United States
Environmental Protection
Agency
Air and Radiation
EPA420-P-99-015
March 1999
M6 STE 003
Determination of Start
Emissions as a Function
of Mileage and Soak Time
for 1981-1993 Model Year
Light-Duty Vehicles
DRAFT
4^ Printed on Recycled Paper

-------
EPA420-P-99-0]
March 199
Determination of Start Emissions as a Function of Mileage and
Soak Time for 1981-1993 Model Year Light-Duty Vehicles
M6.STE.003
DRAFT
Ed Glover
Penny Carey
Assessment and Modeling Division
Office of Mobile Sources
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 which 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

-------
EPA420-P-99-015
- Draft -
Determination of Start Emissions
as a Function of Mileage and Soak Time
for 1981-1993 Model Year Light-Duty Vehicles
Report Number M6.STE.003
March 4,1999
Ed Glover
Penny Carey
U.S.EPA Assessment and Modeling Division
NOTICE
This technical report does not necessarily represent final EPA decisions or positions It is intended to present
technical analysis of issues using data which 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
M6.STE.003
March 4, 1999

-------
-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 artifical 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 summer conditions,
when the extra emissions from a start end after only 1 or 2 mmutes 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 will be addressed in a separate document (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.
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 contain both cars and trucks. Table 1 gives a breakdown by vehicle type,
model year, and technology for the three datasets combined.
M6.STE.003
March 4, 1999

-------
-3-
Table 1
Distribution of Vehicles by Model Year and Technology*

Cars
Cars
Cars
Cars
Cars 1

Trucks Trucks
Trucks
Trucks Trucks
MYR
OPLP
CL
Carb
TBI
PFI
ALL 1
B MYR
OPLP
CL
Carb
TBI
PFI
ALL
81
367
657
15
29
1068 1
I 81
124



124
82
71
71
74
8
224 I
¦ 82
45



45
83
63
57
127
62
309 1
1 83
8
3


11
84
5
30
46
35
116 1
1 84
26
22

1
49
85
24
74
56
66
220 1
I 85
33
30
13
6
82
86
7
34
60
92
193 1
I 86
14
9
23
41
87
87
1
17
76
106
200 1
I 87


6
4
10
88

15
69
113
197 I
¦ 88




0
89

22
38
103
163 I
I 89




0
90


160
250
410 I
¦ 90


144
1
145
91


91
426
517 1
1 91


141
144
285
92


57
347
404 I
1 92


92
92
184
93


29
366
395 1
¦ 93


90
93
183
ALL 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
HC/CO/NOX
1988-93 PFI
1988-93 TBI
1983-87 FI
1986-93 Carb
1983-85 Carb
1981-82 FI
1981-82 Carb
TRUCKS
MY Groups / Technology Type
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 injection (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 tmy 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 m-use fleet.
M6 STE.003
March 4, 1999

-------
-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 penod. 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-
M6.STE.003
March 4, 1999

-------
-6-
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 HR505 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 nommal distance traveled in the 505 driving
cycle:
Basic Start Emission Rate (grams) = [Bag l(g/mi) - HR505(g/mi)] * 3.59 miles
M6.STE.003
March 4,1999

-------
-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. This probably
is due to intermittent emission control system defects. Except in some cases of very small
samples, the negative values were retained in the analysis.
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
M6.STE.003
March 4, 1999

-------
-8-
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 soak) Start (after 10 min soak)	Composite FTP


grams



grams


grams/mile

MYR
HC
CO
NOx
MYR
HC
CO
NOx
MYR
HC
CO
NOx
81
4 002
46 419
1.373
81
0.610
5 115
0.041
81
0 706
9 667
0 897
82
2.445
36 378
1 237
82
0 373
4.843
-0 045
82
0 789
8 318
0.872
83
2.399
26 112
1 264
83
0 400
3 827
0 150
83
0.431
5 073
0.806
84
2.950
34 827
1 190
84
0513
3 418
0 047
84
0.756
9 968
0 893
85
3 468
30 353
1 204
85
0 506
4 737
0.095
85
0 533
6 935
0 770
86
2 526
26 639
1432
86
0 298
2 082
0.241
86
0 926
10.43
0 713
87
2.712
20 030
1 376
87
0 597
2.104
0 170
87
0 656
8 366
0.790
88
2 831
19 716
1 419
88
0 406
1.147
0 223
88
0 406
4 574
0.668
89
2.254
18.610
1 434
89
0 379
2 524
0216
89
0311
3.911
0.652
90
2 169
18 677
1.930
90
0.332
2219
0.611
90
0.274
3 614
0 633
91
2 183
19 494
1 443
91
0 275
2 132
0 530
91
0.237
3 145
0.525
92
2 271
18.878
1 645
92
0.304
2 595
0 485
92
0 267
4 328
0 508
93
2312
21.030
1 801
93
0.310
2 564
0 392
93
0 225
2 551
0 466
M6 STE 003
March 4, 1999

-------
-9-
Table 3
Mean Estimated Start and FTP Emission Levels by Model Year for Light-DutyTrucks
in the FTP Dataset
Basic start (after 12 hour soak)
grams
Start (after 10 min
grams
soak)
MYR
HC
CO
NOx
MY
R
HC
CO
NOx
MY
R
HC
CO
NOx
81
7 342
107 501
1.055
81
1 212
14 211
0 385
81
1 275
18.158
1.752
82
7 909
116 584
-0 119
82
1489
14 189
0 006
82
1 732
16.774
1.732
83
6 537
104 817
0 796
83
1 577
18 657
-0 209
83
1 361
13.226
1436
84
5219
95.893
0 299
84
1.098
20.057
0.004
84
0.802
10.633
1 405
85
4 766
84 621
0.457
85
0.854
7 742
0 102
85
1 281
14 465
1 388
86
3.752
41 196
0 729
86
0 607
2.148
0 128
86
0 823
8.789
1.057
87
3 352
26 635
1 266
87
0 566
1433
0017
87
0 401
4 610
0.605
88



88



88



89



89



89



90
4 705
45 331
4 683
90
0 930
7 037
0 765
90
0 800
9510
0 885
91
3 521
41.128
2 761
91
0 878
7 129
0519
91



92
3 656
41446
3.054
92
0 654
5.746
0 656
92



93
3 644
40 557
2 736
93
0 589
4 634
0 676
93
0 420
5 363
0.847
Composite FTP
grams/mile
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.
Table 4a
Regression Coefficients for START Emissions from Normal Emitter CARS
MY
Group
Tech
Group
HC Coefficients
CO Coefficients
NOx Coefficients


ZML
DET
ZML
DET
ZML
DET
1988-93
PFI
1 9987
0 006830
18 972
0 00703
1444
0 00220
1988-93
TBI
1 9019
0 002679
19 233
0 00000
2 300
0 00000
1983-87
FI
2 3589
0 001388
19 949
0 00000
1461
0 00141
1986-89
Carb
1 4934
0 018238
24 698
0 10947
1 405
0 00000
1983-85
Carb
1 5892
0 009408
24 442
0 10577
0 748
0 00524
1981-82
FI
2 3543
0 008533
20 038
0 22673
1 530
0.00059
1981-82
Carb
2 1213
0 013610
28 637
0 22673
1 601
0 00000
M6.STE.003
March 4, 1999

-------
-11-
Table 4b
Regression Coefficients for START Emissions from
Normal Emitter Li^ht Trucks
MY
Group
Tech
Group
HC Coefficients
CO Coefficients
NOx Coefficients


ZML
DET
ZML
DET
ZML
DET
1988-93
PFI
2 873
0 00000
32 178
0 0168
1 597
0 00000
1988-93
TBI
4 073
0 01309
42 456
0 1411
4 294
0 00324
1981-87
FI
2 599
0 00964
23 497
0 0613
1 384
0 00000
1984-93
Carb
3 916
0 00854
78.286
0 2564
0 143
0 00436
1981-83
Carb
6817
0 00154
98 432
0 3240
1 082
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 then-
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.
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 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 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.
M6 STE.003
March 4, 1999

-------
-12-
Table 5a
Mean START Emissions of High Emitter CARS
MY Group
Tech
Group
HC Mean
CO Mean
NOx Mean
1988-93
PFI
4 829
38 06
Same as Normals
1988-93
TBI
3 293
27 16
Same as Normals
1983-87
FI
5 313
65 31
Same as Normals
1986-89
Carb
10 520
92 82
Same as Normals
1983-85
Carb
10 520
92 82
Same as Normals
1981-82
FI
5 313
92 82
Same as Normals
1981-82
Carb
10 520
92 82
Same as Normals
Table 5b
Mean START Emissions of High Emitter Trucks
MY Group
Tech
Group
HC Mean
CO Mean
NOx Mean
1988-93
PFI
5 212
83 862
Same as Normals
1988-93
TBI
5 212
83 862
Same as Normals
1981-87
FI
5.826
60 319
Same as Normals
1984-93
Carb
9 406
162 115
Same as Normals
1981-83
Carb
17 865
179 549
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
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
M6.STE.003
March 4, 1999

-------
-13-
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. The fraction of high emitters in the fleet is the weighting factor The
fraction of high start emitters is the same fraction as the one used for the running
emissions calculations 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 the
Normals and Highs are assumed to have the same emission rate (no start NOX highs are
assumed to exist).
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.
START = ST_High_ave * Highs + STNormave * 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
M6.STE.003	March 4, 1999

-------
-14-
5.0 Start Emissions Versus Soak Time
Start emissions will be a function of soak time so that MOBILE6 will be able to
account for the entire distribution of soak times observed in the fleet. This ranges from
a minimum soak time 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.
To develop the relationship between start emissions and soak time, the FTP
database was used only to determine the engine start emissions after a 10 minute soak
and a 12 hour (720 minute) soak (these are the only data points available). To predict
start emissions for the entire range of soak duration, a model was developed from the
two FTP points, and from testing done by California of the effect of soak time on engme
start emissions (see CARB report "Methodology for Calculating and Redefining Cold
and Hot Start Emissions")
The model which was developed uses the FTP start emission data from the two
FTP soak times to adjust the curves presented in the California report. The start
emission data points at 10 minutes and 720 minutes are derived from the FTP dataset
described earlier and are a function of pollutant, technology, model year group, and
mileage. The California interpolation curves (California Soak Function) is a function of
pollutant and catalyst type
Mathematically, the start emissions of a given pollutant (in grams) as a function
of soak time is shown as:
Start Emissions (@ soak time) = Basic Start Emissions (@ 12 hour soak) * Soak
Function
where the Soak Function is a multiplicative factor used to calculate start emissions for
other soak times. The Soak Function is calculated by dividing the grams for the soak
time of interest by the grams for a soak time of 12 hours. Therefore, at a 12 hour (720
minute) soak time, the Soak Function is equal to 1.0.
Mathematically, for the first domain of the California Soak Function (see Table 6 for the
two domains for each pollutant and catalyst type) the Soak Function is defined as:
Soak Function = California Soak Function * [Ratio+(l-Ratio)*((SoakTime-10)/(X-10))]
For the second domain of the California Soak Function (i e., HC for catalyst equipped
vehicles the second domain is 90 minutes through 720 minutes) the Soak Function is
M6.STE.003
March 4,1999

-------
-15-
Table 6
Coefficients for Adjusting Engine Start Emissions for Soak
Time
(from "Methodology for Calculating and Redefining Cold and Hot Start Emissions", CARB)
Non-Catalyst Vehicles

HC
Curve 1
HC
Curve 2
CO
Curve 1
CO
Curve 2
NOX
Curve 1
NOX Curve 2
Constant
0 38067
0 43628
0 43803
-0 08541
1 31568
2 48061
minutes
-0 00163
0 00078
-0 00998
0 00303
0 02752
-0 00018
minutes2
6 64E-05
0
7 01E-05
-2 11E-06
-0 00015
-2 6E-06
domain(min)
0-52
53-720
0-119
120-720
0-119
120-720
Catalyst Equipped Vehicles

HC
Curve 1
HC
Curve 2
CO
Curve 1
CO
Curve 2
Nox
Curve 1
NOx Curve 2
Constant
0
0 57130
0
0 70641
0 11796
112983
minutes
0 01272
0 00072
0 01195
0 00033
0 02967
2 21E-05
minutes2
-6 30E-05
-1 76E-07
-4 76E-05
1 0OE-O7
-0 00021
-3 04E-07
domain(min)
0-89
90-720
0-116
117-720
0-61
62-720
Electrically Heated Catalyst Equipped Vehicles

HC
Curve 1
HC
Curve 2
CO
Curve 1
CO
Curve 2
Nox
Curve 1
NOx Curve 2
Constant (a)
0
0 50641
0
0 44733
1 05017
1 37178
minutes (b)
0 00561
0 00069
0 00707
0 00162
0 00362
0 00027
minutes2 (c)
-5 09E-06
0
-1 33E-05
-1 18E-06
-5 57E-06
-1 09E-06
domain(min)
0-117
118-720
0-107
108-720
0-113
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 12 hours
M6.STE.003
March 4, 1999

-------
-16-
defined as:
Soak Function = California Soak Function
where-
California Soak Function: The values developed by the California Air Resources
Board to adjust the start emissions for soak times other than 12 hours. This is a function
of soak time in minutes. The coefficients for catalyst vehicles, non-catalyst vehicles,
and electrically heated catalyst vehicles are given in Table 6. The coefficients for
catalyst-equipped vehicles are for the model year/technology groups examined in this
report. For example, for HC on a catalyst equipped vehicle at a soak time of 100
minutes, the value is
0 57130 + 0.00072*100 + (-1.76E-07)*(100)2 = 0.64154
Ratio: This parameter is calculated by dividing the EPA ratio of start emissions at 10
minutes to start emissions at 12 hours by the California Soak Function at 10 minutes.
Mathematically, it is given by:
Ratio = (Start @10 minutes / Start @720 minutes) / California Soak Function @10
mmutes
The numerator in the above equation (Start @10 minutes/ Start @720 minutes) was
developed from FTP data using the equations in Sections 3.3, and dividing the Start
@10 minutes by the Start @ 720 minutes. One value for each pollutant was developed
that included all technologies and vehicle types These values, used m the numerator of
the equation, are: HC= 0.160, CO = 0.112, and NOx = 0 204. The California Soak
Function is the value obtained from the coefficients in Table 6 at a 10 minute soak
point. The California Soak Function values at a 10 minute soak point for catalyst-
equipped vehicles, used in the denominator of the equation, are. HC=0.1209,
CO=0 1147, and NOx=0 3937. Therefore, the Ratios obtained are: HC=1.3234,
CO=0 9765, and NOx=0.5182.
SoakTime: The time duration in minutes of the soak which is to be calculated (range
zero minutes to 720 mmutes).
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 m Table 6. These are 0-89 minutes and 90-720
M6 STE.003
March 4, 1999

-------
-17-
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.
For all three pollutants, the difference between the MOBILE6 soak function and
the California soak function will generally be small for the entire range of soak times. If
any difference exists it will reach its highest magnitude in the range of 10 minutes to
about 60 minutes.
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 mcludes 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
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.
M6.STE.003
March 4, 1999

-------
-18-
Start Emissions (@88min) = Basic Start Emissions (@12hr) * Soak Function
From Tables 4a, 5 a 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 g
HC
Soak Function = California Soak Funct * [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 m 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)j] =
0.63407
Start Emissions(@90min) = 2.647 * 0.63407 = 1.679g HC
M6 STE.003
March 4, 1999

-------
-19-
Appendix A
Fraction of High and Normal Emitters in the Fleet
The fractions of high and normal emitters in the fleet were calculated based on the
running emission estimates. For consistency, these fractions were also used for start
emissions. The fractions were based on the running estimates, because the running
estimates were corrected to account for the recruitment bias inherent m the FTP data.
A large amount of IM240 data from the Dayton, Ohio I/M program (211,000 initial
tests) were used to correct the recruitment bias inherent in the FTP type data The
recruitment bias is believed to exist in the EPA and AAMA FTP samples because of (1)
their relatively small size in comparison to the likelihood of selecting a high emitter, (2) the
belief that motorists with poorly maintained or tampered vehicles will be reluctant to
volunteer them to a government entity like the EPA or even the vehicle manufacturer, and
(3) the relatively low mileage levels of the EPA and AAMA sample (relatively few over
100,000 miles).
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. Unless
they have intentionally tampered the vehicle and have an expectation of failure, it is
unlikely that they would seek to avoid the program since the data is from the first year of
I/M testing m Dayton.
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. Thus, they should provide a more representative cross-
section of the m-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.
Tables A-l and A-2 show the fraction of HC and CO high emitters in the fleet. The use
of these to develop start emission factors are discussed m Section 4 4 and Section 4.5. The
derivation of these high emitter fractions is discussed in detail in EPA document
M6 IM 001. 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.
M6.STE.003
March 4, 1999

-------
-20-
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 IM240 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 = Highave * 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
M6 STE.003
March 4, 1999

-------
-21-
Table A-l
Estimated Fraction of HC High Emitters in the Fleet

HC
HC
HC
HC
HC
HC
HC
MILEAGE
88-93 PFI
88-93 TBI
83-87 Fl
86-89 Carb
83-85 Carb
81-82 Fl
81-82 Cart>
2142
0 0184
0 0239
0 0223
0 0052
0 0232
0 0203
0 0282
12 823
0 0227
0 0251
0 0157
0 0197
0 0158
0 0654
0 0543
29 335
0 0422
0 0270
0 0406
0 0526
0 0047
0 1613
01580
50
0 0800
0 0386
01003
01042
0 0917
0 2861
0 2906
60 006
0 0987
00458
01298
01296
01348
0 3485
0 3560
74 239
01260
0 0561
01723
0 1661
01972
0 4393
0 4503
87 786
01525
0 0661
0 2078
0 2012
0 2578
0 5275
0 5416
100 01
01770
0 0753
0 2346
0 2334
0 3135
0 6094
06253
112 948
0 2036
0 0851
0 2634
0 2678
0 3737
0 6986
0 7152
124 625
02280
0 0940
0 2898
0 2992
0 4290
0 7812
0 7976
135 738
0 2518
0 1026
0 3153
0 3295
0 4826
0 8620
0 8772
146 315
0 2748
01110
0 3400
0 3586
0 5345
0 9407
0 9539
156 38
0 2972
01190
0 3638
0 3866
0 5847
1 0000
1 0000
165 96
0 3189
01267
0 3868
0 4135
0 6332
1 0000
1 0000
175 077
0 3398
01341
0 4089
0 4393
0 6801
1 0000
1 0000
183 753
0 3601
0 1412
0 4303
0 4641
0 7253
1 0000
1 0000
192 01
0 3798
01480
0 4508
0 4879
0 7690
1 0000
1 0000
199 869
0 3988
0 1546
0 4706
0 5108
08111
1 0000
1 0000
207 349
04171
01609
0 4896
0 5327
0 8516
1 0000
1 0000
214 466
0 4348
01669
0 5079
0 5537
0 8907
1 0000
1 0000
221 241
04519
01727
0 5255
0 5738
0 9284
1 0000
1 0000
227 688
0 4683
01782
0 5425
0 5931
0 9646
1 0000
1 0000
233 823
0 4842
01836
0 5587
0 6116
1 0000
1 0000
1 0000
239 663
04994
01887
0 5743
0 6293
1 0000
1 0000
1 0000
245 22
0 5141
0 1936
0 5893
0 6462
1 0000
1 0000
1 0000
250 509
0 5283
01982
0 6036
0 6624
1 0000
1 0000
1 0000
M6.STE.003
March 4, 1999

-------
-22-
Table A-2
Estimated Fraction of CO High Emitters in the Fleet

CO
CO
CO
CO
CO
CO
CO
MILEAGE
88-93 PFI
88-93 TBI
83-87 Fl
86-89 Carb
83-85 Carb
81-82 Fl
81-82 Carb
2 142
0 0093
0 0552
0 0180
0 0103
0 0130
0 0119
0 0508
12 823
0 0082
0 0553
0 0123
0 0388
0 0093
0 0511
0 1102
29 335
0 0241
0 0553
0 0357
0 0929
0 0473
0 1334
0 2441
50
0 0458
0 0554
0 0889
0 1741
0 1783
0 2466
0 4138
60 006
0 0566
0 0555
0 1150
0 2140
0 2430
0 3024
0 4969
74 239
0 0721
0 0555
0 1496
0 2715
0 3364
0 3830
0 6163
87 786
0 0872
0 0556
0 1765
0 3270
0 4271
0 4611
0 7312
100 01
0 1010
0 0556
0 2012
0 3778
0 5102
0 5327
0 8360
112 948
0 1159
0 0557
0 2276
0 4323
0 5998
0 6097
0 9479
124 625
0 1296
0 0558
0 2518
0 4822
0 6819
0 6802
1 0000
135 738
0 1429
0 0558
0 2751
0 5302
0 7614
0 7484
1 0000
146 315
0 1556
0 0559
0 2976
0 5764
0 8381
0 8141
1 0000
156 38
0 1680
0 0559
0 3193
0 6210
0 9121
0 8775
1 0000
165 96
0 1799
0 0560
0 3402
0 6638
0 9836
0 9387
1 0000
175 077
0 1914
0 0560
0 3602
0 7050
1 0000
0 9976
1 0000
183 753
0 2025
0 0561
0 3795
0 7445
1 0000
1 0000
1 0000
192 01
0 2132
0 0561
0 3981
0 7825
1 0000
1 0000
1 0000
199 869
0 2235
0 0561
0 4159
0 8191
1 0000
1 0000
1 0000
207 349
0 2334
0 0562
0 4330
0 8541
1 0000
1 0000
1 0000
214 466
0 2429
0 0562
0 4495
0 8877
1 0000
1 0000
1 0000
221 241
0 2521
0 0562
0 4653
0 9200
1 0000
1 0000
1 0000
227 688
0 2609
0 0563
0 4804
0 9510
1 0000
1 0000
1 0000
233 823
0 2693
0 0563
0 4949
0 9806
1 0000
1 0000
1 0000
239 663
0 2774
0 0563
0 5089
1 0090
1 0000
1 0000
1 0000
245 22
0 2852
0 0564
0 5222
1 0363
1 0000
1 0000
1 0000
250 509
0 2927
0 0564
0 5350
1 0623
1 0000
1 0000
1 0000
M6.STE.003
March 4, 1999

-------