United States      Air and Radiation      EPA420-P-99-017
           Environmental Protection              May 1999
           Agency                   M6.EXH.009
vvEPA     Determination of
           CO Basic Emission
           Rates, OBD and
           I/M Effects for Tier 1 and
           Later LDVs and LDTs

           DRAFT
                              > Printed on Recycled Paper

-------
                                    - Draft -

  Determination of CO Basic Emission Rates, OBD and I/M Effects
                   for Tier 1 and later LDVs and LDTs

                       Report Number M6.EXH.009
                                   May, 1999
                                 Edward L. Glover
                                   John Koupal

                          U.S.EPA Office of Mobile Sources
                         Assessment and Modeling Division
                                     NOTICE

These reports do not necessarily represent final EPA decisions or positions.  They are intended to present technical
analysis of issues using data which are currently available. The purpose of releasing these reports is to facilitate the
exchange of technical information and to inform the public of technical developments which may form the basis for a
final EPA decision, position or regulatory action.
M6EXH009.WPD                               1                        May 20, 1999

-------
1.0    Introduction

       This document (M6.EXH.009) describes the methodology, data analysis and results of the
process used to develop basic CO emission factors, OBD (On-board Diagnostics II) effects, and I/M
(Inspection / Maintenance) credits for Tier 1, TLEV (Transitional Low Emitting Vehicle), LEV (Low
Emitting Vehicle), ULEV (Ultra Low Emitting Vehicle) vehicles (referred collectively throughout
the document as Future Standards Vehicles (FSV) for MOBILE6. The document "Determination
of NOx and HC Basic Emission Rates, OBD and I/M Effects for Tierl and Later LDVs and LDTs"
(M6.EXH.007) shows the analogous methodology and the results for NOx and HC pollutants for the
FSVs.   These vehicles are current and future model year cars and light trucks which must meet
increasingly stringent tailpipe exhaust emission limits. Of the four types, the first three types (Tierl,
TLEV and LEV) are held to the same numerical CO standards. The fourth type (ULEV) meet a more
stringent CO standard which is 50 percent of the Tierl  standard.

       Table  1 shows the emission standards limits in grams per mile for each vehicle type and
standard. The limits in units of grams per mile are shown for the 5 year / 50,000 mile certification
point.  The other important parameter in regards to the certification standards is the model year
implementation schedule for all of the standards. This is the percentage of a particular model year
and vehicle class that will be required to be certified to the lower standard. This schedule can be
found in EPA document EPA420-B-98-001  "Exhaust Emission Certification Standards".  It is
available on the EPA-OMS website. The current website address is "http://www.epa.gov/oms/stds-
lh.htm"

       This document (M6.EXH.009) also provides a brief explanation of the derivation of CO start
and running emissions for all FSV classes (i.e., Tierl,  TLEV, LEV and ULEV). Results are also
provided in tabular and graphical  form. A more detailed explanation of the concept of start and
running emissions and their derivation can be found in "Determination of Running Emissions as a
Function  of  Mileage for  1981-1993 Model  Year  Light-Duty  Cars"  (M6.EXH.001)  and
"Determination of Start Emissions as a Function of Mileage and Soak Time for 1981-93 Model Year
Light-Duty Vehicles" (M6.STE.003).

       This document (M6.EXH.009) also explains the I/M credit methodology for 1994 and later
model year vehicles, and discusses the algorithm used to predict OBD effectiveness. This OBD
algorithm assumes: (1) that 1996 and later cars are equipped with an onboard electronic diagnostic
(OBD) system, but that 1994 and 1995  model year vehicles are not equipped;  (2) that the OBD
system continuously monitors the performance of the  car's emission control system, and detects
serious problem(s) which cause the vehicle's FTP emissions to exceed 1.5 times its applicable
certification standards, and (3) when  such problems are detected, a code is registered in the car's
computer and a dashboard warning light is turned on to notify the owner.

       The I/M methodology presented in this document for FSVs (M6.EXH.009) is similar to the
I/M methodology for 1981-93 model yearvehicles presented inEPAdocumentM6.IM.001. Thetwo
methodologies differ because of the presence  of OBD and the different emission standards for the
    M6EXH009.WPD                      2                                May 20, 1999

-------
FSVs.  Also, a central I/M assumption for the OBD equipped vehicle fleet (1996+ model years) is
that OBD tests will be performed in I/M stations, and that these OBD tests may eventually replace
traditional emission tests for the 1996 and later model years.  In contrast, the I/M methodology for
the 1981-93 model years is based strictly on the use of traditional exhaust measurement tests.
Table 1
Tierl. LEV and ULEV CO Emission Certification Standards
(gram/mile units at 50K miles except where noted)
Veh Type
LDV
LDT1
(<6000GVW
<3750 Loaded Wt)
LDT2
(<6000GVW
>3750 Loaded Wt)
LDT3
(>6000GVW
<3750 Loaded Wt)
LDT4
(>6000GVW
>5750 Loaded Wt)
TierO
3.4
3.4
10.0*
9 o**
9 o**
Tierl
3.4
3.4
4.4
4.4
5.0
LEV
3.4
3.4
4.4
4.4
5.0
ULEV
1.7
1.7
2.2
2.2
2.5
* 100K mile Standard
** 50K mile Standard
       Most Tier 1 and later vehicles will be equipped with an OBD system and, if in an I/M area,
subject to OBD-based I/M rather than traditional exhaust I/M. However, some Tier 1 vehicles will
not be equipped with OBD (model years 1994 and 1995), and some OBD-equipped vehicles will be
subject to an exhaust I/M program (prior to calendar year 2001). Basic emission rates are therefore
required for the following cases, and are presented in this report:
    M6EXH009.WPD
May 20, 1999

-------
       No OBD/No IM (Base)  applies to pre-OBD Tier 1 vehicles (1994 and 1995 model years).
       It is also used as a basis for the computations of BERs with OBD and/or OBD-based I/M.

       OBD-Only (OBD/ No IM) applies to 1994 and later OBD equipped vehicles where an I/M
       program is not present.

       OBD/IM applies to 1994 and later OBD equipped vehicles where an I/M program which
       conducts OBD checks is present.  An exhaust test may or may not be performed; it is not
       differentiated from  an I/M program with both OBD checks and exhaust testing, since
       additional I/M reductions are not given for exhaust testing if OBD checks are performed.

       Exhaust I/M / OBD represents a situation in which an exhaust I/M test is conducted in an I/M
       program on vehicles which are equipped with OBD, but the OBD I/M test is not performed.
       This situation can only occur on 1996 and later model year vehicles equipped with OBD, and
       subject to an exhaust I/M test. It is a likely scenario until at least calendar year 2001 when
       the EPA approved OBD I/M test is expected to be available.

       Exhaust I/M / No OBD represents a situation in which only an exhaust I/M test is conducted
       in an I/M program (an EVI240, ASM, or Idle test). This scenario will apply only to the 1994
       and 1995 model year vehicles which are not equipped with OBD systems.
2.0    Basic Emission Rate Derivation Concept

       The basic concept underlying the generation of Tier 1 and later BERs is similar to the
approach used to develop the I/M credits for 1981 through 1993 vehicles. For the No OBD/No EVI
case, this concept segregates in-use vehicles into "normal" and "high" emitters.  High emitters are
those vehicles which have emission control systems which are malfunctioning in some way, and are
producing average emission levels which are considerably higher than the overall mean emission
levels; the threshold for defining a high emitter for CO is 3.0 times the intermediate life (50,000
mile) certification emission standard.   The remainder of the fleet are  considered to be properly
functioning, and are considered normal  emitters; by definition, these vehicles are below 3.0 times
the intermediate life certification standard.  It is important to note that all pollutants are considered
independently when determining whether a vehicle is a high emitter. Thus, a vehicle could be a high
CO emitter, but a normal HC emitter.

       Although the segregation of vehicles into the "normal" and "high" categories (and  their
thresholds) is a somewhat arbitrary modeling method, the concept that average in-use emissions are
driven by a group of vehicles emitting well above the applicable standard is supported by data from
many years of EPA vehicle test and repair programs, as well as the bimodal nature of emission
control technology functioning (e.g., EGR valves, air pumps, and oxygen sensors generally function
correctly or not at all).  Two important assumptions in the development of BERs for Tier 1 and later
    M6EXH009.WPD                      4                                May 20, 1999

-------
vehicles are a) the rate at which vehicles malfunction and become high emitters is independent of
the certification standard level, and b) the average emission levels for high emitters becomes higher
relative to the standard as the certification standard becomes lower.

       The overall fleet emission factor is computed as a weighted average of the high and normal
emitters.  Figure lisa general graphical view of the concept with the y-axis representing emissions
in grams per mile (grams for start emissions), and the x-axis representing mileage and/or age. Age
and mileage are related in that a specific mileage is associated with each age.  Three lines are shown
in Figure 1 which show a) the average or basic emission rate, b) the normal emitter emission rate,
and c) the high emitter emission level.
                                        Figure 1
               GENERAL EMISSION FACTOR SCHEMATIC
        EMISSIONS
                                         MILEAGE
       The basic emission rate is shown as Line A.  This line represents the average emissions of
the fleet as a function of both the normal vehicles and the high emitting vehicles.

       Line B represents the average emissions of the normal-emitting vehicles.  These are the
vehicles which have emission control systems which are generally performing as designed. The line
is shown as a linear function of mileage and/or age to reflect the gradual deterioration that normal
vehicles experience, primarily due to catalyst degradation over the life of the vehicle. It was derived
from a least squares linear regression of emissions versus mileage.

       Line C represents the average  emissions of high-emitting vehicles.   This line is a flat
horizontal line because emissions from these vehicles do not appear  to be a strong function of
mileage and/or age, based on previous analysis of Tier 0 data.  The underlying phenomena expressed
    M6EXH009.WPD
May 20, 1999

-------
here is that emission control malfunction will lead to high emissions regardless of vehicle mileage;
as discussed  in subsequent sections, what changes  as the vehicle ages is the probability of
malfunction, rather than the emission levels resulting from a malfunction.

       Line A represents the weighted average of lines B and C, based on appropriate weighting
factor for normal and high emitters.  On a fleet-wide basis, these weighting factor represent the
fraction of high emitters in the fleet, as a function of vehicle age; on a per-vehicle basis, this
weighting factor can be considered to be the probability the vehicle will be a high emitter at a given
age. This weighting factor can be derived at any given vehicle age A by transforming Equations 1
and 2 into Equations 3 and 4.

Where:

Highs = fraction of High emitters, age =A
Normals = fraction of Normal emitters, age = A
AVE is the average emission rate, age = A
High_ave is the high emitter emission average (independent of age)
Norm_ave is the normal emitter emission average, age = A

       Highs + Normals = 1                                                Eqn 1

and

       AVE = High_ave * Highs + Norm_ave * Normals                      Eqn 2

Solving for the variables Highs and Normals produces:

       Highs = (AVE - Norm_ave) / (High_ave - Norm_ave)                  Eqn 3

       Normals = 1 - Highs                                                Eqn 4
3.0    CO Emission Methodology



3.1    Normal Emitter Emission Level

3.1.1   Overview

       This section discusses the methodology and assumptions used to generate the basic CO
emission factors and CO I/M credits for FSVs. A brief overview of the methodology is as follows:

    M6EXH009.WPD                      6                                May 20, 1999

-------
1.      Normal emitter running and start CO emission levels were calculated from 1994 and later
       vehicle FTP test results collected in an EPA test program. These data were used to calculate
       the zero mile emission level of normal vehicles.

2.      CO emission deterioration of 1994+ model year normal emitters was assumed to be the same
       as the deterioration of the 1988-93 model year PFI normal emitting vehicles.

3.      The average CO emissions of a 1994+ High emitter was assumed to be the same as the
       average CO emissions from a 1988-93 model year High emitter.

4.      The fraction of high emitters in the fleet for NO OBD / NO I/M, and OBD and I/M cases
       were calculated.

5.      The OBD and I/M assumptions were applied to generate a 'with I/M' emission line.

6.      The ratio of certification standards between Tier 1 and ULEV was calculated, and applied to
       the zero mile normal emitter CO emission level to reduce the zero mile level for ULEVs.

7.      A summary of the important parameters such as normal emitter zero mile and deterioration
       rates, high emitter emission levels and after repair emission levels are shown in Table A-l.

       Each of these  steps is discussed in detail in the sections below.
3.1.2   Normal Emitter Data
       The data sample used to develop the normal emitter emission levels consisted of test results
froml28 cars and 108 light trucks with model years 1994 through 1998.  It was collected by EPA
as part of its regular emission factor testing program. This program recruits in-use vehicles from the
general public, and subjects them to a battery of tests that include the FTP. This sample is not truly
random, because of the recruitment process used by EPA and  the potential for outliers (High
emitters) to self-select themselves out of the program. This self selection bias has been thought to
exist in past test programs, and can lead to understating of average emission levels.  In this case
however, the effect of any selection bias should be quite minor due to the young age of the vehicles,
the low emission level of the vehicles (only normal emitters were selected), and the difficulty of the
motorist to create a bias by determining outlier status in advance  of recruitment and testing.  As a
result, this sample is thought to be reasonably representative of in-use emission behavior of vehicles
with low mileage.

       This sample contained a mixture  of Tierl and TierO certified vehicles (mostly Tierl).
Unfortunately, it contained no TLEV, LEV or ULEV certified vehicles because these are not
currently present  in the regular fleet.   Thus, in the MOBILE6 model, LEV  vehicles will be

    M6EXH009.WPD                      7                                 May 20, 1999

-------
considered to be equivalent to Tierl vehicles in terms of CO emissions.  ULEVs will be given a
slightly different treatment since their certification standard limits are half (more stringent) than
those of Tierl and LEV vehicles.

       The EPA data sample was split into normal emitters and high emitters based on FTP CO
emission levels. Vehicles with CO emission levels less than three times their certification level were
judged to be normal emitters. Because most of the tested vehicles were new, virtually the entire
sample consisted of normal emitters. The FTP CO emissions as  a function of mileage are shown
graphically in Figures 2 and 3 for the car and truck data samples.

       The approach of reducing only the normal emitter emission levels between TierO, Tierl and
LEV is fairly conservative.   It assumes only small CO emission benefits will  accrue  due  to
technology changes designed to reduce HC emissions.  It also assumes that high CO emitters
(vehicles with malfunctions)  will have as high CO emissions as the high TierO CO emitters.
    M6EXH009.WPD                      8                                 May 20, 1999

-------
                       Figures 2 and 3
         1994+ MY LDV Normal Emitters
IU.UU
9nn i
.UU '
8nn i
~7 nn i
/ .UU '
6nn i
.UU '
5nn i
:*""*• A nn •
C 4.UU1
O) o nn i
• — • o.UU1
^j o nn i
o
n i nn i
L-l- 1 . UU '
LL- 0.00.







!••
-jf^l
.•'°Uod






o
•
• . * 	 =-*
"• _ f
$? *%
> o"° •



•
•
•
•
^tr-~
/>'







— — -—
•• »
%
i
•







_-- — —



MODEL YR
1998
0 1997
• 1996
1995
• 1994
Total Population
               20000
         ODOMETER
40000
60000
80000
100000
         1994+ MY LDT1 & LDT2 Normal Emitters
^U.U1
18.0 '
ID.U '
14.0 '
Iz.U '
IU.U '
p o.U '
O A ni
o 40:
n o n i
LL 0.0 !






•
2 •
.«*
A
'^






O
>
'*&
\ V





•

•
_— — —
•







•
	 — '
•
*•


I




_- — • — •

A







. — • — — '








	 • —









—- -—-



MODEL YR
" 1999
1998
0 1997
• 1996
" 1995
• 1994
Total Population
        0   25000 50000 75000 100000 125000 150000 175000 200000




         ODOMETER
M6EXH009.WPD
                              May 20, 1999

-------
3.1.3   Tierl and LEV Vehicles

       The goal of this analysis was to use the 1994 and later vehicle data to develop the basic CO
emission factors for the normal emitting Tierl, LEV and ULEV vehicles. Originally, a simple least
squares regression of the emission results versus mileage was to be used. This would produce a zero
mile emission level and a deterioration rate versus mileage that could be applied in the MOBILE6
model.

       Vehicle data  from the FTP cycle were available, and were separated into running LA4
(Running emissions) and start emission using  the  method  described in EPA document "The
Determination of Hot Running Emissions from FTP Bag Emissions"(M6.STE.002).  The running
and start emission data  were least squares regressed versus mileage to produce the zero mile
emission level and deterioration rate (slope) for normal  emitters. These values are  shown in
Appendix B in the Statistical Diagnosis Section.  The results from this regression were not used in
the MOBILE6 model. The zero mile emission values were not used in the analysis because the
running emissions were found to be statistically not significant at a 95% confidence level or even
an 80% confidence level.  The  CO start emission  zero mile regression level  was statistically
significant, but was not used to keep consistency between the running and start emissions. The
regression slope was not  used to predict the deterioration of the Tierl and later vehicle because of
the relatively low mileage levels of the vehicles in the sample. Its use would have required complete
extrapolation above 80,000 miles for cars and 100,000 miles for trucks.

       Instead the data were sub-grouped into a sample of normal emitters which had  odometer
readings less than 25,000 miles,  and the mean CO values for each car and truck sub-group were
determined. These mean  CO values were used as the zero mile emission level of the Tierl and LEV
normal emitting vehicles. The values are shown below in Equations 5a-d and 6a-d, and in summary
form in Table A-l, and in statistical output form in Appendix C. For comparison, the difference
between the zero mile CO emissions obtained from the two  methods is fairly slight in absolute
emission terms. For example, for running LA4 CO emissions from cars the regression  zero mile
level is 0.117 g/mi CO, and the mean CO of vehicles with less than 25,000 miles is 0.282 g/mi. This
compares with a value of 0.48 g/mi running LA4 CO for the 1988-93 model year PFI (ported fuel
injection) vehicles.

       The CO deterioration rate for the 1994+ model year vehicles was assumed to be the same as
the deterioration rate for the 1988-93 model year  PFI vehicles  (see EPA report - MOBILE6
"Inspection / Maintenance Benefits Methodology for 1981-93  Model Year  Vehicles"). This
assumption was made because EPA believes that the broader mileage range and larger sample size
of 1988-1993 PFI vehicles better represents fleet CO deterioration than the newer low mileage
sample. The combined effect of using the mean CO emission level of low mileage vehicles in the
sample to represent the zero mile emission level of the FSVs, and the deterioration rate of the 1988-
93 model year PFI vehicles to represent the in-use deterioration estimate of FSVs is to: (1) assume
that the lowering of HC standards in response to Tierl and later requirements (CO standards were
not changed) will reduce new vehicle CO emissions slightly, but (2) that the long term deterioration
levels will remain the same as the 1988-1993 model year vehicles.

    M6EXH009.WPD                      10                                May 20, 1999

-------
       Equations 5a-d are the equations used to model the CO running emissions from 1994 and
later model year normal emitters.  The variable 'odom' is in units of ten-thousand miles. Equation
5a shows value of the normal emitter CO running emissions from cars; Equation 5b is the analogous
one for trucks.  Both are in units of grams per mile. Equation 5c shows value of the normal emitter
CO start emissions from cars; Equation 5d is the analogous one for trucks.  Both start emission
equations are in units of grams per start.
Running Cars:        Norm_Ave(g/mi)    =      0.2821 + 0.2293 * odom            Eqn 5a

Running Trucks:      Norm_Ave(g/mi)    =      0.3219 + 0.2678 * odom            Eqn 5b

Start Cars:           Norm_Ave(g/strt)    =      15.176 + 0.0703 * odom            Eqn 5c

Start Trucks:         Norm_Ave(g/strt)    =      21.884 +0.1680 * odom            Eqn 5d


3.1.4   ULEV Vehicles

       The normal emitter running and start CO emission levels of ULEVs are the same as the
normal emitter running and start emission levels of the Tierl and LEVs, except the zero mile levels
are 50 percent less. The 50 percent reduction was used because the ULEV standards are numerically
half of those of the LEVs.  The actual values are shown in Equations 6a through 6d.


Running Cars:        Norm_Ave(g/mi)    =      0.1411 + 0.2293 * odom            Eqn 6a

Running Trucks:      Norm_Ave(g/mi)    =      0.1609 +0.2678 * odom            Eqn 6b

Start Cars:           Norm_Ave(g/strt)    =      7.588 + 0.0703 * odom             Eqn 6c

Start Trucks:         Norm_Ave(g/strt)    =      10.942 +0.1680 * odom            Eqn 6d
       The Truck4 category also has a different certification standard than the Truck2 and Trucks
vehicles.  Since no Truck4 data were available, its zero mile normal emitter CO emission level was
determined by applying the ratio of the Truck4 and Truck3 certification standards to the Truck2/3
zero mile CO emission value.  The values for the heavier Trucks are shown in Table A-l

       The rationale for reducing the zero mile level of the ULEVs proportionally to the standard
is that basic emission levels for properly operating vehicles should receive some benefit of reduced
standards for lower standard levels.  This approach assumes that brand new normal emitters for the
ULEV standards will on average achieve the same compliance margin ("headroom") as the normal
emitters observed in the sample.

    M6EXH009.WPD                      11                               May 20, 1999

-------
3.2    High Emitter Emission Level

       High Emitter basic CO emission rates (BERs) are meant to estimate emissions from vehicles
that significantly exceed their certification standards due to malfunctioning emission control systems.
A key assumption in the development of high-emitter CO BERs is that, as HC and NOx emission
standards are lowered for Tier 1 and LEV classifications (CO standards are not changed until ULEV),
emission levels for CO high emitters will not change relative to the performance of the TierO High
CO emitters.  The rationale for this assumption is since emission control equipment on these high
emitters will be degraded or completely malfunctioning, they will emit largely independent of their
certification standard level. To illustrate,  compliance with the 3.4 g/mi CO standards is primarily
a function of the effectiveness of a vehicle's fuel delivery system control. Thus, vehicles which have
become High CO emitters due to a loss of fuel delivery system control are expected to have similar
CO emission performance regardless of whether they are  TierO,  Tierl,  or LEV certified.  This
assumption may not apply to ULEV certified vehicles, since their certification level is reduced by
50 percent from the LEV level, and different technologies may be used to control emissions on these
vehicles. Nevertheless, this assumption is being extended to the ULEVs since there are no in-use
data available on high emitting ULEVs.

       The FTP high emitter average CO emission levels used for the 1994 and later model years
are the same values used for the 1988-93 model year PFI vehicles (cars and trucks have separate
values). These values for Running and Start emissions for cars and trucks are shown in Equations
7a through 7d. These values will be used for all 1994 and later model years including Tierl, LEV
and ULEV vehicles.
Running Cars:       High_Ave(g/mi)     =      36.106                            Eqn 7a

Running Trucks:     High_Ave(g/mi)     =      33.283                            Eqn 7b

Start Cars:           High_Ave(g/strt)     =      38.060                            Eqn 7c

Start Trucks:         High_Ave(g/strt)     =      83.862                            Eqn 7d
       These extrapolations from TierO vehicles could not be reliably verified from the relatively
small 1994+ data sample, since only one High emitter was present in the 1994-1998 model year
sample. However, this vehicle's FTP emissions were quite high at 215 g/mi CO.
3.3    High Emitter Fractions - No OBD / No I/M

       This section describes the fraction of high emitters in the fleet given the base case of no OBD
and No I/M. The numerical values are the same as the no I/M high emitter fractions of the 1988-93
PFI vehicles, and were derived from the general Equation 3 with TierO values. The rate at which

    M6EXH009.WPD                      12                               May 20,  1999

-------
vehicles become high emitters under the No OBD / No IM scenario was also assumed constant for
all standard classes.  The age-based high emitter fractions for Tierl and later vehicles are shown in
Appendix A.  These values are referred mathematically as HighBASE(i) in subsequent calculations.

       Reduced CO certification standards are not expected to influence the rate at which emission
control technology malfunctions, because a) manufacturer's design and durability practices are not
expected to differ between TierO and later standards, and b) many cases  of emission  control
degradation and/or malfunction are owner-induced, outside the manufacturer's liability for in-use
emission performance, and unlikely to change due to the new standards.  Separate rates of emission
control technology malfunction were used for LDVs and LDTs, only because separate car and truck
rates were used for the 1988-93 model year PFI vehicles.
3.4    High Emitter Fraction - OBD and OBD I/M
       Separate BERs were developed for all standards and vehicle classes to account for the effects
of OBD and OBD-based I/M programs.  The methodology used to account for these programs is
based on reducing the fraction of high emitters in the fleet from the No OBD/No EVI case.  Thus,
emission levels for normal and high emitters were not changed under these programs, only the
fraction of highs in the fleet. This methodology introduces a new category of vehicle into the fleet:
"Repaired" emitters.  These vehicles are high emitters that are flagged by an OBD  system and
undergo successful repair. For the OBD-only and OBD/EVI cases, these vehicles are treated distinctly
from normal and high emitters.

       OBD effectiveness  is defined by three parameters: a) the probability the OBD  system will
detect a failure (MIL-on Rate), b) the probability an owner will respond to a MIL-on (Response
Rate), and c) the average after-repair emission level for responding vehicles (Repair Level). At the
time of this analysis, sufficient in-use data were not available to empirically determine in-use
patterns for these parameters.  Thus, assumptions were developed for the projected likelihood of
malfunction detection, owner response and repair level.
3.4.1   MIL-on Rate

       For all vehicle  classes and standards, it is  assumed that OBD will detect (i.e., set an
appropriate code and illuminate the MIL) 85 percent of the CO high emitters. Because high emitters
are defined independently for all pollutants, this response rate is assumed to apply equally to all
pollutants. The remaining 15 percent of the high emitters  will not be identified, and thus will
remain in the fleet as high emitters. No deterioration in the ability of the OBD system to correctly
identify high emitters is assumed. Because this parameter is solely dependent on the vehicle's OBD
system, it is the same for I/M and non-I/M areas.

    M6EXH009.WPD                      13                                May 20, 1999

-------
3.4.2   Response Rate

       In order to obtain emission reductions from a vehicle equipped with an OBD system, not only
must the system correctly identify the vehicle, but the motorist must also respond to the MIL and
take corrective action in a timely manner. A key assumption for this analysis is that the response rate
in OBD-based I/M areas (where repair is required) will be much higher than in non-I/M areas. In
OBD-based I/M areas, the MIL-on response rate is assumed to be 90 percent over the entire life of
the vehicle. Response was not set at 100 percent to account for waivers from, or evasion of, or delay
in responding to the I/M program.

       For non-DVI areas, OBD response is assumed to be a function of vehicle warranty.  It is
assumed that an owner is much more likely to respond to a MIL-on when repairs will be paid for by
the manufacturer. Three mileage bins were therefore developed: 1) 0 through 36,000, the standard
bumper-to-bumper warranty period; 2) 36,001  to  80,000, for which federal law mandates that
catalysts and electronic control modules (ECMs) remain under warranty; and 3) above 80,000, for
which no warranty is in effect (extended warranties are not accounted for in this methodology).

       Under 36,000 miles, it is assumed that 90 percent of MIL-on vehicles will be repaired. This
is based on the judgment that for new vehicles still under warranty, owners will have little hesitation
in addressing a MIL-on. The 10 percent loss accounts for a delay in the response rate and the small
percentage of owners who will not respond to a MIL-on even with the warranty incentive.

       Between 36,000 miles and 80,000 miles, it is assumed that 10 percent of MIL-on vehicles
will be repaired. This response rate is greatly reduced from the pre-36,000 mile level to account for
the discontinuation of warranty coverage on several emission-related components (e.g. secondary
air,  EGR, oxygen sensors, fuel injectors),  and reduced willingness of owners to make emission-
related repairs on an aging vehicle in the absence of an I/M program.

       Above 80,000 miles, it is assumed that no MIL-on vehicles will be repaired.  This assumption
reflects the end of warranties, the lower economic value of the vehicle,  and the (further) reduced
willingness of owners to make emission-related repairs in the absence of an I/M program.
3.4.3   Repair Level

       The emission level after an OBD-induced repair is assumed to be 1.5 times the applicable
50,000 mile certification standard.  This creates a third emitter category - "repaired" emitters.
Repaired emitters are assumed to have constant emissions at the repaired level, although a subset of
these vehicles "migrate" back to the high emitter category.  The estimated post-repair level of 1.5
times the standard is the required threshold for illuminating the MIL.  This level was chosen as the
after repair emission level (as opposed to a lower value) after considering  three issues.  First,
although many vehicles will likely be repaired  below this level, others undergoing repair  will

    M6EXH009.WPD                       14                                 May 20, 1999

-------
continue to be higher than 1.5 times the standard due to synergistic effects of multiple malfunctions
which  considered independently would not trigger the MIL.   Second, the OBD malfunction
threshold requirement of 1.5 times the standard applies to the applicable full useful life standard;
thus, repairing to 1.5 times the 50,000 mile standard is below the required threshold between 50,000
and 100,000 miles (in fact, ARB's most recent modification to OBD for LEVs requires detection at
1.75 times the full-useful standard (100,000 mile standard) at all mileages, thus increasing the gap
between the repair level and MIL threshold).  Third, these vehicles are assumed not to deteriorate
for the remainder of their life, unless they migrate into the high emitter category. Thus, the repair
levels can be lower than average normal emitter emissions at higher mileages.

3.4.4   High Emitter Fractions

       Equations 8 through 10 were used to calculate the high emitter growth rate under the OBD
and OBD-based I/M scenarios (HighOBD and HighhOBDIM). Overall, the high emitter fraction in a
given year is a function of a) the number of high emitters in the previous year, b) the base high
emitter "growth rate" in the absence of OBD or I/M, and c) the OBD effectiveness assumptions
outlined in Section 3.4.1 and 3.4.2.  The subscript 'i' is the vehicle age. High(O) is assumed to be
zero. MOBILE6 will assign a value of 'odom' for each age 'i'.

       Nonhigh(i)           =      1.0-HighBASE(i)                                 Eqn 8

       Delta_High(i)        =      HighBASE(i)-HighBASE(i-l)                        Eqn 9

       Growth_High(i)      =      Delta_High / Nonhigh(i)                         Eqn 10
HighOBD(i)  = HighOBD(i-l) + [(l-OBD)*MIL*Growth_High(i)*(l-HighOBD(i-l))] +
             [(l-MIL)*Growth_High(i)*(l-HighOBD(i-l)]                           Eqn lla
HighOBDIM(i)  =  HighOBDIM(i-l) + [(!-OBDIM)*MIL*GrowthJIigh(i)*(l-HighOBDIM(i-l))] +
                [(l-MIL)*Growth_High(i)*(l-HighOBDIM(i-l)]                       Eqn 1 Ib
Where:

HighOBD(0)   =     0.0

HighOBDIM(0)  =     0.0

'MIL'        =     0.85

'OBD' is the OBD response rate; 0.90/0.10/0.0 for mileage bins (0 - 36K), (36K - 80K), and (80K+),
respectively.  These are used in conjunction with Equation lla.

    M6EXH009.WPD                       15                                May 20, 1999

-------
'OBDIM'  =  0.90

It is the effectiveness of the OBD I/M program at identifying High emitters, and is set at 90%. It is
used in conjunction with Equation 1 Ib.
       HighOBD is the fraction of highs in the vehicle fleet equipped with OBD if no I/M program
is present. HighOBDIM is the fraction of highs in the vehicle fleet equipped with OBD if an OBD I/M
program is present.

       A further description of Equations 8 through 1 Ib for a given vehicle age is listed below. The
fraction of high emitters is the sum of the following:

1.      The number of highs from the year before, plus;

2.      The number of MIL-on highs added in that year due to OBD non-response (a function of
       "non-response" rate, MIL-on rate, and the high emitter growth rate applied to the available
       pool of "non-highs" - normals and repaired vehicles), plus;

3.      The number of highs added in that year that the OBD system did not detect (a function of
       MIL-"off'  rate and the high emitter growth rate applied to the available  pool of "non-
       highs"). The high emitter growth rate for a given year is the absolute increase in high
       emitters under the No OBD / No EVI case from the previous year divided by the fraction on
       Nonhigh - i.e., the available pool of vehicles which can become high emitters.
3.4.5   Repaired Fractions

       Once the high emitter fraction is calculated for the OBD or OBD/EVI cases, the fraction of
repaired emitters can be calculated as the difference between the fraction of high emitters that would
occur without OBD or I/M (HighBASE)and the fraction of high emitters with OBD and/or I/M from
Equation 11. In equation form,

Repaired(i)    =     HighBASE (i) - HighOBD (i)                                      Eqn 12
3.4.6   Normal Fractions

       The normal emitters are those vehicles which are not high or repaired emitters. At a given
age, the rate of normal emitters remains constant between the No OBD / No I/M, OBD-only and
OBD/EVI case; only the fraction of high emitters decrease, directly replaced by repaired emitters.
However, over time (as age increases) the fraction of normal emitters decreases due to the growth
in the fraction of high emitters.
    M6EXH009.WPD                       16                                May 20, 1999

-------
       The CO emission emitter fractions for normal, high and repaired emitters for the OBD Only
and OBD/IM cases are shown in Appendix A by vehicle class.
3.5    High Emitter Fraction - Exhaust I/M and OBD and Exhaust I/M

       For Tierl and later vehicle standards classes, this option contains three possibilities which
depend on vehicle model year and the presence of an exhaust test I/M program and an OBD I/M
program. An exhaust test I/M program refers to any type of exhaust I/M test such as the Idle test,
the Two Speed Idle test, the ASM test or the EVI240 test.

3.5.1   Exhaust I/M/NO OBD

       For Tierl and later vehicles this combination can only occur in the 1994 and 1995 model
years where OBD is not present on vehicles, and an exhaust I/M program is in place.  Since this
scenario is really just an extension of the 1981-93 model year (no OBD is present), the same I/M
treatment of high emitters before and after I/M as is used for the 1981-93 model years will be used.
The reader is encouraged to refer to EPA document M6.EVI.001 for more details.

3.5.2   Exhaust I/M with OBD

       For Tierl and later vehicles this combination can only occur on 1996 and later model year
vehicles equipped with an OBD system. Under this scenario the I/M program conducts an exhaust
I/M test, but does not conduct an OBD I/M test. This is a likely scenario until at least calendar year
2001 when the EPA OBD I/M guidance is expected to be implemented. The fraction of fleet high
emitters remaining after the I/M test under this scenario is the same as that presented in Section
3.5.1. It is a function of the stringency of the exhaust I/M test, and uses the same general treatment
discussed in EPA document M6.IM.001. This scenario differs from the one presented in Section
3.5.1 in the 'before'I/M failure rates or base failure rates. These are lower than the exhaust I/M / No
OBD because of the presence of the OBD  systems on the vehicles.  They are calculated using
Equation 11 a.
3.5.3   Exhaust I/M and OBD I/M

       This scenario can exist for 1996 and later model year vehicles equipped with an OBD system,
and subject to both an OBD I/M inspection and an exhaust I/M test. If an I/M program should
choose to implement this type of program (both OBD and exhaust I/M on the same vehicles), the
after I/M high emitter rate will be the same as the after OBD I/M high emitter rate calculated in
Equation 1 Ib. This assumes that a combined exhaust I/M and OBD I/M program will be as effective
as the OBD I/M only program.  This is not an unreasonable assumption since the  OBD I/M only
program is assumed to identify 90 percent of the high emitters, and the exhaust  I/M  tests have
identification rates which are in most cases considerably lower.
    M6EXH009.WPD                      17                               May 20, 1999

-------
4.0    Basic Emission Factor Calculations for Various I/M and OBD Scenarios

       This section describes how the information described in the previous sections is consolidated
to compute emissions for each of the possible OBD and I/M scenarios.
4.1     NO OBD and NO I/M

       The No OBD and No I/M emission level for the Tierl and later vehicles is the basic CO
emission level. For CO emissions, it is essentially the same value as the  1988-93 model year PFI
vehicle CO emission level, except basic emission rates for normal emitters ar calculated with a lower
zero mile emission level. The No OBD and No I/M emission level is the basis for the subsequent
CO emission levels; however, it only appears directly in the MOBILE6 CO output for the 1994 and
1995 model year vehicles with no installed OBD  system and no I/M program.

       The average No OBD and No I/M in-use running and start CO emission levels are calculated
for  each vehicle, standard and age  combination using  the general Equation 2 repeated here as
Equation 13. The parameters in Equation 13 such as "High_ave" and "Norm_ave" are calculated
from Equations (5a-d or 6a-d) and 7a-d. The fraction of high emitters which is used in Equation 13
is the value of HighBASE(i)) found in  Appendix A in Table A-2 in column "Base". Figures 4 and 5
show the CO emission levels for the cars and trucks for Tierl and LEV, and ULEV vehicles. The
line on each figure labeled "Base" is the NO OBD and NO I/M CO emission line shown in terms of
FTP emissions (running and start are combined together).

       Base© =      HighBase(i) * High_ave + Normal(i)*Norm_ave(i)                       Eqn 13
4.2    OBD and NO I/M

       The OBD and No I/M emission level for the Tierl and later vehicles is used in MOBILE6
as the CO emission level for 1996 and later vehicles if no I/M program is present. It differs from the
NO OBD and NO  I/M case because of the ability of OBD to identify high emitters and induce the
vehicle owner to repair them absent an I/M program. Calculation of the average CO emission rate
for OBD and NO I/M uses Equation 14. It is similar to the calculation described above in Section
4.1 for the No OBD and No I/M case, and uses the same values for the normal and high emitter
emission level average parameters.  The differences are the use of HighOBD rather than HighBASE for
the high emitter fraction parameter, and the after repair emission term. The after repair emission
term is the product of the after repair emission level (Rep_ave) and the fraction  of high emitters
which are repaired ("Repaired" from equation 12). The value of "Rep_ave" is the CO emission level
which is 1.5 times  the applicable 50,000 mile FTP certification standard (See Section 5.0 for more
details on Rep_ave).
    M6EXH009.WPD                      18                               May 20, 1999

-------
OBD(i) =     HighoBD(i) * High_ave + Normal©*Norm_ave(i) + Repaired(i) * Rep_ave           Eqn 14

       The FTP CO emission levels for cars and trucks and Tierl and LEV and ULEV for the case
of OBD and NO I/M are shown in Figures 4 and 5.  These are indicated by the dotted line and are
under the legend heading of "OBD".  In general, they are very similar to the "Base" NO OBD
emission levels. This reflects the relatively small impact of OBD on overall fleet CO emissions in
the absence of an I/M program.
4.3    OBD and OBD I/M

       The OBD and OBD I/M emission level for the Tierl and later vehicles is used in MOBILE6
as the CO emission level for 1996 and later vehicles if an OBD I/M program is present.  It differs
from the OBD and NO I/M case because of the ability of OBD I/M to identify and force the vehicle
owner to repair a high emitter. Contrasting the two, the OBD and NO I/M case can be viewed as
a voluntary OBD I/M program, and the OBD and OBD I/M case can be viewed as a mandatory I/M
program. Calculation of the average CO emission rate for OBD and NO I/M is done using Equation
15. It is similar to the calculation done in Equation 14 for the No OBD and No I/M case, and uses
the same values for the normal and high emitter emission level average parameters. The difference
is the use of HighOBDIM rather than HighOBD for the high emitter fraction parameter.
OBDIM(i)     =      HighoBDIM(i) * High_ave + Normal(i)*Norm_ave(i) + Repaired(i) * Rep_ave   Eqn 15
       The CO emission levels for cars and trucks and Tierl and LEV and ULEV for the case of
OBD and OBD I/M are shown in Figures 4 and 5.  These are indicated by the line with the circles
on it and are under the legend heading of "OBD+EVI". In general, these emissions are considerably
lower than the "Base"  and "OBD" lines.  This difference reflects the relatively large impact of a
mandatory OBD I/M program on overall  fleet CO emissions after the main warranty ends.  The
difference between the OBD line and the  OBD+EVI line is the I/M benefit.
4.4    Exhaust I/M / NO OBD

       As mentioned in Section 3.5.1, for Tierl and later vehicles this combination can only occur
in the 1994 and 1995 model years where OBD is not present on vehicles, and an exhaust I/M
program is in place. Since this scenario is really just an extension of the 1981-93 model year (no
OBD is present), the same general calculation of before and after I/M CO emissions levels will be
done for the Tierl  and later vehicles as was done for the 1981-1993 model year vehicles.  The
"before" I/M CO emission levels are those calculated using Equation 13 and labeled as "Base" in
Figures 4 and 5. The "after" I/M CO emission levels "ExhBase" are calculated using Equation 16.
Except for the use of the HighBASE variable, this is the same equation that is used to calculate the after
I/M emission levels for the  1981-1993 model year vehicles.  However, for more details  on the

    M6EXH009.WPD                      19                               May 20, 1999

-------
development of 1981-1993  model years I/M credits, the reader is encourage to  refer to EPA
document M6.IM.001.
ExhBase(i) = Norm_ave*(l-HighBASE) + High_ave*HighBASE*(l-IDR) + HighBASE*IDR*W*High_ave*RW +
              Norm_ave*R*HighBASE*IDR*FIX + High_ave*HighBASE*IDR*NC                  Eqn 16

IDR is the identification rate of high emitters using an exhaust emission test.
R is the after repair emission level of vehicles repaired to pass an exhaust I/M test.
Fix is the fraction of vehicles which are repaired.
NC is the fraction of vehicles which are in non-compliance following their I/M test
W is the fraction of vehicles which receive a cost or other type of waiver.
RW is the after repair level of the vehicles which get waivered. It is shown as a fraction of the high emitter level.

4.5    Exhaust I/M with OBD

       As mentioned in Section 3.5.2, for Tierl  and later vehicles this combination can only occur
on 1996 and later model year vehicles equipped with an OBD system.  Under this scenario the I/M
program conducts an  exhaust I/M test, but does not conduct an OBD I/M test.  This is a likely
scenario until  at least calendar year  2001 when the EPA OBD I/M  guidance is expected to be
implemented.  The only difference between this scenario and the one discussed in Section 4.4 above
is the different "before" I/M high emitter fractions. For example, this case uses the value of HighOBD
and the previous case uses the value of HighBASE for high emitter fraction. This difference can be
seen by comparing Equation 16 with Equation 17.  Another comparison is that this case uses the
before I/M CO emission line labeled in Figures 4 and 5 as "OBD", and the previous case of Exhaust
I/M and NO OBD uses the line labeled as "Base".

ExhOBD(i) = Norm_ave*(l-HighoBD) + High_ave*HighoBD*(l-IDR) + HighoBD*IDR*W*High_ave*RW +
              Norm_ave*R*HighOBD*IDR*FIX + High_ave*HighoBD*IDR*NC                   Eqn 17
4.6    Exhaust I/M and OBD I/M
       As mentioned in Section 3.5.3, this scenario can exist for 1996 and later model year vehicles
equipped with an OBD system, and subject to both an OBD I/M inspection and an exhaust I/M test.
If an I/M program should choose to implement this type of program (both OBD and exhaust I/M on
the same vehicles), the before and after I/M emission levels will be the same as the before and after
OBD I/M emission level.  This emission level is shown in Figures 4 and 5 under the legend label
"OBD+EVI". This is the lowest CO emission level which can be obtained for a given standards class,
vehicle type, and I/M program. It assumes that the additional exhaust I/M program in conjunction
with an effective OBD program is redundant.
    M6EXH009.WPD                       20                                 May 20, 1999

-------
5.0    OBD Repair Effects

       Separate running and start CO emission factors for normal and high emitters were developed
directly from start and running test data.  However, development of the after  OBD repair CO
emission levels are more problematic because there are no after repair test data, and because of the
OBD requirement for MIL illumination at 1.5 times a vehicle's FTP standard (see  Section 3.4.3 for
a description and justification for the 1.5 times standard after repair level). The requirement of MIL
illumination if the vehicle's emissions exceed 1.5 times the standard is a difficulty in the context of
this analysis because it is expressed in terms of FTP emissions rather than separately in terms of the
running or start emissions which are used in MOBILE6.

       Two simple multiplicative factors to express the 1.5 times FTP standard MIL illumination
in terms of running and start emissions were developed based on a sub-sample of 1994 and later
model  year test results.  The sub-sample  consisted of 17 cars  and trucks  which had FTP CO
emissions  greater than 5.0  g/mi  (approximately  1.5 times standards) and less than 3.0 times
certification CO standards. The factors were developed by taking the ratio of the running and/or start
emissions and the FTP emissions. They are shown below as ratios of running emissions over FTP
emissions, and start emissions over the FTP emissions. The values are:

       Running CO (g/mi)/FTP CO(g/mi)         =      0.338                     Eqn 18

       Start CO(grams/start) / FTP CO (g/mi)      =      4.149                     Eqn 19
       The approach of using a sub-sample of 17 vehicles rather than the overall 1994+ sample was
selected because the ratios are a function of the FTP CO emission level. The overall sample
contains mostly low normal emitters rather than the sub-sample vehicles which are close to the 1.5
times certification standards emission level that is being modeled. The limited data and simple
analysis suggests that the low emitters have smaller running / FTP ratios and larger start / FTP ratios
than the vehicles which are close to the 1.5 times the standards level. For example, use of the entire
sample would produce a running/FTP CO ratio of 0.164 and a start/FTP CO ratio of 9.74 versus
the values of 0.338 and 4.15.  These "full sample ratios" would likely improperly raise the after
repair start emission level and lower the after repair running emission level versus the ratios based
on the sub-sample.

       For the  statistically minded, the 90 percent confidence interval  around the sub-sample
running / FTP ratio is from 0.277 to 0.398.  The corresponding sub-sample start / FTP ratio
confidence level ranges from 2.82 to 5.48.

       These ratios are used to calculate the value of Rep_ave (After repair emission level for OBD
vehicles) for start and running emissions used in Equations 14 and 15. For example, the after OBD
repair CO running emission level for a Tierl or LEV car is:
    M6EXH009.WPD                      21                                May 20, 1999

-------
       1.5X * High Emitter Running Fraction  *  CO Standard




       1.5 * 0.338 * 3.4 g/mi        =      1.724 g/mi running LA4 CO




and




       1.5 X *  High Emitter Start Fraction  *  CO Standard




       1.5 * 4.149 * 3.4 g/mi        =      21.160 grams CO / start
    M6EXH009.WPD                      22                                May 20, 1999

-------
Table A-l: CO Basic
Vehicle
Class
LDV/T1
LDT2/3
LDT4
Standard
Class
Tierl / LEV
ULEV
Tier 1 / LEV
ULEV
Tier 1 / LEV
ULEV
50K
Standard
(g/mi)
3.4
1.7
4.4
2.2
5.0
2.5
Mode
Running
Start (grams)
Running
Start (grams)
Running
Start (grams)
Running
Start (grams)
Running
Start (grams)
Running
Start (grams)
Emission Rates
"Normal" BER
(g/mi)
ZML
0.282
15.176
0.141
7.588
0.322
21.884
0.161
10.942
0.366
24.868
0.183
12434
DR
0.229
0.070
0.229
0.070
0.268
0.168
0.268
0.168
0.268
0.168
0.268
0 168
"High"
BER
(g/mi)
36.11
38.06
36.11
38.06
33.28
83.86
33.28
83.86
33.28
83.86
33.28
83 86
"Repaired"
BER
(g/mi)
1.724
21.160
0.862
10.580
2.228
27.383
1.114
13.692
2.532
31.118
1.266
15559
M6EXH009.WPD
23
May 20, 1999

-------
Table A-2: CO Emitter Fractions
Age
(Years)
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
LDV/LDT1
All
Normal
0.991
0.992
0.976
0.954
0.943
0.928
0.913
0.899
0.884
0.870
0.857
0.844
0.832
0.820
0.809
0.798
0.787
0.777
0.767
0.757
0.748
0.739
0.731
0.723
0.715
0.707
Base
High
0.009
0.008
0.024
0.046
0.057
0.072
0.087
0.101
0.116
0.130
0.143
0.156
0.168
0.180
0.191
0.202
0.213
0.223
0.233
0.243
0.252
0.261
0.269
0.277
0.285
0.293
OBD Only
High
0.002
0.002
0.006
0.026
0.036
0.051
0.066
0.080
0.095
0.109
0.123
0.136
0.149
0.161
0.173
0.184
0.195
0.205
0.216
0.225
0.235
0.244
0.252
0.261
0.269
0.276
Repair
0.007
0.006
0.018
0.020
0.021
0.022
0.021
0.021
0.021
0.020
0.020
0.020
0.019
0.019
0.019
0.019
0.018
0.018
0.018
0.018
0.017
0.017
0.017
0.017
0.017
0.016
OBD/IM
High
0.002
0.002
0.006
0.011
0.014
0.017
0.021
0.025
0.028
0.032
0.035
0.039
0.042
0.045
0.048
0.051
0.054
0.057
0.060
0.063
0.066
0.068
0.071
0.073
0.075
0.078
Repair
0.007
0.006
0.018
0.035
0.043
0.055
0.066
0.076
0.088
0.098
0.107
0.117
0.126
0.135
0.143
0.151
0.159
0.166
0.173
0.180
0.186
0.193
0.199
0.204
0.210
0.215
LDT2 / 3
All
Normal
0.988
0.988
0.976
0.955
0.946
0.933
0.921
0.911
0.899
0.888
0.878
0.868
0.858
0.849
0.840
0.831
0.823
0.814
0.807
0.799
0.792
0.785
0.778
0.772
0.766
0.760
Base
High
0.012
0.012
0.024
0.045
0.054
0.067
0.079
0.089
0.101
0.112
0.122
0.132
0.142
0.151
0.160
0.169
0.177
0.186
0.193
0.201
0.208
0.215
0.222
0.228
0.234
0.240
OBD Only
High
0.003
0.003
0.006
0.025
0.034
0.046
0.058
0.069
0.081
0.092
0.102
0.113
0.123
0.132
0.141
0.150
0.159
0.167
0.175
0.183
0.190
0.197
0.204
0.210
0.217
0.223
Repair
0.009
0.009
0.018
0.020
0.020
0.021
0.021
0.021
0.020
0.020
0.020
0.020
0.019
0.019
0.019
0.019
0.019
0.018
0.018
0.018
0.018
0.018
0.018
0.017
0.017
0.017
OBD/IM
High
0.003
0.003
0.006
0.011
0.013
0.016
0.019
0.022
0.025
0.027
0.030
0.033
0.035
0.038
0.040
0.042
0.045
0.047
0.049
0.051
0.053
0.055
0.057
0.059
0.060
0.062
Repair
0.00$
0.00$
0.01?
0.03'
0.04'
0.05-
0.06C
0.06?
0.07*
0.08!
0.09^
0.10C
0.107
0.11'
0.12C
0.127
0.13:
0.13$
0.14'
0.15C
0.15i
0.16C
o.ie;
0.16$
0.17'
0.17?

-------
                            Figures 4 and 5
         CO Emission Factors from Cars and Truckl Vehicles
                         LDV/LDT1 Tier! & LEV CO
                                 150        200

                            Mileage/1000
                                 -OBD '
                                        -OBD+IM
20 T
18
            50
                          LDT2/3 Tierl & LEV CO
                      100
                                 150

                            Mileage/1000
                                           200
                          •Base '
                                 -OBD'
                                        -OBD+IM
                                                      250
                                                                300

-------
                         Figures 6 and 7
     CO Emission Factors from Truck2 and Trucks Vehicles
                          LDV/LDT1 ULEV CO
           50
                     100
                                150         200

                           Mileage/1000
                        •Base '
                               -OBD '
                                      -OBD+IM
                                                    250
                                                               300
                            LDT2/3 ULEV CO
18
                      100
                                150

                            Mileage/1000

                           ;e	OBD   •
                                          200
                                                     250
                                                               300

-------
                                   APPENDIX B
                   Statistical Diagnostics for CO Running and
                           Start Normal Emitter Levels
       This Appendix contains the statistical diagnostics from the regression of the normal emitter
CO emission versus odometer.  Standard least squares regressions were done on running CO
emissions versus odometer for cars and trucks separately, and start CO emissions versus odometer
for cars and trucks separately.
           ****   MULTIPLE   REGRESSION

Regression for 1994+ cars

Listwise Deletion of Missing Data

Equation Number 1    Dependent  Variable..   HR505_CO

Block Number  1.   Method:   Enter      ODOMETER
                                                            * * * *
Variable(s)  Entered on Step  Number
   1..     ODOMETER
Multiple R           .37098
R Square             .13763
Adjusted R Square    .13079
Standard Error       .61807
Analysis of Variance
                    DF
Regression           1
Residual           126
F =
         20.10894
                Sum of Squares
                       7.68187
                      48 . 13358

            Signif F =   .0000
Mean Square
    7.68187
     .38201
Variable

ODOMETER
(Constant)
	 Variables in the Equation 	

          B        SE B     95% Confdnce Intrvl B

 1.24335E-05  2.7727E-06  6.94648E-06  1.79206E-05
     .117206      .091991     -.064842      .299254
                    Beta

                 .370985
Variable

ODOMETER
(Constant)
        T  Sig T

    4.484  .0000
    1.274  .2050
           ****   MULTIPLE   REGRESSION   ***
Listwise Deletion of Missing Data

Equation Number 1    Dependent  Variable..   STRT_CO

-------
Block Number  1.  Method:  Enter
                                      ODOMETER
Variable(s) Entered on Step Number
   1..     ODOMETER
Multiple R            .07287
R Square              .00531
Adjusted R Square   -.00258
Standard Error      8.82307
Analysis of Variance
                    DF
Regression           1
Residual           126
F =
            .67256
                 Sum of Squares
                       52.35654
                     9808.65890

             Signif F =  .4137
Mean Square
   52 .35654
   77 .84650
Variable

ODOMETER
(Constant)
	  Variables in the Equation 	

           B        SE B     95% Confdnce Intrvl B

 3.24599E-05  3.9580E-05 -4.58687E-05  1.10788E-04
   14.753554    1.313187    12.154795    17.352312
                    Beta

                  .072866
Variable

ODOMETER
(Constant)
        T  Sig T

     .820  .4137
   11.235  .0000
End Block Number
            All  requested variables entered.

-------
           ****   MULTIPLE   REGRESSION   ****


Listwise Deletion of Missing Data

Equation Number 1    Dependent Variable..   CO

Block Number  1.  Method:  Enter      ODOMETER
Variable(s) Entered on Step Number
   1..     ODOMETER
Multiple R            .41750
R Square              .17430
Adjusted R Square     .16775
Standard Error      1.23853
Analysis of Variance
                    DF
Regression           1
Residual           126
F =
         26 .59849
             Sum of Squares
                   40.80123
                  193.27996

         Signif F =  .0000
Mean Square
   40.80123
    1.53397
Variable
      -  Variables in the Equation 	

       B        SE B     95% Confdnce Intrvl B
ODOMETER    2.86548E-05  5.5561E-06  1.76595E-05  3.96502E-05
(Constant)     1.187217      .184338       .822418     1.552017
                                                                    Beta

                                                                  .417497
Variable

ODOMETER
(Constant)
    T  Sig T

5.157  .0000
6.440  .0000
End Block Number
        All requested variables entered.

-------
           ****   MULTIPLE   REGRESSION   ****


Listwise Deletion of Missing Data

Equation Number 1    Dependent Variable..   HR505_CO

Block Number  1.  Method:  Enter      ODOMETER
Variable(s) Entered on Step Number
   1..     ODOMETER
Multiple R            .34670
R Square              .12020
Adjusted R Square     .11190
Standard Error      1.05037
Analysis of Variance
                    DF
Regression           1
Residual           106
F =
         14 .48171
             Sum of Squares
                   15.97724
                  116.94662

         Signif F =  .0002
Mean Square
   15.97724
    1.10327
Variable
      -  Variables in the Equation 	

       B        SE B     95% Confdnce Intrvl B
ODOMETER    1.14258E-05  3.0025E-06  5.47314E-06  1.73785E-05
(Constant)      .173842      .150593     -.124724       .472408
                                                                    Beta

                                                                  .346696
Variable

ODOMETER
(Constant)
    T  Sig T

3.805  .0002
1.154  .2509
End Block Number
        All requested variables entered.

-------
           ****   MULTIPLE   REGRESSION    ****


Listwise Deletion of Missing Data

Equation Number 1    Dependent Variable..   STRT_CO

Block Number  1.  Method:  Enter      ODOMETER
Variable(s) Entered on Step Number
   1..     ODOMETER
Multiple R            .00946
R Square              .00009
Adjusted R Square   -.00934
Standard Error     13.38235
Analysis of Variance
                    DF
Regression           1
Residual           106
F =
           .00948
              Sum of Squares
                     1.69800
                 18983.24665

          Signif F =  .9226
Mean Square
    1.69800
  179.08723
	 Variables in the Equation 	

Variable              B        SE B     95% Confdnce Intrvl B       Beta

ODOMETER   -3.72482E-06  3.8253E-05 -7.95656E-05  7.21160E-05    -.009457
(Constant)
              21.096190
                           1.918655
                                       17.292269
                                                    24.900111
Variable

ODOMETER
(Constant)
     T  Sig T

 -.097  .9226
10.995  .0000
End Block Number
         All requested variables entered.

-------
           ****   MULTIPLE   REGRESSION    ****


Listwise Deletion of Missing Data

Equation Number 1    Dependent Variable..   CO

Block Number  1.  Method:  Enter      ODOMETER
Variable(s) Entered on Step Number
   1..     ODOMETER
Multiple R            .33380
R Square              .11142
Adjusted R Square     .10304
Standard Error      2.65967
Analysis of Variance
                    DF
Regression           1
Residual           106
F =
         13 .29152
                            Sum of Squares
                                  94.02251
                                 749.83013

                        Signif F =   .0004
                                    Mean Square
                                       94 .02251
                                        7.07387
Variable

ODOMETER
(Constant)
                      - Variables in the Equation  	

                      B        SE B     95% Confdnce  Intrvl  B
2.77174E-05
   1.778813
                         7.6026E-06
                             .381323
                                     1.26444E-05  4.27903E-05
                                        1.022802      2.534823
                                                                     Beta

                                                                  .333797
Variable

ODOMETER
(Constant)
                   T  Sig T

               3.646  .0004
               4.665  .0000
VEHICLE  by  MODEL_YR

                    MODEL YR
            Count
VEHICLE
                                                          Page  1  of  1
                                                                 Row

1. 00

2 . 00

1994
32

20

1995
21

10

1996
22

30

1997
26

24

1998
27

20

1999


4

Total
128
54.2
108
45. 8
            Column      52     31     52      50      47       4     236
             Total    22.0   13.1   22.0   21.2    19.9     1.7   100.0

-------
         Appendix C



1994+ Normal Emitters < 25,000 Miles
VEHICLE
CAR
TRUCK
Total
Mean
N
Std. Deviation
Median
Mean
N
Std. Deviation
Median
Mean
N
Std. Deviation
Median
Running CO
(g/mi)
.28205
77
.64849
.13200
.32188
50
.64250
6.0500E-02
.29773
127
.64387
.10300
Start CO
(g/start)
15.17562
77
9.26452
13.89300
21.88350
50
12.51306
19.85450
17.81652
127
11.11251
15.67400
FTP CO
(g/mi)
1.55696
77
1.20706
1.27800
2.14848
50
1.44522
1.74850
1.78984
127
1.33239
1.43700
    1994+ Normal Emitters All
VEHICLE
CAR
TRUCK
Total
Mean
N
Std. Deviation
Median
Mean
N
Std. Deviation
Median
Mean
N
Std. Deviation
Median
Running CO
(g/mi)
.44910
128
.66294
.19100
.59868
108
1.11458
.16000
.51755
236
.89929
.18450
Start CO
(g/start)
15.62002
128
8.81168
13.93250
20.95769
108
13.32026
18.42950
18.06269
236
11.39519
15.45700
FTP CO
(g/mi)
1.95212
128
1.35763
1.68150
2.80940
108
2.80829
1.94450
2.34443
236
2.18407
1.75150

-------