United States Air and Radiation EPA420-R-03-011
Environmental Protection September 2003
Agency M6.EXH.012
&EPA Update to the
Determination of CO Basic
Emission Rates, OBD and
I/M Effects for Tier 1 and
Later LDVs and LDTs
> Printed on Recycled Paper
-------
EPA420-R-03-011
September 2003
Update to the Determination of CO Basic Emission
Rates, OBD and I/M Effects for Tier 1 and
Later LDVs and LDTs
M6.EXH.012
Edward L. Glover
John Koupal
David Brzezinski
Assessment and Modeling Division
Office of Transportation and Air Quality
U.S. Environmental Protection Agency
NOTICE
This technical report does not necessarily represent final EPA decisions or positions.
It is intended to present technical analysis of issues using data that are currently available.
The purpose in the release of such reports is to facilitate the exchange of
technical information and to inform the public of technical developments which
may form the basis for a final EPA decision, position, or regulatory action.
-------
1.0 Introduction
This document (M6.EXH.012) updates the CO emission factors originally described in
the document M6.EXH.009 (April 2001). The update is necessary to account for the changes
made in MOBILE6 to account for the effects of Low Emitting Vehicle (LEV) and Tier 2 HC and
NOx standards on CO emissions. This document completely supercedes the previous document
(M6.EXH.009) and contains all of the descriptions contained in that document, but the
descriptions have been updated to reflect the subsequent changes in CO emission rates for
MOBILE6.
The purpose of this document is to describe the methodology, data analysis and results of
the process used to develop basic light duty vehicle CO emission factors, OBD (On-board
Diagnostics II) effects, and I/M (Inspection / Maintenance) credits for Tierl, 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 (M6EXH.012) 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.012) 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
September 2003 2
-------
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.012) is similar to
the I/M methodology for 1981-93 model year vehicles presented in EPA document M6.IM.001.
The two methodologies differ because of the presence of OBD and the different emission
standards for the 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
September 2003
-------
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:
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/EVI 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 IM240, 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
IM 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.
September 2003
-------
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 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 1 is a 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
September 2003
-------
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 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^4 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
September 2003
-------
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:
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 test result between Tier 1 and LEV was calculated, and applied
to the normal emitter Tier 1 CO emission rates to be used for the normal emitter CO
emission rates for LEVs.
7. The ratio of certification standards between Tierl and ULEV was calculated, and applied
to the normal emitter Tier 1 CO emission rates to be used for the normal emitter CO
emission rates for ULEVs.
8. 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.
September 2003
-------
3.1.2 Normal Emitter Data
The data sample used to develop the normal emitter emission levels consisted of test
results from 128 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
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.
September 2003
-------
Figures 2 and 3
1994+ MY LDV Normal Emitters
10.00
.00'
.00'
.00'
.00
.00
c 4.00'
O) o nn i
- 3.00
o zo°
u- 1.00'
LL- 0.00.
V
* o
* #'<«
o
o-'o
1 *_^ rr~"^
^"^ 9
'<& **
> o"°
^r"
/>*
^=^^
*
*
i
_--
MODEL YR
1998
0 1997
1996
A 1995
' 1994
Total Peculation
20000
ODOMETER
40000
60000
80000
100000
1994+ MY LDT1 & LDT2 Normal Emitters
ZU.U'
lo.U '
-i c n i
I^.U '
-ion.
1 n n i
O) R n i
O 4 ni
o 40:
n o n i
LL 0.0,
* m
6
A
« rf
o idi
"ip
o
irV
^°».
| V »
_ '
__
*l
I
- '
A
. '
_ -
_- "
MODEL YR
* 1999
1998
0 1997
1996
A 1995
1994
Total Population
0 25000 50000 75000 100000 125000 150000 175000 200000
ODOMETER
September 2003
-------
3.1.3 Tierl 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
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 the EPA report -
MOBILE6 "Inspection / Maintenance Benefits Methodology for 1981-93 Model Year Vehicles"
(M6.IM.001)). 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
September 2003 10
-------
model year vehicles.
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
The LDT1 truck category (see Table 1) has the same CO emission standard as passenger
cars and will use the car CO emission rates. The LDT4 truck category has a different
certification standard than the LDT2 and LDT3 vehicles. Since no LDT4 data were available, its
zero mile normal emitter CO emission level was determined by applying the ratio of the LDT4
and LDT3 certification standards to the zero mile CO emission value determined for
LDT2/LDT3. The deterioration rate used for LDT4 is the same as used for the LDT2 and LDT3.
The values for the LDT4 trucks are shown in Table A-l
3.1.4 Low Emitting Vehicles (LEV}
The normal emitter running and start CO emission levels of LEVs are determined from
the normal emitter running and start emission levels of the Tierl (shown in Equations 5a through
5d), except the emission levels are reduced by the ratio of LEV to Tier 1 certification test results.
The ratio was determined by using the unadjusted test results submitted for CO emission
certification to the 3.4 CO standard for Tier 1 from the 1998 model year to represent Tier 1
vehicles. The unadjusted test results submitted for CO emission certification to the 3.4 CO
standard for LEVs from the 2003 model year to represent LEVs. The average CO emission from
the certification test results are shown in Table 2. The test results from each model year were
averaged and the ratio was determined to be 0.505 (0.4811/0.9529).
Table 2
Comparison of Average CO Emissions from Certification Test Results
September 2003 11
-------
Emission Standard Group (Model Year)
Tier 1 (1998)
Low Emission Vehicle (2003)
CO (g/mi)
0.9529
0.4811
Number of Tests
318
211
Standard Deviation
0.5754
0.3518
This ratio is applied to the resulting emission rates calculated by Equations 5a through 6b
for Tier 1 vehicles to determine the LEV CO emission rates. The ratio affects both the zero mile
level and the deterioration rate. This adjustment of the LEV CO emission rates reflects the
observed effect of emission control enhancements necessary to achieve the tighter HC and NOx
standards for LEVs on the CO emission rates from LEV vehicles even though the CO emission
standard has not changed.
3.1.5 Ultra-Low Emitting Vehicles (ULEV)
The normal emitter running and start CO emission levels of ULEVs begin with the
normal emitter running and start emission levels of the Tierl (shown in equations 5a through
5d), 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 Tier 1 vehicles. The actual values are
shown in Equations 6a through 6d.
Running Cars:
Norm_Ave(g/mi)
Running Trucks: Norm_Ave(g/mi)
0.1411 +0.2293 *odom
0.1609 + 0.2678 * odom
Start Cars:
Start Trucks:
Norm_Ave(g/strt) = 7.588 + 0.0703 * odom
Norm_Ave(g/strt) = 10.942 + 0.1680 * odom
Eqn 6a
Eqn 6b
Eqn 6c
Eqn 6d
These emission rates are further reduced by the ratio of the ULEV CO emission standard
(1.7 g/mi) to the Tier 1 CO emission standard (3.4 g/mi), which reduces the resulting emission
rate calculated by Equations 6a through 6b by 50%.
The LDT1 truck category has the same CO emission standard as passenger cars and will
use the car CO emission rates. As with Tier 1, the LDT4 truck category zero mile normal
emitter CO emission level was determined by applying the ratio of the LDT4 and LDT3
certification standards to the zero mile ULEV CO emission value determined for LDT2/LDT3.
The deterioration rate used for LDT4 is the same as used for the LDT2 and LDT3. The values
for the LDT4 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
September 2003
12
-------
emitters for the ULEV standards will on average achieve the same compliance margin
("headroom") as the normal emitters observed in the sample.
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 Tierl 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.
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 year Tierl and
LEVs.
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
ULEV certified vehicles have lower high emitter rates for CO emissions. The high
emission rates for ULEVs are reduced by the ratio of the ULEV CO emission standard (1.7 g/mi)
to the Tier 1 CO emission standard (3.4 g/mi), which results in high emitter CO rates for ULEVs
which are 50% the values used for Tier 1 and LEVs shown in Equations 7a through 7d.
LDT4 trucks, although they have different standards than the lighter trucks and are
assigned different normal emitter CO emission rates, will use the same high emitter CO emission
rates as the LDT2 and LDT3 trucks.
September 2003 13
-------
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 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 IM
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
September 2003 14
-------
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.
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-IM 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.
September 2003 15
-------
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 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 T is the vehicle age. High(O) is
assumed to be zero. MOBILE6 will assign a value of 'odom' for each age T.
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) + [(l-OBDIM)*MIL*Growth_High(i)*(l-HighOBDIM(i-l))] +
[(l-MIL)*Growth_High(i)*(l-HighOBDIM(i-l)] Eqn lib
September 2003 16
-------
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 1 la.
'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,
September 2003 17
-------
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/IM 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.
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.IM.001 for more details.
September 2003 18
-------
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 1 la.
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.
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
September 2003 19
-------
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 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©*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).
OBD(i) = HighoBD(i) * High_ave + Normal(i)*Norm_ave(i) + Repaired(i) * Rep_ave Eqn 14
The FTP CO emission levels for cars and trucks and Tierl 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
September 2003 20
-------
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 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+IM". 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 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
September 2003 21
-------
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.
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
September 2003 22
-------
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:
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
September 2003 23
-------
Table A-l: CO Basic
Vehicle
Class
LDV/T1
LDT2/3
LDT4
Standard
Class
Tierl
LEV
ULEV
Tierl
LEV
ULEV
Tierl
LEV
ULEV
50K
Standard
(g/mi)
3.4
3.4
1.7
4.4
4.4
2.2
5.0
5.0
2.5
Mode
Running
Start (grams)
Running
Start (grams)
Running
Start (grams)
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
0.142
7.664
0.071
3.794
0.322
21.884
0.163
11.501
0.081
5.471
0.366
24.868
0.185
12.558
0.092
6.217
DR
0.229
0.070
0.116
0.035
0.115
0.035
0.268
0.168
0.135
0.085
0.134
0.084
0.268
0.168
0.135
0.085
0.134
0.084
"High"
BER
(g/mi)
36.11
38.06
36.11
38.06
18.06
19.03
33.28
83.86
33.28
83.86
16.64
41.93
33.28
83.86
33.28
83.86
16.64
41 93
"Repaired"
BER
(g/mi)
1.724
21.160
1.724
21.160
0.862
10.580
2.228
27.383
2.228
27.383
1.114
13.692
2.532
31.118
2.532
31.118
1.266
15559
September 2003
24
-------
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'
0.041
0.051
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?
25
-------
Figures 4 and 5
CO Emission Factors from Passenger Cars and Trucks
LDV/LDT1 Tierl CO
150 200
Mileage MOOO
-OBD + IM
LDT2/3 Tierl CO
150 200
M ileage / 1000
26
-------
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
Regression
Residual
F =
20.10894
DF Sum of Squares
1 7.68187
126 48.13358
Signif F = .0000
Mean Square
7 .68187
.38201
Variable
ODOMETER
(Constant)
Variables in the Equation
3 SE B 95% Confdnce Intrvl B
1.24335E-05
.117206
2.7727E-06
.091991
6.94648E-06
- .064842
1.79206E-05
.299254
Beta
.370985
Variable
ODOMETER
(Constant)
T Sig T
4.484 .0000
1.274 .2050
**** MULTIPLE REGRESSION ***
Listwise Deletion of Missing Data
27
-------
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
Variables in the Equation
Variable B SE B 95% Confdnce Intrvl B Beta
ODOMETER 3.24599E-05 3.9580E-05 -4.58687E-05 1.10788E-04 .072866
(Constant)
14.753554
1.313187
12.154795
17.352312
Variable
ODOMETER
(Constant)
T Sig T
.820 .4137
11.235 .0000
End Block Number
All requested variables entered.
28
-------
* * * *
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
3 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.
29
-------
* * * *
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
3 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.
30
-------
**** 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 Sum of Squares Mean Square
Regression 1 1.69800 1.69800
Residual 106 18983.24665 179.08723
F = .00948 Signif F = .9226
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 T Sig T
ODOMETER -.097 .9226
(Constant) 10.995 .0000
End Block Number 1 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
Variables in the Equation
3 SE B 95% Confdnce Intrvl B
ODOMETER 2.77174E-05 7.6026E-06 1.26444E-05 4.27903E-05
(Constant) 1.778813 .381323 1.022802 2.534823
Beta
.333797
Variable
ODOMETER
(Constant)
T Sig T
3.646 .0004
4.665 .0000
VEHICLE by MODEL_YR
MODEL YR
VEHICLE
Count
Page 1 of 1
Row
19941 19951 1996 1997 19981 19991 Total
1. 00
2.00
32
21
22
26
27|
20
10
30
24
20
128
54 .2
108
45. 8
H 1 1 1 1 1 1.
Column 52 31 52 50 47 4 236
Total 22.0 13.1 22.0 21.2 19.9 1.7 100.0
32
-------
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
33
-------
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
34
------- |