United States Air and Radiation EPA420-P-99-008
Environmental Protection M6.FUL.001
Agency March 1999
&EPA Fuel Sulfur Effects on
Exhaust Emissions
> Printed on Recycled Paper
-------
EPA420-P-99-008
March 1999
on
for 6
Venkatesh Rao
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.
-------
Objective
This document describes EPA's effort to estimate empirical relationships between fuel
sulfur content and exhaust emissions of hydrocarbons (HC), non-methane hydrocarbons
(NMHC), nitrogen oxides (NOx), and carbon monoxide (CO) as a function of vehicle technology
and vehicle emitter classification for gasoline-powered vehicles. MOBILE6 will use these
relationships to adjust exhaust emission rates in response to fuel sulfur levels input by the user.
The vehicle technologies addressed in this analysis include: Tier 0, Tier 1, Low-Emitting
Vehicles (LEVs), and Ultra Low-Emitting Vehicles (ULEVs). Where possible, the vehicle
technology data are further stratified by passenger car (PC), Light-Duty Truck Class 1 (LDT1),
and heavier Light-Duty Truck classifications (LDT2/LDT3/LDT4). Table 1 below defines the
weight classifications for the different truck classes.
Table 1
Light-Duty Truck Weight Definitions
Truck Category
LDT1
LDT2
LDT3
LDT4
Gross Vehicle Weight in Ibs.
(GWVR)
0-6000
0-6000
6001-8500
6001-8500
Loaded Vehicle Weight in Ibs
(LVWR)
0-3750
3751-5750
3751-5750
5751-8500
Wherever possible, normal and high emitters are also addressed and analyzed separately. Diesel-
powered heavy-duty vehicles are not considered in this analysis as very little data exists to
describe sulfur's effect on emissions from these vehicles.
Background
Since the early 1970s it has been recognized that gasoline sulfur levels impact the
conversion efficiency of automotive three way catalysts. A significant amount of test data has
been generated in the recent past to investigate this phenomenon. Data were collected first
during the development of EPA's Complex Model, and more recently in response to questions
about the in-use performance of advanced technology (i.e., low-emission and Tier 2) vehicles.
The Complex Model was constructed in response to the Clean Air Act's Reformulated Gasoline
(RFG) requirements and is used to certify reformulated gasolines. The Complex Model is an
empirical model designed to predict emissions as a function of fuel properties. The exhaust
portion of the Complex Model is based solely on data from a number of different emission
testing programs and includes the effect of aromatics, olefms, RVP, distillation characteristics,
sulfur, and oxygen content on emissions of VOC, NOx, and air toxics. The exhaust Complex
Model also applies strictly to only "1990 technology" vehicles (which are, in general, vehicles of
Model Year 1987-1992). The reader is referred to the RFG web site
-2-
-------
(http://www.epa.org/oms/reformulated gasoline) for further information on the RFG regulations
and the Complex Model. In this report, any reference to the "Complex Model" will only refer to
the exhaust portion.
Many studies have been conducted on the effects of fuel sulfur levels on vehicle exhaust
emissions. In 1991, the US Auto/Oil Air Quality Improvement Research Program (AQIRP)
published the first1 of its six studies on the effects of sulfur. The first study concluded that
reducing sulfur levels from 450 to 50 ppm reduced exhaust emissions in 1990 Tier 0 technology
vehicles by 16, 13, and 9% respectively, for total hydrocarbon (THC), carbon monoxide (CO),
and nitrogen oxide (NOx) emissions. A subsequent study was published by the US AQIRP in
1993 2 jjjjg stu(jy confirmed the previous studies' results for emission benefits accrued from
reducing sulfur from 450 to 50 ppm and also investigated effects of reducing sulfur down to 10
ppm. In 1995, the US AQIRP published a study3 investigating sulfur's effect on emissions from
vehicles certified to Tier 1 standards. In addition, EPA has conducted several studies to
investigate sulfur's effect on emissions4'5'6. Recently, the American Petroleum Institute also
sponsored a "Sulfur Reversibility" study7 which examines sulfur's effect on exhaust emissions
from newer vehicles. All data sources used in this analysis are described below in more detail in
the "Data Sources" section.
Based on much of this data, EPA proposes to include an adjustment for in-use gasoline
sulfur levels in the next version of its highway motor vehicle emission factors model, MOBILE6.
Consistent with the new start/running methodology11 being implemented in MOBILE6, this
report will present separate estimates for the impacts of gasoline sulfur levels on running exhaust
emissions and start emissions. MOBILE6 will use these start and running emission rates to
calculate composite effects when necessary. However, to facilitate comparisons to results from
previous studies and other existing emission models, estimates have been developed for
FTP/composite (the composite emissions calculated from appropriate weighting of bag data)
emissions directly from the data. FTP/composite emissions will be identified by "FTP
emissions" in the remainder of this report. It should be noted that all FTP emissions correlations
shown and discussed in this paper are for illustrative purposes only: sulfur correction factors in
MOBILE6 will be solely based on the correlations developed for start and running emissions.
The structure of this report is as follows:
Brief overview of EPA's previous sulfur adjustment proposal made at
10/1/97 MOBILE6 workshop
Summary of major comments received on the proposal made at 10/1/97
workshop
EPA's proposed final methodology for MOBILE6
Data sources
Analysis of sulfur data
Definition of emitter classes
Data sources
-3-
-------
Regression methodology and mathematical fits
Start, running, and FTP emissions
Analysis of Tier 0 normal emitters
Analysis of Tier 1 normal emitters
Analysis of LEV and ULEV normal emitters
Analysis of high emitters
Valid sulfur range
Summary
EPA Preliminary Proposal for MOBLE6 on 10/1/97
As a first-cut approach to model the impacts of fuel sulfur content on exhaust emissions,
EPA first segregated all available vehicle data on the effects of sulfur on exhaust emissions
(mostly Auto/Oil and EPA testing data) into normal and high emitter categories. High emitters
were defined, as they were in the Complex Model, as vehicles emitting more than two times the
particular vehicle's HC standard on a base fuel (0.82 grams/mile for Tier 0 vehicles). Vehicles
were not grouped by fuel injection technology as the type of fuel injection was shown not to
heavily effect sulfur's impact on emissions. For each data set, an average gram/mile value was
calculated at each individual sulfur level. Then, a nonlinear curve of the following form was fit
through the resulting averages:
Emissions = A* In(Sulfur) + B
This fit was used for all vehicle technologies, emitter classifications, emission categories, and
pollutants. In this initial analysis for the 10/1/97 MOBILE6 workshop, the effects of sulfur on
start emissions were found to be small in magnitude and statistically insignificant in most cases
when compared to FTP emissions and running emissions effects. So, the effect of sulfur on start
emissions was assumed to be zero in all cases. The reader is referred to the 10/1/97 workshop
presentation materials and handouts for the exact correlations developed based on the
relationship between sulfur and exhaust emissions shown above. Table 2 below illustrates the
effects calculated from the correlations developed using the above equation:
Table 2
Sulfur Reductions Based on Correlations Presented at 10/1/97 MOBILE6 Workshop
Emis-
sions
Mode
FTP
Vehicle
Tech
TierO
Tier 1
Percent Reduction in HC
Emissions when Sulfur (in
ppmW) Changed From:
700->
400
4.8
2.8
400->
200
6.0
3.4
200->
50
12.8
7.1
Percent Reduction in NOx
Emissions when Sulfur (in
ppmW) Changed From:
700->
400
1.64
1.87
400->
200
2.08
2.40
200->
50
4.20
4.80
Percent Reduction in CO
Emissions when Sulfur (in
ppmW) Changed From:
700->
400
5.28
7.50
400->
200
6.92
5.23
200->
50
14.8
11.1
-4-
-------
^^^"
running
RFG
Model*
TierO
Tier 1
14.0
8.6
10.1
9.9
11.7
13.4
7.5
26.3
32.5
2.90
2.66
1.84
5.70
3.38
2.30
7.40
7.00
4.71
14.4
7.80
7.20
9.60
10.5
9.72
7.26
23.3
21.4
High Emitters
FTP
running
TierO
RFG
Model*
TierO
0.55
-1.60
4.00
0.72
-1.00
5.04
1.43
-0.80
10.8
0.45
7.00
2.91
0.55
4.80
3.70
1.10
3.60
7.66
0.85
11.8
5.37
1.10
6.90
6.97
2.20
6.10
15.2
* "RFG Model" refers to EPA's Complex Model for certifying reformulated gasolines. The Complex Model applies
strictly only to "1990 technology" vehicles. The numbers in the "RFG model" rows are for comparison purposes only.
After the workshop, EPA received comments indicating that the methodology used to compute
the correlations and obtain the results shown in Table 1 may not be the most accurate way to
model sulfur's effect on emissions for this set of data. The following concerns were identified:
modeling sulfur impacts as a non-linear effect for all pollutants may not be appropriate in
all cases;
a simpler approach should be used for modeling higher emitters since the data available is
so sparse and comes from different test programs;
statistical techniques that adjust for the different fleets and base fuels used in the testing
programs must be employed;
the definition of a high emitter should be re-examined (it is not clear that the Complex
Model definition of 0.82 g/mi HC is appropriate for MOBILE6);
an approach should be developed that allows sulfur effects to be estimated for both
conventional gasoline and reformulated gasoline (RFG) areas.
Based on these comments, and on additional test data from Tier 1, LEV, and ULEV vehicles that
has become available since the workshop, EPA now proposes a revised methodology to estimate
sulfur's effect on exhaust emissions for gasoline-powered vehicles.
EPA's Final Proposed Methodology for MOBILE6
Data Sources
Auto/Oil Phase I Sulfur Study^In this portion of this extensive testing program, ten 1989 model
year light-duty gasoline vehicles (representing a subset of the "Current" fleet (Tier 0 vehicles)
tested in all the other Auto/Oil studies) were tested using two fuels with sulfur levels of 466 and
49 ppm (other fuel parameters were held constant). The results of that testing indicated that
overall HC, CO, and NOx emissions were reduced by approximately 16%, 13% and 9%,
respectively, when fuel sulfur content was reduced from 466 to 49 ppm.
-5-
-------
Auto/Oil Phase n Sulfur Study2-This portion of the testing expanded on the Phase I study by
testing the same "Current" fleet vehicles over a wider range of sulfur levels with more
intermediate points. This was done to determine non-linear trends in the data. Two fuel sets
were used. The first, termed "Part F, was a five-fuel set ranging from a nominal sulfur level of
450 ppm down to 50 ppm in increments of 100 ppm. The second, termed "Part IF', was a three-
fuel set having sulfur levels of 50 ppm to 10 ppm in increments of 20 ppm. This study confirmed
the results of the Phase I study and further found that reducing fuel sulfur from 50 ppm to 10
ppm, resulted in a reduction in HC of 6% and CO of 10%; there was no statistically significant
effect on NOx emissions in this range.
T50/T90/Sulfur Study3-Testing in this part of the Auto/Oil program was designed to investigate
possible non linear impacts of the fuel distillation parameter T90, interactive impacts of fuel
distillation parameters T50 and T90, and sulfur on emission from light-duty vehicles. Three
vehicle fleets were tested: the Current fleet assessed in the Phase I and Phase II Studies (10
vehicles), a Federal Tier 1 fleet (consisting of six vehicles), and an Advanced Technology fleet
(six production type LEV and ULEV vehicles). Only the Current and Tier 1 fleets were tested
for their response to changes in sulfur levels. Two fuel sets tested in this program can be used to
investigate the impact of fuel sulfur on exhaust emissions: a low T90 set and a high T90 set with
approximate sulfur levels of 33 and 317 ppm.
API Extension Fuel Set8-In this program, the 10-vehicle "Current" fleet from the Auto/Oil
program was tested at sulfur levels of 450 and 900 ppm to investigate the impact of the higher
levels of fuel sulfur observed in U.S. gasoline. Results from this program showed very modest
emission reductions as a result of reducing sulfur from 900 to 450 ppm (5% HC, 2% CO, and 3%
NOx).
EPA RFG Phase I Study4-Phase I was an initial investigation of the impacts of oxygenates,
volatility, distillation properties, and sulfur on emissions. Vehicles included in this program
represented 1990 model year or equivalent technology(Tier 0 vehicles). Two fuels examined in
this program had differing sulfur levels (112 ppm and 371 ppm) with other fuel parameters at
approximately constant levels. Results indicated that decreasing sulfur from 371 ppm to 112
ppm caused an approximate 5% reduction in HC emissions, a 7% reduction in NOx emissions,
and a 9% reduction in CO emissions in the fleet tested.
EPA RFG Phase Ft Studys-Phase Ft was a continuance of Phase I investigating further the effects
of oxygen content, oxygenate type, volatility, sulfur, olefins, and distillation parameters.
Relevant testing included fuels with sulfur levels of 59 and 327 ppm. Again, vehicles with 1990
model year or equivalent technology were tested. For the fleet tested, the results indicated that a
reduction in sulfur from 327 to 59 ppm caused an approximate 7% reduction in HC, a 5%
reduction in NOx emissions, and a 8% reduction in CO emissions.
API "Reversibility" Study7- API tested a series of vehicles in response to the issue of sulfur
reversibility in LEV and advanced technology vehicles. Sulfur has an almost immediate effect
-6-
-------
on catalyst performance, with the sulfur level of the fuel primarily impacting the speed with
which the catalyst is affected. One tankful of fuel containing high levels of sulfur will inhibit
catalyst performance to essentially the same degree as several tankfuls of fuel with somewhat
lower sulfur content. However, the return of catalyst performance upon refueling on low sulfur
fuel is not as prompt with the higher sulfur fuel. Therefore, the potential "reversibility" of the
sulfur effect could have substantial consequences for the design of a commercial sulfur control
program. That being the case, API undertook this study to begin to investigate this phenomena.
When this EPA report was being drafted, only very few vehicles had finished testing. Only one
of the vehicles tested that had accumulated 100K mileage (Ford Taurus-VIN #). This vehicle
was used in this analysis as part of the LEV emissions data set (all of which had approximately
100K mileage). The other vehicles in this test program were not included in the analysis either
because: 1) they did not have the 100K mileage accumulation or, 2) the testing was not
completed in the time frame during which the analysis for this report was completed. See
discussion below on why only vehicles with 100K mileage were thought to me most appropriate
for the purposes of this EPA report.
CRC Sulfur/LEV Study9- This testing involved 6 LDV models certified for sale in California as
LEVs in 1997. Two vehicles from each model type were tested on 7 fuels. Two fuel sets were
investigated: one fuel set was a California RFG with two sulfur levels (nominally 40 ppm and
150 ppm); the other set of five fuels consisted of a base, conventional fuel with five different
sulfur levels (nominally 40, 100, 150, 330, and 600 ppm). The same base gasoline was used for
all five of these fuels and the sulfur levels were varied by adding representative sulfur-containing
hydrocarbons. The vehicles were first tested in an "as-received" condition (average vehicle
mileage of 10,000 miles) and with the catalysts bench-aged to simulate 100,000 miles of
operation (although the oxygen sensors were original, low mileage sensors). The 10,000 mile
emissions data will hereafter be referred to as the "10K data" and the 100,000 data will be
referred to as the "100K data." The conclusions from this study included:
For the 10,000-mile catalysts, reducing sulfur from 600 to 40 ppm resulted in a fleet
emission FTP composite emission reduction in NMHC of 46% , in NOx of 63%, and in
CO of 57%.
For the aged 100,000-mile catalysts, reducing sulfur from 600 to 40 ppm resulted in a
fleet emission FTP composite emission reduction in NMHC of 32%, in NOx of 61%, and
in CO of 46%.
The fleet response to fuel sulfur changes was found to be linear for the 10,000-mile
catalysts and nonlinear for the 100,000-mile catalysts. With the aged catalysts, the effect
of sulfur was more pronounced at lower sulfur levels.
In this EPA analysis, only the 100K data was used (since the other major LEV/ULEV testing
program only tested vehicles with aged components to simulate 100,00 miles of driving-see the
next section below). Emissions data from both fuel sets (conventional and RFG gasoline) were
used in this analysis.
AAMA/AIAM Sulfur/LEV Studv10-This study tested 21 vehicles, each of different design: 9
-7-
-------
LEV LDVs, 1 LEV LDT1, 7 LEV LDT2s, and 4 ULEV LDVs. Some of the vehicle designs have
been certified for sale in California, while others were designs which were deemed ready for
certification and production. The vehicles were equipped with emission control components that
were aged to mimic 100,000 miles of on-road driving. The base fuel used in the program was a
California RFG with a nominal sulfur level of 40 ppm. The base fuel was then doped with sulfur
compounds to obtain nominal sulfur levels of 100, 150, 330, and 600 ppm. Based on the 21-
vehicle fleet, AAMA/AIAM reached the following conclusions:
The emissions benefits of low-emission vehicle hardware are diminished as fuel sulfur
level is increased above 40 ppm;
The LEVs and ULEVs tested in this program showed a larger detrimental effect from fuel
sulfur increases than the Tier 0 or Tier 1 vehicles tested in the Auto/Oil program; and
The emissions response of LEVs and ULEVs to fuel sulfur is non-linear for all pollutants
and is more pronounced at lower sulfur levels.
Valid Sulfur RangeThe range of sulfur data available for this analysis varied from 10 ppm to
900 ppm. Table 3 summarizes the range of actual data available to estimate sulfur's effect on
emissions by vehicle category:
Table 3
Range of Available Sulfur Data by Vehicle Technology and Type
Vehicle Technology/Type
Tier 0/Normals/LDVs
Tier 1/Normals/LDVs
LEV/Normals/LDVs
LEV/Normals/T rucks
High Emitters (Data
Available only for Tier 0
Vehicles)
Studies Available
Auto/Oil, EPA RFG, and
API Extension-Set Studies
Auto/Oil T50/T90 Study
AAMA/AIAM , CRC, and
API Reversibility Studies
AAMA/AIAM Study
EPA RFG Studies
Approximate Sulfur* Range for which Data is Available
and Over which Regressions were Based
10->900 ppm
30->350 ppm
30->600 ppm
30->600 ppm
40->450 ppm
All sulfur values in the databases are "actual" sulfur values (as opposed to "nominal" values)
While the regressions were based on all the data, the valid range for MOBILE6 will be limited to
30 ppm on the low end and 600 ppm on the high end. The main reasoning for these limits is that
LEV emissions data is only available over that range. Because LEV emissions are most
sensitive to sulfur fluctuations, it was thought that extrapolations would be both speculative and
possibly very inaccurate, especially below 30 ppm. Thus, in MOBILE6, sulfur's effect on
emissions will be limited to a range of 30 ppm on the low end and 600 ppm on the high end. For
consistency within MOBILE6 and for ease of use, this valid range will apply to all vehicle
-------
technologies and emitter classifications.
Analysis of Sulfur Data
Definition of Emitter Classes
Several comments since the October 1, 1997 MOBILE6 workshop have indicated that
sulfur's effect on emissions is a strong function of emission levels of individual vehicles and that
the Complex Model definitions of emitter classes are too broad and are based on hydrocarbon
emissions only. The comments further suggested that predictive equations based on normal
emitters with relatively low emissions (e.g., vehicles in the Auto/Oil program) may not be
directly applicable to the entire fleet of vehicles with HC emissions below the Complex model
normal-emitter definition of < 0.82 grams/mile HC. Some of the suggestions further indicated
that a "moderate emitter" category be established12. While this type of approach has merit in
estimating the effect of sulfur control on emissions, MOBILE6 is structured with only two
emitter classes, normals and highs due to lack of data. Thus, for this work, emitter categories are
defined in the following manner and are slightly different than the original proposal made at the
10/1797 workshop:
Table 4
Emitter Categories
Normal
Highs
< Two times the
> Two times
emission standard for NOx, and HC, and < Three
standard for CO
the emission standard for either NOx, or HC, or >
emission standard for CO
times the emissions
Three times the
Table 5 below shows the number of vehicles in each category that are normal and high emitters
according to the definition in Table 4. Note that there are no Tier 1 or LEV high emitters (as
defined in Table 4) in the database.
-9-
-------
Table 5
Distribution of Number of Vehicles in Each of the Emitter Categories
Defined in Table 3 for Studies Used in this Analysis
Study
All Auto/Oil
(all Tier 0 Vehicles)
EPA RFG Phase I
(all Tier 0 Vehicles)
EPA RFG Phase II
(all Tier 0 Vehicles)
Tier 1 T50/T90 Study
(all Tier 1 Vehicles)
CRC Sulfur/LEV Study
(LEV and ULEV Vehicles)
AAMA/AIAM Sulfur/LEV Study
(LEV and ULEV Vehicles and
Trucks)
TOTALS:
Number of Normal Emitters
10
20
24
6
12
21
93
Number of High Emitters
0
19
15
0
0
0
34
Regression Methodology and Mathematical Fits
Unless otherwise specified, all data sets were analyzed using the following regression
methodology. Individual fuel/vehicle data points were analyzed using a regression procedure in
SAS termed "ABSORB." Please consult the SAS manual for details on this procedure. Dummy
variables were used to absorb the vehicles' effect on emissions thereby allowing the fuel sulfur
effect to be isolated and better approximated. This approach is rather similar to the approach
used in the development of the reformulated gasoline Complex model in which a "dummy"
variable was created for each vehicle in the data set. Repeat tests on vehicles (and for the same
vehicle(s) used in different testing programs) at a given sulfur level were averaged to represent
one data point. Emissions were regressed against raw sulfur values (all sulfur values in this
report are in ppmW).
In all cases, two different mathematical fits were used to represent the emissions vs.
sulfur data: a log-linear fit (ln(emissions(in grams/mile)) = (Regression Coefficient) * S (in
ppm) + Constant}, and a log-log fit (In (emissions(in grams/mile) = (Regression Coefficient) * In
(Sulfur (in ppm)) + Constant}. The final decision on which fit to use in the different sulfur
regimes were based on accuracy of fit, previous knowledge of how sulfur affects emissions in the
regime being considered, quantity of data available, and other published work. Some of the
comments (resulting from the methodology proposed at the workshop on 10/1/97) suggested use
of a polynomial (quadratic) fit seemed to represent the emissions data. However, quadratic fits
and the non-linear logarithmic fits shown above yielded nearly the same level of accuracy and fit,
thus polynomial regressions were not included in the final analysis. The correlations developed
-10-
-------
in this analysis relating sulfur to emissions for use in MOBILE6 will be used only to estimate
percent changes in emissions when changing sulfur from one level to another. Whenever a
tabular entry in this report indicates "Ln-Ln" fit, it refers to the following relationship:
ln(Emissions (g/mile)) = [(Regression Coefficient) * In (S (in ppm))]+ C
Whenever an entry indicates a "Ln-Linear" fit, the corresponding mathematical relationship is:
ln(Emissions (g/mile)) = [(Regression Coefficient) *S (in ppm)]+ D
Note that C and D in the above equations are constants but are never used in the calculations. As
discussed previously, all regressions in this report can only be used for comparing emission
effects (i.e., percent change in emissions resulting from sulfur variation) and not for estimating
absolute or relative g/mile numbers.
Start. Running, and FTP-Composite Emissions
Start and running emissions were calculated from bag data using the methodology
outlined in the MOBILE6 EPA report entitled "The Determination of Hot Running and Start
Emissions from FTP Bag Emissions." It is report number M6.STE.002 and can be found on the
MOBILE6 web site11. Though this report used only Tier 0 vehicle data to generate the bag-to-
emissions mode correlations, the correlations were used for Tier 0, Tier 1, and LEV vehicles and
trucks (for both normal and high emitters) in this analysis. For regression purposes, FTP-
composite emissions were used as reported in the individual databases.
In most cases, the effect of sulfur on start emissions was statistically insignificant;
however, the magnitude of the effect, especially for NOx emissions, was similar to the sulfur
effect for running emissions. Since MOBILE6 will use start and running emissions to recalculate
FTP emissions whenever necessary, it is very important that these start effects be included.
Thus, regressions for start emissions will be developed and used for all vehicle technologies and
pollutants. It should be noted that, in a few cases, the sulfur effect on start emissions is negative.
While this is counterintuitive, it is supported by the data and assures that the composite
emissions are not too high.
Analysis of Tier 0 Normal Emitters
The sulfur impacts for normal-emitting Tier 0 vehicles will be based on analysis of the
entire Auto/Oil database, the API extension fuel set, and the EPA Phase I and Phase n RFG data
sets. The "ABSORB" procedure was applied to the Tier 0 data and regressed using the two non-
linear schemes discussed above. It was found that the log-log fit was consistently better than the
log-linear fit. The resulting correlations are shown below in Table 6 and the emission effects
resulting from these correlations are shown in Table 7:
Table 6
-11-
-------
Regression Analysis for Tier 0 Normal Emitting Vehicles
Pollutant
HC
NMHC
CO
NOx
HC
NMHC
CO
NOx
HC
NMHC
CO
NOx
Emissions Mode
Composite
Composite
Composite
Composite
Running
Running
Running
Running
Start
Start
Start
Start
Type of Regression
Fit
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Regression
Coefficient
0.06126
0.05502
0.07596
0.03077
0.15262
0.15187
0.19086
0.02083
0.0027436
0.0037181
-0.01792
0.04772
R2
0.963
0.959
0.950
0.939
0.947
0.918
0.886
0.944
0.959
0.961
0.860
0.862
Table 7
Emission Effects from Varying Sulfur for Tier 0 Normal Emitting Vehicles
Pollutant
HC
NMHC
CO
NOx
HC
NMHC
CO
NOx
HC
NMHC
Emissions Mode
Composite
Composite
Composite
Composite
Running
Running
Running
Running
Start
Start
% Increase in Emissions when Sulfur is Increased from 30 ppm to:
75
5.77
5.17
7.21
2.86
15.0
14.9
19.1
1.93
0.25
0.34
150
10.4
9.26
13.0
5.08
27.8
27.7
36.0
3.41
0.44
0.60
330
15.8
14.1
20.0
7.66
44.2
43.9
58.0
5.12
0.66
0.90
600
20.1
17.9
25.6
9.66
58.0
57.6
77.1
6.44
0.83
1.12
-12-
-------
CO
NOx
Start
Start
-1.63
4.47
-2.84
7.98
-4.21
12.1
-5.23
15.4
The continuum of composite emission effects from the regression equations in Table 6 are shown
in graphical format in Figure 1 in Appendix A. The Tier 0 analysis summarized in Tables 6 and
7 will apply to all normal emitters of Tier 0 or earlier (pre-Tier 0) categorization (all vehicles
equipped with a catalyst) since very little data is available to support an individual evaluation of
sulfur's effect on pre-Tier 0 vehicles. Pre-catalyst vehicles cannot be considered because sulfur
will have no direct effect on exhaust emissions from those vehicles.
As a comparison, Table 8 shows estimated emission effects of reducing sulfur from 450->50
ppm using the regressions listed in Table 6 for Tier 0 normal emitters and the effects computed
from the Complex Model12 for normal emitters. The results are similar for CO but the HC and
(especially the) NOx effects estimated in this EPA analysis are smaller when compared to the
NOx and HC effects predicted by the Complex Model. This is most likely due to inclusion of the
T50/T90 heavy-hydrocarbon Auto/Oil data set in this analysis. The T50/T90 heavy-hydorcarbon
Auto/Oil data was not available at the time the Complex Model was constructed. Inspection of
the T50/T90 heavy-hydrocarbon data shows muted somewhat muted HC effects and much lower
NOx effects for sulfur variations. Thus, the overall HC and NOx effects of reducing sulfur are
much lower in this analysis than that estimated by the Complex Model.
Table 8
Comparison of Composite Emission Effects for Normal Emitting
Tier 0 vehicles estimated from this Analysis to those Estimated from the Complex
Model when Sulfur is Reduced from 450 to 50 ppm
Approach
This EPA Analysis
Complex Model*
Percent Reduction in HC
13.0
19.0
Percent Reduction in NOx
6.6
13.6
Percent Reduction in CO*
15.4
18.5
* CO emissions were not part of the original RFG Complex Model. The CO model estimates are based on the CO model
developed separately (using the same statistical techniques used to construct the RFG Complex Model) from the RFG
rulemaking and discussed in SAE paper 96121413.
Analysis of Tier 1 Normal Emitters
Only one set of data3 has examined the effects of sulfur on emissions from certified Tier 1
vehicles. Two sulfur data points were tested in this analysis (~ 30 ppm and ~ 330 ppm) at high
and low levels of T90. Because only two sulfur data points were available, the log-linear version
of the fits were chosen to represent the data. Semi-log regressions were run using the procedures
outlined earlier and the regression coefficients and effects obtained are shown respectively in
Tables 9 and 10. It is interesting to note from comparing Tables 7 and 10 that Tier 1 emission
benefits (in percentage reduction space) from reducing sulfur are generally larger than Tier 0 for
CO and HC and about the same for NOx.
-13-
-------
Table 9
Regression Analysis for Tier 1 Normal Emitting Vehicles
Pollutant
HC
NMHC
CO
NOx
HC
NMHC
CO
NOx
HC
NMHC
CO
NOx
Emissions Mode
Composite
Composite
Composite
Composite
Running
Running
Running
Running
Start
Start
Start
Start
Type of Regression
Fit
Ln-Linear
Ln-Linear
Ln-Linear
Ln-Linear
Ln-Linear
Ln-Linear
Ln-Linear
Ln-Linear
Ln-Linear
Ln-Linear
Ln-Linear
Ln-Linear
Regression
Coefficient
8.053E-4
7.223E-4
6.295E-4
3.181E-4
2.457E-3
2.897E-3
1.746E-3
6.337E-4
9.516E-5
9.172E-5
-2.338E-4
8.023E-4
R2
0.765
0.748
0.907
0.903
0.818
0.785
0.911
0.853
0.941
0.936
0.820
0.692
Table 10
Emission Effects from Varying Sulfur for Tier 1 Normal Emitting Vehicles
Pollutant
HC
NMHC
CO
NOx
HC
NMHC
CO
NOx
HC
NMHC
CO
Emissions Mode
Composite
Composite
Composite
Composite
Running
Running
Running
Running
Start
Start
Start
% Increase in Emissions when Sulfur is Increased from 30 ppm to:
75
3.69
3.30
2.87
1.44
11.7
13.9
8.17
2.90
0.43
0.41
-1.05
150
10.1
9.05
7.85
3.89
34.3
41.6
23.3
7.90
1.15
1.11
-2.77
330
27.3
24.2
20.8
10.0
109.0
138.5
68.8
20.9
2.90
2.79
-6.77
600*
34.8
30.7
26.6
12.6
143.0
181.7
91.4
26.3
3.65
3.47
-8.41
-14-
-------
NOx
Start
3.68
10.1
27.2
34.6
* Please see explanation below about how the effects at 600 ppm were estimated.
Since the available Tier 1 data only extends to 330 ppmW sulfur, it would be inaccurate to use
the semi-logarithmic regression equations listed in Table 6 all the way out to 600 ppm, which
will be the valid high end of the sulfur range in MOBILE6. Instead, the regressions listed in
Table 6 will be applicable only for sulfur values between 30 and 330 ppm; and, for any sulfur
level above 330 ppm, the following equation will be used to estimate Tier 1 effects for a given
pollutant and a given emissions mode:
Tier 1 Effect at any sulfur level "X" above 330 ppm = [(TierOx)/(Tier0330)]*(Tierl330)
where,
TierOx = Tier 0 percent effect at level X using a 30 ppm as baseline
(can be estimated from Table 4)
Tier0330 = Tier 0 percent effect at 330 ppm using 30 ppm as baseline
(available in Table 5)
Tierl330 = Tier 1 percent effect at 330 ppm using 30 ppm as baseline
(available in Table 7)
For example, the Tier 1 effect of increasing sulfur to 600 ppm from 30 ppmw on running HC
emissions according to the above equation would be: (58.0%/44.2%)* 109.0% = 143.0 %. The
values 58.0% and 44.2% were obtained from Table 7 and 109.0% was obtained from Table 10.
The continuum of composite emission effects from the regression equations in Table 9 and the
approach described here are shown in graphical format in Figure 2 in Appendix A.
Analysis of LEV and ULEV Normal Emitters
As discussed in the "Data Sources" section above, 100K data from the recently completed
AAMA/AIAM and CRC testing programs will be used to develop sulfur impacts for LEVs and
ULEVs. Emissions from both the conventional and RFG set of fuels in the CRC testing program
were used in this analysis. The CRC 10K data was omitted because: (1) it was felt that is was
not representative of true in-use conditions and, (2) there was no accurate way to combine the
10K with the 100K data common to both testing programs. Because the AAMA/AIAM testing
program contained data on trucks, the impacts will be stratified by light-duty vehicles (passenger
cars and light trucks) and LDT2 trucks. This combination of passenger cars and LDT1 trucks
will be referred to as Light-Duty Vehicles (LDVs) hereafter. Emissions data from passenger cars
and LDT1 trucks were combined due to the technical similarities in their catalyst systems that
result in similar emission responses. Past EPA analyses14 have also traditionally combined these
two categories of vehicles. As suggested by the authors of the CRC and AAMA/AIAM reports,
log-log fits were found to be better for most of the pollutants and emission modes. Tables 11 and
12 show the regression statistics and the emission effects of changing sulfur for LDVs,
-15-
-------
respectively:
Table 11
Regression Analysis for LEV and ULEV Normal Emitting LDVs
Pollutant
HC
NMHC
CO
NOx
HC
NMHC
CO
NOx
HC
NMHC
CO
NOx
Emissions Mode
Composite
Composite
Composite
Composite
Running
Running
Running
Running
Start
Start
Start
Start
Type of Regression
Fit
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Regression
Coefficient
0.16845
0.13992
0.23746
0.35392
0.42809
0.49561
0.48626
0.57085
0.05067
0.05552
0.04847
0.11240
R2
0.947
0.944
0.917
0.889
0.879
0.859
0.915
0.904
0.958
0.954
0.941
0.723
-16-
-------
Table 12
Emission Effects from Varying Sulfur for LEV & ULEV Normal Emitting LDVs
Pollutant
HC
NMHC
CO
NOx
HC
NMHC
CO
NOx
HC
NMHC
CO
NOx
Emissions Mode
Composite
Composite
Composite
Composite
Running
Running
Running
Running
Start
Start
Start
Start
% Increase in Emissions when Sulfur is Increased from 30 ppm to:
75
16.7
13.7
24.3
38.3
48.0
57.5
56.1
68.7
4.75
5.22
4.54
10.8
150
31.1
25.3
46.5
76.8
99.2
122.0
118.7
150.6
8.50
9.35
8.11
19.8
330
49.8
39.9
76.7
133.6
179.1
228.2
220.9
293.1
12.9
14.2
12.3
30.9
600
65.6
52.1
103.6
188.7
260.5
341.4
329.2
453.0
16.4
18.1
15.6
40.0
Note that Table 12 shows ULEV and LEV normal emitting vehicles to have a much stronger
emissions response to sulfur changes than did Tier 1 vehicles (Table 10) or Tier 0 vehicles (Table
7). The continuum of composite emission effects upon changing sulfur on LEV LDVs are shown
in Figure 3 in Appendix A.
While the Tier 0 and Tier 1 analysis is based only on LDV data, truck data on the effect
of sulfur on emissions were also available in the AAMA/AIAM study. A total of 7 LDT2 trucks
were tested in the AAMA/AIAM program in addition to the testing conducted on LDVs. These
data were analyzed in the exact same manner as described above using "ABSORB" in SAS to
arrive at the regression analysis and emission effects shown in Tables 13 and 14, respectively:
-------
Table 13
Regression Analysis for LEV and ULEV Normal Emitting LPT2 Trucks
Pollutant
HC
NMHC
CO
NOx
HC
NMHC
CO
NOx
HC
NMHC
CO
NOx
Emissions Mode
Composite
Composite
Composite
Composite
Running
Running
Running
Running
Start
Start
Start
Start
Type of Regression
Fit
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Ln-Ln
Regression
Coefficient
0.12549
0.08956
0.15084
0.14625
0.31818
0.25326
0.38379
0.29491
0.02551
0.02846
0.07030
0.04130
R2
0.985
0.983
0.980
0.951
0.939
0.960
0.887
0.934
0.990
0.989
0.968
0.901
Table 14
Emission Effects from Varying Sulfur for LEV & ULEV Normal Emitting LPT2 Trucks
Pollutant
HC
NMHC
CO
NOx
HC
NMHC
CO
NOx
HC
NMHC
CO
NOx
Emissions Mode
Composite
Composite
Composite
Composite
Running
Running
Running
Running
Start
Start
Start
Start
% Increase in Emissions when Sulfur is Increased from 30 ppm to:
75
12.2
8.55
14.8
14.3
33.8
26.1
42.1
31.0
2.36
2.64
6.65
3.86
150
22.4
15.5
27.5
26.5
66.9
50.3
85.5
60.7
4.19
4.68
12.0
6.88
330
35.1
24.0
43.6
42.0
114.5
83.5
151.0
102.8
6.31
7.05
18.4
10.4
600
45.6
30.8
57.1
55.0
159.4
113.5
215.7
141.9
7.94
8.89
23.4
13.2
-------
Note that the sensitivity of emissions to changes in sulfur is much lower for LEV trucks than for
LEV LDVs. Figure 4 in Appendix A shows the emissions response to changes in sulfur for LEV
trucks.
Analysis of High Emitters
The emissions criteria for high emitters are listed in Table 3. Actual data on the effects of
sulfur on emission from high emitters is available only for Tier 0 vehicles as indicated in Table 4.
These data were used to determine regression coefficients for high emitters. A log-linear fit was
used since the amount of high emitter data available was small and only two sulfur levels were
tested in the EPA RFG programs. The regression coefficients for high emitters are shown in
Table 15 and the corresponding emission effects are shown in Table 16.
Table 15
Regression Analysis for Tier 0 High Emitting Vehicles
Pollutant
HC
NMHC
CO
NOx
HC
NMHC
CO
NOx
HC
NMHC
CO
NOx
Emissions Mode
Composite
Composite
Composite
Composite
Running
Running
Running
Running
Start
Start
Start
Start
Type of Regression
Fit
Ln-Linear
Ln-Linear
Ln-Linear
Ln-Linear
Ln-Linear
Ln-Linear
Ln-Linear
Ln-Linear
Ln-Linear
Ln-Linear
Ln-Linear
Ln-Linear
Regression
Coefficient
3.727E-5
3.727E-5
6.317E-6
3.046E-4
1.138E-4
9.614E-5
1.111E-4
2.848E-4
-2.227E-4
-1.824E-4
-5.336E-4
2.519E-4
R2
0.997
0.997
0.997
0.996
0.996
0.996
0.993
0.998
0.985
0.989
0.962
0.889
-------
Table 16
Emission Effects from Varying Sulfur for High Emitting Tier 0 Vehicles
Pollutant
HC
NMHC
CO
NOx
HC
NMHC
CO
NOx
HC
NMHC
CO
NOx
Emissions Mode
Composite
Composite
Composite
Composite
Running
Running
Running
Running
Start
Start
Start
Start
% Increase in Emissions when Sulfur is Increased from 30 ppm to:
75
0.17
0.17
0.03
1.39
0.51
0.43
0.50
1.29
-1.00
-0.82
-2.37
1.14
150
0.45
0.45
0.08
3.72
1.37
1.16
1.34
3.48
-2.64
-2.17
-6.20
3.07
330
1.12
1.12
0.19
9.57
3.47
2.93
3.39
8.92
-6.46
-5.32
-14.8
7.85
600
2.15
2.15
0.37
19.0
6.70
5.63
6.54
17.6
-11.9
-9.87
-26.2
15.4
The effects in Table 16 are in good agreement with the Complex Model which showed that NOx
effects were much more sensitive than HC effects to sulfur variation in Tier 0 high emitting
vehicles14. As an example, the Complex Model indicates that the effect of reducing sulfur from
450 to 50 ppm on high emitting vehicles to be an approximate 10% decrease in NOx emissions.
This EPA analysis shows the same effect to be approximately 11%. Table 17 is a comparison of
the emission effects estimated in this EPA report to the Complex Model estimates for high
emitters.
Table 17
Comparison of Composite Emission Effects for High Emitting
Tier 0 vehicles estimated from this Analysis to those Estimated from the Complex
Model when Sulfur is Reduced from 450 to 50 ppm
Tool
This EPA Analysis
Complex Model
Percent Reduction in HC
1.5
-5.00
Percent Reduction in CO*
0.3
1.4
Percent Reduction in NOx
11.2
10.0
* The same footnote as in Table 8 applies here
Note that in some cases, in Table 16, reducing sulfur may actually increase start emissions.
Despite these counterintuitive effects, MOBILE6 will apply the start effects exactly as reported in
Tables 15 and 16 . This is because MOBILE6 will combine running and start emissions to
estimate composite emissions whenever necessary and any manual adjustments to start emissions
-------
(such as "zeroing out" the counterintuitive effects) would result in skewing of the composite
emission estimates made by MOBILE6.
The regression coefficients shown in Table 15 will be used for Tier 0 vehicles. However, no
high emitter test data relating sulfur to emissions exists for any other category of vehicles. Thus,
a simple algorithm is required to estimate the emission sensitivity to sulfur changes in vehicles
certified to Tier 1 and cleaner standards. The following methodology will be used to estimate
high emitter effects for other-than-Tier 0 vehicles. Analysis of the Complex Model15 indicates
that the NOx sensitivity of high emitters is approximately 60% of the sensitivity for normal
emitters in percent-emissions-change space. Thus, for vehicles and trucks certified to Tier 1 and
cleaner standards (LEV, ULEV, Tier 2, etc.), this correction factor will be applied to estimate a
NOx effect for high emitting vehicles. A high emitter in any given vehicle category is defined by
the criteria listed in Table 3. For example, if the normal emitter NOx emissions effect of
reducing sulfur from 150 to 50 ppm is 25% for Tier 1 (or LEV or ULEV) vehicles, then high
emitters in this same category would get a NOx benefit of (0.6)*(25%), or 15%.
The Complex Model does not show nearly as great or as consistent a CO or HC effect1344 and
sensitivity to changes in sulfur from high emitting vehicles; thus, the high-emitter regression
coefficients listed as-is for HC and CO in Table 15 will be used for all vehicle categories. Table
18 summarizes the high emitter effects to be used in MOBILE6:
-------
Table 18
Summary of High-Emitter Effects
Vehicle Category
Tier 0-LDVs
Tier 0-LDVs
Tier 0-LDVs
Tier 0-LDVs
Tier 1-LDVs
Tier 1-LDVs
Tier 1-LDVs
Tier 1-LDVs
LEV & ULEV-LDVs
LEV & ULEV-LDVs
LEV & ULEV-LDVs
LEV & ULEV-LDVs
LEV & ULEV-LDT2s
LEV & ULEV-LDT2s
LEV & ULEV-LDT2s
LEV & ULEV- LDT2s
Pollutant
HC
NMHC
CO
NOx
HC
NMHC
CO
NOx
HC
NMHC
CO
NOx
HC
NMHC
CO
NOx
High Emitter Effect
Use Appropriate Regression
Coefficient listed in Table 15
Use Appropriate Regression
Coefficient listed in Table 15
Use Appropriate Regression
Coefficient listed in Table 15
Use Appropriate Regression
Coefficient listed in Table 15
Use HC Regression Coefficient
listed in Table 15
Use NMHC Regression Coefficient
listed in Table 15
Use CO Regression Coefficient
listed in Table 15
Use (0.60* Normal Emitter Tier 1
Effect (calculated from Table 9))
Use HC Regression Coefficient
listed in Table 15
Use NMHC Regression Coefficient
listed in Table 15
Use CO Regression Coefficient
listed in Table 15
Use (0.60* Normal Emitter LEV
LDV Effect (calculated
from Table 1 1))
Use HC Regression Coefficient
listed in Table 15
Use NMHC Regression Coefficient
listed in Table 15
Use CO Regression Coefficient
listed in Table 15
Use (0.60* Normal Emitter LEV
LDT2 Effect (calculated
from Table 13))
Summary
-------
Empirical relationships were developed for relating sulfur to exhaust emissions from
Tier 0, Tier 1, and LEV vehicles based on available data. Separate correlations were obtained for
light-duty vehicles and trucks wherever possible. Analysis was conducted on two separate emitter
classifications, normal and high. The valid sulfur range for MOBILE6 was estimated to be
30
-------
12. EPA's Complex Model for certifying Reformulated Gasolines. Please consult the RFG
rulemaking documents.
13. Rao. V., "Development of an Exhaust Carbon Monoxide Emissions Model" SAE Paper
No. 961214, 1996.
14. EPA Final Regulatory Impact Analysis for Reformulated Gasoline, December 13, 1993.
15. Private communication with EPA's Rick Rykowski, National Expert, Engine Programs
and Compliance Division, August 1998.
-------
APPENDIX A
-------
Figure 1: COMPOSITE EMISSIONS
Tier 0 Normal Emitters
60 120 180 240 300 360 420 480 540 600
Sulfur (ppm)
Figure 2: COMPOSITE EMISSIONS
Tier 1 Normal Emitters
60 120 180 240 300 360 420
Sulfur (ppm)
480 540 600
-------
E
Q.
Q.
E
o
.C
CD
ro
CD
Figure 3: COMPOSITE EMISSIONS
LEV+ULEV Normal Erritting LDVs
200
150 -
100 -
0 I«i
0 60 120 180 240 300 360 420 480 540 600
Sulfur (ppm)
Figure 4: COMPOSITE EMISSIONS
LEV+ULEV Normal Emitting Trucks
60 120 180 240 300 360 420 480 540 600
Sulfur (ppm)
------- |