad:
I.rt Inspection frequency:
I.M program cyp:
;?0L t iater MtR test type:
Vehicle type* covtrtd:
:/M valvar race (pre-1901):
I/M waiver rate lpot-1981):
I-H ecrrpllance rate:
Cost Asaufrpclons
"TTT3	
: ?93
:?68-2o:o fixed
Annual
~entrall:ed
Idle
LOCV
0.0*
0.0%
100.0%
Jt rinqency: 7j
:lec Assunptlonj
I.M inspection cost: (pec Inspj 3 9.00
Avq. Gasoline Costfper gallon): $ 1.25
P.epalr Costs	Pre-01 ?
IM240 test erode 1 ynr crraraac	2 0 2 0 
Furqe check trcae 1 year covraa: 2320*
Preseure check xede 1 year covecaaet 2020*
LD<3V LDT1 LPT2 HCKTV LDDV LCDT HDOV
;jt:t~trrr
.'hlcLe pcct\tais *>3 . 2 1}. 7 z. . 1 ". o : . 2 T~
;rcwcn rate: 0.0 % per year
"leec site Is 1000000. vehicles in base year 19P6
'/q. I.M repair cose	: i 50.	"3. ; !!'.240 recall  i ISO. ffO* rapalt" S 0.
3en(lc> <1000 tons/yet and Coses ' 1000 S/yri are ivcciaed over Che calenaar y*r 2000 thru .'COO.
jo refuel
j lender ieac ^000
Fleet Sire 1000000
Baseline No-Proqcun Emissions:
:2.929
TTzvoc
21.955 12.3.217
Program Benefits and Costs
V6C
;. 02*
2.067
i LOs s
J . 597
? ZZ L.93
:. 2je
To
Average Annual ATP Benefit,fee
Cvap/RunLoa* Benefit, fee
vap/RunLo MPC Benefit
Avtj Annual I/M Benefit, Fee
Avo.Annual ATP Repair Cose
Avq Ann. Prq/Prese Repair Cost
Avg Ann. IM Repair Cost^fallRt
Average Annuel IM Fuel Savings
Exhaust TaiTparlni) Deterrence
Cvap
0.005
PxiLoas
0.001
ExhHC
1.175
Ann Avq
<
j.aao
o
o
o
3.
0.006

0.


t 0.
1.033
17.762
5074.


0.


0.


3 5 36.


( 219S.
0.142
1 . 646

' 1.101
15.JOB
? 412.
2936
48237
Mloorama |
0.0483
3

-------
 L : ?,'jn flQ Hlah Option 3lennlli
;cal ["^riirataci Selectea:
Vnblenc	i 7 , 5
Tparaclna Mode: 20.6
''aion: L;w
Altitude! *00. Ft.
?aSe RVP 11.5
-use RVP s J. 7
Minimum Terror
:taxirnmn ferr:
{F)
IF)
.^hLcle Type: LOCV
_wOTl

'.IDV _^DT .5C0V MC

'"h. Speeds; &

->-b
: Pttcr
..^rD ..^.6 . ^. 6
am Selected*.
I M stact year {January 1);
.

itrlnaency: -0

Tairpecina detertence atact:
:?83



Eldest medal year covered:
;7G9-20ZQ
 1 xed


l.H Inspection tre<7uency:
alennial



r.'M prooram typei
.'antral 1 zed



i901 ( later MYR cast typ*:
..500 rpra /
idle
;.4240 cec rn^el year ccveraoe?
: ?84>
'.'chicle type* covered!
-0CV, UXT1
. -d^t:
Furqe checK mcdei year c?veraae:
1 ?94-*-
I'M waiver rate fur lnsp cose (per lnspf :
!.'C<0 tnsp Increwnt ever Pufde:
Avq. Gasoline Cost(per gallonj:
:o
?. jJ
0. e?
rleat Assurroticns
_TGV LDT1 LTT:
rV LTDT HDOV
jnizie c*r:*ncaoj: ? :. -
:>cn race: - - 0 * F  c y  a c
rr
?Leec sue 19 1000000. vehicles in base year 1996
Repair Costs
Avq. I/H repair coat 
-577
"5.  :m2
and
Costa (1000 3/yri
S 2*0. :iQx repair*
Calendar fear 2000
Fleet Site L000000
Baseline No-Proqraa tr\issionst
Avcro Annual ATP Benefit,Fee
vap/Runios* Benefit, Fee
vap/ftunLoss HPG Bane It
Avq Annual I/M Benefit, Fee*
Avq Annual ATP Repair Cost
Avq Ajin. Pcq/Presa Repair Cot
Avq Ann. IM Repair Cost,FaiiRt
Averaae Annual IM Fuel Savings
BXftauet Tarrperlng Deterrence
TotVod 
"2
ixhHC ivap PCul
:i.95S
r-
 1
T 1
10.92S i.024 ;.067
Pcoaram
Genet Its
and Costs


'; t
0.
~ . o 4 0

3. 503

2057.


' 3363.1
1. 92 5
.>0.973
1006.


998.


3577.


"015. 0.0894

i
[ ^79.)
0.210
:.Jis

Svap
1 - 673
P.nLosa
1,710
CxhHC
2.631
? jtLoj
1 .0 38
15443
?4290 kilograms per day f
4

-------
Lew Occion
.ccal Facaractrs Selected:

'jtolanc ro: $ '. j
Operating node: ^O.u
 V 5 -7.5 ' F
z~.)  :o.s
.-gion: Lew
ALtitude: j00.
. ft .
Base PVP: 1L.5
^uia RVP : ? . 7
In-uja scare Vr: 1??2

.'!lnlnura
I'-a x lrnum
Tarr: 7^.
Terr*?: 72.
(F)
:
Tanprlno daCrrenc start;
Oldait nodal year covarad:
I/M Inspection frequency:
E/M program cypa:
L991 4 lacar HYP tasc Cypa?
VehicL* cypa* covirtd:
I/M valvar raca 53
L?S3
1? 8-7020 Fixad
Annual
Centrm i itad
Idla
locv. ldcti, :tct:
1 . 0*
1 . 0%
? .0%
I .'IT 40 test nodal yaar c;vrio:
r'icoa cfiec* rrodel year ccverace:
?riiar* eneefc. moo*l year covraa
:o20*
:o:o*
2020*-
Ant l-Tarnoerlna Prcacira Selected:
"T5T3	r	
Annual
1 ?91-2020 Fixed
L2CV , LCCTl. LPCT2
Tacalyjc
\I"P ec-mllance raca
ATP program cypa:
.'oners
HDCV LTD'/ LCDT HOOV
ATP start year (January 1):
AT? Inspection frequency:
Oldest modal yaar covered:
Vehicle types covered:
Parameters covirtd by ATP:
Test Assuiroclons
Inspeccicn ce*c: (per In.ipi
'.TP ifUMctlcn csit: (per Lnaoj
\vq. 'iasollne Cscnicia rercencaaes
LTCV LDT1
.rrvtn rate:
rzrrr,rr~
J \ cer vear
Avg. LJM rapair cosc
j
53.
~.
Air Puirp rapair cost
%
15.
15.
CacalytC raplacamant cost:
S
ISO.
165.
Mlsfualad catalyse cost
s
175.
L?0.
Zvap. jystam rapair co*e :
$
5.
5.
PCV syacara rapair cost
s
5.
5.
Gas cap rapair cose
s
5.
5.
."'.at si.te Is L-00000. vnicles in fca.se year 1?S6
( IM240 repair - S 150. NO* repair* S 0. i
Benefits tlOOO tons/yd and Cose* (1000 5/yri ara averaqad ovtr tha calanoar yaar* 2000 ttiru 2000.
txhRC tvi
TocVOC
P.eiuel PjiLoa
Calendar Yaar 2000
Pleat Slza 1000000
B**llne No-frogr am Emissions!
Average Annual ATP BaneClt.Fae
vap/Runl>ose Benefit, Fee
t.vap/Runboss KPG Benefit
Avg Annual I/M BanetlC, Fee
Avq Annual ATP Repair Coat
Avg Ann. Pra/Preaa Repair Ccst
Avg Ann. IK Repair Coat.FallRt
Average Annual IM Fual Savings
Exhaust Tan*>erlng DaCarranea
21.955 123.217 10.92.9
Ptc-gram Benefits-and Cotti
Evap
0.032
P.nLoaa
0. 003
ExhHC
:.i54
Ann Avg
T3T
CO
-OJt

1. b94
i}9.
o;03

0 .


i 7.
1.697
Z6.634
*:so.


5^4.


5192-,


[ 3 6 0~9 .
o.:io
2.315

2.229
3A.533
' 995 .
5539
76136
kilograms |
5

-------
IM4.1 P.un *5 :! ^

if)
'"ahicla
_CCV "w3GTl
1TZ
aOCV
jzor -.is
w

.'oft. Spaads :
.. b . .* . *
.. to
- . c . to
Prraraw Slaccd:
..~o
. to
' ^ . 0
.H j C a c z yea rI January Tj:
r%a*>arinq Q*ctcc*nc start*.	- 2 S i
Oldaac rroo 1 yaar cevacad:	-.?C02020 ? 3
I/'H Inspection fraquancy:	Annual
t/H program typa:	"antral 1xd
19fll 6 latar M*R taac typa:	:dla
Vehicle cypaa covtrd:	LTCV, LD3T1.
I/H valvar rata  S
Purga lnapacc cose (par inspi : J	f. 5 3
Pressure map cot {par lnapi :	l.?4
IM240 Insp lneraoant ovar Puroa: 5	3.97
Avq. CajolXna Costlper aallcni: 5	1.C3
Rapalc Costs Pra-fil	9
Avg. I/H ripalr coat !  50.	"" S.
Air Purp rapalr cost $ IS.	IS.
Catalyat replacement co*c: S ISO.	165.
Hlifald catalyat eoac 5 L?S.	190.
Enp, systet* tapair coat t S 5.	5.
PCV systaea repair cote : 5 5.	5.
Cat cap rapalc coat S 5.	5.
Purge repair coac S 70.	70.
Prasaura rapalc coat $ 30.	30.
: >93
Annual
o:q
L2CV , LDCT1, LZCTZ
rjcaiYJt	Tual Iniac
r? ccrroiianca racer
*7F prcgrajo type:
\-5.G*	
."antral! tad
LI-DT HCDV
r. pat yac
rleet size 19 1000000. vanicLas in base yaac 1996
IM2 4 0 repair - $ ISO. UOx repair* S
Benefits (1000 tonsfyr) and Coita (1000 5/yr) act averaged ovar Cha calendar yaara 2090 thru 2000.
	fotV6d
Calendar tfaar 2000
Flaat Slta 1000000
Baaeiina Ko-protjrana Lt&lafllonsj
ExhHC
E vaf>
Cua 1 PnLoaa
Rj C Loi"
TO
21. 955 i:j. 211 1U.929
Prc^r&a BanaClta and Coaca

YOC
;o
wOJt r
Avaraoa Annual ATP Sanafit.Faa
>y.i 67
1 . b&4
i!5.
E%rap/RunLcaa BanaClt, f
:.:62

1765..
Evap/RunLoaa MPG 0anait



Avq Annual l/H BanaClt, f
: . 697
26.634
7 290.
Annual AT? Rapalz Cct


59-4.
Av<7 Ann. Pr3/?rata Papal* Coac


1743.
Ann. IH Rapair Ccac.fallRt


5192,
A/aca-qa Anrtual LH Foal Savinqa


i.
txhauat Tan^ring Oatarranca
i.no
2. J15

Ewap RnLoaa EzhHC Ann Avq
iUK
JO.fJ3
11*35-
1.383 0.676 2. IS* (
11075
"613
klloqrama par
6

-------
Mobile output from CEM 4.1 1/24/92
"in *10 Hiah Occicn iBlennialJ
:^BXLE4.1< 4Nov9U
:iL20 Warning:
M0BILE4.1 d:s net moaei rrcjt 1?93 ana later Clean Air Act
requirements; Emission Factors icr 1 ??3 cr later are affected.
I'M proa rain selected:
itart year 
\ . >
'antral! zee
Slennlal
LTGV - Yes
LDCT1 - :>s
LDGT2 - s
HDGV - No
1300 rpm / Idle
M0BILS4.1 ZM240 Tranaient Biennial I/M Credits .5/15 (?/5/911
Mdest model year covered by:
IK240 Transient Test
Purge System Check
Fuel System Pressure Cheex
1394
1384
1271
.-.ntl-ta/rperino program selected:
Jtarc year (January 1>:
first model year covered:
Last model year covered:
Vehicle types covered!
?9 3
?75
o:o
LDCV

Type:
Frequency:
rocTpllanc* Ratei
Centra 11 ted
Biennial
99.0*
Air punp system disablements i	Yes
Catalyst removals:	Yes
Fuel inlet restrlctct disablements* Yes
Tailpipe lead deposit test:	Yes
CGR disablement]	No
Evaporative syrtera disablements:	Yet
PCV system disablementsi	t9
Hissing gas capsi	Yea
Period 1 RVpt 11. 5
Minimum Tenp: 12.
Period 2 RVPt B.7
Maximum Tempi 92. (F)
Period 2 Stare Yr: 1992-
VCC HC emission factors Include evaporative HC emission factors.
7*1. Years 2000
I/M
Proarant
Yes
Ajrblent.
Tenv r 97

Antl-tam.
Program!
Yes
Operating
Modei 20
"'eh. Typet
LDCV
LDGT1
LSG72
LDGT
H0GV
Veh. Speedsi
19.6
19.6
19.6

1^5
VHT HI x 1
0. 584
0. 199
0.080

0.03:
Conposite Emission Factors
(Gra/Hlle)



VOC HCs
1.16
1.47
1 .67 "
1.53
4.54
Exhaust HC:
0.46
0.00
0.89
9.62
:. 07
Cvapocat HC:
0.16
0.15
0.23
0.18
1.44
Refuel L HC:
0.19
0.25
0.25
0.25
0. 40
Run log L HC:
0.23
0.10
0.20
0.19
J.51
Rstlng L HC:
O.ll
0.10
0.10
0.10
0.12
Exhaust CO:
6.32
7.06
9.20
7.36
36.27
Exhaust NOX:
0.72
1.05
1.13
1.07
4. 51
6T.S
:7.3
51.5	JD Region: Low
20.6	Altitude c 500. Ft.
19.
0.65
0.65
1. 60
1.37
19.?
0.001
0.84
0.84
1.74
1.50
1975
0 . 09 J
:. is
2 .15
11.21
? . 30
1 ?. 6
0 . 007
4 .96
1. 495
1.99
0.796
2.S2
0.206

0.192

0.203
0.54
0.096
4.79
0 . 223
0. 77
i .750
Cal. Year: 2000
I/M
Program?
No
A/rtblent
T to : 9 7.5 .
' a".S /
?7.5 (F) Pealcn: L:w



Anti-Cam.
Program:
No
Operating
Mode: 20.6 
' 27.3 '
20.6 Altitude: l-OO.
Ft .

'/eh. Type:
LDCV
LDCT1
LDCT2
LDCT
HPCV
LTDV
LOOT HDOV
:r
ML V.h
Veh. Speeds:
I?,6
19.6
19.6

L ?. 6
1 v. 6
L ? . 6 ' L ?  6


VKT Mix:
0.584
0.199
0.090.

. 0 35
S.H02
,l.00l
, ' 1 (i T

Cottposlte Emission Factors
(Cm/ Ml le 1







VOC KC:
1.77
2.25
:.S9
: .35
;.5*
). 65
0.54 -.15
4 . -'6
. .094
Exhaust HC:
0.71
1.20
1. 37
1 . c J
J . 'J 7
>.6 5
IJ  3 4 : . 1 5
y : o
 . 'J49
Evaporat H'Z:
0.36
0.35
0.46
J?
1.44


2. - 2
. .'92
Refuel L HC:
0. 19
0.25
0.25
0.25
0.40



J. 192
Punlng L HC:
n. 42
0.36
0. 40
''.37
\ 51



<). J64
Rstlnq L HC:
0.11
0. 10
0.10
0.10
7. 12


0.24
0.099
Exhaust CO:
10.OS
U .56
14.40
12.66
K .27
I .60
1.74 11.21
24.78
11.374
Exhaust HOX:
0.75
1.15
1.25
1.19
J.51
1.37
1.50 ?. 30
0.77
I . 799
7

-------
Mobile output from CEM 4.1 1/24/92
.-un 40 Medium Optica
:^BIL4.1(4NOV91I
M12G Warning:
MOBILE*.1 <1:cj net rodel Jtoat 1993 and later Clean Air Act
requirements.* Emission factors tec C( X ?93 or Later ic affected.
X proacirn selected?
j^art year (January 1>:
:7--1981 MYR jcrinancy rata:
"lrse moaei year covereoj
Last model year covtrco:
Waiver race (pce-1991):
'Waiver race (191 and neer>:
"rffpllance Race:
Inspection type:
l.nsoectlcn frequency
Vehicle types covered:
91. 6 Liter HYR test type:
1993
L?e
:o:o
;. %
 . *
1. *
:entrilli#d
Annual
LDCV
LDGT1
LDCT2
-	:6J
-	:> 9
-	:>
HDGV - No
idle
Eldest model year covered by:
IM2 40 Transient Test
rurge System Check
fuel System Preiur< Chec*
Ant 1-cajToerino program selected:
:o:o
:o:o
1971
.Start year Uanuary II:
first model year covered:
Last mroel. year covttco:
Vehlele type* covered:
L ??3
1991
:o:o
Lr>cv
 VP* *
' reaviency:
~:rrplijQce Rate:
" 9.0%
Ale pwrp system dls1 emnc* :
"ataLyst cnotlst
f'iei inlet restrictcr <31 a ablctrant* :
Tailpipe lead deposit test:
EGR disablement!
Evaporative system dlsaciemencj:
PCV system disablements:
Missing gas cap*i
No
Ye i
Yes
No
No
No
!0
No
Period 1 RVp: 11.5
Ml ft lamia Terrpi
Period 2 RVPi
t:. tri
0.7
MaxlAtm Tenpt 92. (FJ
Period 2 Scare Yr: 1992
VOC HC emission factors Include evaporative HC emission factors.
Cal. Yeari 2000
I/H Programs ^es
Anti-Cam. Program: Yes
AefeleQC Teffpi 87.5 / 87,5
Operating Modet 20.6* / 27.3
87.5
20.6
IF) Region: Lev
Altitudes 500. Ft.
ah. Typei
Veh. Speeds 
VHT Klxt
IS.?
0.504
15.S
0.199
19.1	
0 .080
19.6
0.035
13.6
n .002
19.6
0 .001
19 TV
19.6
0.007
"rrrposlte Emission Factors . J?
1. 44


J . r: 2
"1. J Q. 2
Refuel L HC:
0.19
0.25
o.:s
rj.25
0. 40



^. 1 ?2
P.unlna L HC:
0.42
. 36
J. 40
}. J7
0.51



). Jf 4
Pstina l HC:
0.11
0. LO
). 10
'.'.10
12


".SI
\i. 0 i o
Exhaust CO?
10.05
11 . ?6
N. 40
i:. 66
>?.27 1.00
1 ."4
n.zi
i . " ?
Ji.?74
Exhaust NOX:
0.75
. 15
L.25
. L 
4.-31 1.37
L .'j0
 JO
- 7
. 7 ? 9
8

-------
Mobile output from CEM 4.1
1/24/92
?un *3 : l?w Ootlon
MOBlI.4.li4Nov9i>
M120 Warning: M0BXLE4.1
doaa not ircdal
caqulramanta: ml4*16n
factdfd tec CY
I M prcxjrxra jalactad:

Start yeac (January I):
1983
?r-19 81' HtR atrtnaaftCy rata:
:o%
first modal yaar cav#rd 1 yar covrd:
1901
Laat modal yaar covrtd:
:o:o
'.>hlcla typaa covaradi
U5CV , LTCT1
VP**
.'antra 11 tad
"requancy:
Annual
rrrrpllanca Rata:
?e.o%
Alt pusp ayatam dlsablamancax
MO
Tatalyat ramovli
:*
fual lnlar raatrlctor dlaa&lamancaj im
Tallplpa laad dapoait taats
No
ECR dlsablaantt
Ho
Evaporatl?* ayataa dlaablaraaotat
No
PCV gystmm dlaablapvat* >
No
Kljalng gaa capat
No
Miftlnaua Tanp i
Parlod 2 RVPi
7:.
9.7
Partod 1 RVPt 11.5
VQC HC aniaalon tactoca include avaporatlva HC amAaaton Cactoca.
Kaxlaaia Taspi 92.
Paelod 2 Start tc: 1992
Cal. Yoan 2000
I/H
Program
Yaa
Aflfeianc
TtflOi 87.5 /
87.5 /
87.5 (F)
Ragloni U



Ancl-tart.
Pcogrami
*as
Cparating
Mod* t 20.6 /
27.3 /
20.6 Altltudat 500.
re.

Vah. Typat
LDCV
LDCT1
L0CT2
LDCT
HDCV
ID0V
LOOT
KDDV
MC
All Vah
Vah. Spaadat
19.5
19.6
19.6

19.6
19. i
19.6
19.6
19.6

VffT Ml X X
0. 584
0.199
0.080

0.035
0.002
0.001
0.093
0.0 07

Corrpoaita Cmiaalon Factor*
(Cm/Hlla)








VOC HCI
l.SB
I .90
:.ie
1 .99
4.54
0 . 65
0. 94
: .15
4 - 96
1.070
Extiauat HCi
0.52
0.86
0. 97
0.89
?:o7
0,65
0 . 04
2 .15
1. 09
0.038
Evaporat HC:
0. 36
0.34
0.45
0. 31
1.44



2.52
0.379
Rafual t HC:
0.19
0.25
0.25
0.23
0. 40




0.192
PunlDQ L HC:
0.41
0.36
0. 40
0.37
0.51




0.363
Pjtlnq L HC:
0. 11
0. 10
0.10
0.10
0.12



0.54
0.098
Sxhauat CO:
7.01
8.07
9. 48
8.47
36.27
1.60
1.74
11.21
24.78
8.927
Exhaaat NOXi
0.72
1.05
1.13
1.07
4.51
1.37
1.50
9.30
'0.77
1.752
Cal. Yaaci 2000
r/M
Proqraai
No
Antolaftc
Tairpt 87. 5 / 87. 5 /
87.5 (F) Ragloni Low



Antl'taa.
Proqramt
NO
Operating
Modal 20.6 / 27.3 /
20.6 AltltudaI 500.
Ft.

Vah. Typa t
LOGV
LDCT1
 LDCT2
LDCT
HDCV LDDV
LDOT HD0V
MC
All Vah
Vah. Spaadat
19.5
19.6
19.6

19.6 1?.6
:?,6 i?. 6
19.6

VHT HlXi
0. 504
0.199
0. 080

0.035 . 0.002
o
o
o
o
>0
0.007

Conooalta Emlaaion Factora
(Cm/HIla)






VOC HC:
1-77'
2.25
2.59
: .35
4.54 5.65
0.94 :.15
1-96
2 .0*4
Exhauat HC:
0.71
1.20
1.37
1.25
2.07 ?.65
?4 2.15
1.99
I .049
Evaporat HCt
0.36
0.35
0. 46
39
1. 44

:. 52
n. 302
Patual L HC:
0.19
0.25
0.25
0.25
0. 40


>. 1?2
Punlnq L HC:
0.42
0. 36
0. 40
0.37
0.51


n. 364
Rating L HC:
0.11
0,10
0. 10
?. 10
o. i:

0,54
',
Exhaust CO:
10.05
11.96
14.40
i: .C
36.27 1.60
:."4 11.21
:i.7q
11 .974
Exhauat NOX:
0.75
1.15
1.25
1.19
4.51 1.37
i.50 -.30
0.77
I . 7?9
9

-------
APPENDIX J
IDENTIFYING EXCESS EMITTERS WITH A REMOTE SENSING DEVICE
PRELIMINARY ANALYSIS

-------
911672
Identifying Excess Emitters with a Remote
Sensing Device: A Preliminary Analysis
Edward L. Glover and William B. Clemmens
U.S. Environmental Protection Agency
ABSTRACT
There has been considerable interest in
applying remote measuring methods to sample
in-usc vehicle emissions, and to characterize
fleet emission behavior. A Remote Sensing
Device (RSD) was used to measure on-road.
carbon monoxide (CO) emissions from
approximately 350 in-use vehicles that had
undergone transient mass emission testing at a
centralized I/M lane. On-road hydrocarbon
(HC) emissions were also measured by the RSD
on about 50 of these vehicles. Analysis of the
data indicates that the RSD identified a
comparable number of the high CO emitters as
the two speed I/M test only when an RSD
cutpoint much more stringent than current
practice was used. Both RSD and I/M had
significant errors of omission in identifying
High CO Emitters based on the mass emission
test. The test data were also used to study the
ability of the RSD to characterize fleet CO
emissions.
INTRODUCTION
Researchers at the University of Denver
and elsewhere [1]*, [2], [3J, [4] are in the
process of developing systems to remotely
measure the concentration emissions from
vehicles operating on the public roads. Thpse
* Numbers in brackets denote references listed at the
end of the paper.
systems work by focusing a beam, or in some
cases multiple beams, of infrared light across
the roadway into an infrared detector. The
instrument determines the concentration of
the pollutant in the path of the beam based on
the amount of infrared light absorbed by the
detector at specified wavelengths, and the
theoretical relationship of carbon monoxide to
carbon dioxide in auto exhaust. Early
equipment had only carbon monoxide (CO)
capability, while later equipment may have
the potential to measure other emissions
species (primarily hydrocarbons - HC).
Two potential uses of remote sensing
devices (RSD) have been proposed by various
sources. They are: (1) the on-road
identification of gross polluting vehicles
(commonly called High Emitters) for
subsequent repair, and (2) the monitoring of
on-road vehicle emissions in a specified area
over a period of time for program evaluation
purposes. Such potential uses can only come to
fruition, however, if the remote sensing
techniques are shown to provide accurate and
reliable results that reflect the vehicle's true
emission levels over a variety of normal
operating conditions (e.g.. acceleration, cruise,
hot/cold operation, etc.).
Several test programs with remote sensing
devices have been conducted by a variety of
organizations including; the University of
Denver (by Professor Donald Stedman),
General Motors [4], the EPA Environmental

-------
2
911672
Monitoring and Support Laboratory (EMSL) in
Las Vegas [2], and the California Air Resources
Board (CARB) [5]. Many of these studies have
involved measuring the RSD concentration
emissions from thousands of vehicles [1], [3].
Others were concerned with verifying the
accuracy of tfre RSD measurements by
directing small samples of exhaust with a
known concentration into the RSD beam and
comparing the results to more traditional
analyzers [2]. Another involved a roadside
pull-over program conducted by CARB which
compared RSD results to I/M tailpipe and
tampering checks on in-use vehicles [5].
However, none of these programs provided a
quantitative comparison between RSD results
and corresponding transient dynamometer
mass emission results on. the same set of cars.
The goal of this study was to fill this gap
and to provide a comparison between RSD
results and dynamometer results on the same
cars. The driving schedule which was utilized
for the comparison was the IM240. It is a new
transient driving schedule developed by the
EPA [8] that consists of the first two "hills" of
the new car certification procedure, the
Federal Test Procedure (FTP), with some
modifications (further details are discussed
later in the paper). Although linear
correlations between the RSD and IM240
emissions were performed, most of the
investigation was focused on whether the RSD
Figure 1
REMOTE SENSING DEVICE SO
emissions from a given vehicle would
categorically match IM240 mass emissions
from the same vehicle. For example, it was of
interest to know if a vehicle with a high RSD
concentration would also have high IM240
mass emissions. and vice versa. The
effectiveness of the RSD system in identifying
gross emitters and excess emissions was also
evaluated relative to the capability of the
standard Indiana I/M test using the IM240 as
the yardstick for excess emissions. Finally, the
effect of external variables, such as vehicle
operating mode, owner response, weather
conditions. and siting factors were also
investigated as to whether they affect the
ability of the RSD system to correctly identify
vehicles with high emissions.
All RSD and IM240 testing was conducted in
Hammond. Indiana, by Automotive Testing
Laboratories Inc. (ATL) under contract to the
U.S. Environmental Protection Agency.
TEST EQUIPMENT
Figure 1 shows a schematic of the RSD
system in a roadside setting! The principal
parts of the system include the IR detector and
source; a video camera to record the license
plates of passing vehicles; a modified police
radar gun: a personal computer equipped with
an A/D board; and special software developed
by the University of Denver researchers. Also
included in the system are
the special calibration
equipment and several gas
EMAT1C	bottles
IR SOURCE
VIDEO
CAMERA
ETECTOR
RADAR
GUN
COMPUTER
The RSD system operates
by continuously monitoring
the intensity of the IR
source. When a vehicle
breaks the beam path, the
reference voltage drops to
zero which signals the
presence of the vehicle.
Span voltages collected
before the beam is blocked
and zero voltages during the
blockage are recorded. As
the vehicle exits the beam,
samples are taken over one
second at 125 Hertz. The C02
spectral region is isolated by
filtering at 4.3 |im and the CO
spectral region is filtered at

-------
911672
3
4.6 M-m. The instantaneous CO and C02 values
are then regressed using a linear least squares
procedure. From the slope of this regression
the average C0/C02 molar ratio (Q) and the
average HC/C02 molar ratio (Q') are obtained
and reported. The HC ratio is calibrated and
reported in terms of propane.
The average Q and Q' molar ratios over the
one second sample time are the only emission
measurements that are recorded during an RSD
test. Only these are recorded because the size,
position and distribution of the exhaust plume
is not known. Therefore, to help safeguard the
quality of the subsequent emission
calculations which are based on these
theoretical relationships, the system employs a
built in feature to evaluate the reasonableness
of the observed molar ratio. If an
unreasonable molar ratio is observed, the RSD
reports a 'non-linearity error'.
The majority of testing was conducted using
ihe original RSD design developed by Dr.
Stedman. This unit (designated in the paper as
RSD #1) had a liquid nitrogen cooled detector,
and measured only carbon monoxide (CO). It
used two indium antimonide photovoltaic
detectors. A second unit was used in the later
stages of testing. This second RSD unit
(designated as RSD #2) was air cooled, and had
both HC and CO measurement capability. The
two systems also had slightly different
versions of software controlling the data
acquisition and processing.
DESCRIPTION OF TESTING
Two different RSD testing formats were"
employed during this study. These formats
included track testing at the ATL (Bendix) test
facility in New Carlisle, Indiana, and on-road
testing of in-use vehicles in Hammond.
Indiana. The RSD track testing was very
limited, lasting only three days at the
beginning of the test program, and employed
the first RSD unit (RSD #1). The purpose of the
track testing was to compare RSD results
collected under controlled track conditions
with FTP results on the same set of vehicles. A
second purpose was to evaluate the
repeatability of the RSD emissions by replicate
testing of the same vehicle under Controlled
conditions. A total of ten (10) cars were tested
at two test-track sites. At the first test-track
site, five cars received multiple RSD tests on a
level roadway at speeds of 5. 10. 20. 30, and 4.0
MPH. At the second track site, five other cars
were . lested several times on an inclined
roadway with a 3 percent grade at the same
speeds. All ten of the vehicles which
participated in this part of the RSD test
program had been recruited to the ATL
laboratory far other emission testing
programs and were simply selected for the RSD
testing based on availability. As a result of
participating in the other programs, these
vehicles underwent FTP testing and repairs.
However, none of the vehicles received an RSD
test after repair.
RSD testing at roadside sites of in-use
vehicles. which had received a transient
dynamometer test, was the major effort of this
study. This RSD testing was conducted at two
expressway ramp sites, and a secondary street
with two-way traffic. The majority of this on-
road testing was conducted at the expressway
sues, and employed the first design RSD unit
(RSD #1). The second design unit (RSD #2) was
used at the second test site, a fairly lightly
traveled, two-lanc expressway service drive
with two-way traffic. Both RSD test sites were
located less than one mile from the specially-
modified centralized I/M facility in Hammond,
Indiana. which administered the transient
IM240 dynamometer test.
The IM240 is a new two bag mass emission
test conducted on a dynamometer over a
transient driving schedule. The driving
schedule for the IM240 test consists of the first
two "hills" of the new car certification test, the
Federal Test Procedure (FTP), with some
modifications to include more transient
operation. However, the driving schedule does
not include a vehicle engine-start. The
dynamometer inertia weights and horsepower
settings were selected based on a consolidated
list of the new car certification settings for the
vehicle model. The emissions were measured
using a CVS-CFV system with laboratory grade
emissions analyzers. For further information
on the IM240 consult EPA technical report
number EPA-AA-TSS-91-1 [8],
The vehicles which participated in the RSD
testing were a subset of vehicles selected for a
larger program designed to monitor the
emission performance of in-use vehicles using
the IM240 dynamometer test. These vehicles
were selected for IM240 testing as they entered

-------
4
911672
the Hammond I/M facil iiy based on the
availability of the dynamometer, and the
owners willingness to participate. If the
vehicle owner declined to participate, that car
went through the normal Indiana I/M
procedure (idle and 2500 RPM no load), and the
next car in line was chosen to receive an
IM240. Only after the vehicle had completed
the IM240 test was the owner approached about
participating in the RSD drive-by test. Drivers
were asked to drive by the RSD site and were
given a monetary incentive if they indicated
that they would participate (cars were not
stopped at the RSD test site itsel 0 - This
additional recruitment procedure for the RSD
testing was needed only for the expressway
sites, in that the secondary street site was
located so that owner participation was blind.
After receiving the IM240 test, the car was
inspected for catalyst tampering and
misfueling. It then received a pressure test of
its evaporative emission control system. This
procedure took only a few minutes. However,
during this period the vehicles were tumed-
off to conduct the evaporative pressure test.
The effect of this vehicle shutdown on
subsequent RSD emissions is unknown.
However, since the shutdown was brief, it is
expected to be minimal.
The two expressway test sites were on the
northeast cloverleaf of the Cline Avenue and
1-94 Interchange in Hammond, Indiana. The
cloverleaf was an entrance ramp leading onto
Westbound 1-94 from Northbound Cline
Avenue. The geometry of the ramp was
circular with a slightly upward grade,
throughout.
On the third day of testing at this site, the
problem of vehicles slowing down was solved
by moving to a second site on the ramp about
25 yards from the ramp exit. The road grade at
this point was also 2.75 degrees. Because the
vehicles were about to enter the freeway, it
was more difficult for the the vehicle owners
to slow down to look at the equipment while
passing through the RSD beam. As a result, the
average vehicle speed was more than 25 MPH
and vehicle operation was believed to be fairly
representative of ordinary expressway ramp
driving.
After receiving the new RSD unit (RSD #2),
it was decided to move to a test site on a
secondary road with two-way traffic, about
one-quarter of mile from the I/M lane. The
site was a straight, flat road with a posted speed
limit of 35 MPH. It was situated such that
somewhat more than half of the cars exiting
the I/M lane would pass through the test
section. Thus, it was possible to conduct the
RSD tests at this site without involving the
vehicle owner. The RSD device was set up so
that the beam crossed both lanes of traffic.
OPERATIONAL EXPERIENCE
The original goal of this test program was
to conduct RSD tests on all the vehicles which
received IM240 tests. However, for various
reasons not all the test vehicles received the
RSD test. As shown in Tables 1 and 2. described
more fully in the following sections, there
were several reasons for vehicles not
receiving the RSD test, including recruitment
problems. RSD equipment problems, and
inclement weather.
Two different test sites on the expressway VEHICLE RECRUITMENT - As noted earlier,
ramp were used. The first site was about 50	the vehicle recruitment procedures differed
yards from the ramp entrance. The road grade	between test sites. At the expressway site it
at this point was about 2.75 degrees. Also,	was necessary to inform the owner of the RSD
because the ground surrounding the roadway	test and offer a monetary incentive for
was relatively level, this site was convenient	participation. In some cases, the owners
for setting up the equipment and parking a	declined to participate, or in some instances,
supporting van. However, it proved to be	agreed to participate (accepted the monetary
unsatisfactory because it allowed curious	incentive), but never showed up for the RSD
vehicles owners to slow down while passing	test. In other cases, they showed up, but were
through the RSD beam. Where it was estimated	missed, because of communication problems
that typical average speeds on that section of	from the IM240 lane to the roadside RSD
the ramp were around 15 to 25 MPH, most of	operator, alerting the operator that a test
the test vehicles were found to be travelling at	vehicle was on its way.
speeds of 5 to i 5 MPH.

-------
911672
5
Site
X-wav
2-lane
Combined
Table 1
Vehicle Recruitment
Owner Declined
1M240 /Vehicle Missed
407
266
673
71
iifl
191
Actual RSD
Participation
336(83%)
U6f55%>
J82 02%)
Most of these RSD refusals to participate
were due to the vehicle owner citing time
constraints or other reasons.. Many of these
refusals occurred at the end of the day when
the I/M lane was backed-up. Unfortunately, it
was noticed that a few of the owners of
vehicles with very high IM240 emissions, or
with signs of tampering (fuel inlet restrictor
disabled) also tended to decline. tn addition,
several vehicle owners declined because they
did not want to drive on an expressway.
As a result of owner refusals and operator
misses, only 336 vehicles out of 407 vehicles
actually drove by the expressway site where
an RSD measurement was attempted (see Table
I).
The second site was partially chosen so as
to avoid the need to inform the vehicle owners
about the. RSD test. Thus, direct recruitment
was not necessary, and vehicle owner
involvement was minimized. This helped
insure that vehicle owners would not change
their driving patterns in the RSD test section,
and avoided the issue of some owners with
high emitting vehicles declining to participate
in the RSD test. The other reason for choosing
a second site was to obtain results ai a location
that would be expected to have different
vehicle operation.
Unfortunately, as shown in Table 1, a high
RSD test rate was not achieved at this site
either. This was due to the fact that the
location of the second site allowed IM240
vehicles to exit the 1/M test lane in two
directions. Only one of these directions took
the vehicle by the RSD site. Therefore, as
indicated in Table 1, only 146 vehicles out of
266 (55%) participated in the RSD testing. The
other 120 vehicles were not involved because
they exited from I/M lane in the opposite
direction from the RSD test site, or they missed
the RSD test because of lack of communication
between the lane and the RSD operator.
RSD EQUIPMENT OPERATION . Operation
of both RSD systems was fairly straightforward
with the set-up and operating procedures
being nearly identical for both systems. The
only differences were due to software
upgrades, and the absence of cryogenic
cooling for the RSD #2 (two-lane testing)
system. Most of the basic set-up and operating
problems were addressed during the first few
days at the first on-road site (the initial
expressway location). These problems
involved the typical learning curve,
experienced when operating unfamiliar
equipment. Beyond that. testing was
reasonably routine, interrupted only by
occasional equipment problems, and inclement
weather.
Severe equipment problems with either
RSD units were infrequent. However, one
major problem did occur with the RSD #1. The
problem was the failure of a computer board
which controlled the video acquisition and
storage, and it resulted in a few days of down
time. Once diagnosed, the board was replaced
with a new one (provided by the supplier of
the RSD equipment). The second equipment
problem which resulted in a couple days
downtime was the theft of the roadside power
generating equipment. This occurred at the
two-lane site while RSD #2 was operating. It
illustrates a potential practical problem with
this type of emission testing. Other more
minor equipment problems which were
resolved by the equipment operator included:
(1)	the heating element used to produce the
collimated RSD beam burned out several times.
(2)	rain entered the detector and had to be
dried out using a heat gun. (3) high winds
potentially causing the instrument . to
erroneously make a reading were present at
times.
The number of vehicles lost to RSD testing
due to equipment problems with each of the
RSD units is shown in Table 2. Overall,
approximately 16 percent of the RSD tests were
missed. However, many of these problems
were related to the emerging nature of this
technology, and as it matures, the number of
RSD tests lost to equipment problems would be
expected to decline drastically.
WEATHER EFFECTS - One of the objectives
of this study was to determine if the RSD could
operate in inclement weather such as rain or

-------
6
911672
Site
X-way
2-lane
Combined
Actual RSD
Participation
336
li&
-182
snow. Track testing was
excluded from this
requirement. However,
the on-road RSD testing
was attempted when
inclement weather was
present during the
expressway testing in
August, November, and
December. Testing in inclement weather led to
some missed RSD tests. At the expressway site.
27 out of the 336 participating vehicles (8
percent) were clearly missed due to rain or
snow (see Table 2). Also, as indicated in the
instrument description, the RSD unit has a
built-in quality control algorithm that flags
illogical results as 'invalid readings'. Some of
the RSD results discussed in the next section,
and labelled 'invalid' may have been tested on
a rainy day. However, it was not possible to
segregate the recorded data in a manner that
would allow an answer to that question.
Therefore, the actual loss due to weather
conditions could have been higher than
indicated.
Testing of the RSD #2 unit at the two-lane
site was done in late March. During this period
no inclement weather testing was done due to
the generally fair weather conditions on all
but two days. During these two days, RSD
testing was not attempted because of
hailstorms, and the vehicles which received
1M240 tests on those days are not included in
Table 2 under weather losses.
In addition to lost test time, inclement
weather also resulted in many of the
equipment problems which were mentioned
above. Nevertheless, it was found that testing
could occur when very light rain was present,
although the number of non-linear C0/C02
errors increased if the pavement was wet and a
large water "rooster tail" was created behind
the car that interfered with the IR beam. In
addition. the intermittent rain became a
nuisance since all the equipment had to be
covered and uncovered.
Table 2
Vehicle Capture Rates
RSD SAMPLE - The true RSD
sample, shown in Table 3. was
computed by removing the
number of vehicles that were not
tested because of recruitment
problems, equipment problems,
or inclement weather. This true
Site
X-way
2-lane
Combined
iquip
Malfunction
28
li
Weather
Losses
27
27
True RSD
Sample
281
;
J03
Percent of w/o Equip
Participation Malfunct.
(84%)
<34%)
(91%)
1100%)
(89%)
sample (403 vehicles overall) contains all the
vehicles which were attempted to be tested by
the RSD under, as best we could determine,
reasonable conditions. The table includes both
ihe valid and 'invalid' readings based on these
measurement attempts. Invalid reading being
those flagged by the unit's internal quality
control algorithm.
Subsequent analysis of the RSD's capability
to identify High Emitters used only the valid
readings because traditionally, only valid I/M
and IM240 tests are used for such analyses. If
necessary, invalid I/M or IM240 tests can be
repeated. However, unlike I/M or IM240 tests
which can be easily repeated, RSD tests might
not be conveniently repeated. A RSD second
chance test would require the vehicle to drive
past the instrument again under as nearly as
identical conditions as possible. Since repeat
RSD testing was not in the program plan, it was
not done. Therefore, in assessing the RSD's
capability to identify High Emitters, an-
argument could be made that the entire True
RSD Sample' (in Table 3) should be used, since
the 'invalid' sample may include some High
Emitters.
Based on consistency with past practices,
only the valid RSD readings in Table 3 were
used for subsequent analyses. However, this
decision should be reviewed in 'any future
programs relative to the type of vehicle
sampling planned, since the effect of using
the 'True RSD Sample' (as opposed to only valid
RSD tests') would have the effect of reducing
the effectiveness of the RSD unit in
identifying High Emitters.
Table 3
RSD Sample
True RSD
Sample
281
122
403
Valid
CO RSD
257
22
356
Invalid
CO RSD
24(9%)
uizm
47(12%)
Valid
HC RSD
12
53
Invalid
HC RSD
69(57%)
69

-------
911672
7
To clarify, an invalid test result as tabulated
in Table 3, could have resulted for two reasons.
First, if the standard deviation of the C0/C02
ratio, which is measured more than one
hundred times per second, was greater than 20
percent, a flag would be set. This prevents
rapidly changing C0/C02 ratios from being
recorded as -valid. Second, if the instrument
could not detect a CO plume, a flag would be set.
This prevents pedestrians, multiple axle
trucks, and other accidental beam blockages
from being recorded as valid. Wet conditions
could cause both these RSD errors to occur,
however, as indicated, this potential source for
'invalid results' could not be separated from
other causes in the recorded data.
The number and percentage of valid and
invalid RSD CO results for each site suggest
different levels of performance between the
two instruments. For example in Table 3. RSD
tt 1 shows a fairly low rate of invalid readings.
Whereas. RSD #2 at the two-lane site shows a
high rate of invalid readings (approximately
25 percent). This higher rate of invalid
readings suggests that RSD #1 is more efficient
and sensitive to low CO concentrations than
RSD #2. possibly due to the cryogenic cooling
of RSD #1. However, the longer RSD beam
length at the two-lane road (greater than 30
feet) versus the expressway site (19 feet) may
also have contributed to the higher rate of
invalid readings.
The rate of RSD HC invalid readings made
by RSD #2 (56 percent in Table 3) is
unacceptably high, and suggests that the HC
capability of the RSD in its current state of
evolution needs much improvement.
Potentially one of the problems with the RSD
HC channel is the very tight instrument
sensitivity required to measure relatively
dilute HC concentrations. According to the
manufacturer of the RSD unit, a difference of
100 ppm HC produces a difference of only 1 mV
by the detector. In addition, the fairly long
RSD beam length at the two-way road site made
the already weak signal even weaker, and the
detection more difficult. Based on the
experience in this program with an admittedly
early design unit, the performance and the
results from the HC RSD channel should be
viewed with caution.
TEST RESULTS
A description of each vehicle tested along
with its test scores from the IM24Q. RSD, and
the Indiana I/M tests can be found in
reference [9],
HIGH EMITTER IDENTIFICATION
The primary goal of this project was to
evaluate the ability of the RSD concept to
consistently identify vehicles wiih excessive
CO emissions as determined by transient
dynamometer testing. In other words, could
the RSD properly categorize, vehicles (i.e.. pass
or fail) based on their IM240 mass emission test
scores. RSD hydrocarbon measurements were
added later in the program, and were similarly,
but less extensively, analyzed.
To achieve this goal, the analysis focused
on several areas. The first addresses a brief
comparison and correlation of the IM240 with
the Federal Test Procedure (FTP) results.
Logically, the next would be to address the
ability of the RSD to categorically identify
vehicles as gross or high emitters versus
normal emitters. However. before the
categorical analysis can take place, it is
necessary to define the ground rules for High
Emitters, and from this, identify the number of
High Emitters in the valid RSD sample. Only
then can the ability of the RSD to properly
categorize vehicles be assessed.
IM 240 VERSUS FTP - In order to better
understand the usefulness of the IM240 as a
vehicle to evaluate the RSD results, a brief
comparison of the IM240 to the Federal Test
Procedure (FTP) is appropriate. All of the
testing' to compare the IM240 to the FTP was
conducted in a separate test program, and a
more extensive evaluation of those results will
be the subject of some future report. However,
for the purposes of this RSD report, a simple
regression of the data available should be
sufficient to make the point.
A regression of the available IM240 and FTP
data from the 300 vehicles in Figure 2 indicates
reasonably good correlation, as demonstrated
by the correlation coefficient (r2 = 0.76) and
the slope (b = 0.99). More importantly, a
review of the specific vehicte data indicates
that the IM240 is excellent at identifying
vehicles with high excess FTP emissions. In

-------
8
911672
Figure 2
FTP Versus IM240
3.043 + 0.990X rt  0.76
O 120
0 20 40 60 80 100 120 140 160 160 200
IM240 CO (g/mile)
fact, for virtually all vehicles tested. where the
model year ranged between 1976 and 1989,
there were zero errors of commission by the
IM240 relative to the new car standards on the
FTP (i.e., all cars that failed the IM240 at the
new car standards for HC. CO. or NOx. also failed
the FTP for the same pollutant). The almost
perfect categorical relationship between the
IM240 and the FTP was a little surprising
vecause the IM240 does not include the cold
art operation found in the FTP. Only
warmed-up operation occurs on the IM240, and
the results do not reflect the contribution of
any engine-start emissions (cold or hot). The
lack of cold operation on the IM240 should,
however, not pose a significant problem, since
all RSD measurements and all I/M test results
in this study were made on warmed up cars.
Therefore, based on the good linear
correlation of the IM240 to the FTP, and its
success at identifying gross CO and HC emitters,
it can be viewed as a representative substitute
by which to judge the RSD results and base the
effectiveness of the RSD systems.
DEFINITION OF A HIGH EMITTER  A
necessary part of the analysis needed to
identify High Emitters was to determine the
appropriate level with which to ascertain that
a vehicle was a High Emitter. Historically, the
specific levels have been related to a vehicle's
FTP emission levels, and EPA's modeling
programs for in-use fleets (MOBILE4) [10] have
used a factor of the "mean plus two times the
standard deviation of the mean fleet emissions
the High Emitter level. To determine High
niiter cutpoints for the IM240, this same
procedure was applied to 135 vehicles which
received before and after repair IM240 tests
in another EPA test program. The resulting
cutpoints are shown in Table 4. As in
previous analyses of FTP results, this testing
showed that many vehicles whose before-
repair emissions exceeded these cutpoints
usually achieved substantial reductions in
emissions from repair. Vehicles whose
emissions were less than the cutpoints in
Table 4. often had no reductions or a very
small percentage reduction in emissions from
repair. In short, our practical experience has
been that vehicles with mass emissions above
such cutpoints can usually be repaired to a
value under the cutpoint quite effectively,
while those below the cutpoint have been
very difficult to repair in a cost effective
manne r.
Our focus in this analysis was to evaluate
the ability of the RSD to find High Emitters
(HE) which could realistically be expected 10
yield substantial emission reductions from
repair. A vehicle is defined as a HE if its IM240
emissions exceed the cutpoints shown in Table
4. The potential emission reductions from the
repair of these vehicles was defined as the
emissions in excess of me Table 4 standards
(i.e. greater than 10 g/mile CO for 1983+ Model
Year vehicles), and such reductions were
termed Repairable Excess Emissions (RPEE).
HIGH EMITTERS IN THE SAMPLE - The
cutpoints in Table 4 were used to identify High
Emitters (HE's) in the sample of valid RSD tests.
For instance, at ihe two sites there were a total
of- 356 valid RSD CO tests conducted. Of that
total, 97 vehicles, or 27 percent, exceeded the
IM240 cutpoints in Table 4. Whereas, only 53
valid HC test were conducted at the two-lane
site. The analyses in the following sections use
the number of valid RSD test in Table 5 as the
base value to compute certain RSD
effectiveness rates.
Table 4
IM240 High Emitter* Cutpoints
Model YearGrp
1976- 1980
1981 - 1982
1983 and Later
m
30 g/mile
20 g/mile
10 g/miie
HC
2.0 g/mile
1.0 g/mile
1.0 j^mile
^ Includes vehicles termed both "HIGH" and"SUPER"
Emitters in MOBILE4

-------
911672
9
Table 5
High Eminers
Valid RSD IM240 CO
CO Sample Highs
Valid RSD IM240 HC
HC Sample Hijths
X-way
2-lane
Comb
257
22
356
72 (28%)
25(25%)
97 (27%)
52
53
fi
8
The model year distribution of the vehicles
in the 'combined sample' of valid CO tests in
Table 5 is shown in Figure 3. This figure
provides a finer breakdown of the overall
distribution and of the distribution of High CO
Emitters (HE). It also shows several interesting
points. First, the 1982 through 1985 model
years comprise almost 50 percent of the
sample, and they also include a large portion
of the HE vehicles. Second, the HEs make up a
discernible portion of most model year
samples, even some late model years. For
example, for the 1978 through 1985 model
years, typically 25 to 35 percent of the model
year sample are HE vehicles. For the 1989
model year, the percentage is between 5 and 10
percent.
To further simplify subsequent analyses,
the three model year groups in Table 4 were
combined into two groups -- a 1976 to 1982
group, and a 1983 and later group. However,
.he outpoints in Table 4 were still used to
define High Emitters within even the
consolidated groups. .
By grouping the 97 HE CO vehicles
(combined sample in Table 5) into these two
model year groups, it can be shown that the
Figure 3
Model Year Distribution
c
o
3
a
Q
20
18
16
14
12
10
8
6
4
2
0
~ Total Sample
| IM240 High Emitters
PI
76 77 78 79 80 81 82 83 84 85 86 87 88 89 90
Model Year
older 1976-82 sample contributes less than half
(45 percent) of the total HEs for CO. This older
vehicle group also only contributes 46 percent
of the vehicles to the overall sample. The
closeness of the fraction of HE's to that for the
overall sample distribution indicates thai the
older vehicles did not contribute a
disproportionate share of the HE vehicles in
this program.
RSD IDENTIFICATION OF HIGH EMITTERS
 The analysis of the RSD's ability to properly
identify High Emitters focused primarily on CO
emissions, since a combined total of 356 valid
RSD CO tests were conducted at the two sites. A
lesser emphasis was placed on HC emissions,
because only 53 valid HC measurements were
obtained at the one site. These analyses in most
instances evaluated the CO sample in the two
distinct model year groups (i.e., 1976  1982
model years, and 1983 and later model years).
However, due to the small size of the HC sample,
all model years were grouped together.
The primary data set used in the analysis
was formed by combining the expressway (X-
way) and two-way street (2-lane) data sets
together. This combination was done in order
to produce a larger, and potentially more
representative database. However, limited
separate analyzes were conducted on the
results from the two principal roadside sites to
illustrate any potential differences, and to
determine if combining the data masked any
important results.
RAW DATA - The first analysis was to scatter
plot the RSD concentration data versus the
1M240 mass emission data. Figures 4(a) and
4(b) show the combined sites CO results for
each model year group. Figure 4(c) shows
the HC results for all model years. In
viewing these scatter plots, there appears to
be more of a relation between the RSD CO
data and the IM240 values for the older 1976
- 1982 model year group than for the 1983
and later group. In any case, the regression
correlation coefficient is . marginal in the
case of CO. and extremely poor in the case of
HC. Scatter plots, of course, provide only an
overall relationship, and Figure 4 does not
provide any information on the ability of
the RSD to properly categorize a vehicle as a
High Emitter on the IM240, or as a Normal
Emitter.

-------
10
911672
Figur 4
RSD  IM240 Scatter Plots
(Combined Sites}
(a)
i o.o
1983 and Latar VthUUi
0 20 40 60 80 100 120 140 160 180 200
IM240 CO
(b)
(l976 1982 VhtcUJ-j	 	r	I
: 
/
Y - 049 * (MM9X ^
;		

r2  0-50 A
0 20 40 60 80 100 120 140 160 180 200
III240 CO (g/mlU)
(C)
3300
3000
2300
2000
I 300
1000
All Modtl	(Two-Lah* Sitt) ^
- : 1 I	t
-t	
-i	I	\	I	i
  I ! !
	i		I	J"'
 i i i >-
21	r	ri	rf
ii:t-v-
Y  544.0  14.4X
rl . 0.001
d n
0" 1 2 3 4 S 6 7 8 9 10 11 12
1M240 KC
Figur* 5
RSD Identification Rate
100 J
i io.o
m i ui i ui t.m m
CO LmI (%)
<}
im \.m ta \m am iod lb i.m too
COLWM
(b)
Ltrt Slttj
RSD HIGH EMITTER IDENTIFICATION RATE - The
abiliiy of the RSD to correctly identify vehicles
with high IM240 mass emissions as High
Emitters is a central issue in the evaluation of
the per/ormance of the RSD concept. The
ability of the RSD to identify such vehicles is
shown in Figure 5 as a function of RSD CO
level. A detailed table of these values can be
found in reference [9]. The identification rate
in Figure 5 is defined as the ratio of the
number of IM240 High Emitting vehicles
identified by the RSD to the total number of
High Emitting vehicles identified by the
IM240. If the RSD were to identify all of the
High Emitters identified by the IM240, the
identification rate would be 100 percent.
The CO High Emitter rates are separated by
site to show any differences that might be
apparent between the results. Differences
were possible because a different RSD
instrument was used at each site, and both sites
had different traffic characteristics, (e.g..
vehicle speeds and accelerations).
Nevertheless, a comparison of the RSD CO High
Emitter identification curves from both sites
(Figures 5a and 5b) illustrates that roughly the
same results at both sites were obtained over a
complete range of RSD CO levels. This is
especially true in the case of the 1983 and later
vehicles, where the difference in
identification rates between sites is typically
only 5 or 10 percent (Figure 5a). The site
effect in Figure 5(b) for the 1976 through 1982
model years shows a slightly larger
difference between sites. However,
because the results from both sites were
	 rather consistent for each model year
group, the RSD data from both sites were
combined for most of the" analyses
throughout the remainder of the paper.
Combining the RSD results by site
produced a larger data set for analysis. A
further combining of the database across
the two model year groups would produce
a still larger data set. Based on the
proportional distribution of High
Emitters by model year in Figure 3, such
combination would seem appropriate,
except possibly for some late model years.
However. Figure 5(c) shows that the RSD,
at all CO levels, systematically identifies a
higher proportion of 1976 though 1982
High Emitters than 1983 and later High
Emitters. Typically, the difference is

-------
911672
11
around 10 or 20 percent. This is evidence that
although the distribution of High Emitters is
similar for most individual model years, the
ability of the RSD to find a High Emitter is no;
the same, and is a function of model year, or
model year group. Thus, combining the two
model year groups for subsequent analysis is
probably not appropriate.
Unlike the CO results. no sampie
stratification was done for the HC results
because the RSD HC sample was relatively small
(only S3 valid readings), and because there
were very few High HC Emitting vehicles in
the 1976 through 1982 model year range. For
example, only one vehicle in the 1976 through
1982 model year group was found to be a High
Emitter.	Instead, the High Emitter
identification rate as a function of HC level was
computed using an 'all' model year sample.
This information is shown in Figure 5(d).
RSD CUTPOINT ANALYSIS - The RSD CO and HC
levels shown in Figure 5 can also be used as a
standard to pass or fail vehicles. The exact CO
or HC values (called cuipoints) which are
chosen, would reflect the desired severity of
the test, and would determine the number of
High Emitting vehicles which would fail. The
selected values are usually chosen based on the
results of a cutpoint analysis designed to
maximize the number of High Emitters found,
and minimize the number of Normal Emitters
(i.e., not High Emitters) which might be falsely
failed. For this cutpoint analysis, the RSD CO
cutpoints range from the very tight 0.05% RSD
CO cutpoint to the loose 7.50% RSD CO cutpoint.
For RSD HC, the cutpoints ranged from 50 ppm
HC to 2500 ppm HC (Hexane equivalent).
The first part of this cutpoint analysis
simply compares the number of IM240 High
Emitting vehicles identified by the RSD to the
total number of vehicleis with an RSD CO level
above a given cutpoint. As seen in Figure 6.
the RSD in all cases identifies more cars than
are truly High Emitters on the IM240.
At cutpoints above 3% CO in Figure 6(a) for
late model cars, there appears to be reasonably
close agreement in identifying High Emitters
between the RSD and the IM240. However, at
the higher cutpoints fewer cars were
identified. As the CO cutpoints are tightened,
the the RSD and IM240 curves begin to diverge
amatically, indicating thai a great
percentage of these1 new vehicles captured by
the lower RSD cutpoints are, in fact, low IM240
emitters which should not be failed. For
example, at the extremely tight 0.10% RSD CO
cutpoint the number of 1983 and later RSD
failures was more than 140 vehicles, while the
number of truly High Emitting vehicles was
only about 35.
Similar comments can be made for the 1976
1982 model year group, except that the
divergence between the RSD identification and
the IM240 occurs at a higher RSD cutpoint, in
this case 5% CO (see Figure 6b).
Figure 6
Failures versus High Emitters
(C)
{ 1983 ~ Vehicle* (Combined Sitea) )
(a)
f AU RSD cars )
RSD cars
IM240

mo -r
*

2?
BO -
S



n
60 -
(b) <5

O
40 -



20 -
z
0 -
0.00 1.00 2.00 3.00 4.00 9.00 6.00 7.00 B.00
CO Cut Point (%)
f 76 -'82 Vehiclee (Combined Sites) ")
	i	!"'
( Alt RSD cars"^)
RSD care with
IM240 > lOg/mi
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00
CO Cut Point (%)
SO
"8
c
2
m
o
2

3
z
: f All Vehicle* (Two-Lane Site) )
7	I	r
(All RSD cars )
RSD cart with
IM240 > lOg/mi
500 1000 1 500 2000 2500 3000
HC Cut Point (%)

-------
12
91 1672
The ability of the RSD to identify HC High
Emitters is extremely poor. The curves in
Figure 6(c) do not have similar shapes, and
nd to diverge even at high HC cutpoints.
urthermore. the number of High HC Emitter
Identified is extremely low at all cutpoints
above 1000 ppm.
vehicles
missed.
that should be identified would be
6, the RSD can falsely
are not High Emitters,
is the opposite issue of
NOT identified by the
As seen in Figure
identify vehicles that
Not shown in Figure 6
High Emitting vehicles
RSD. Such errors, generally referred to as
errors of commission (Ec) or False Failures,
and as errors of omission (Eo) or False Passes,
are inherent in the selection of a cutpoint to
maximize the number of High Emitters found,
while minimizing the number of Normal
Emitters falsely failed.
False Failures were determined by
subtracting the number of IM240 High
Emitting vehicles that were identified by the
RSD from the total number
identified by the RSD. and then
divided by the total number
of vehicles identified by
the RSD. In essence, this
~tlculation is a measure of
e fraction of cars
identified by the RSD that
are NOT High Emitters. A
large value would indicate
thai many of the vehicles 1M
identified are not High
Emitters, and should not
have been identified.
of
the
vehicles
sum was
The effectiveness of the RSD test in terms of
High Emitter Identification (RSD), False
Failures (Ec). and False Passes (Eo) is shown in
Figure 7 for both model year groups. A
detailed table of combined site results for each
model year group can be found in reference
[9].
As indicated in Figure 7(a) relatively few
1983 and later High Emitters were identified by
the RSD system (less than 30 percent) at high
to moderate cutpoints (7.5% RSD CO to 2.0% RSD
CO), while the False Pass rate was extremely
high (greater than 70 percent). However, on
the plus side, the False Failure rate was
generally quite low, with the exception of the
high rate around 6.0% RSD CO. This exception
occurred because the sample of vehicles with
very high RSD CO scores was so small thai one
False Failure dominated the
probably more representative
rate pattern can be seen in
results. A
False Failure
Figure 7(c).
Figure 7
RSD Identification Accuracy
Eo * Fall* High Emitter PASS Rate by RSD
Ec  Flit* High Emitter FAIL Rale by RSD
RSD  RSD Vehicle Identification Rate	
1983+ Model Yean (Combined Sites")
*.o
 *0.0 *
False Passes were
determined subtracting the
number of IM240 High
Emitting vehicles that were
identified by the RSD from
the total number of High
Emitters identified by ihe
IM240, and then the sum
was divided by the total
number of High Emitters
identified by the IM240.
This calculation is a
measure of the fraction of
High Emitters NOT
identified by the RSD. A
large value, in this
istance, would indicate
at many High Emitting
CC M.
20,9
100.0
*0.0
1976  1982 Model Years (Combined Sites)
100.0
0.00 V.OO 2.00 3.00 4.00 S.00 6.00 7.00 8.00
RSD CO Cut Point
()
19*3+ Model Years (X-wmj Site)
90$
t

M.I
20.0
100.0
io.o-
60.0
20.0
0.04 1.00 3.00 3.00 4.00 5.00 6.00 7.00 8.00
RSO CO Cut Point (%)

All Model Yean (Two-Lane Site)
0.00 1.00 2.00 3.00 4.00 S.00 9.00 7.00 8.00
RSO CO Cut Point (%)
(C>
500 1000 ISOO 2000 2300
RSD HC Cut Point (%)
'(d)
3000

-------
91 1672
13
constructed only from the RSD results at the
express way site. It shows that the false failure
rate for High Emitting vehicles is virtually
zero at the higher outpoints.
Despite the exceptions, the overall results
from the 1983 and later vehicles suggest that
the RSD will not identify the majority of the
High Emitting late model vehicles. However,
in the instances when a vehicle is identified as
a High Emitter by the RSD (at the higher
cuipoints), it is usually correctly identified.
For example, at the frequently cited cutpoint of
4.5% RSD CO. less than 15 percent of the 1983
and later High Emitters were identified. This
means that 85 percent of the High Emitters
were missed by the RSD at 4.5% RSD CO causing
a False Pass rate of 85 percent. However, the
False Failure rate was only 10 percent of the
RSD failure rate, suggesting that virtually all
the late model RSD failures at 4.5% will be High
Emitters.
The 1976 through 1982 model years show
results in Figure 7(b) generally similar to the
1983 and later vehicles. However, they
typically had higher identification rates of
High Emitters, lower rates of False. Passes, but
higher rates of False Failures at identical
cutpoints (to the 1983 and later model years).
These somewhat different results are likely
due to the inherently higher emission levels
of the older vehicles (because they were
certified to higher new car standards), and a
longer period of normal deterioration due to
age. They suggest that in order to be fair to all
model year vehicles, RSD cutpoints should be
based on model year or technology type,
instead of one single value such as 4.5% RSD
00.
Like the HC RSD results shown in Figures
5(d) and 7(d). the High Emitter Identification,
and the False Failure and False Pass rates
shown in Figure 7(d) were quite negative.
They indicate that at present, the RSD HC
channel cannot be used to accurately and
repeatedly identify High HC Emitters without
simultaneously falsely identifying many low
emitting vehicles. For example, even at the
extremely tight cutpoint of 25 ppm HC (Hexane
equivalent) the RSD identified about 80
percent of the HC High Emitters. At higher
and more reasonable cutpoints, the HC High
Emitter Identification rate dropped to around
10 percent; Despite this low rate, the False
Failure rate was extremely high.
The results also show that at virtually all
reasonable HC cutpoints the false failure rate
exceeded 70 percent of the total failure rate.
In fact, no cutpoint was identified which could
reduce the false failure rate below 70 percent
and still produce a High Emitter identification
rate of more than 20 percent. In addition,
during the test program one vehicle had IM240
HC emissions exceeding 11 g/mile. but recorded
an RSD HC score of 0 ppm HC. Therefore, based
on these findings in conjunction with the fact
that more than half of the RSD HC
measurements were invalid, strongly suggests
that at the present time, the RSD HC system
cannot make repeatable or accurate
measurements of in-use vehicle hydrocarbon
emissions.
Viewing the cutpoint analysis for CO on an
overall basis would indicate that the
percentage of High Emitters identified versus
the overall number of failures is fairly high at
cutpoints greater than 3.0% RSD CO. This
suggests that typically if a vehicle tests above
3.0% RSD CO it is likely to be a High Emitter.
Overall this is a positive sign, since it suggests
that a moderate RSD CO cutpoint can be found
that will identify at least some High Emitters
without falsely failing a large percentage of
the fleet. On the negative side, however such a
cutpoint would also likely miss a large portion
of the true High Emitters. More High Emitting
vehicles could be identified by using lower
cutpoints. but that would increase the numbers
of False Failures to possibly unacceptable
levels. Such trade-offs could seriously affect
the use of the RSD as a screening tool for High
Emitters. One can only speculate, but it may be
that requiring a high RSD reading to have
been observed on several different days, or at
several different sites, before a vehicle was
classified as a failure would improve the
balance between proper and improper
failures.
REPAIRABLE EXCESS EMISSIONS IDENTIFIED BY
RSD - The previous section on cutpoint
analysis described the effectiveness of the
RSD system in terms of the number of High
Emitting vehicles identified by- the RSD at
particular CO or HC cutpoints. An equally
important consideration is the amount of
excess emissions represented by the vehicles

-------
14
identified by the RSD (i.e.. which High
Emitters are failed). As indicated, this paper
addresses excess emissions as emissions above
the levels defined in Table 4. Defined here as
Repairable Excess Emissions, or RPEE. these
excess emissions represent the potential
emission reductions from those vehicles
identified by the RSD. if they were properly
repaired.
The RPEE was calculated for each cutpoint
as the sum of the excess IM240 emissions
from all High Emitting vehicles identified by
the RSD (i.e., those above the levels in Table
4). divided by the sum of the excess IM240
emissions from all of the High Emitting
vehicles. In short, it is the fraction of excess
emissions from all cars with mass emissions
sbove the levels in Table 4 that were
identified by the RSD.
The effectiveness of the RSD was analyzed
in terms of RPEE because the RPEE
identification rate reflects the actual
emission level of each High Emitter, whereas
the High Emitter identification rate accounts
only for the number of. High Emitters
without regard for emission level. The
difference between the two effectiveness
parameters arises because of the effect that
vehicles with very high mass emissions can
have on the total emissions of all the vehicles
in a given sample. Typically, these vehicles
(called Super Emitters in MOBILE4) comprise
a large share of the excess emissions in . a test
sample (or fleet), although they are usually
few in number. Further, such vehicles can
be readily repaired to moderate emission
levels. For these reasons, their correct
identification is of prime importance in the
control of repairable excess emissions, and in
the evaluation of the RSD system's ability to
identify vehicles with repairable excess
emissions.
The first order of analysis was to evaluate
the RPEE identification rate of the RSD. The
difference in model year groups can be seen
in Figure 8. The results show that at
cutpoints greater than 4.0% CO, the model
year differences are pronounced, with a
greater amount of RPEE identified for 1976
through 1982 vehicles than for 1983 and
later vehicles. The trend of identifying more
->m the old cars than the newer cars was
eviously noted in the rate comparisons of
911672
High Emitter identification shown in Figure
5(c). However, this trend occurred at all CO
cutpoints relative to the number of vehicles
identified. This previous analysis suggested
that the RSD was more effective at finding
1976 through 1982 High Emitters than 1983
and later ones. In contrast to this previous
analysis. Figure 8 shows similar RPEE rates
for both groups at tighter cutpoints.
suggesting that if appropriate cutpoints are
chosen, the RSD can be equally effective at
identifying the excess emissions from new
cars as well as old cars.
To better understand the true relationship
between the number of High Emitters
identified by the RSD, and the Repairable
Excess Emissions from those vehicles, the
data were plotted in Figure 9. A careful
examination shows that for the 1983 and later
vehicles, at cutpoints between 1% to 4% CO.
the RSD identified less than half of the High
Emitters. However, those few vehicles which
were identified contribute significantly to
the total repairable excess CO emissions. At
relatively light cutpoints, the RSD seems to
find the worst of the 1983 and later High
Emitters, although, not all of them, since
even at extremely tight cutpoints, the RPEE
rate was only around 80 percent.
For the 1976 through 1982 vehicles
(Figure 9b), the story is a little different, and
possibly a little more positive. For this model
year group, the RPEE and High Emitter
identification rates appear to proportionally
track each other better. In addition, for
these vehicles, the RPEE rate seems to be a
more linear function of CO cutpoint,
suggesting a stronger relationship between
number of vehicles identified by RSD
emissions and repairable excess emissions.
FIgur* 8
Repairable Excom CO Emissions
(Modtl Year Effect)
o.o
1.00 2.00
9.00 .00 7.00 t.OO
o.oo
4.00
RSO CO Cut Pofnt

-------
911672
15
Figure *
Vehicle vs Excess CO Emissions
RPEE = RSD Identification R-ilc of Repairable Excess Emissions
R5D : RSD Vehicle Identification Rat*
'.^3 md Liter Modal Yun
1!t. 1J Model Yet/v
RSO CO Cut Point
()
 Lao  la uo
RSO CO Cut Point
(61
missed by the RSD. Also, as
mentioned previously, the False
Failure rate (Ec) in identifying
the number of High Emitters was
around 10 percent of the
identification rate, and the False
Pass rate (Eo) was more than 85
percent of all of the High Emitters
in the model year group. This
means that 10 percent of the RSD
failures identified contribute
nothing to the RPEE. However, 85
percent of the High Emitters
which do contribute something to
the RPEE were not identified.
Like the High Emitter identification rate
shown in Figure 7. the RPEE rates also need to
be compared with the False Failure and False
Pass rates. This is done to see if a potential
balance between high RPEE rates and low
False Failure rates exists for RSD. To make
the comparison easier, the False Failure and
False Pass rates from Figures 7 were plotted
along with the RPEE rate as a function of CO
and HC cutpoint. The results of
this effort are shown in Figure
10. stratified by model
group, site, and pollutant
analogous arrangement to
7.
An optimal balance between a
high RPEE and a low False Failure rate is
more difficult to find in the 1976 through
1982 sample. Here, both the RPEE rate and
False Failure rate seem to be linear functions
of RSD CO cutpoint. However, if the goal is to
hold False Failures to less than 20 percent
then a 3.0% RSD CO cutpoint which identifies
60% of the RPEE seems reasonable. If
Failures need to be almost zero, then a
False
6.0%
year
in an
Figure
Figui* 10
Repairable Excess Emissions Identification
[ Eo = Fibc High Emitter PASS Rate by USD
Ee r False High Firrittrr FAIL Rate by RSD
[ RPEE s RSD Identification Rate of Repairable Execs Emissions
For 1983 and later vehicles
(Figure 10a), the RPEE rate
(indicated as RPEE in the figure)
ranged from 10 percent at very
loose cutpoints to 70 percent at
about 2.0% RSD CO. Further
tightening of the RSD cutpoint
produced little additional RPEE,
but dramatically increased the
False Failure rate (Ec). This
increased False Failure rate
suggests that the 2.0% RSD CO
cutpoint may be the lowest
practical RSD cutpoint that
should be considered for the
later model year vehicles
(assuming that the False Failure
rate of 20 to 25 percent is
acceptable at the 2% cutpoint).
In contrast, the widely reported
4.5% RSD CO cutpoint had a RPEE
rate of about 25 percent,
meaning that at this cutpoint, 75
percent of the RPEE would be
D
19S3* Model Yean (Combined Sites)
19-76- 1982 Model Yean (Combined Sites)
0.00 1.00 2.00 1.00 4.00 IM 1.00 7.00 *.00
RSD CO Cut Point (%)
()
19B3+ Model Yean (X-way SlU)
0.00 1.00 2.00 1.00 4.00 UO *.00 7.00 1.00
RSO CO Cut Pofnt (%)
<*>)
All Model Yean (Two-Lane Site)
li.t
0.00 1.00 2.00 1.00 4.00 5.00 S.00 7.00 .0Q
RSD CO Cut Point (%)

-------
16
911672
RSD CO cutpoint is more appropriate. For
comparison, the 4.5% RSD CO cutpoint
lentified about 40 percent of the RPEE, and
urprisingly had a False Failure rate of about
20 percent.
Like the HC High Emitter identification
rates, the HC RPEE rates were extremely poor.
!n fact they seemed to be worse in the sense
ihat the maximum HC RPEE rate was less than
50 percent of the total HC RPEE. However,
ihis was principally due to the 11 g/mile
IM240 HC emitter which had an RSD HC test
result of 0 ppm. Nevertheless, the HC RPEE
identification was very low for most
outpoints, yet the False Failures were nearly
30 percent of the total failures. In all.
coupling these results with all the other low
High Emitter identification rates, and high
False Failure and False Pass rates, leads to the
conclusion that the HC RSD in its current
state will not accurately measure the HC
concentration in a vehicle's tailpipe with
sufficient accuracy to be able to determine if
a vehicle is truly a High Emitter and in need
of repair.
RSD and I/M
As indicated, the primary goal of this study
was to evaluate the ability of the RSD to
properly categorize High Emitters. However,
because the IM240 vehicles had also received
an official Indiana I/M test, the unique
opportunity to compare the RSD results with
I/M scores on the same vehicles was exercised.
The Indiana I/M test is a centralized
biennial contractor run I/M program [11]. A
two-speed idle test (2500 RPM and idle) is used
for 1981 and later vehicles. The 207(b)
cutpoint of 1.2% CO is used for the idle portion
of the test, and a 1% CO value is used for 2500
RPM. The 207(b) HC standard of 220 ppm is
employed at both speeds. Only the idle test is
used for 1976 through 1980 vehicles. The 1980
vehicles are tested against a 2% CO cutpoint.
while the 1976 through 1979 model years use a
3.5% CO cutpoint.
Because the Hammond. Indiana I/M site was
also used to evaluate second chance tests and
preconditioning cycles as part of a previously
completed program, all of the vehicles were
;sted consistent with the recently published
PA guidelines for short test procedures (12).
RAW DATA - Similar to the evaluation of
High Emitter identification by the RSD, the
first analysis here was to scatter plot the data.
Figures 11(a) and 11(b) address the I/M CO test
for the late model cars, while Figures 11(d) and
U(e) provide an HC comparison on these same
vehicles. The idle test for the older cars is in
Figure 11(c). About the best that can be said
about these data is that the relation between
RSD and I/M scores is poor for CO, and
nonexistent for HC. However, it must be
remembered that the relationship of short
tests to the certification test is not based on a
relationship of scores, but a relationship of
pass/fail categories.
PICURB 11
RSD - I/M Scatter Plot
IMf	L>IW M*l V
I O

7
4
3
2
\
O
RPM TmI  CO (%)
L/M
Y
I O
 i ^~5eeccspBcsr
.< - CO
41
8
2
190	Lw	V
500
OOO
O
2.0
0 . 1.0	1,4 1. 1.0
RPM Tt . HC (ppm a lOOO)
4000
3SOO
T 3000
A 29QO
()  aooo
S3 o
06 i ooo
soo
o
All Modri Ytara
I/M Ml* TMI . HC (ppm M tOOO)

-------
911672
1
K
1
/ / ^
RSD VERSUS I/M - This
analysis consisted of
comparing the performance
parameters previously used to
evaluate the ability of th$ RSD
to identify High Emitters and
excess emissions with those
same parameters calculated for
the I/M tests. For simplicity,
RSD results are shown for only
three cutpoints, which include
the frequently referenced
1.5% RSD CO cutpoint. a
moderate 3.5% RSD CO cutpoint,
and a more stringent 2.0%
cutpoint. RSD HC results faired
so poorly in the scatter plots
that no attempt was made to compare them
with the HC I/M results.
The effectiveness of the RSD and I/M tests
in identifying High CO Emitters is shown in
Figure 12 for both the 1983 and later vehicles
and for the 1976 - 1982 vehicles. Examination
of Figure 12(a) for late model cars shows that
the RSD at moderate and high CO cutpoints is
generally no better than the two-speed I/M
test (i.e.. Both' in Figure 12a) in identifying
High Emitters. In the extreme case, the RSD
found less than 15 percent of the High
Emitters at the highly reported 4.5% CO
cutpoint. while the two-speed I/M test
identified almost 25 percent of the High
Emitters. However, when the more stringent
RSD cutpoint of 2% CO was used, about 35
percent of the High Emitters were found.
For the 1976 through 1982 vehicles in
Figure 12(b), the I/M identification of High
Emitters was marginally better than the RSD
identification rate (at 4.5% CO). Relative to
the I/M identification, the RSD identification
improved substantially ai tighter RSD
cutpoints. For example, at the 4.5% RSD CO
cutpoint. the identification rate of 1976
through 1982 High Emitters was only around
25 percent, slightly less than I/M rate,
however, at the 2.0% RSD cutpoint, almost 60
percent of the High Emitters were found.
(Note that the 2500 RPM test results are not
shown for the 1981 and 1982 model years in
the figure for the sake of consistency,
although, they would boost the High Emitter
identification rate of the I/M test somewhat
for the 1976 - 1982 model year group.)
figuheu
High Emitter Identification Rate
(CO. Combined SIlm)
[l97|tWVh>elH 1

Em
&
& & &
%- -j" V
<)

Mathematically, the RSD and I/M tests
were combined to determine if the
combination of tests would identify additional
High Emitters. As seen in Figure 12, there is
an increase in the identification rate,
suggesting that the two tests identify
different cars. This would seem reasonable
since the tests are conducted under
significantly different vehicle operating
conditions, and would be expected to find
problems which are particular to each
operating mode. From the analysis shown in
Figure 13. it is clear that a subset of different
vehicles are identified by each test. For the
1983 and later group, the additional RSD
identification is very marginal until the 2%
RSD cutpoint is applied. However, for the
older model year group, the additional RSD
identification is substantial at all cutpoints.
In fact, adding the I/M and 4.5% CO tests
together . nearly doubles . the identification
Figure 13
High Emitting Vehicle identification Split
(CO onlj)
4.5% RSD
1.5% RSD
10* RSD
1981 fx) Latar MY"
Hdlc/25M I/M Tot)
I/M RSD
tot*h	tyttah
" ($ 
197S-lW2MV*a
Cldk 1/M Tot)
I/M
RSD
13
13 [ 3 ( 10 ]  1 II
TfTTAI-f 				v TOT AH
 (75?i 
15
19) 25

-------
18
911672
rate of High Emitters, further attesting that
different older vehicles fail each of the tests.
It should be pointed out, though, that the
RSD test for the newer cars was compared
against the iwo-speed idle test in Figure 13.
whereas the older cars were compared to
only the single speed idle test. The
differences in I/M test type could possibly
explain the improved identification of
additional older cars by the RSD. In addition,
the Indiana I/M CO cutpoint of 3.5% CO for
the older cars (1976-1979 models) may not be
as stringent as it should be. Tightening this
I/M cutpoint for older cars would increase
the number of High Emitters identified by
1/M such that the identification
improvements exhibited by the RSD in Figure
12(b) could possibly vanish.
Another important factor in any
comparison of I/M with RSD are False Failure
rates. False Failure rates were computed as a
percentage of the failure rate, and they
indicate the percentage of improper failures.
Combined with the High Emitter identification
rate, they are a measure of a test's ability to
find High Emitters, yet avoid failing
low and marginal Emitters.
described RSD CUTPOINT ANALYSIS.' With this
method, a 1983 and later vehicle failing the
RSD or I/M test, would only need an IM240 test
score less than 10 g/mile. rather than 3.4
g/mile. to be considered a False Failure. Also,
with this method, cars with IM240 scores
between 3.4 and 10 g/mile (labeled Marginal
Emitters) would be considered False Failures.
Both methods are shown in Figure 14. the
first in (a) and (b),and the second in (c) and
(d). The False Failure rates calculated in terms
of only the High Emitters (second method) are
substantially higher for both RSD and I/M
relative to the False Failure rates calculated by
method 1. However, the False Failure rate
percentages for the two methods were based
on different sample sizes. Therefore, a direct
mathematical comparison can not be made.
Even so, a qualitative comparison can be
made. In those cases where the the False
Failure rates for method 1 (Figures 14a and b)
are zero, it is probably safe to assume that all
of the False Failures determined by method 2
(Figures 14c and d) were Marginal Emitters
with IM240 scores between 3.4 and 10.0 g/mile.
Rgun 14
Falsa Faliura Rata*
For the comparison of False
Failures between RSD and I/M, two
methods were used, The difference
was in the level of IM240 score below
which a failed car would be defined as
a False Failure. The first method used
new car certification standards
applied to the IM240 results as the
boundaries for False Failures. For
example, if a 1983 and later vehicle
failed either the RSD or the I/M test,
and had an IM240 test score of less
than 3.4 g/mile (1983 new car
standard), it was considered a False
Failure. This method of determining
False Failures is similar to that used in
M0BILE4, except thai MOBILE4 uses FTP
test scores (as opposed to IM240 test
scores).
The second method used to compute
the False Failure rate utilized the
IM240 High Emitter cutpoints shown
in Table 4 as the boundaries for False
Failures. This method is identical to
ihe method used in the previously
J Margin*) na Mtgh fcirtltOT )
z
1
s
J
1
1963 and Lur Modl Ytar*
197* ihnjuh \9t2 MexM Yur
///
^ ^ f
m.

S S S f / /
^ ^
// /
()
(b)
( High EffWIlf Only ]
1913 and	Mo

-------
911672
19
Given (his assumption, it is likely that
a majority of the False Failures
identified by method 2. where the
method 1 False Failure rate is greater
than zero, were .also Marginal
Emitters. As implied previously.
Marginal Emitters are more difficult
to repair, and as a result, less likely to
contribute to overall emission
reductions.
Interestingly, the combination of
the RSD and I/M tests shown in
Figure 14(a) shows a lower False
Failure rate than the RSD test alone.
This apparent anomaly is a result of
the sample size increasing, when
combining the RSD and 1/M samples, faster
than the number of failures. In this
particular case, the 1/M-only False Failure
rate was zero, therefore, the combined rates
were the same as that for the RSD. Thus, the
fraction of False Failures was lower for the
combined sample than for the individual
samples.
Another possible concern is the similarity
in magnitude of the High Emitter
identification rates in Figures 12(a) and 12(b)
to the False Failure rales in Figures 14(c) and
14(d). The similarity might suggest that both
tests tend to falsely identify a vehicle as a
High Emitter almost as often as they correctly
identify a High Emitter. However, it should be
remembered that the calculations for' the High
Emitter identification rate were based on the
total number of vehicles identified as High -
Emitters by the IM240, whereas the False
Failure rate was based on the number of
vehicles failing the RSD or I/M test. Because
the number of IM240 High Emitters and the
number of RSD or I/M failures were
significantly different, any numerical
similarity of identification rates and False
Failure rates is merely coincidental.
A characteristic that must also be evaluated
when comparing RSD to I/M are the False
Passes or missed vehicles. The False Pass rate
is the fraction of High Emitters not identified
by the RSD or I/M test relative to the total
number of High Emitters identified by the
IM240 (also defined near Figure 7). An
alternative calculation, in this case, would be
to subtract the identification rates in Figure
12 from unity.
Fipn IS
Falsa Pass Bates
( Hlgli EmitUrt Only)
1 >Q and Latff Modtl Yort
M I*7* ttroixgfl 13 Afadd Yart
f S S
<>
(O)
False Pass rates are. shown in Figure 15.
and as would be expected, more stringent tests
(i.e.. two-speed idle test), or tighter RSD
cutpoints reduce the number of vehicles
missed. In examining both model year groups
in this figure, it is evident that the Indiana
I/M test missed fewer vehicles than the RSD at
a 4.5% cutpoint. Only at the most stringent
cutpoints did the RSD substantially reduce the
number of High Emitters missed by the I/M
test.
Finally, as in the analysis of the RSD's
ability to identify High Emitters, the
comparison of RSD to i/M needs to look at the
amount of repairable excess emissions (RPEE)
identified by both tests. The RSD values in
Figure 16 are the same as the individual RPEE
values in Figure 10. The I/M RPEE values were
calculated in a similar fashion, and are shown
in Figure 16. Clearly, the Indiana I/M test for
the newer vehicles identified more repairable
excess emissions than the RSD test at a 4.5% CO
cutpoint. However, at the most stringent
cutpoint, the RSD did better.
For the older cars, the RSD consistently
identified more excess emissions than the I/M
test. In this case, though, it should be noted
that the I/M test for these cars consists of only
a single speed idle test, and it is likely that the
identification rate of the I/M test would
increase if a two-speed idle test were used.
In reviewing the comparisons between RSD
and I/M in Figures 12 though 16, it is difficult
to see any substantial improvement over the
I/M test with the RSD. At the 4.5% RSD
cutpoint. the the I/M test for the new cars (i.e..

-------
20
911672
Figui* 1*
Repairable Excess Emissions
fl07* Iftfi VNcl*
V	ailUEJI	L.L.UJJ
Sff /// Jf
^
f f T
V	V
(>

the two-spccd lest) identified more High
Emitters, had lower False Failures, had lower
False Passes, and identified substantially more
excess emissions. At the tightest cutpotnt. the
RSD did identify more cars and excess
emissions, with a correspondingly lower False
Pass rate, than the I/M test, however, it did so
at the expense of an increase in False Failures.
Lowering the cutpoint on the I/M test to
achieve the same False Failure rate would be
expected to produce similar increases in the
identification of High Emitters and excess
emissions with the I/M test.
Clearly, the fictitious
combination of I/M and RSD test
did better with the older cars
than than the newer cars, and
was better than either test alone.
However, even with the
combination, only 30 to 40
percent of the late model High
Emitters and 60 to 70 percent of
the older ones were found. This,
suggests that neither test is as
.  .	effective as it should be in
r v	finding High Emitters. In the
case of I/M, there are several
possible methods to improve the
capture of High Emitters. As previously
indicated, tighter I/M cutpoints could be
used. or. a single speed idle test could be
replaced with a two-speed test - a two-speed
test could be replaced with a loaded test. etc.
The ultimate replacement, of course, would
be the IM240. However, for RSD, the
improvements in the identification of High
Emitters are less obvious, and seem to be
limited to decreasing the cutpoint. with a
near exponential increase in False Failures
(see Figures 6 and 7), or to reducing
measurement variability (discussed next).
For the older vehicle group, the RSD faired
liitle better. At the 4.5% cutpoint. the single
speed I/M lest and the RSD were roughly
comparable in number of cars identified, false
passes, and excess emissions. At this cutpoint,
the RSD did have a measurably lower False
Failure rate. At the lower cutpoints. the RSD
performed better than the single speed idle
test on the older cars with comparable False
Failures.
In reviewing the fictitious combination of
the I/M and RSD tests in these figures, some
interesting observations can be made. For late
model cars, the combination increased
identification somewhat over the I/M test, but
with the expected increase in False Failures.
However, for the older model years, the
combination drastically increased the
identification (number of vehicles, and excess
emissions) over the I/M lest without a
substantial increase in False Failures. In fact,
the combined tests at the highest RSD cutpoint
(4.5% CO) increased the identification over
both the I/M and RSD tests while reducing the
cMse Failure rate relative to the I/M test.
REPEATABILITY ANALYSIS
A . key requirement of any measurement
system is that it provide consistent results to
the same stimuli. In the case of RSD. Stephens
and Cadle [4], and Lawson et.al. [5], have shown
that under relatively normal conditions, the
RSD can measure with reasonable accuracy, CO
frOm a specially controlled vehicle. However,
other elements in the measurement-
consistency equation include ihe vehicle
characteristics, its operation, and weather
conditions.
A previous analysis of RSD data collected in
Chicago [6] suggested that there could be
considerable difference between RSD reading
on the same car on different days (at about the
same time of day). To further explore this
area, several of these other elements that could
cause variability in the RSD results, and the
effect that these factors could have on the
ability of the RSD to accurately identify gross
emitters were analyzed. The track testing
primarily addressed the vehicle and its
operation. The analysis of the on-road data
addresses vehicle operation, weather effects.

-------
911672
21
o
o
Q
en
cc
and a possible combination of the two on
ihe exhaust plume behind a vehicle.
TRACK TESTING - The track testing
was conducted over a three day period at
the Bendix test track in New Carlisle.
Indiana by Automotive Testing
Laboratories (ATL). The track testing was
the first opportunity for the ATL
personnel to operate the RSD without the
assistance of the developer.
The track testing consisted of driving
several vehicles by the RSD multiple
limes at the same speed to obtain replicate
RSD readings. Several speeds were used at
each of two sites - one a level road, and
the other a 3 percent uphill grade. A total
of 10 cars were tested, five at each site.
The speeds used during the RSD tests
included 5, 10, 20, 30. and 40 MPH. The data
were spotty at many of the these test speeds
with numerous low voltage errors (i.e.. no RSD
readings). The lack of readings due to the
automatic quality control feature was assumed
to be partly due to the lack of experience with
this equipment by the operator. As a result,
only at 20 MPH was sufficient data collected to
allow reasonable analysis of (he
reproducibility of the RSD measurements.
Figures 17 and 18 were plotted using the
RSD data collected at 20 MPH and the as-
received FTP from each vehicle. Figure 17
shows the repeatability of the RSD of tests done
on the level road, and Figure 18 shows those
from the 3 percent uphill grade. Both
figures provide a visual indication of the
variance among RSD scores on the same
vehicle at a constant speed. The
accompanying as-received FTP result
provides a convenient qualitative
comparison of the FTP mass emissions
with RSD concentration emissions.
Figure 17
RSD Reproducibility
Bm4. Jfl MPH. ATL T-Tr*ck)
7 5J
2 .oo -
S.OO 
1.00
FTP CO
3.2 &wi
FTP CO
203 yml
FTP CO
18.6 8/mI
FTP CO
12 yml
~
......
*
4-
rrpco '
3.1 g/ml
 73a
*742
*601
*
*680
TEST VEHICLE NUMBER
percent CO all the way to 3% CO. This large
variation in RSD readings could be a potential
problem if a pass/fail cutpoint were to be
selected that happened to be in the upper pan
of this range.
Conversely, vehicle #680. a very high FTP
vehicle, had a similar spread in RSD readings,
but the lower range of #680's readings
overlapped the high end of #742's readings.
Based only on this observation, it would be
difficult to discriminate the dirty car from the
clean car under these operating conditions
with an RSD cutpoint in the 2% to 4% CO range.
If a typical vehicle showed the kind of RSD
emissions distribution evident in this example,
the outcome of an RSD test could be based
considerably on chance, or be a strong
function of specific test conditions, instead of
Figure 18
RSD Reproducibility
(Uphill Bw4,2 MPH. ATX TabTnci
 5.00
Examination of Figure 17 shows
extremely high variability between RSD
measurements made under very similar
conditions on the level track. Based on
these RSD results, it is likely that' for some
vehicles, the range of RSD results will fall
on both sides of a likely cutpoint. For
example, the RSD emissions of vehicle
#742, which had a low FTP score, varied
from a low reading of essentially zero
o 4.00
C 3.00
(5 2.00
8
O 1.00
1A
s


rrpco
434 0mt

r rrpco
ijiidJ-
KMOCO
1.0 0/il
FTP CO
i,4 ymi
FTP CO
kO q/mI
 743
*744
1671
671
TEST VEHICLE NUMBER

-------
22
911672
being based on ihc vehicle's FTP emissions or
ievel of repair.
Examination of the uphill RSD results in
Figure 18. show thai they contrast very much
with those from, the level road testing in
Figure 17. In terms of test variability the
uphill results show a fairly high level of
repeatability. For instance, the individual test
results usually differed by less than 0.5% CO.
with many showing much better repeatability
than that. Whereas, the RSD readings from the
level site frequently varied by as much 1.0 to
3.0% CO. none of uphill vehicle had RSD scores
which varied to the extent that an individual
vehicle's pass/fail status would likely change
between tests.
The reasons for the inconsistency between
the uphill and the level sites is not completely
known. One possibility is that since the level
road RSD testing was done on the first day of
testing in this study, the lack of repeatability
may have been due to the operator's lack of
experience or some initial non-recurring
start-up problem. If this were the case, then
these results would not be representative of
normal use.	However. subsequent
conversations with the operators, have ruled-
out equipment problems a possible cause. Lack
of experience could still have been a factor,
though.
A second reason for the inconsistency may
have been that cars can vary more on a level
road. This could easily be the case, since the
load on the engines is very light under these
conditions. Any, even imperceptible change,
in operating conditions during the split second
that the RSD lest was done could be a large
percentage of the required load for the level
road, possibly resulting in a large change in
CO concentration level. If this were truly ihe
case, then the results between tests, even on
the same car. could be very' different. On the
other hand, the tests done at the uphill site
placed a much heavier load on the engine. The
higher load possibly led to less variability in
the load demanded by the driver, since any
small changes in load which did not cause a
noticeable change from the specified test
speed would be a considerably smaller portion
of the higher uphill load. The smaller load
variability may have resulted in more
tpeatable RSD data.
An obvious third reason is ihat different
cars were used at each site, and the observed
variability was simply a matter of coincidence.
This possibility is somewhat discounied
because of the consistency of the results
within each road site. Furthermore, the
difference in the False Failure rates in Figures
7(a) and 7(c) might also imply that the site
selection can affect the variability of the RSD
reading. Figure 7(a) included both level road
operation and uphill operation (with
acceleration), whereas Figure 7(c) included
only uphill operation. As previously indicated,
the False Failure rate of IM240 High Emitters
for the combined site (Figure 7a) was higher
than that for the uphill site. It is possible that
some vehicles at the two-lane site exhibited the
behavior demonstrated by vehicle #742, and
were identified as High Emitters by the RSD.
when in fact they were not. If this scenario
truly occurred, it would suggest that only sites
that impose some moderate load on the engine
should be selected for future testing.
ON-ROAD DATA - RSD variability was also
investigated using the on-road vehicle speed
data collected at the expressway ramp site.
Figure 19 shows a plot of the RSD CO
concentration emissions versus vehicle speed
at this site. No speed measurements were
recorded ai the two-lane site.
The vehicle speeds shown in this figure
were measured manually by the RSD operator
holding a radar gun. They reflect as closely as
possible a vehicle's speed at the instant of the
RSD test. The plot shows that the vehicle
speeds ranged from 5 to 33 MPH with a mean of
23 MPH. The lower speeds (5 to 15 MPH) were
generally recorded at the first site on the
expressway ramp, and the higher speeds (20 to
35 MPH) at the second site on the expressway
ramp.
The scatter and the linear regression
results in Figure 19 indicate no relationship
between speed and RSD CO concentration. The
linear regression line -is almost horizontal and
the correlation coefficient is 0.0025. However,
the lack of correlation between vehicle speed
and RSD results is not surprizing. The more
critical variables, as mentioned above, are
likely to be vehicle acceleration and load.
However, because of problems with the
interface between the RSD system and the

-------
911672
23
Figure 19
On-Road Speed Effects
Y  0.016X ~ 0.49
0.0025
10 15 20 25
VoMciaSpcsd (MPH)
radar gun, acceleration
in this study.
could not be measured
Considering these results, along with those
from the test track, vehicle speed (at least over
this range) does not appear to be useful as a
site selection criteria. As suggested, other
criteria are probably more useful, but possibly
more difficult to measure or obtain.
PLUME INTERFERENCE -A potential
problem exists where the residual exhaust
plume from a leading vehicle could possibly
create an interference in the RSD reading of a
closely following second vehicle. Such
Figure 20
Vehicle Separation Effect
interference could cause erroneous
results if the separation time between
successive cars was short, and the
difference in emission levels of the
successive plumes was large. The
potential problem of plume
interference during the RSD
measurements was anticipated by its
developers. To minimize the effects of
this problem, the researchers
developed a criteria to detect and
eliminate rapidly changing C0/C02
ratios when two exhaust plumes mix
together. Under such test conditions,
the RSD discards readings when the
standard deviation of the linear fit of
CO and C02 exceeds 20 percent.
Despite these precautions.
Stephens and Cadle [4] found potential
RSD measurement problems if ihe
remnants of a" previous vehicle's exhaust
plume were present during the current RSD
measurement.	From their work they
concluded that (1) a plume effect does occur.
(2) the RSD test criteria identifies plume
nonlinearity. (3) the residual plume effect
seems to last on average at least one second and
maybe longer, and (4) plume nonlinearity
only occurs when sequential plumes are of
greatly different concentrations.
In this study, a plume interference
analysis, similar to one performed by Stephens
and Cadle was conducted using a large sample
of 21,000 vehicles. These vehicles were
coincidentally measured by the RSD
0.70
 0.60
C
8
c
o
o
O
o
a
CO
ec
c
flj
t5
0)
o.so
0.40
0.30
0.20
0.10
0.00
(5  15% CO
Previous
Vehicle
(1-5% CO
/
23 4 56789
Time Between Measurements (Seconds)
at the expressway site, in the process
of measuring the 257 vehicles which
received the both the 1M240 test, and
the RSD test. The results were similar
to those obtained by Stephens and
Cadle. For example. Figure 20
compares the median RSD emissions
stratified into three levels based on
the previous vehicle's RSD emissions.
The figure shows a definite likelihood
of measuring a higher RSD value if
the time separation is two seconds or
less and the previous RSD CO value
was more than 5 percent. Based on
the results in Figure 20. the RSD CO
emissions of a typical vehicle will be
about 0.5% CO higher if it is following
(within one or two seconds) a High
RSD Emitter (greater than 5%). than

-------
24
911672
if it is following a low RSD High (less than
1.0%). For separation times of three seconds or
more, the CO medians show less relationship
with the previous vehicle's CO. indicating the
previous vehicle's plume has dispersed.
IN-USE FLEET ANALYSIS
A second goal of this project was to evaluate
the ability of the RSD units to predict average
in-use emissions in a specified area. This
evaluation consisted of three parts: (1)
comparison of estimated average RSD CO mass
emissions with IM240 mass emissions. (2)
comparison of the median RSD and IM240
emissions collected in Indiana with RSD
emissions collected in other states, and (3) use
of- the Indiana RSD emissions from the
expressway site to simulate the collection of
RSD emissions at a number of independent
sites.
All three parts of this analysis required
converting the RSD CO concentrations to CO
mass emission estimates. . This was done in a
two step process following the
by Stephens and Cadle [4].
concentrations were converted
gallon units using molar ratios
in
method outlined
First, the CO
to grams per
from idealized
fuel combustion
economy equation
second step
gallon units
fuel economy estimate. Two estimates for an
individual vehicle's fuel economy were
available. The first was the actual fuel
economy measured during the IM240 test. It
could be viewed as the best estimate available,
since instantaneous fuel economies at the
precise RSD test conditions would generally
not be available. The second was an average
by-model-year and by-manufacturer fuel
economy estimate based on CAFE numbers [4],
ESTIMATED MASS EMISSIONS - RSD mass
emission estimates in grams per mile were
calculated using both of the above fuel
economy estimation methods so that they could
be compared directly with IM240 mass
emission results. The individual model year
results, and overall average results from the
combined sample are shown in Table 6 along
with the IM240 mass emissions. Also, shown
for comparison are the average by-model-year
RSD and I/M concentration measurements.
equations, the EPA fuel
and other constants. The
involved converting the grams per
to grams per mile units using a
Table 6
Average Emissions by Model Year
(Combined Site)
A comparison
estimates at the
IM240 mass CO
good agreement.
IM240 based fuel
g/mi is only 4
IM240 value.
of the overall RSD mass
bottom of Table 6 with the
emissions shows reasonably
The difference between
economy RSD value of 15.51
percent below the measured
When the CAFE based fuel




Indian* l/M Raulrt
Uodtl
Samel*
CO
HC
NO*
CO*
CO"
CO


lm
SIZE




fa/mdrt
ou
IrftaCO
25QQ CQ
1976
9
42.19
2.67
2.32
4049
34 J6
1.71
: a

1977
7
11.7)
IJ7
4.40
36.74
31J6
2.02
1.36

1678
19
40 J 2
2JO
3.49
37.77
3345
1.90
1.61

1979
19
22.96
uo
4.41
19.25
20-02
1.26
2.11

1960
18
2-4.77
1.46
3.27
26.79
19.77
1.40
0.43

1981
IS
3506
Ml
1.15
nit
14J1
1.10
0.91
 l.U
198?
30
a.17
1.72
2.65
25.17
2117
1.66
0.7J
1.49
1963
22
27.99
1J2
2.80
24.01
1113
1.36
1.23
1.69
1364
37
10 0
0.64
1.79
16.34
14J5
. 1.14
0.30
0.66
1085
31
9~51
0.64
1.69
5.32
4.24
0.32
0.2?
0.11
1968
31
10.77
0.62
1.33
11.64
10O5
0.81
0.77
0.67
1 987
3]
7J3
0.43
1.67
5.35
5.15
0.42
0.1J
0.31
1988
27
3.11
0.41
0.19
4^4
4.49
0.39
0.16
0.13
1989
32
4.44
0.21
0.19
6.36
6.11
0.48
0.09
0.12
1990
20
3.13
0.15
0.61
5.62
SJ1
0.44
o.u
0.13
199-)
1
Ijl
mi
Q-39
U

0 36
Qffl
Q.qp
1976-82
117
29J2
un
3.20
27 44
24.13
1.54
1.24

1983 ~
239
9.63
0J7
1.47
9.67
 J7
0.66
0.36
0.43
ALL
396
16.1?
1JJ5
2.04
15 Jl
13J5
0.95
0.63
0.45
Grams aalon RSO CO rwuls
oorwud to mriiM grmrm par mis uang th* vttvcfc tuf 
>ao noma*

dtfvnngtft
IIH240 eye*.







economy values were substituted. the
difference between the RSD estimate and the
IM240 increased to 16 percent. A likely reason
for the larger difference when using the CAFE
values is that the IM240 fuel
economy is based on the
actual performance of the
vehicles, whereas the
is simply a rough model
and manufacturer
fuel economy based
vehicles tested over
in-use
CAFE
year
average
on new
~ Giro pm 0*4ion RSD CO a** ooffwtod to Amciad grama p* uung	CAFE lui coremy vaiuM.
the FTP. In any case, both
estimating methods resulted
in an average RSD value less
than the IM240 value
In examining ihe
individual model year
results. however, the
relationship between the
estimated RSD emissions and
the IM240 emissions was not
as consistent, even though
both measurement methods
showed a generally

-------
911672
25
decreasing pattern of emissions from older to
newer model vehicles. For example, the RSD
estimates for the 1984 model year with 37
vehicles were significantly higher than the
IM240. while the 1985 RSD estimates with 38
cars were significantly lower than the IM240
value. This dramatic switch in RSD estimates,
approximately a 70 percent decrease, occurred
while the IM240 values changed only about 10
percent between the 1.984 and 1985 model
years. Although this example represents the
most extreme case in Table 6, the change in
RSD estimates from year to year, in general,
seemed to vary more than the IM240 average
emissions.
Relative to two methods of computing the
RSD estimates, the similar results indicate that
substituting the CAFE fuel economy estimates
seems to be a reasonable approach. This is
fortunate. since actually measured fuel
economy values for individual in-use vehicles
would probably never be available in any
application of RSD to on-road measurements.
OTHER RSD TEST PROGRAMS - The CO
concentration results from other RSD
programs, which did not measure mass
emissions, had previously been analyzed by
the author [6] relative to their estimated fleet
mass emissions. To put these other estimates in
perspective, they were compared to the
Indiana results.
. Since the other results were analyzed in
terms of the median values, the RSD and IM240
median values from the express way site were
computed, and plotted in Figure 21(a)..
Excellent directional agreement between
medians can be observed. For example, both
curves seem to follow the same inconsistent
emissions pattern for the older vehicles.
Suggesting that the medians are similar, and
that the same outliers and small sample sizes
effect the RSD and the IM240 medians equally.
This is a positive result, and suggests a good
relationship between average RSD and IM240
results. However, total agreement between the
two is not present, and the magnitude of both
curves suggests that the RSD, tends to under
predict the IM240 over all model years.
The RSD data in Figure 21(a) were compared
to the median data from the previous analysis
[6] in Figure. 21(b). This previous analysis
included test data collected in Chicago. Illinois
[1], Denver. Colorado, and Ute Pass Colorado by
Dr. Stedman. Of the three sites, the vehicle
operation and test site topography at the
Denver location was probably the most similar
to the Hammond. Indiana expressway on-ramp.
The Ute Pass location at 8000 feet with a steep
uphill grade and typically strong headwinds
was probably the least similar. The Chicago
express way ramp was marginally similar in
[hat it likely included vehicle acceleration, but
was a straight entrance ramp instead of a
curvy one like Hammond and Denver.
Given the site differences, there is a
general similarity in the model year pattern at
all sites except Ute Pass. The difference
between the general trend and a specific
model year comparison may be due to the
difference in sample size. The Denver and
Illinois data included thousands of vehicles,
where the Hammond data included only
hundreds of vehicles. Nevertheless, the
Hammond data is in the same ballpark as the
Denver and Chicago data.
FIGURE 21
Median By-Model-Year CO Emissions
RSD (IndUnTT)
IM240
? f BSPdfciw) ]
:@D
(Indi*
O 20
76 77 71 79 M II 12 tt 14 ftS M I? 8S S0 79 77 71 79 80 It 12 U M 19 16 17 H 19
MODELYEAR	MODEL YEAR
<)	
MULTIPLE FLEET
SIMULATION - A serious
shortcoming of previous RSD
testing is that they did not
include any mass emission
data on matched vehicles.
This test program addressed
that need on an individual
vehicle basis. but in
evaluating fleet emissions, it
is only one program with
only one fleet average
emission value (see Table 6).
To determine how well the
RSD fleet average in Table 6
might apply to on-road

-------
:
911672
easurcmcnts. many programs similar to this
would need to be run.
In order to obtain a qualitative assessment
iihout running many test programs. a
3mputer program w-as developed to simulate
:e effect of many sites. Additionally, because
e analysis in this paper seems to suggest that
SD may track broad trends' in fleet averages
:asonably well, but does not do so well on
idividual vehicles, it would be useful to
mulate many sites with different cars, but
ith identical test conditions for each
idividual vehicle.
The correlated variables in this simulation
-rogram were FTP/IM240 mass emissions,
M240/RSD estimated mass emissions, and
M240/RSD vehicle speed. Only expressway
lata was used for this simulation. It was
xpected that the FTP/IM240 simulation would
how good correlation as in Figure 2 earlier,
nd that the IM240/RSD speed would show little
orrelation.
The first step in the FTP/IM240 simulation
vas to randomly partition the data in Figure 2
-*t two equal sized samples, arbitrarily called
;te 'A' and sample 'B'. Next, the average
. r* and IM240 CO mass emissions for each
ample group were calculated. The difference
letween the average FTP emissions of sample A
nd sample B was then determined. Likewise
he difference in the average IM240 emissions
)f sample A and B was calculated. This
.alculation process created a unique FTP
:missions difference between sample A and B,
ind a unique IM240 emissions difference
between the samples. This unique data could
hen be plotted on an x - y plot as a point.
The process of randomly partitioning the
data in Figure 2, and calculating the FTP and
IM240 emission differences was repeated 1000
times, generating the 1000 paired FTP and
1M240 data points in Figure 22(a). In this way.
the mix of vehicles in each sample could be
different, representing different fleets, but
each vehicle would be tested under identical
conditions. Therefore, only the effect on
different fleet mixes would be evaluated.
An analogous procedure was repeated with
the 256 vehicles tested at the expressway site
which received both the RSD and the IM240
test. These CO differences were plotted in
Figure 22(b). The same process was then used
with IM240 CO data and the RSD speed data
resulting in Figure 22(c).
The best agreement, as expected, of the
three comparisons in Figure 22 is between the
FTP and the IM240. This is noted by the high
regression correlation coefficient, and the
relative absence of data points in the upper
left quadrant and the lower right quadrant.
Points in these quadrants would occur if the
difference between the FTP means of sample A
and sample B were negative, and the
difference between the IM240 means of sample
A and sample B were positive, or vice versa. If
this had occurred, it would have indicated
there was little agreement between the paired
FTP and IM240 results for that particular
simulation.
In contrast to the FTP/IM240 simulation is
the .IM240 versus RSD speed simulation. As
indicated, this simulation was only conducted
as a control parameter of poor correlation,
because theoretically, their should have been
Flgiu* 22
Fleet Simulations
(Emtotoo or Speed OUTcrcoca between Riodomly Selected Simple
[Y  0.83X-0.0H rlmOM ]
V . 0.005X  0.032 rl m 0.0004
Y  0.72X - 0.084
L.S.
I	i	
.10 <5 0 9 10
FTP - CO (0fmO
0 s
IM240  CO (jj/ml)
10 -J 9 I 10
IM240  CO (g/ml)
(

-------
911672
27
no relationship between these variables since
they came from non-overlapping,
independent tests. The generally circular
pattern shown in Figure 22(c) shows that this
was the case. Note that the RSD speed
parameter is the speed difference between
sample A artd B, not the average speed in
either group.
Finally, comparison of the RSD and IM240
mass emission simulation showed fairly
positive results, although not as positive as the
FTP and IM240 results. As in the other figure,
these positive results were evidenced by the
reasonable correlation and the relative
absence of the points in the two quadrants. In
general, these results generally show that if
the RSD and the IM240 were tested at many
different test sites under IDENTICAL
conditions, their mean emissions would
typically be similar to one another.
As a final point, despite the good overall
average and median emission agreement
between IM240 and RSD. care must be taken as
to not conclude that RSD estimates will
consistently represent individual vehicle mass
emissions over a complete driving cycle, or in-
use operation over a wide range of conditions.
This fact can be seen in the scatter plots of
Figure 4, the repeatability plots of Figures 16
and 17, and in the individual model year
emission results in Table 6. These results
indicate that RSD measurements can vary
considerably, for a number of different
reasons which probably can never be
completely accounted for in a one second test.
Therefore, the utility of the RSD results in
terms of fleet predictions is best if limited to a
specific consistent test condition, and based on
a large RSD sample.
CONCLUSIONS
The primary goal of this test program was
to evaluate the capability of the RSD to
properly identify and categorize High Emitters
as determined by the IM240. A spin-off
analysis evaluated the performance of the RSD
relative to a typical I/M test procedure, and the
ability of the RSD to identify average fleet
emission levels was investigated. Specific
observations and conclusions from this test
program are as follows.
1. The RSD did not, even at the most
stringent cutpoints, identify all of the High CO
Emitters. At the most commonly reported
cutpoinc of 4.5 percent CO, the RSD identified
less than 15 percent of the late model High
Emitters, which accounted for only about 25
percent of the repairable excess emissions, and
only identified about 20 percent of the older
model High Emitters. Approximately 85
percent of the late model High Emitters
accounting for 75 percent of the repairable
excess emissions were missed by the RSD in
this test program.
2.	Of the vehicles identified by the RSD as
High CO Emitters, a moderate portion of them
were, in fact, not High Emitters (i.e.. False
Failures). For cutpoints above 2 percent CO.
RSD errors of commission ranged from zero to
as high as 35 percent (combined site) for late
model year cars, and from zero to
approximately 25 percent for older cars. These
ranges of False Failure rates, however, did
confirm that at least 65 to 75 percent of the
vehicles identified as High Emitters by the RSD
truly are High Emitters.
3.	The RSD was able to operate in mildly
inclement weather conditions including light
rain. Only about 10 percent of the CO tests
were lost due to weather. Of this reduced total,
about another 10 percent was lost due to
improper RSD measurements detected by the
RSD's internal quality control algorithm.
Neglecting equipment malfunctions, slightly
more than 15 percent of the sampled fleet was
not measured.
4.	The RSD CO performance looked more
like a traditional I/M test than a loaded,
transient, mass emission test. At the most
reported cutpoint of 4.5 percent CO, the RSD
was not nearly as effective in identifying High
Emitters or repairable excess emissions as the
two-speed idle I/M test for late model cars.
Lowering the RSD cutpoint to 2 percent,
increased the performance of the RSD to a
level marginally better than the two-speed
test, but at the expense of higher errors of
commission (False Failures) relative to the the
I/M test. Although the RSD identified fewer
older model year cars than the single speed
idle I/M test, it identified slightly more excess
emissions with a measurably lower False
Failure rate.
5.	The RSD and the I/M CO tests identified
different population groups. The RSD only

-------
28
911672
portion was generally smaller for the newer
model group than for the older model year
group.
6.	Combining the RSD CO tesi and the I/M
CO test improved identification of late model
High Emitters, but at the expense of increased
False Failures. For older cars, the combination
substantially increased High Emitter
identification with little change in False
Failures relative to the I/M test. In
combination with I/M, reasonable increases in
identification of older vehicles were even
observed at the 4.5 percent RSD cutpoint with a
lower False Failure rate than with I/M alone.
7.	Selecting an RSD cutpoint below 2
percent CO did not substantially improve the
identification of repairable excess CO emissions
from the late model vehicles. Selecting a
cutpoint above 4 percent resulted in around 80
percent of all High Emitting vehicles falsely
passing the RSD test. A recommended CO
cutpoint range for future test programs would
be from 2% to 3.5% CO, depending on the
program objectives, and the acceptable False
Failure rate.
8.	Repeatability of the RSD CO emission
measurements on track-tested vehicles was
found to be extremely poor under steady-state
conditions on a level road. However, much
improved repeatability in RSD CO
measurements were observed on vehicles
operating on a 3% uphill grade. It is thought
that the reasons for these variable results
were due to individual vehicle differences, and
its precise operating characteristics at the
lime of the RSD test, rather than the
instrument itself. The reasons for the
inconsistent measurement resuits are less
important than the fact that they existed, and
that they cannot currently be accounted for
by the RSD system, since the current system
measures neither vehicle speed nor
acceleration accurately. With the current
state of RSD development, it is recommended
that future site selections lean toward ^hose
sites with light to moderate acceleration or
uphill loads.
9.	Improving the CO identification
performance of the RSD may be difficult
because reducing RSD variability and
;nc reas ing identification performance
jppears to be influenced by vehicle operation.
which is largely uncontrolled. Because of the
likely uncontrolled vehicle operation, RSD
identification improvements might only be
achieved by lowering cutpoints, whereas I/M
programs can lower cutpoints, add program
features (e.g., functional checks), or change
test type.
10.	The RSD HC performance was abysmally
poor. However, the unit used was one of the
first prototypes developed, and future units
would be expected to improve.
11.	Reasonable agreement seemed to exist
between the IM240 overall fleet-average
emission levels and those estimated from the
RSD measurements. However, substantial
variations between the two measurement
methods for individual model years was
observed. Further analysis simulating
multiple fleets, tends to imply that there is a
rough relationship between the RSD fleet
estimates and the IM240 when identical test
conditions and very Jarge samples are used.
12.	The RSD plume interference analysis
indicated that there is a definite likelihood of
measuring a higher RSD value if the time
separation between vehicles is two seconds or
less, and the previous RSD CO value is more
than 5 percent. However, even if these two
conditions are met. the median effect is less
than 0.5% CO.
As a final note, the inability of the RSD to
identify even a significant fraction of the
High Emitters was the most striking
observation of this study. The cause for this
inability to identify a large fraction of the
High Emitters is probably more a reflection of
the fact that an instantaneous RSD emission
measurement is not likely to be completely
representative of an overall driving cycle,
rather than an inability of the RSD to
accurately measure the instantaneous CO
emissions in an exhaust plume. Compounding
this problem is the current lack of ability of
the RSD system to measure the exact operating
mode of the vehicle during the test. Therefore,
not only is the RSD system not capable of
completely measuring the emissions from
overall driving, the emission results which are
measured are diminished in usefulness
because there is no exact way to relate them to
a vehicle driving mode.

-------
911672
29
Even so, this may not be a fatal handicap,
since the RSD seems to have the ability to
identify a portion of the High Emitters in a
roadside environment quickly and non-
obtrusively. As such, this is probably the RSD
system's principle asset, and may be best used
on a random basis in conjunction with a
traditional I/M program. For example,
identifying and repairing even a. portion of
the High Emitters in a CO 'hot spot' (outside of
their periodic I/M cycle) may be sufficient to
lower the ambient levels in the hot spot area.
ACKNOWLEDGEMENTS
Recognition is given to Mr. Joe Patterson of
Automotive Test Laboratories (ATL) for the
many hours he spent alongside the expressway
during the cold winter days in Hammond.
Indiana. Recognition is also given to Mr.
Pradeep Tripathi of (ATL for his roadside
efforts, and especially for the hours he spent
processing RSD data and reading license plates
from video tapes. Special acknowledgement
also goes to Dr. Donald Stedman and Dr. Gary
Bishop for all their valuable advice and
support. Finally, thanks goes to Mr. Dennis
McClement of ATL for his patience, and helpful
suggestions throughout this process.
REFERENCES
1.	Stedman, D.H., and Bishop. G.A., "An
Analysis of On-road Remote Sensing as a Tool
for Automobile Emission Control". Report
Prepared for the Illinois Department of Energy
and Natural Resources", ILENR/RE-AQ-90/05.
2.	Stedman, D.H., and Bishop, G.A., "Remote
Sensing for Mobile Source CO Emission
Reduction", Finai Report under EPA
Cooperative Agreement, CR-815778-01-0.
August. 1990.
3.	Bishop, G.A.. and Stedman. D.H.,
"Oxygenated Fuels. A Remote Sensing
Evaluation"', SAE Paper No. 891116.
4.	Stephens, R.D.. and Cadle, S.H., "Remote
Sensing Measurements of Carbon Monoxide
Emissions from On-road Vehicles", GM
Research Publication GMR-7030, May, 1990.
5.	Lawson, D.R., et al, "Emissions- from In-
use Motor Vehicles in Los Angeles: A Pilot
Study of Remote Sensing and the Inspection
and Maintenance Program", Journal of the Air
& Waste Management Association, August.
1990.
6.	October 9, 1990, Memorandum from E.
Glover to P. Lorang, "Analysis of Existing
Remote Sensing Data1'
7.	"An Evaluation of Remote Sensing for
the Measurement of . Vehicle Emissions",
Report Prepared for the California Air
Resources Board by Sierra Research. Report
No. SR90-08-02, August, 1990.
8.	Pidgeon, W.. and Dobie. N. "IM24Q
Transient I/M Dynamometer Driving Schedule
and the Composite I/M Test Procedure". Report
No. EPA-AA-TSS-I/M-91.-01.
9.	June 20. 1991, Memorandum from N.
Brown to W. Clemmens. "Compilation of
Remote Sensing Data Collected in Indiana".
10.	Glover. E.. and Brzezinski. D. "Exhaust
Emission Factors and Inspection/Maintenance
Benefits for Passenger Cars". Report No. EPA-
AA-TSS-I/M-89-2.
11.	Inspection/Maintenance Program
Implementation Summary, EPA Mobile Source
Document, July, 1990.
12.	EPA Technical Report, "Recommended
I/M Short Test Procedures For the 1990's: Six
Alternatives, " Report No. EPA-AA-TSS-I/M-90-
3.

-------
APPENDIX K
MODEL YEAR FAILURE RATES BY TEST TYPE

-------
HAMMOND LANE 1/M FAILURE RATES

HC 2500
...
220 ppm
220 ppm
...
...




CO 2500
...

1.2 percent
...


IM240


HC Idle
220 ppm
220 ppm
220 ppm
250 ppm
350 ppm

HC:
0.8 g/mi

CO idle
1.2 percent 1.2 percent 1.2 percent 2.0 percent 3.5 percent

CO:
15 g/mi
Model
Lane





Model
Lane

Year
Sample





Year
Sample

76
102
65
N.A.
NA
57
2d
76
121
1 04
77
255
202
NA
NA
187
116
77
281
255
78
355
252
NA
NA
226
140
78
385
338
79
431
266
NA
NA
230
119
79
456
354
ao
347
140
NA
NA
m
7B
80
351
239
81
353
57
66
98
50
37
81
366
219
82
392
86
99
141
67
47
82
403
243
83
548
70
76
105
53
35
83
467
1 99
84
804
106
-
156
86
49
84
677
263
85
819
66
74
105
56
35
85
658
194
86
792
52
- 56
85
39
24
86
643
121
87
696
24
aa'
39
19
1 3
87
546
5 S
ee
755
15
id
22
12
4
66
632
37
89
791
8
'9 "
1 2
6
3
69
632
27
90
784
2
3
8
2
1
90
604
1 4
91
89
0
0
0
0
0
91
36
0
92
1
0
0
0
0
0

Note: Shaded area indicates official Indiana l/M test
Page 1

-------
HAMMOND LANE l/M FAILURE RATES
HC 2500:	 220 ppm
CO 2500:
HC Idle:	220 ppm 220 ppm
CO Idle:	1.2 percent 1.2 percent
Model
Year
220 ppm
1.2 percent
220 ppm 250 ppm 350 ppm
1.2 percent 2.0 percent 3.5 percent
76
63.7%
NA
NA
55.9%
27.5%
77
79.2%
NA
NA
73.3%
*5.5%
78
71.0%
NA
NA
63.7%
39.4%
79
61.7%
NA.
NA
53.4%
27.6%
80
40.3%
NA
NA
, 29.4%
22.5%
81
16.1%
1.7%
27.8%
14.2%
10.5%
82
21.9%
25.3% "
36.0%
17.1%
12.0%
83
12.8%
13.9%
19.2%
9.7%
6.4%
84
13.2%

19.4%
10.7%
6.1%
85
8.1%
:
12.8%
6.8%
4.3%
86
6.6%
7,1%
10.7%
4.9%
3.0%
87
3.4%
4.0%
5.6%
2.7%
1.9%
88
2.0%
2,4%
2.9%
1.6%
0.5%
89
1.0%
1.1%
1.5%
0.8%
0.4%
90
0.3%
0.4%
1.0%
0.3%
0.1%
91
0.0%
0.0%
0.0%
0.0%
0.0%
92
0.0%
0.0%
0.0%
0.0%
0.0%
IM240 i
H: 0.8 g/mi
00: 15 g/mi
Model
Year
76

86.0%
77

90.7%
78

87.8%
79

77.6%
80

68.1%
81

59.8%
82

60.3%
83

42.6%
84

38.8%
85

29.5%
86

16.8%
87

10.8%
88

5.9%
89

4.3%
90

2.3%
91

0.0%
Note: Shaded area indicates official Indiana l/M test
Page 2

-------
APPENDIX L
COMPARATIVE PURGE FLOW DATA

-------
Figure L- - 1
Purge Flow Strategies
Q.
e
o
V
v>
0)
g>
3
0.
40.0
200	250
Test #238
0.40
200	250
Test #393
0.50
0.45
0.00
200	250
Test #354
0.5
0.4
0.40
8:33
0.25
o:?s
8:6
0.00
Test #118
0.50
0.45
8:!8
0.30
0.25
8:?e
0.10
0.05
0.00
5
200	250
Test #236
0.50
0.45
0.40
8:35
0.25
0.20
0.15
m
0.00
ti AaAw b a *. [.<
200	250
Test #427
0.50
0.45
0.40
0.35
8:2
0.20
0.15
50
100
Seconds
150
200
250

-------
Figure L - 2
Cumulative Purge Flow
30
25
Test#427
20
Test#393
15
Test #236
5
0
250
150
200
50
100
0
Time (seconds)

-------
Figure L- 3
Vehicle Loading versus Speed
(IM240, ASM, and Arizona)
50.00
ETW = Curb Wt. + 300 lbs
Curb Wt. = 2200 lbs.
45.00
40.00
ASM 2525
Hp=(ETW/300)
35.00
a 30.00
AZ 3-4 Cyl
Range
AZ 5-6 Cyl
Range
2 25.00
" AZ > 8 Cyl
Range
ASM 5015
' Hp=(ETW/250)
20.00
15.00
50MPH
Road Load
10.00
5.00
0.00
10
0
20
30
40
Speed (MPH)

-------
50
45
40
35
30
25
20
15
10
5
0
Figure L - 4
Vehicle Loading versus Speed
(IM240, ASM, and Arizona)
ETW = Curb wt +300 lbs
Curb Wt. = 3000 lbs.
ASM 2525
Hp={ETW/300)
A ASH load
 AZ Stdy State
AZ 3-4 Cyl
Range
AZ 5-6 Cyl
Range
AZ > 8 Cyl
Range
ASM 5015
Hp=(ETW/250)
50 MPH
Road Load
*
1
-oA
1 A
60
50
30
40
10
20
0
Speed (MPH)

-------
Figure L - 5
NOx Driving Cycle Emissions
30.0
Test #343
12.00
V
10.00
2.00
Test #393
12.00
V.V.
I 2.00
10.00
8.00


250
Test #238
12.00
I 0.00
8.00
6.00
4.00
2.00
0.00

A
SSa,
250
Test #461
12.00
to.oo
: ;
/-rfrrh
Seconds

-------

-------