United States         Air and Radiation       EPA420-R-01-058
          Environmental Protection                 November 2001
          Agency                       M6.STE.003
&EPA    Determination of Start
          Emissions as a Function of
          Mileage and Soak Time for
          1981 -1993 Model Year
          Light-Duty Vehicles
                                 yŁu Printed on Recycled
                                 Paper

-------
                                                       EPA420-R-01-058
                                                         November 2001
 Determination of Start Emissions as a Function of
  Mileage and Soak Time for 1981-1993 Model Year
                     Light-Duty Vehicles

                          M6.STE.003
                              Ed Glover
                             Penny Carey
                    Assessment and Standards 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.

-------
                                      -2-

1.0    INTRODUCTION

       MOBILE6 will allocate vehicle exhaust emissions to either the "extra" emissions
associated with engine start (start emissions) or the "base" emissions associated with travel
(running emissions).  This distinction is to some extent an artificial constraint since in
reality "start" emissions in part will be released while the vehicle is in motion.  However,
it is a useful constraint and not too far from physical reality, in  at least warm weather
driving conditions, when the extra emissions from a start end after only 1 or 2 minutes of
driving. This split allows the separate characterization of start and running emissions for
correction factors such as fuel effects and ambient temperature.  It also allows a more
precise weighting of these two aspects of exhaust emissions for particular situations such
as morning commute,  parking  lots  and  freeways.  This  document describes  the
methodology used to calculate start emissions as a function of mileage and soak time for
use in MOBILE6.  The results for model year 1981-1993 light-duty cars and light-duty
trucks are presented.  The deterioration of running emissions are addressed in a separate
document (EPA Report Number M6.EXH.001).

       This document is organized into six sections.  The first section is the short
introductory section.  Section 2 describes the FTP data sources and the  model year and
technology groups which are used.  Section 3 provides a definition of start emissions in
mathematical terms and shows the relationship between start emissions  and  the Federal
Test Procedure (FTP) emissions. This includes a description of the FTP cycle, the Hot505
cycle, and  a definition of cold start and hot start emissions.  Section  4  describes  the
methodology used to predict start emissions as a function of soak time. Section 5 shows
the algorithm used to predict start emissions versus mileage. Section 6 contains an example
of the final start emission results as used in MOBILE6 as a function of both deterioration
and soak time.  Appendix A briefly discusses the methodology and rational for using a high
emitter correction factor and presents the fleetwide high emitter penetration fractions.
2.0    DATA SOURCES USED

       The basic datasets used to determine in-use deterioration are based on FTP testing.
(I/M data from Dayton, Ohio were also used to correct the results for recruitment bias
which EPA believes affects the FTP test samples - see Appendix A). Three FTP data
sources were used: 1) the test results from the EPA laboratory in Ann Arbor, Michigan, 2)
the data received from AAMA (American Automobile Manufacturers Association) based
on testing conducted in Michigan and Arizona, and 3) the API (American Petroleum
Institute) data collected in Arizona.  The model years in the dataset range from 1981
through 1993,and containboth cars and trucks. Table 1 gives a breakdown by vehicle type,
model year, and technology for the three datasets combined.

-------
                                                    -J-
     ALL
                                                 Table 1
                         Distribution of Vehicles by Model Year and Technology*

MYR

81
82
83
84
85
86
87
88
89
90
91
92
93
Cars
OPLP

367
71
63
5
24
7
1






Cars
CL
Carb
657
71
57
30
74
34
17
15
22




Cars
TBI

15
74
127
46
56
60
76
69
38
160
91
57
29
Cars
PFI

29
8
62
35
66
92
106
113
103
250
426
347
366
Cars
ALL

1068
224
309
116
220
193
200
197
163
410
517
404
395
Trucks Trucks Trucks
MYR

81
82
83
84
85
86
87
88
89
90
91
92
93
OPLP CL
Carb
124
45
8 3
26 22
33 30
14 9







TBI





13
23
6


144
141
92
90
Trucks Trucks
PFI




1
6
41
4


1
144
92
93
ALL

124
45
11
49
82
87
10
0
0
145
285
184
183
538    977
898  2003   4416
ALL
250     64    509
382   1205
         No entry indicates no data available for that model year/technology type in the FTP dataset used for this analysis.
M6.STE.003
                                                                                        March 4, 1999

-------
                                      -4-
       In general, most of the 1990+ model year vehicle data were supplied by AAMA,
and most of the pre-1990 data were supplied by the EPA laboratory testing. The API
sample is a relatively small sample (99 cars and trucks). Its chief appeal is that the vehicles
all have generally higher mileage readings than the rest of the sample (all over 100,000
miles). The other general trend in the data is toward PFI technology, and away from the
others. This is seen in the 1990+ vehicles which are predominately PFI with some TBI still
present. The 1981 -1989 model years start with a high percentage of carbureted and some
open loop, but end with mostly TBI and PFI technology. Although not explicitly shown
in the tables, new catalyst technology was phased slowly into the fleet starting in the mid
1980's.

       For analysis, the cars and trucks were placed into the model year/technology groups
shown below.  Trucks were separated into five different groups by pollutant due to
differences in certification standards.
                CARS
       MY Group / Technology Type
       TRUCKS
MY Groups / Technology Type
             HC/CO/NOX
             1988-93 PFI
             1988-93 TBI
             1983-87 FI
             1986-93 Carb
             1983-85 Carb
             1981-82 FI
             1981-82 Carb
       HC/CO/NOX
       1988-93 PFI
       1988-93 TBI
       1981-87 FI
       1984-93 Carb
       1981-83 Carb
The technology groups are closed-loop ported fuel inj ection (PFI), closed-loop throttle body
injection (TBI), and carbureted (CARB).  FI refers to a combination of PFI and TBI.
CARB includes both closed-loop and open-loop vehicles which are carbureted.  These
model  year/technology grouping boundaries were selected on the basis of changes in
emission standards  or  the development/refinement of new  fuel metering or catalyst
technologies. It is assumed that as of 1990, carbureted technology had a very tiny market
share, and are included with the previous carbureted group. Because of the relatively large
amount of 1988-93 fuel injected data, the category was split into PFI technology and TBI
technology for both cars and trucks.  This produces separate deterioration functions based
on this fuel delivery technology and allows the modeling of the future penetration of PFI
technology into the in-use fleet.

-------
                                      -5-
3.0    DEFINITION OF START EMISSIONS

3.1    Overview of the Federal Test Procedure (FTP)

       The Federal Test Procedure (FTP) is a test cycle which is used to certify new
vehicles to emission  performance standards (see 40  CFR Part 86,  Subpart B, Section
86.144). The FTP consists of a cold start segment (Bag 1), a hot stabilized segment (Bag
2), and a hot start segment (Bag 3). Initially, the vehicle is stored for a minimum of 12
hours before testing to simulate a 12 hour overnight soak period. The vehicle is then driven
over the cold start segment, which lasts 505 seconds over a length of 3.59 miles, and the
emissions collected as Bag 1.  The latter part of the driving in Bag 1 occurs with the engine
and catalyst in a hot stabilized condition. Bag 2 emissions are then immediately collected
from the hot stabilized segment, which lasts 867 seconds over a length of 3.91 miles. After
a 10 minute soak, the 505 seconds of the start segment is then repeated and the emissions
collected as Bag 3.

       The FTP composite emission rate is a weighted combination of the three measured
bags to represent two trips. The first trip is a cold start trip after a 12 hour soak, and the
other is a hot start trip after a 10 minute soak.  Each trip is a "LA4"  cycle, which is a
combination of the 505 cycle (either Bag 1 or Bag 3) and the Bag 2 cycle. In a typical FTP
test, the Bag 2 is only measured once and the results are used for both trips.  Since the 505
cycle is 3.59 miles long and the Bag 2 cycle is 3.91 miles long, each LA4 trip is 7.5 miles
long. The cold start trip is weighted at 43% and the hot  start trip weighted 57%. If the cold
start trip is 43% of the driving, then the vehicle miles traveled (VMT) in Bag 1 (containing
the cold start) is:

FTP Bag 1 VMT Weighting = 43% * ( 3.59 miles / 7.5 miles ) = 0.206

The hot start trip is 57% of driving, and the VMT weighting for Bag 3 (containing the hot
start) is:

FTP Bag 3 VMT Weighting = 57% * ( 3.59 miles / 7.5 miles ) = 0.273

The remaining VMT is from stabilized driving, represented by Bag 2.  Since Bag 2 is used
for both the cold start and hot start trips, it uses VMT weighting from both.

FTP Bag 2 VMT Weighting = ( 43% + 57% ) * ( 3.91  miles / 7.5 miles ) = 0.521

The standard VMT weighting of the bags reported in grams per mile for the full FTP are:

-------
FTP = (Bag 1 * 0.206) + (Bag 2 * 0.521) + (Bag 3 * 0.273)

where the fractions represent the amount of vehicle miles traveled within the three modes
during the FTP trip, and Bagl, Bag2, Bag3 and the FTP emissions are in grams per mile
(g/mi).

3.2   Overview of the Hot Running 505 and Its Use

       The FTP testing method outlined above does not allow the precise separation of
start and running emissions, since Bags 1 and 3 contain both start and running emissions.
Bag 2 of the FTP does not contain an engine start; however, the driving cycle used in the
second bag is  significantly different from the cycle used for Bags 1 and 3.   Thus, to
estimate the amount of FTP emissions that can be allocated to engine start, the concept of
the Hot Running 505 (HR505) is needed.

       The FIR505 is an extra 505 cycle performed immediately following bag 3 of the
FTP.  It uses an identical driving cycle as the first and third bags of the FTP, but does not
include an engine start. For more information, refer to the document "The Determination
of Hot Running Emissions from FTP Bag Emissions", report number M6.STE.002. With
a HR505  emission result, it is possible to compare the results obtained from the HR505 to
the results from Bags 1 and 3  of the FTP to determine the  portions of Bags  1  and 3
attributable to start emissions following a 12 hour soak and start emissions following a 10
minute soak, respectively.

       Since the HR505 has not historically been included in FTP test programs, a method
of estimating the HR505 was developed, as  described in report M6.STE.002.  Briefly,
HR505 emissions were measured from a sample of 77 vehicles tested under EPA contract.
The results from this vehicle sample were used to develop a correlation between the HR505
and FTP bag data. This correlation was then used to estimate HR505 results for the FTP
dataset used for this analysis.
3.3   Basic Start Emission Rate

       For MOBILE6, the basic unit of engine start emissions is defined as a start after a
12 hour soak. The units for engine start emissions will be grams, instead of grams per mile,
since start emissions will not be allocated by vehicle miles traveled.  The engine start basic
emission rate can be determined by subtracting the HR505 emission rate from the  Bag 1
emission rate (in grams per mile) using the nominal distance traveled in the 505 driving
cycle:

Basic Start Emission Rate (grams) = [Bag l(g/mi) - HR505(g/mi)]  * 3.59 miles

-------
                                      -7-

       For illustration purposes, the average basic start emission rates (in grams) after a
12 hour soak were calculated for each model year and are shown in Tables 2 and 3 for cars
and trucks.

       Start emissions after a 10 minute soak can also be estimated from the Bag 3 and
HR505 emission rates, analogous to the basic start emission rate:

Start Emissions after 10 minute soak (grams) = [Bag 3(g/mi) - HR505(g/mi)] * 3.59 miles

       The average start emissions after a 10 minute soak are also shown in Tables 2 and
3 for each model year and for cars and trucks.

       For some FTP tests of some cars, the predicted HR505 emissions are higher than
Bagl and/or Bag3 emissions. This causes the start emissions to be negative. Thisprobably
is due to intermittent emission control system defects.  Except in some cases of very small
samples, the negative values were retained in the analysis. Sample sizes are shown in Table
1.

4.0     Basic Start Deterioration with Mileage
4.1    Definition of Categories

       The basic modeling concept behind the start emission factor methodology is that
the fleet can be represented as two types of vehicle emitter categories. These two types are
termed "high" emitters, and "normal" emitters. The "high" emitters have FTP average
emission levels which are considerably higher than the overall mean emission levels, and
significantly higher than their FTP standards, indicating that they have problems with their
emission control systems. The "normal" emitters are low and average emitting vehicles
with emission control systems which are generally functioning properly. The overall fleet
emission factor is a weighted average of the high and normal emitters.

       The high/normal emitter modeling concept is also used in the estimation of running
emissions, and is discussed in depth in other reports. For both start and running emissions,
the high/normal concept allows for corrections due to recruitment bias against higher
emitting vehicles, and for more  accurate estimates of the effects of I/M programs, fuel
effects, temperatures, etc.

       In the data analysis, vehicles were defined as normal emitters for a specific pollutant
if their  FTP  HC  emissions were  less than  twice  the applicable new car certification
standard, or their FTP CO emissions were less than three times the applicable new car

-------
                                            Table 2
          Mean Estimated Start and FTP Emission Levels by Model Year for Light-Duty Cars
                                       in the FTP Dataset
Basic start (after 12 hour

MYR
81
82
83
84
85
86
87
88
89
90
91
92
93

HC
4.002
2.445
2.399
2.950
3.468
2.526
2.712
2.831
2.254
2.169
2.183
2.271
2.312
grams
CO
46.419
36.378
26.112
34.827
30.353
26.639
20.030
19.716
18.610
18.677
19.494
18.878
21.030
soak)

NOx
1.373
1.237
1.264
1.190
1.204
1.432
1.376
1.419
1.434
1.930
1.443
1.645
1.801
Start (after

MYR
81
82
83
84
85
86
87
88
89
90
91
92
93

HC
0.610
0.373
0.400
0.513
0.506
0.298
0.597
0.406
0.379
0.332
0.275
0.304
0.310
10 min soak)
grams
CO
5.115
4.843
3.827
3.418
4.737
2.082
2.104
1.147
2.524
2.219
2.132
2.595
2.564
Composite FTP
grams/mile
NOx
0.041
-0.045
0.150
0.047
0.095
0.241
0.170
0.223
0.216
0.611
0.530
0.485
0.392
MYR
81
82
83
84
85
86
87
88
89
90
91
92
93
HC
0.706
0.789
0.431
0.756
0.533
0.926
0.656
0.406
0.311
0.274
0.237
0.267
0.225
CO
9.667
8.318
5.073
9.968
6.935
10.43
8.366
4.574
3.911
3.614
3.145
4.328
2.551
NOx
0.897
0.872
0.806
0.893
0.770
0.713
0.790
0.668
0.652
0.633
0.525
0.508
0.466
M6.STE.003
                                                                               March 4, 1999

-------
                                               -9-

                                            Table 3
         Mean Estimated Start and FTP Emission Levels by Model Year for Light-Duty Trucks
                                       in the FTP Dataset
Basic start (after 12 hour soak)

IYR

81
82
83
84
85
86
87
88
89
90
91
92
93

HC

7.342
7.909
6.537
5.219
4.766
3.752
3.352


4.705
3.521
3.656
3.644
grams
CO

107.501
116.584
104.817
95.893
84.621
41.196
26.635


45.331
41.128
41.446
40.557

NOx

1.055
-0.119
0.796
0.299
0.457
0.729
1.266


4.683
2.761
3.054
2.736
Start (after 10 min soak)

MY
R
81
82
83
84
85
86
87
88
89
90
91
92
93

HC

1.212
1.489
1.577
1.098
0.854
0.607
0.566


0.930
0.878
0.654
0.589
grams
CO

14.211
14.189
18.657
20.057
7.742
2.148
1.433


7.037
7.129
5.746
4.634
Composite FTP
grams/mile
NOx

0.385
0.006
-0.209
0.004
0.102
0.128
0.017


0.765
0.519
0.656
0.676
MY
R
81
82
83
84
85
86
87
88
89
90
91
92
93
HC

1.275
1.732
1.361
0.802
1.281
0.823
0.401


0.800


0.420
CO

18.158
16.774
13.226
10.633
14.465
8.789
4.610


9.510


5.363
NOx

1.752
1.732
1.436
1.405
1.388
1.057
0.605


0.885


0.847
M6.STE.003
                                                                                March 4, 1999

-------
                                      -10-

certification standard, or their FTP NOx emissions were less than twice the new car
certification standard.

       Vehicles were defined as high emitters for a specific pollutant if their FTP HC
emissions or FTP CO emissions exceeded twice or three times the applicable new car
certification standard, respectively, or their FTP NOx emissions were two times the new
car certification standard.  Because high NOx emissions often occur with low HC and/or
low CO emissions, and sometimes even HC can be high and CO normal, the three
categories were kept separate. Thus, a vehicle could be a high HC emitter, but a normal
CO and NOx emitter.
4.2    Calculation of Start Emission Rates for Normal Emitters

       Emission rates for normal emitters were calculated by least squares regression of
the emissions of the normal emitters versus mileage. The regression was done for each
pollutant / model year / technology group. The start emission regression coefficients for
cars are shown in Table 4a and for trucks in Table 4b. The column labeled ZML
contains the zero mile coefficients, and the column DET contains the deterioration
coefficients (slope) from the regressions.  Appendix B Tables B-3  through B-8 show the
regression statistics for normal emitter cars and trucks for each pollutant.
Table 4a
Regression Coefficients for START Emissions from Normal Emitter CARS
MY
Group

1988-93
1988-93
1983-87
1986-89
1983-85
1981-82
1981-82
Tech
Group

PFI
TBI
FI
Carb
Carb
FI
Carb
HC Coefficients
ZML
1.9987
1.9019
2.3589
1.4934
1.5892
2.3543
2.1213
DET
0.006830
0.002679
0.001388
0.018238
0.009408
0.008533
0.013610
CO Coefficients
ZML
18.972
19.233
19.949
24.698
24.442
20.038
28.637
DET
0.00703
0.00000
0.00000
0.10947
0.10577
0.22673
0.22673
NOx Coefficients
ZML
1.444
2.300
1.461
1.405
0.748
1.530
1.601
DET
0.00220
0.00000
0.00141
0.00000
0.00524
0.00059
0.00000

-------
                                      -11-

MY
Group

1988-93
1988-93
1981-87
1984-93
1981-83
Table 4b
Regression Coefficients for START Emissions from

Tech
Group

PFI
TBI
FI
Carb
Carb
Normal Emitter Light Trucks

HC Coefficients
ZML
2.873
4.073
2.599
3.916
6.817
DET
0.00000
0.01309
0.00964
0.00854
0.00154
CO Coefficients
ZML
32.178
42.456
23.497
78.286
98.432
DET
0.0168
0.1411
0.0613
0.2564
0.3240


NOx Coefficients
ZML
1.597
4.294
1.384
0.143
1.082
DET
0.00000
0.00324
0.00000
0.00436
0.00000
4.3    Calculation of Start Emission Rates for High Emitters

       High emitters are the vehicles in the fleet which likely have problems with their
emission control systems, as evidenced by emission levels which are considerably
higher than the FTP standards.  In the analysis they were defined as those vehicles
exceeding either twice FTP  standards for HC or NOX or three times FTP standards for
CO. The  emissions line is a flat horizontal line because the emissions of a high emitter
were not a statistically significant function of mileage.  In addition, the relatively small
sample sizes of high emitters make regression determined mileage coefficients
unreliable indicators of actual behavior. Table 5a shows the average emissions of the
high emitters (cars only) for the pollutant / model year / tech groups, and Table 5b
shows the analogous results for light trucks.  Appendix B Tables B-l and B-2 show the
sample size and standard deviation statistics for the High emitter cars and trucks. It
should be noted by the reader that the data were combined across model year and
technology group in some cases to produce a larger sample of high emitters.

       For NOX start emissions, the normals and the highs were combined together,
and the emissions were regressed versus mileage.  This has the effect of eliminating the
NOX high emitter group for start emissions.  This combination was done for two
reasons. First, for many of the model year / tech groups the average NOX start
emissions of high emitters were found to not be statistically significantly different from
start emissions from the normals. This was found to be the case even when different
definitions of a high emitter were tried (IX FTP NOX, 1.5X FTP NOX, and 2X FTP
NOX). This phenomenon is consistent with the mechanisms of NOX formation - higher
emissions under  lean high temperature / load FTP conditions, and lower during rich and

-------
                                      -12-
cooler start conditions.  Second, the sample sizes for NOX high emitters were smaller in
both an absolute sense, and in comparison to the HC and CO high emitter sample.

MY Group
1988-93
1988-93
1983-87
1986-89
1983-85
1981-82
1981-82
Table 5a
Mean START Emissions of Hish Emitter CARS

Tech
Group
PFI
TBI
FI
Carb
Carb
FI
Carb
HC Mean
4.829
4.829
5.313
10.520
10.520
5.313
10.520
CO Mean
38.06
38.06
65.31
92.82
92.82
92.82
92.82

NOx Mean
Same as Normals
Same as Normals
Same as Normals
Same as Normals
Same as Normals
Same as Normals
Same as Normals

MY Group
1988-93
1988-93
1981-87
1984-93
1981-83
Table 5b
Mean START Emissions of Hish Emitter Trucks

Tech
Group
PFI
TBI
FI
Carb
Carb
HC Mean
5.212
5.212
5.826
9.406
17.865
CO Mean
83.862
83.862
60.319
162.115
179.549

NOx Mean
Same as Normals
Same as Normals
Same as Normals
Same as Normals
Same as Normals
       The other anomally in the results was the HC and CO high emitter emission
levels for the 1990-93 TBI group.  In both of these cases the average start emission
levels of the few cars tested were judged to be unrealistically low. For the case of CO,
the value was -123.84 grams, and for HC it was 0.0356 grams.  These low average
levels are the result of small sample size, and the possibility of negative values when the

-------
                                      -13-

hot running 505 on a particular car is greater than Bagl of the FTP. Rather than insert
negative values in the MOBILE6 model, the HC and CO high emitter emission levels
from the 1988-93 PFI group were substituted in the 1988-93 TBI group. This is a
reasonable assumption since these vehicles are generally about the same age and model
year vintage, and have reasonably similar emission control technology.

4.4    Fraction of High and Normal Emitters in the Fleet

       The basic start emission factor is computed from a weighted average of the highs
and normals, and the fraction of high emitters in the fleet is the weighting factor for the
highs and  normals. The fraction of high emitters in the fleet and the definition of a high
emitter are based on FTP emission results. Since similar definitions of a high emitter
were not available in running and start units, the consistent FTP high emitter definition
and high emitter fraction values were used for both start and running, but only for HC
an CO emissions. For NOx emissions, the high emitter fraction for running emissions is
the same as the FTP fraction, but the start fraction is not.  The start NOx high fraction
is assumed to be zero because NOx emissions rarely occur in high concentration during
start operation. A quick analysis of the data using various definitions of a high emitter
for start and running indicated that this was a reasonable choice.  However, from a
vehicle engineering perspective, start emission problems sometimes are not well
associated with running operation emission problems (i.e., choke system malfunctions).

       Appendix A presents the fraction of HC and CO high emitters in the fleet at
selected mileages / ages for each pollutant (see document M6.IM.001 for further
details). The fraction of NOX high emitters is not shown because for NOX because  the
Normals and Highs are assumed to have the same emission rate (no start NOX highs are
assumed to exist). Appendix A also provides a brief discussion of the high emitter
adjustment factor used to correct for recruitment bias inherent in FTP data type
sampling.  This adjustment factor generally increased the rate of high emitters in the
fleet over  the fraction that would be predicted based on the FTP sampling.
4.5    Calculation of Basic Start Emission Rates

       The basic start emission rate is calculated for each combination of vehicle type /
pollutant / model year group / technology group. The units are start emissions in grams.
Equation 1 is used to calculate the basic (mean) start emission rate from the high and
normal emitter emission values and the rate of high emitters in the fleet. For NOX
emissions a special case of this equation is used where the normal and high emission
rates are set equal to each other.

-------
                                     -14-

       START = ST_High_ave * Highs + ST_Norm_ave * Normals            Eqn 1


Where:

Highs = fraction of High emitters
Normals = fraction of Normal emitters
START is the basic (mean) emission rate
ST_High_ave is the high emitter start emission average
ST_Norm_ave is the normal emitter start emission average

Where:

       Highs + Normals = 1                                               Eqn 2


5.0    Start Emissions Versus Soak Time
       Start emissions will now be modeled as a function of soak time in MOBILE6.
As such, the model will be able to account for the entire distribution of soak times
observed in the fleet rather than just two soak time points (10 and 720 minutes) that
were implicit in the FTP test procedure and in MOBILES.  The model will allow the
soak time to range from a minimum of zero minutes up to a 12 hour soak period (720
minutes). Soak periods exceeding 12 hours will be assumed to be the same as for a 12
hour soak.

       The MOBILE6 relationship between start emissions and soak time was
developed by using the FTP database, and a California soak time and engine start
model (CARB model). The FTP start emission data were available only at the soak
time periods of 10 minutes (hot start), and 720 minutes (cold start). The CARB model
predicted start soak emission effects for the entire range of  soak time lengths (0 to 720
minutes). The details of CARB model are documented in the report "Methodology for
Calculating and Redefining Cold and Hot Start Emissions".

       The start emissions model developed for MOBILE6 is really just an adjusted or
calibrated version of the CARB model. It uses the FTP start emission data from the two
FTP soak times to adjust the CARB model curves at the ten minute level.  This has the
effect of forcing the CARB model curves through these two points, but retaining the
general shape of the CARB model curve.  The start emission data points at 10 minutes
and 720 minutes are derived from the FTP dataset described earlier. The California

-------
                                      -15-

interpolation curves (California Soak Function) is a function of pollutant and catalyst
type.

Basic Start Soak Equation

       The general form of the MOBILE6 start emissions as a function of soak time
calculation is shown in Equation 3.

Start Emissions (@ soak time)  =  Basic Start Emissions (@ 12 hour soak) *
                                Soak Function(@soak time)                 Eqn 3

       At a given soak time, it is the product of the basic 12 hour cold start emission
factor and the soak function.  The Basic Start Emissions are the start emission factors
discussed previously in this document.  The Soak Function is a multiplicative correction
factor that converts the basic 12 hour start emission factor into a cold start emission
factor at any soak time length. Mathematically, the Soak Function is defined in
Equation 4.

Soak Function = California Soak Function * [Ratio+(l-Ratio)*((SoakTime-10)/(X-10))]        Eqn 4

The terms used in the equation re defined below.


California Soak Function:

       The California Soak Function is the basis of the MOBILE6 start soak time
model. It is an empirical relationship derived by the California Air Resources Board
(CARB), and used in the EMFAC model to calculate start emissions as a function of
soak time. Upon review of the CARB document "Methodology for Calculating and
Redefining Cold and Hot Start Emissions", EPA found it to be the only available
empirical model for this purpose.

       Mathematically, the California Soak Function has the following general form:

       California Soak Function  =  a + b*soak length +  c*soak length A 2      Eqn 5

where soak length is the time in minutes since the last engine operation, and a, b, and c
are soak function coefficients. These are shown in Table 6 for each pollutant, three
different catalyst types (the coefficients for catalyst-equipped vehicles are for the model
year/technology groups examined in this report), and by soak time domain. The soak
time domain is divided into two groups (a low domain and a high domain) which are
listed in Table 6.

-------
                              -16-
                            Table 6
 Coefficients for Adjusting Engine Start Emissions for Soak
                             Time
(from "Methodology for Calculating and Redefining Cold and Hot Start Emissions", GARB)
Non-Catalyst Vehicles

Constant
minutes
minutes2
domain(min)
HC
Curve 1
0.38067
-0.00163
6.64E-05
0-52
HC
Curve 2
0.43628
0.00078
0
53-720
CO
Curve 1
0.43803
-0.00998
7.01E-05
0-119
CO
Curve 2
-0.08541
0.00303
-2.11E-
06
120-720
NOX
Curve 1
1.31568
0.02752
-0.00015
0-119
NOX Curve
2
2.48061
-0.00018
-2.6E-06
120-720
Catalyst Equipped Vehicles

Constant
minutes
minutes2
domain(min)
HC
Curve 1
0
0.01272
-6.30E-
05
0-89
HC
Curve 2
0.57130
0.00072
-1.76E-07
90-720
CO
Curve 1
0
0.01195
-4.76E-
05
0-116
CO
Curve 2
0.70641
0.00033
1.00E-07
117-720
Nox
Curve 1
0.11796
0.02967
-0.00021
0-61
NOxCurve2
1.12983
2.21 E-05
-3.04E-07
62-720
Electrically Heated Catalyst Equipped Vehicles

Constant (a)
minutes (b)
minutes2(c)
domain(min)
HC
Curve 1
0
0.00561
-5.09E-
06
0-117
HC
Curve 2
0.50641
0.00069
0
118-720
CO
Curve 1
0
0.00707
-1.33E-
05
0-107
CO
Curve 2
0.44733
0.00162
-1.18E-
06
108-720
Nox
Curve 1
1.05017
0.00362
-5.57E-
06
0-113
NOxCurve2
1.37178
0.00027
-1.09E-06
114-720
California Soak Function = a + b * minutes + c * minutes2 (where minutes is time since last engine operation (i.e., soak time))
The Soak Function is the grams per soak time i divided by the grams per overnight soak (720 minutes or 1 2 hours)

-------
                                      -17-

Ratio

       The Ratio is the variable shown in Equation 4. It is a ratio of two ratios, and is
shown mathematically in Equation 6.

             EPA Start Emissions at 10 minutes
             EPA Start Emissions at 720 minutes
Ratio  =     	                            Eqn 6

             CARS Start Emissions at 10 minutes
             CARB Start Emissions at 720 minutes

or

Ratio  =     EPA Ratio / CARB Ratio
       The EPA ratio for each pollutant was derived empirically from FTP hot and cold
start test data by calculating a hot start emission factor and a cold start emission factor
using the equations in Section 3.3, and dividing the Start @10 minutes (hot start) by the
Start @ 720 minutes (cold start). One value for each pollutant was developed that
included all technologies and vehicle types. These values, used in the numerator of the
equation, are:

EPA Ratio   HC       =         0.160
EPA Ratio   CO       =         0.112
EPA Ratio   NOx     =         0.204.

       The statistics from the data analysis used to generate these ratios is provided in
the attached document STAT.lst.

       The CARB ratios were computed by inserting a soak time of 10 minutes and a
soak time of 720 minute into Equation 5 for catalyst equipped vehicles.  For catalyst
equipped vehicles, the ratios were for each pollutant.
CARB Ratio  HC       =         0.1209
CARB Ratio  CO       =         0.1147
CARB Ratio  NOx     =         0.3937

       Differences in these ratios reflect differences in the hot and cold start data from
EPA and California. Note there are only small differences between the EPA and CARB

-------
                                     -18-

results for CO start emissions.

The final ratios used in Equation 4 is the ratio of the EPA and CARB ratios. These are
shown below for each pollutant.

Ratio HC     =         1.3234
Ratio CO     =         0.9765
Ratio NOx    =         0.5182


SoakTime

       The soak time is the time duration between successive engine starts in minutes.
It can range from zero minutes to 720 minutes.
X term: This term is defined to be zero for soak times from 0-10 minutes. For the range
from 10 minutes to 720 minutes, it is set equal to the highest minute in the domain of
the California Soak Function. For example, HC emissions from catalyst equipped
vehicles have two time domains in Table 6. These are 0-89 minutes and 90-720
minutes.  Thus, for this example, X = 0 for times of 10 minutes or less, and X = 89 for
times from 11 minutes through 89 minutes. No soak adjustment is applied for the
remaining soak period of 90 minutes through 720 minutes.  Only the California Soak
Function is used.
6.0    START EMISSION RESULTS

       Start emissions are both a function of vehicle deterioration represented by
mileage, and soak time. In previous sections the results were shown separately.  In this
section, examples of the results are shown with both effects combined.

       Shown in the linked EXCEL spreadsheets (CAR_BER.xls and TR_BER.xls) is a
sample calculation of the basic emission start factors for the various model year groups
and pollutants. It includes calculations for both start and running emissions. The
calculations in the spreadsheet use Equation 1 in this document, and show the
magnitude of the start emission factors that will be used in MOBILE6.

       The statistical results and output are too voluminous to present directly in this
document.  However, they are available in the linked document (stat.lst). The statistical

-------
                                       -19-

software SPSS was used to perform the linear regressions and compute the means. In
general, the regression correlation coefficients (r-squared) are not high (< 0.10), and
reflect the tremendous scatter in emissions data. However, virtually, all of the
regression coefficients of the normal emitters are significant at least at a 90 percent
confidence level.  On the other hand, the confidence intervals around the average start
emission levels of the high emitters are quite large due to high scatter and small sample
sizes.

       Shown below for illustration purposes is a sample calculation of start emissions.
It illustrates the soak function equations and methodology shown in Section 5.
Example:     Calculate HC start emissions at a soak time of 88 minutes for a 1991
              model year PFI-equipped car with 60,000 miles.

Start Emissions (@88min) = Basic Start Emissions (@12hr) * Soak Function

   From Tables 4a, 5a and A-l:
   Basic Start Emissions = High_ave * Highs + Norm_ave * Normals

   START     =      4.829*0.0987 + (1.999 + 0.00683*60)(1.0 - 0.0987) = 2.647 gHC

   Soak Function = California SoakFunct * [Ratio+(l-Ratio)*((SoakTime-10)/(X-10))]

       From Table 6, using the coefficients for catalyst-equipped vehicles:
       California Soak Funct = 0.000 + (0.01272)*(88) + (-6.30E-05)*(88)2 = 0.63149

       Ratio = (Start@10min / Start@12hr) / California Soak Funct@10min
            = 1.3234 for HC as given in Section 5.0

       SoakTime = 88 minutes

       X = 89 minute HC time domain (from Table 6).

       Soak Function = 0.63149 * [1.3234 + (1-1.3234) * ((88-10) / (89-10))] = 0.63407

       Start Emissions(@90min) = 2.647 * 0.63407 = 1.679g HC

-------
                                      -20-

New Data Update

      No new data have been incorporated into this analysis since the initial Draft release
of this document.


Response to Stakeholder and Peer Review Comments

      Significant stakeholder comments were not received for this document. However,
three separate paid and independent peer reviewers were used, and provided the following
comments. Their comments were either addressed directly in the document or are discussed
below.

1.     One reviewer questioned the inclusion  of the API vehicle test data solely on the
      basis of its  higher mileage nature.  He  felt that the different geographical location
      of the testing and sampling, and potentially different recruitment techniques may
      add uncertainty to the results.

      In response, EPA feels that these geographical differences should be minimal,
      and that API followed reasonable and traditional recruitment and testing
      methods. In addition, the inclusion of these data provide largely unique and
      sorely needed information on higher mileage vehicles.

2.     In regards to the  Start Soak Adjustment, one reviewer expressed concern about
      the use of two data points to adjust the  entire continuum of soak time emission
      effects.

      This is certainty a valid concern.  However, the only data outside of the special
      CARB study which were available are  at the FTP soak time values of 10 minutes
      and 720 minutes. To help overcome this lack of data, the CARB study model is
      used to fill  in these blanks.

3.     One reviewer stressed  the need for the  regression statistics to be reported in an
      easily readable table.   This will allow subsequent reviewers to determine the
      uncertainty in the final results.

      These are now provided in Appendix B in Tables B-l through B-8.

4.     The remaining comments presented by the Peer reviewers which could not be
      incorporated into this document largely pertain to suggestions that more testing
      and data analysis are needed. In particular more information is needed to better

-------
                                -21-

characterize the behavior of high emitters, and develop more robust statistics.
They also stressed that additional data should be obtained on start soak periods
between the extremes of 10 minutes and 720 minutes to better validate the model.

EPA agrees on these points.  However, such data collection efforts  and analysis
will likely have to wait for the next generation of MOBILE models.

-------
                                     -22-

                                 Appendix A
               Fraction of High and Normal Emitters in the Fleet

      This appendix shows the fraction of HC and CO high and normal emitters in the
fleet for start emissions. High emitter fractions for NOx emissions are not shown because
only normal emitters are assumed to exist for NOx start emissions (data suggested that high
NOx emissions rarely occur during cold start). This Appendix also briefly discusses the
methodology and rational for the high emitter adjustment factor for recruitment bias.

      Tables A-l and A-2 show the fraction of HC and CO high emitters in the fleet after
they have been adjusted for recruitment bias.  The use of these to develop start emission
factors are discussed in Section 4.4 and Section 4.5.  The derivation of these high emitter
fractions and the average High and Normal emitter emission factors is discussed in detail
in EPA document M6.EVI.OO 1. However, a brief description and the mathematical equation
is shown below.

      The number of High and Normal emitters is calculated at each age point for each
combination of vehicle type / pollutant / model year / technology group using the following
general equations.

Where:

'Highs'  is the fraction of High emitters.
'Normals'  is the fraction of Normal emitters.
'RLA4' is the average running emission rate, after adjustment based on EVI240 data from
Dayton, OH.
'High_ave' is the high emitter running emission average estimated from the FTP data.
'Norm_ave' is the normal emitter running emission average estimated from the FTP data.
      Highs + Normals = 1                                          Eqn A-l

and

      RLA4 = High_ave * Highs + Norm_ave * Normals             Eqn A-2

Solving for the variables Highs and Normals produces:


      Highs = (RLA4 - Norm_ave) / (Highave - Norm_ave)            Eqn A-3

      Normals = 1 - Highs                                          Eqn A-4

-------
                                      -23-

Recruitment Bias Discussion

      The theoretical justification  for the FTP recruitment bias  rests on  anecdotal
observations from test programs over the years that sample vehicles recruited by EPA or
a major automaker will  be disproportionately lower in high  emitters.  This is because
owners who will participate in a test program are less likely to tamper or deliberately mal-
maintain their vehicle than are owners who decline to participate.  This may result from
the fear exposure by a Federal agency for an act which is illegal (tampering the emission
controls on a vehicle). The other bias which is  likely present in the EPA and automaker
data is the tendency to  recruit relatively low mileage vehicles at the expense of high
mileage vehicles. These low mileage vehicles are frequently a test priority for an automaker
due  to  warranty and recall concerns.   This type of  sampling  bias leads to a well
characterized sample with mileages less than 50,000 miles, but a more poorly characterized
sample with higher mileages.

      To experimentally verify the potential high emitter recruitment bias, a large amount
of in-use IM240 data from the Dayton, Ohio I/M program (211,000 initial tests) were
obtained.  These data were used to investigate and correct the recruitment bias believed to
be inherent in the FTP type data.  Since the Dayton sample contains virtually  the entire
vehicle fleet (or a randomly selected 50 percent) of the city these issues of recruitment bias
should be minimal. For example, the sample size is quite large (211,000 vehicles). Even
if high emitters are just a few percent of this sample, they should  be well characterized.
Also, I/M  is not a voluntary program in Ohio; thus, a motorist cannot simply decline to
participate; however, they can illegally evade the program by registering outside of the
program boundary. It was suggested by some reviewers, that the program evaders could
range from 1 percent to 5 percent of the total fleet.

      Although the mileage  data from Dayton are highly suspect  and not useful,  the
vehicles of a given model year vintage are older than the vehicles of the same model year
vintage in the EPA and AAMA databases because the Dayton testing was done subsequent
to the EPA  and AAMA  testing.   Thus,  the Dayton data should  provide a more
representative cross-section of the in-use fleet.  The derivation of the average running
emission rates, with the adjustments based on the Dayton IM240 data, are discussed in EPA
document  M6.EXH.001.

      The high emitter recruitment adjustment factor is documented  in  EPA report
M6.EXH.001 in detail. In terms of start emissions it had a tendency to  increase the fraction
of high emitters in the fleet by a few percent. The impact of this was slightly higher on start
emissions  factors.

-------
                      -24-
                   TableA-1
Estimated Fraction of HC High Emitters in the Fleet

MILEAGE
2.142
12.823
29.335
50
60.006
74.239
87.786
100.01
112.948
124.625
135.738
146.315
156.38
165.96
175.077
183.753
192.01
199.869
207.349
214.466
221.241
227.688
233.823
239.663
245.22
250.509
HC
88-93 PFI
0.0184
0.0227
0.0422
0.0800
0.0987
0.1260
0.1525
0.1770
0.2036
0.2280
0.2518
0.2748
0.2972
0.3189
0.3398
0.3601
0.3798
0.3988
0.4171
0.4348
0.4519
0.4683
0.4842
0.4994
0.5141
0.5283
HC
88-93 TBI
0.0239
0.0251
0.0270
0.0386
0.0458
0.0561
0.0661
0.0753
0.0851
0.0940
0.1026
0.1110
0.1190
0.1267
0.1341
0.1412
0.1480
0.1546
0.1609
0.1669
0.1727
0.1782
0.1836
0.1887
0.1936
0.1982
HC
83-87 Fl
0.0223
0.0157
0.0406
0.1003
0.1298
0.1723
0.2078
0.2346
0.2634
0.2898
0.3153
0.3400
0.3638
0.3868
0.4089
0.4303
0.4508
0.4706
0.4896
0.5079
0.5255
0.5425
0.5587
0.5743
0.5893
0.6036
HC
86-89 Carb
0.0052
0.0197
0.0526
0.1042
0.1296
0.1661
0.2012
0.2334
0.2678
0.2992
0.3295
0.3586
0.3866
0.4135
0.4393
0.4641
0.4879
0.5108
0.5327
0.5537
0.5738
0.5931
0.6116
0.6293
0.6462
0.6624
HC
83-85 Carb
0.0232
0.0158
0.0047
0.0917
0.1348
0.1972
0.2578
0.3135
0.3737
0.4290
0.4826
0.5345
0.5847
0.6332
0.6801
0.7253
0.7690
0.8111
0.8516
0.8907
0.9284
0.9646
1 .0000
1 .0000
1 .0000
1 .0000
HC
81-82FI
0.0203
0.0654
0.1613
0.2861
0.3485
0.4393
0.5275
0.6094
0.6986
0.7812
0.8620
0.9407
1 .0000
1 .0000
1 .0000
1.0000
1.0000
1.0000
1.0000
1.0000
1 .0000
1 .0000
1 .0000
1 .0000
1 .0000
1 .0000
HC
81 -82 Carb
0.0282
0.0543
0.1580
0.2906
0.3560
0.4503
0.5416
0.6253
0.7152
0.7976
0.8772
0.9539
1 .0000
1 .0000
1 .0000
1.0000
1.0000
1.0000
1.0000
1.0000
1 .0000
1 .0000
1 .0000
1 .0000
1 .0000
1 .0000

-------
                      -25-

                   Table A-2
Estimated Fraction of CO High Emitters in the Fleet

MILEAGE
2.142
12.823
29.335
50
60.006
74.239
87.786
100.01
112.948
124.625
135.738
146.315
156.38
165.96
175.077
183.753
192.01
199.869
207.349
214.466
221.241
227.688
233.823
239.663
245.22
250.509
CO
88-93 PFI
0.0093
0.0082
0.0241
0.0458
0.0566
0.0721
0.0872
0.1010
0.1159
0.1296
0.1429
0.1556
0.1680
0.1799
0.1914
0.2025
0.2132
0.2235
0.2334
0.2429
0.2521
0.2609
0.2693
0.2774
0.2852
0.2927
CO
88-93 TBI
0.0552
0.0553
0.0553
0.0554
0.0555
0.0555
0.0556
0.0556
0.0557
0.0558
0.0558
0.0559
0.0559
0.0560
0.0560
0.0561
0.0561
0.0561
0.0562
0.0562
0.0562
0.0563
0.0563
0.0563
0.0564
0.0564
CO
83-87 Fl
0.0180
0.0123
0.0357
0.0889
0.1150
0.1496
0.1765
0.2012
0.2276
0.2518
0.2751
0.2976
0.3193
0.3402
0.3602
0.3795
0.3981
0.4159
0.4330
0.4495
0.4653
0.4804
0.4949
0.5089
0.5222
0.5350
CO
86-89 Carb
0.0103
0.0388
0.0929
0.1741
0.2140
0.2715
0.3270
0.3778
0.4323
0.4822
0.5302
0.5764
0.6210
0.6638
0.7050
0.7445
0.7825
0.8191
0.8541
0.8877
0.9200
0.9510
0.9806
1.0090
1.0363
1.0623
CO
83-85 Carb
0.0130
0.0093
0.0473
0.1783
0.2430
0.3364
0.4271
0.5102
0.5998
0.6819
0.7614
0.8381
0.9121
0.9836
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
CO
81-82 Fl
0.0119
0.0511
0.1334
0.2466
0.3024
0.3830
0.4611
0.5327
0.6097
0.6802
0.7484
0.8141
0.8775
0.9387
0.9976
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
CO
81 -82 Carb
0.0508
0.1102
0.2441
0.4138
0.4969
0.6163
0.7312
0.8360
0.9479
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000

-------
                            -26-
                        Appendix B
Statistical Detail: Standard Errors, P values and Standard Deviations

Table B-l
Standard Deviations of Means

START Emissions of High Emitter CARS
MY Group
1988-93
1988-93
1983-87
1986-89
1983-85
1981-82
1981-82
Tech
Group
PFI*
TBI
pp**
Carb**
Carb
FI
Carb
HC
Sample
Size
25
25
103
371
371
103
371
CO
Sample
Size
30
30
124
350
350
124
350
HC
Standard
Deviation
5.84
5.84
10.39
23.89
23.89
10.39
23.89
CO
Standard
Deviation
73.73
73.73
92.75
83.21
83.21
83.21
83.21
NOx
Standard
Deviation
N/A
N/A
N/A
N/A
N/A
N/A
N/A
* 1988-93 PFI and TBI combined together for a total sample size of 3
** 1986-89 Carb, 1983-85 Carb and 1981-83 Carb combined together for a total sample size of 350 or 371
*** 1983-87 FI and 1981-82 FI combined together for a total sample size of 124 or 103
Table B-2
Standard Deviations of Means
START Emissions of High Emitter TRUCKS
MY Group
1988-93
1988-93
1981-87
1984-93
1981-83
Tech
Group
PFI
TBI
FI
Carb
Carb
HC
Sample
Size
3*
3
4
12
17
CO
Sample
Size
3
3
18
19
23
HC
Standard
Deviation
5.10
5.10
5.30
8.81
13.72
CO
Standard
Deviation
68.29
68.29
115.30
93.90
116.09
NOx
Standard
Deviation
N/A
N/A
N/A
N/A
N/A
* 1988-93 PFI and TBI combined together for a total sample size of 3

-------
-27-
Table B-3
Regression Statistics from Normal Emitting Cars - CO EMISSIONS
MY
Group

1988-93
1988-93
1983-87
1986-89
1983-85
1981-82
1981-82
Tech
Group

PFI
TBI
FI
Carb
Carb
FI
Carb
Sample
Size
CO
1591
433
641
94
234
108
816
S.E
Slope
CO
0.01618
0.02424
0.01600
0.06987
0.06328
0.09004
0.03257
SE
ZML
CO
0.6688
1.0846
0.9760
4.4534
2.4836
4.4260
1.3539
SigT
Slope
CO
0.1897
0.9624
0.7870
0.1206
0.0960
0.0038
0.0000
SigT
ZML
CO
0.000
0.000
0.000
0.000
0.000
0.000
0.000
Table B-4
Regression Statistics from Normal Emitting Cars - HC EMISSIONS
MY
Group

1988-93
1988-93
1983-87
1986-89
1983-85
1981-82
1981-82
Tech
Group

PFI
TBI
FI
Carb
Carb
FI
Carb
Sample
Size
HC
1583
436
623
92
234
105
839
S.E
Slope
HC
0.00115
0.00205
0.00100
0.00351
0.00376
0.00391
0.00184
SE
ZML
HC
0.0475
0.0914
0.0860
0.2172
0.1469
0.1948
0.0775
SigT
Slope
HC
0.0002
0.1092
0.3450
0.000
0.0130
0.0313
0.000
SigT
ZML
HC
0.000
0.000
0.000
0.000
0.000
0.000
0.000

-------
-28-
Table B-5
Regression Statistics from Normal Emitting Cars - NOX EMISSIONS
MY
Group

1988-93
1988-93
1983-87
1986-89
1983-85
1981-82
1981-82
Tech
Group

PFI
TBI
FI
Carb
Carb
FI
Carb
Sample
Size
NOX
1611
441
694
95
248
108
974
S.E
Slope
NOX
0.00122
0.00328
0.00100
0.00345
0.00328
0.00368
0.00157
SE
ZML
NOX
0.0509
0.1467
0.0760
0.2215
0.1347
0.1808
0.0708
SigT
Slope
NOX
0.1064
0.5472
0.2460
0.5311
0.1113
0.8725
0.0000
SigT
ZML
NOX
0.000
0.000
0.000
0.000
0.000
0.000
0.000

-------
-29-
Table B-6
Regression Statistics from Normal Emitting TRUCKS - CO EMISSIONS
MY
Group

1988-93
1988-93
1981-87
1984-93
1981-83
Tech
Group

PFI
TBI
FI
Carb
Carb
Sample
Size
CO
330
464
76
115
157
S.E
Slope
CO
0.00410
0.00363
0.00477
0.00857
0.01256
SE
ZML
CO
0.1993
0.1550
0.3028
0.4228
0.5563
SigT
Slope
CO
0.2870
0.0003
0.0467
0.3220
0.9020
SigT
ZML
CO
0.0000
0.0000
0.0000
0.0000
0.0000
Table B-7
Regression Statistics from Normal Emitting TRUCKS - HC EMISSIONS
MY
Group

1988-93
1988-93
1981-87
1984-93
1981-83
Tech
Group

PFI
TBI
FI
Carb
Carb
Sample
Size
HC
329
465
90
122
163
S.E
Slope
HC
0.0514
0.0542
0.0844
0.2009
0.1920
SE
ZML
HC
2.496
2.314
5.783
10.434
8.671
SigT
Slope
HC
0.744
0.009
0.407
0.204
0.094
SigT
ZML
HC
0.0000
0.0000
0.0000
0.0000
0.0000

-------
-30-
Table B-8
Regression Statistics from Normal Emitting TRUCKS - NOX EMISSIONS
MY
Group

1988-93
1988-93
1981-87
1984-93
1981-83
Tech
Group

PFI
TBI
Fl
Garb
Garb
Sample
Size
NOX
331
466
93
132
166
S.E
Slope
NOX
0.00282
0.00541
0.00315
0.00498
0.00642
SE
ZML
NOX
0.137
0.233
0.219
0.276
0.293
SigT
Slope
NOX
0.179
0.549
0.020
0.382
0.081
SigT
ZML
NOX
0.0000
0.0000
0.0000
0.6044
0.0003

-------