United States       Air and Radiation      EPA420-R-98-005
             Environmental Protection                May 1998
             Agency
&EPA       EPA Staff Paper on
             Gasoline Sulfur Issues

-------
             United States        Air and Radiation       EPA420-R-98-005
             Environmental Protection                 May 1998
             Agency
vvEPA       EPA Staff Paper on
             Gasoline Sulfur Issues

-------
                                   EPA420-R-98-00
    EPA Staff Paper

            on

  Gasoline Sulfur Issues
         May 1, 1998
U.S. Environmental Protection Agency
     Office of Mobile Sources

-------
                    EPA Staff Paper on Gasoline Sulfur Issues

                                 Executive Summary
       In response to a Congressional mandate contained in the Clean Air Act, the U.S.
Environmental Protection Agency (EPA) recently published a draft study of the air quality nee
for and the feasibility and cost-effectiveness of emissions reductions from light-duty vehicle
(LDVs) and light-duty trucks (LDTs) beyond those required by the existing Tier 1 emission
standards.  EPA plans to follow this study by the end of 1998 with a notice of proposal for thi  '
next level of emission standards, Tier 2 standards.  In evaluating the technical feasibility of
additional emission controls, the draft Tier 2 study identified commercial gasoline sulfur level
as a potentially significant factor in determining the emission control potential of future vehicles.
Hence, as part of establishing Tier 2 standards, the Agency will also establish a gasoline sulfur
control program, to be proposed in the same time frame.

       This paper presents EPA's current understanding of the impact of gasoline sulfur on
emissions from current  and future motor vehicles. The paper also explores what  can be done by
gasoline producers and  automobile manufacturers to reduce sulfur's impact on  vehicle emissions.
The Impacts of Gasoline Sulfur on Emission Control Systems

       Modern gasoline-fueled vehicles use catalysts to reduce emissions of hydrocarbons (HC),
carbon monoxide (CO), and oxides of nitrogen (NOx) by over 90 percent (%).  Sulfur is a
catalyst poison. Sulfur and sulfur oxides inhibit the reaction of HC, CO, and NOx on the
catalytic surface. Sulfur also interferes with the management of oxygen on the catalyst surface,
which must be precisely controlled to maximize NOx emission reductions. The degree of
sulfur's interference appears to depend on many factors, such as the specific design of the
catalyst, the range of exhaust temperatures experienced by the catalyst, the materials used to store
oxygen in the catalyst, and the range of air-fuel mixtures fed to the engine. Of all of the noble
metals used in modern catalysts, palladium appears to be the most susceptible to sulfur
poisoning. Likewise, of oxygen storage media, ceria appears to be the most sensitive to sulfur.
Palladium is currently the catalyst of choice of most vehicle manufacturers for  their emission
control systems designed to meet California's low emission vehicle (LEV) standards. Palladium
is able to withstand higher temperatures, which allows the catalyst to be placed closer to the
engine and start working sooner after a cold start. Ceria allows the catalyst to operate at its
maximum efficiency over a wide range of vehicle operating conditions, which  is critical to
reducing emissions in-use and not just in the laboratory.

       It is possible that future optimization of catalyst and engine designs could reduce the
impact of sulfur on catalyst performance.  For example, the increased use of the precious metal
rhodium in catalysts and the maintenance of consistently high temperatures during typical

EPA Staff Paper:  Gasoline Sulfur                                                 May 1, 199

-------
operation (e.g., above 650° C) may reduce the negative emissions impact of sulfur. Increasing
the total precious metal content of the catalyst can also reduce the impact of sulfur, though at
significant cost. However, due to sulfur's inherent chemical nature, it is highly unlikely that the
impact of sulfur can be completely eliminated. As tighter emission standards require higher
catalyst efficiencies, the impact of sulfur is bound to be more dramatic.

       Automakers and oil companies have both recently completed extensive test programs to
investigate the impact of sulfur on emissions from vehicles meeting California LEV standards.
These vehicles are currently being sold in the Northeast and will be sold nationwide starting in
2001, as part of the voluntary National Low Emission Vehicle (NLEV)  program.  Some of the
results of this testing are summarized in Table El below. Many of the vehicles tested failed their
certified emission standards when using higher sulfur fuels.
Table El. Average Increase in Emissions for LEVs and ULEVs Relative to 40 ppm Sulfur
[Source: CRC and AAMA/AIAM Test Programs]
Pollutant
Fuel Sulfur, ppm
All LDV/LDT1 a
All LDT2/LDT3 b
NMHC
150
26.7%
23.0%
330
43.0%
26.4%
NOx
150
65.7%
33.7%
330
135.6%
65.5%
LDV/LDTls Broken Down by NMHC+NOx Emissions with 40 ppm Sulfur Fuel
9 Lowest Emitters
9 Highest Emitters
26.8%
26.6%
44.0%
42.1%
127%
41.2%
264%
84.2%
LDV/LDTls Broken Down by the Sensitivity of NMHC+NOx Emissions to Sulfur
9 Least Sensitive
9 Most Sensitive
32.2%
25.1%
49.5%
42.2%
31.0%
122%
61.0%
251%
"LDV = light duty vehicle (passenger car); LDT1 = light duty truck, class 1
"LDT2 = light duty truck, class 2; LOT 3 = light duty truck, class 3
       These emission impacts are very significant, particularly for NOx emissions. For
example, the data indicate that if a LEV LDV emits 0.30 grams per mile (g/mi) NOx with 40
parts per million by weight (ppm) sulfur fuel at 100,000 miles (which is the LEV emission
standard for NOx), this vehicle would emit 0.71 g/mi NOx with typical non-California fuel.  This
level exceeds the NOx standard for Tier 1 LDVs, which these LEVs replaced. Even in RFC
areas, after implementation of the Phase n RFG standards in 2000 (which are expected to result
in average sulfur levels of 150 ppm), this LEV LDV would emit 0.50 g/mi.
EPA Staff Paper: Gasoline Sulfur
May 1, 1998

-------
       Sulfur increased the emissions from every vehicle in these test programs (i.e., not one
vehicle was completely insensitive to sulfur). However, some of the vehicles were less sensitive
to sulfur than others. This variability is not surprising, given that these vehicles were primarily
designed to be operated on California Phase 2 reformulated gasoline, which contains less than 40
ppm sulfur on average. As mentioned above, vehicle manufacturers may be able to directionally
reduce the impact of sulfur in future vehicles.

       To illustrate this variability, the impact of sulfur on four subsets of the 18 unique LDVs
and LDTls tested in the two industry programs are also summarized in Table El. The impact of
sulfur on NMHC emissions from the nine vehicles with the lowest NMHC+NOx emissions on
California-like 40 ppm fuel is essentially the same as the impact of sulfur on the remaining nine
vehicles. However, the impact of sulfur on NOx emissions from the nine vehicles with the
lowest NMHC+NOx emissions on California-like 40 ppm fuel is much greater than the impact of
sulfur on the remaining nine vehicles. This is consistent with the overall trend observed that
sulfur's impact, particularly on NOx, increases as the overall stringency of emission control
increases.

       At the same time, NOx emissions from the nine vehicles showing the least sulfur
sensitivity (in terms of NMHC+NOx emissions) are much less  sensitive to sulfur than the
remaining nine vehicles. The opposite is true for NMHC emissions, because the change in NOx
emissions with higher sulfur is much greater than that for NMHC.

       EPA expects that auto manufacturers should, in general, be able to reduce the sulfur
sensitivity of LEVs and ULEVs to that of the nine least sensitive vehicles. However, one factor
not reflected in these data is the need to meet stringent emission standards, applicable during
periods of aggressive driving and use of the air conditioner, starting in 2001.  These "off-cycle"
standards will place additional limits on the fuel management strategy and catalyst designs used
by manufacturers.  In particular, more precise air-fuel mixture control will likely be required.
This could eliminate one method which appears to reduce sulfur's emission impact, wide swings
in the air-fuel ratio. Future Tier 2 vehicles may also have to meet more stringent emission
standards than today's LEVs, which could further limit the manufacturers' choices for emission
control system designs. EPA plans to further analyze the emission data produced in these test
programs to determine the degree that off-cycle standards would affect manufacturers' abilities to
design emission control systems with lower sulfur sensitivity.

       While sulfur has a large impact on emissions from current low-emission vehicles,
technological advances are also being made to improve the fuel efficiency of the nation's
vehicles. Gasoline-direct injection engines and fuel cells are two of the advanced powerplants
being developed. Both of these technologies may be more sensitive to sulfur than current
vehicles; current designs of these powerplants can tolerate very little gasoline sulfur. Therefore,
gasoline sulfur removal is not only important to maintain the emission control potential of
current vehicles, but is being highlighted by many as an  important technology enabler in the
future.
EPA Staff Paper: Gasoline Sulfur                  iii                              May 1,1998

-------
       Finally, the automotive industry has expressed concern that sulfur will also interfere with
the on-board diagnostics (OBD) systems designed to monitor catalyst performance.. OBD
catalyst monitors utilize oxygen sensors which incorporate catalytic materials which can be
compromised by sulfur much like the exhaust catalyst. EPA has evaluated the impact of sulfur
on OBD catalyst monitors in the past and determined that any significant impacts should be
limited to a few specific model lines. EPA is willing to address any sulfur-related OBD
compliance problems on a model-specific basis during certification.
Fuel and Vehicle Emission Controls as a System

       Vehicles should be designed to meet emissions standards based, in part, upon the range of
fuels that are used to certify the emissions performance as well as the range of fuels likely to be
encountered in-use. Currently, the test fuels used by industry and EPA for federal vehicle
certification reflect a narrow range of permitted fuel specifications and can differ substantially
from the gasolines actually sold nationwide, particularly with respect to sulfur content. The
sulfur level of the test fuel generally used in federal certification is typically less than 100 ppm,
often approaching that of California certification fuel, which averages about 30 ppm sulfur.
However, while California restricts the sulfur content of commercial gasoline to about 30 ppm on
average, gasolines sold outside of California average over 300 ppm sulfur.  Even federal Phase 2
reformulated gasoline, which will not be available until 2000, is expected to average 150 ppm,
still well above that of typical federal certification test fuel.

       For Tier 1 and earlier vehicles, this difference in sulfur levels between certification test
fuel and commercial gasolines probably did not affect the emission control technology selected
for use by automakers, since sulfur's effect on emissions from these vehicles was smaller.  With
advances in emission control technology, this is no longer the case.  Therefore, the arguments in
favor of changing the sulfur content of certification test fuel to match that of commercial gasoline
are increasing.  The only issue is whether the sulfur content of certification test fuel should be
increased or that of commercial gasoline reduced, or both.

       Increasing the sulfur content of the certification test fuel would either increase the
stringency of the Tier 2 standards (by requiring the same level of emission reduction with a
higher sulfur fuel), or result in numerically higher standards than would otherwise be established
(to accommodate the sulfur impact). Reducing the sulfur content of commercial gasoline would
do the opposite. It would also reduce emissions from the current fleet of vehicles, reduce sulfur
dioxide and sulfate emissions from all vehicles, and potentially enable advanced low emission
and fuel efficient technologies.  These are the options available for addressing the current
difference in sulfur levels between typical certification test fuels  and commercial fuel.
EPA Staff Paper: Gasoline Sulfur                 iv                              May 1,1998

-------
Reducing the Sulfur Content of Commercial Gasoline

       Sulfur occurs naturally in crude oil. As crude oil is refined, some of the sulfur ends up in
gasoline. The amount of sulfur in the gasoline from any particular refinery can vary widely
depending on the specific crude oil processed and the processing capability of the refinery.

       While a number of refinery processing options are available to reduce the sulfur content
of gasoline, the most likely technique would be to remove most of the sulfur from the products of
the refinery's fluidized catalytic cracker (FCC) which typically end up in gasoline. These FCC
products are the source of the great majority of sulfur in gasoline. Just as current sulfur levels
vary between refineries, the cost of removing sulfur varies, as well.

       Significant sulfur reductions would likely require investment in new refinery processing
hardware, as well as increased operating costs. EPA, with help from the Department of Energy
(DOE), has developed preliminary estimates of the cost to control sulfur from current average
levels of all the gasoline produced in the Northeast and the Gulf Coast of the U.S. (see Table E2).
MathPro, Inc. has also projected sulfur reduction costs for these refiners, plus those in the
Midwest, under contract to the American Petroleum Institute (API). Both of these sets of
projections are based on current sulfur removing technologies. As can be seen, the projected
costs of sulfur removal increase substantially as sulfur is reduced from 100 ppm to 40 ppm.  The
MathPro/API projections are higher than those of EPA and DOE for control to 150 ppm and 100
ppm sulfur, but below the low end of EPA/DOE's projected cost range for 40 ppm sulfur.  EPA
and DOE are currently pursuing estimates of the cost of reducing sulfur levels for Midwest,
Rocky Mountain, and Far West refineries.
Table E2. Preliminary Sulfur Reduction Costs (cents per gallon)
Sulfur Control
Level (Average),
ppm
150
100
40
EPA/DOE: Northeast and Gulf
Coast Refiners With Current
Technology
1.1-1.8
1.9-3.0
5.1-8.0
API/MathPro: Northeast, Midwest
and Gulf Coast Refiners With
Current Technology
2.7
3.4
5.1
       While conventional technology is capable of reducing the sulfur content of gasoline, EPA
has recently learned of emerging sulfur removal technologies which purport to be much less
costly. Companies which license these technologies estimate that the cost of reducing sulfur to
40 ppm (on average) is about 1-2 cents per gallon. EPA and DOE are currently evaluating these
technologies to assess whether these projected costs are reasonable.  If so, then these new
processes would dramatically reduce the projected cost of sulfur control.
EPA Staff Paper: Gasoline Sulfur
May 1, 1998

-------
Sulfur Control Program Considerations

       Domestic and overseas automakers recently jointly petitioned EPA to implement an 80
ppm sulfur cap on all gasoline sold nationwide starting in 2004. (This cap is expected to result in
an average sulfur level of about 40 ppm.)  Several state organizations as well as automotive
catalyst manufacturers have made similar recommendations. In contrast, the trade organizations
representing the  majority of gasoline refiners have proposed a regional sulfur reduction program.
Under this proposal, gasoline sulfur would be limited to 150 ppm on average in a 22-state region
during the summer, with a 300 ppm average sulfur restriction on gasoline sold elsewhere during
the summer.

       Clearly, a broad consensus now exists thai: the sulfur content of commercial gasoline must
be reduced to protect the current investments being made in vehicle emission control
technologies and to facilitate the achievement of even more stringent emission standards in the
future. Questions remain regarding the appropriate level and form of the sulfur limit.  The level
of sulfur reduction, and whether the limit is based on a refinery average or a per-gallon
maximum, will depend primarily on the factors already discussed. Once the technical concerns
of the emission control systems are answered, the greatest difference between a sulfur standard
based on average levels and one based on an absolute cap on sulfur content lies in the
implications for compliance and enforcement of the requirement. An averaging standard  would
increase the recordkeeping and reporting requirements, and may necessitate the implementation
of periodic surveys to ensure compliance not just at the refinery, but at the retail level. In
addition, the identification of the appropriate level of sulfur control will help determine if
gasoline sulfur levels must be fairly consistent over time and over different covered areas.
Otherwise, compliance and enforcement will have to be designed to ensure that there are not
great variations in sulfur levels between cities.

      While the specific sulfur limit will be based largely on vehicle technology needs, the form
of the sulfur program will be based, in part, on two additional factors: regionality and
seasonality.  If the ozone benefits of Tier 2 standards are of primary importance, as assumed by
the refiners' associations proposal, a regional program could be defined to encompass most of the
projected future ozone non-attainment areas, and could be designated only for the time of year
when ozone is most problematic.  Limiting sulfur control to specific regions of the country or to
a specific  season would likely reduce  the cost of the program, as the cost of removing sulfur
increases as more gasoline is affected. Under a regional program, refiners who find it less
expensive to reduce sulfur will supply the majority of low sulfur gasoline, while others refiners
would still have  a market for higher sulfur fuel. Compliance costs under a program involving
multiple sulfur levels would be somewhat more costly, however, since fuels would likely  have to
be segregated  and compliance from the refinery to the retail level would have to be ensured.

       A  key requirement for a regional sulfur program is that the impact of sulfur be reversible.
That is, emissions would  return to their original low levels once use of high sulfur fuel ceased.
While the available test data are limited, some vehicles' emissions appear to be reversible, while
EPA Staff Paper: Gasoline Sulfur                  vi                              May 1, 1998

-------
others' are not easily reversible. The same factors described above which determine the
sensitivity of a catalyst to sulfur (catalyst design, air-fuel ratio, and temperature) also appear to
affect the ease of reversibility.  Likewise, the reversibility of the sulfur effect could be impacted
in the future by off-cycle emission controls.  While the existence of sulfur reversibility would not
in itself be sufficient reason to focus on a regional or seasonal sulfur program, its absence would
be sufficient to only consider national and year-round sulfur limits. Also, any degree of
irreversibility effectively increases the stringency of the emission standards for auto
manufacturers and/or increases their in-use liability, since vehicles tested in an in-use recall
program (even if tested with low sulfur fuel) could have been operated on higher sulfur fuel at
some point in their lives.

       Just as there are many benefits to a regional control program, there are numerous reasons
for a single national sulfur standard. The Tier 2 vehicle standards  are expected to apply
nationally (e.g., all states except California); a single commercial gasoline sulfur standard
applicable nationwide (all states except California) would allow vehicle manufacturers to design
these vehicles for a single in-use federal fuel, instead of a range of in-use fuels. A single
nationwide sulfur standard (particularly a per-gallon limit) would simplify the enforcement of the
requirement and may decrease the possibility of sulfur-induced OBD problems. Local areas with
inspection and maintenance (I/M) programs which lay outside of a sulfur control region would
also avoid the need for special procedures and cut-points to  compensate for their higher fuel
sulfur levels.  Furthermore, many of the emission reductions associated with the potential Tier 2
standards and gasoline sulfur reduction would be beneficial  nationwide (e.g., reductions in
particulate matter (PM) formed from VOC, NOx, and oxides of sulfur (SOx) emissions,
reductions in air toxic emissions, etc.). A national sulfur limit would also be desirable if lower
sulfur levels were needed to permit the future introduction of vehicle technologies which provide
significant reductions in emissions or fuel consumption. California and Japan have imposed
stringent sulfur limits and Europe will decide on a stringent sulfur  limit shortly. Thus, without
similar sulfur limits throughout the U.S., advanced technology designed for these markets may
not be available in the U.S.

       One additional argument for a national sulfur standard is that many  areas outside of the
region covered by the refining industry proposal are projected to only marginally attain the ozone
standard in the future. If these areas experience economic and population growth, additional
reductions in VOC and NOx emissions will likely be necessary to ensure maintenance of the
ozone standard.  Thus, it is not clear that a regional strategy similar to the region identified by th
refiner associations is sufficiently large to ensure that ozone-related benefits occur in every area
where they are needed.

       The costs of a national program, however, are likely to be substantially greater than for a
regional program.  Controlling every gallon of gasoline will require that all refiners  take action;
there would be no  market to accommodate the gasoline that is most expensive to control.  Given
that there are some regional variations in the sulfur content of gasoline, and that some refiners are
better positioned than others to reduce gasoline  sulfur, a national program could result in
EPA Staff Paper: Gasoline Sulfur                 vii                             May 1, 1998

-------
significant variations in costs between regions, and individual refiners in some regions could
experience greater economic hardship than the average refinery. The higher production costs of a
national program could be offset in part by lower relative compliance and enforcement costs, but
that depends on how the program is designed.
EPA's Next Steps

       Based on the data and information obtained to date and the consensus among a wide
variety of stakeholders, EPA staff recommend that the Agency develop proposed gasoline sulfur
standards to be implemented coincidentally with the implementation of the planned Tier 2
emission standards for LDVs  and LDTs. There are many issues, such as the health and welfare
benefits of gasoline sulfur control, which will need to be addressed as EPA proceeds. This Staff
Paper is intended to further dialog on these issues, and foster further research.

       To identify the appropriate actions, EPA will work closely with all stakeholders. In May,
1998, a public workshop will  be held to discuss the technical issues raised in this Staff Paper.
Throughout the coming months, EPA will continue to interact with individual companies, trade
organizations,  the states, and environmental interests to determine the best approach to dealing
with this issue.
 EPA Staff Paper:  Gasoline Sulfur                 viii                             May 1, 1998

-------
                    EPA Staff Paper on Gasoline Sulfur Issues
I.     Introduction

       The U.S. Environmental Protection Agency (EPA) is considering setting limits on the
sulfur content of gasoline at this time primarily because advanced automotive catalysts are more
susceptible to poisoning by higher sulfur levels.' This susceptibility to poisoning has become a
critical issue as motor vehicle emission standards have become more stringent. Setting a limit on
sulfur content would also reduce emissions from current technology vehicles. The series of
reports published by the joint efforts of the automotive and oil industries (Auto/Oil)2 and other
studies in the early 1990's demonstrated that higher fuel sulfur levels increased non-methane
hydrocarbon (NMHC), carbon monoxide (CO), and oxides of nitrogen (NOx) emissions. With
Tier 0 and Tier 1 vehicles, sulfur levels up to 500 parts per million (ppm) increased emissions by
up to  15-20 percent (%) over emissions with sulfur levels around 40 ppm.  While the effect of
sulfur was significant, in-use levels of sulfur did not play a substantial role in defining the types
of catalysts used on motor vehicles meeting Federal Tier 0 or Tier 1 standards.

       Recent advances in both engine controls and catalyst designs have allowed a number of
current production models to meet either California's low emission vehicle (LEV) or ultra-low
emission vehicle (ULEV) standards without having to resort to novel emission control
technology, such as electrically heated  catalysts or hydrocarbon (HC) adsorbers, or to fuels other
than gasoline.  However, very recent test data show that emissions from vehicles designed to
meet these LEV and ULEV standards can more man double with the use of higher sulfur
gasoline. The sensitivity of these current LEVs and ULEVs to sulfur is a great cause of concern.
Some LEVs have already begun to be sold in Northeast states under state-adopted California
LEV programs and the. recently established National LEV (NLEV) program.  More LEVs will be
sold nationwide beginning in 200! under NLEV. Higher sulfur levels have the potential  to
significantly reduce the benefits of the  NLEV program if left unchanged.

       EPA is also in the process of evaluating whether potential federal Tier 2 standards should
replace the NLEV program in 2004. EPA recently published a draft study mandated by Congress
of the air quality need for and the feasibility and cost effectiveness of Tier 2 Standards.3  EPA
plans  to follow this study by the end of 1998 with a notice of proposal for Tier 2 standards.  As
the emission control technology used to meet potential Tier 2 standards could be very similar in
design to that used on LEVs and ULEVs, emissions from potential Tier 2 vehicles could also be
very sensitive to sulfur and impede progress in developing emission controls capable of meeting
       1  Sensitivity is used herein to refer to the percentage change in emissions for a specified change in fuel
sulfur content (e.g., 50% emission increase for an increase in sulfur from 40 to 300 ppm).

       2  Auto/Oil Air Quality Improvement Research Program, various publications, 1992-1996.

       3  U.S. EPA, "Tier 2 Study," Draft Report to Congress, April 23, 1998.

EPA Staff Paper: Gasoline Sulfur                                                 May 1,1998

-------
more stringent emission levels.  Thus, EPA has decided to evaluate the need for limits on the
sulfur content of commercial fuel which would be developed in conjunction with the potential
Tier 2 standards and proposed in the same time frame. This paper is the first step in the process
of developing these commercial fuel sulfur limits.

       The remaining sections of this paper discuss:

1)     The interactions between sulfur and vehicle technology, including the effect of sulfur on
       emissions from LEVs and ULEVs, the ability of these vehicles to recover their emission
       performance after temporary operation on high sulfur fuel, and the effect of fuel sulfur on
       advanced vehicle technology designed for high fuel efficiency or extremely low
       emissions;

2)     The impact of sulfur control on refinery operations, including the technology used to
       reduce sulfur levels, the cost of sulfur removal, and the ability of the industry to make the
       necessary investments;

3)     Other fuel quality issues which could affect refiners' approaches to sulfur control,
       including limits on certain distillation properties of gasoline and limits on onroad or
       nonroad diesel fuel properties (particularly sulfur);

4)     Actions which have been taken to reduce the sulfur content of gasoline in other countries;

5)     Recommendations regarding levels of sulfur control which have been made by
       stakeholders to date; and

6)     The paper's conclusions.

       Before moving to these six remaining sections, it would be helpful to provide an
overview of the key factors which affect the need for gasoline sulfur control and the direction
these factors provide regarding the design of a sulfur  control program for the U.S.

       Tremendous progress has been made over the past several years in developing technology
to reduce emissions from gasoline-fueled vehicles. This progress has focused on two inter-
related areas: the exhaust catalyst and management of the air and fuel mixture inducted into the
combustion chamber (e.g., the air-fuel ratio).  Automotive catalysts are now far  more efficient in
removing NMHC, CO, and NOx emissions than they were just five years ago. These catalysts
can also withstand much higher temperatures,  which  allow them to be placed closer to the
engine. This in turns shortens the time it takes for the catalyst to begin operating ("light-off)
after the engine is started, dramatically reducing emissions.  Precise control of the air-fuel ratio is
important, because the catalyst can only achieve its maximum efficiency when the air-fuel ratio is
within a narrow range.  Advanced computerized engine controls are able to maintain  the proper
air-fuel ratio over a wide range of operating conditions.  At the same time, improved catalyst
EPA Staff Paper: Gasoline Sulfur                  2                              May 1, 1998

-------
promoters within the support material (washcoat) have improved the ability of the catalyst to
withstand temporary swings in oxygen levels in the exhaust when the engine controls are not able
to precisely control the air-fuel ratio.

       Sulfur interacts chemically with both the catalyst and the washcoat which manages
oxygen on the catalyst surface. Sulfur compounds adsorb onto the surface of these materials and
prevent pollutants, oxygen, and other exhaust compounds from interacting with the catalyst
surfaces. This inhibits the ability of the catalyst to convert pollutants into less harmful
emissions. Emission control technology designed for California, which has the most stringent
vehicle emission standards in the world, is often used eventually on vehicles sold in the rest of
the U.S., Europe, and Japan, so those developing emission control systems have a tremendous
incentive to develop technology for California. Since California restricts the sulfur content of
gasoline statewide to no more than 80 ppm, catalyst developers have focused on developing
catalysts for this low level of sulfur. They also evaluate their prototype catalysts with slightly
higher 100 ppm sulfur fuel, as this is the typical upper limit of the sulfur level of the test fuel
used by industry and EPA in the federal motor vehicle compliance programs (commonly referred
to as Indolene).4 This higher sulfur level can significantly reduce catalyst performance.
However, this difference (100 ppm versus 30 ppm) is small relative to the sulfur levels of current
U.S. gasoline outside of California, which averages over 300 ppm sulfur and can reach levels as
high as 1000 ppm.

       In the past, this difference in federal test fuel and commercial fuel sulfur levels appeared
to cause only a small increase in emissions (e.g., 5-10% on average). This effect was considered
in the development of the NLEV program, which involves the nationwide sale of vehicles
certified on California low-sulfur fuel. Given the recent test programs which have demonstrated
that LEV emissions increase to a much greater degree (in percentage terms), EPA  staff believes
that it is not appropriate to continue testing vehicles on gasoline with 50-100 ppm sulfur knowing
that they will be operated on high sulfur fuels in-use.  The issue is whether to perform
certification and compliance testing with higher sulfur levels or whether to reduce the sulfur
content of commercial gasoline to levels more consistent with those prevalent in California (or a
combination of the two changes).

       Key factors affecting this decision include: 1) the effect of sulfur on LEV and ULEV-like
vehicles which will also meet other emission control requirements effective early in the next
century; 2) the cost of reducing catalyst sensitivity; 3) the cost of removing sulfur from gasoline;
4) the effect of sulfur on emissions from the current fleet (Tier 0, Tier 1, and NLEV vehicles)
prior to the turnover to Tier 2 vehicles, and the overall health and environmental benefits
achieved by these emission reductions. These factors will be considered together, and can have
significant impacts on the ultimate level of emission controls achievable through Tier 2 vehicles.
For example, if they lead to a decision to select a commercial fuel sulfur limit above California's,
       4 Indolene has typically averaged a sulfur level of 50-60 ppm, although recent batches at EPA's Ann
 Arbor, Ml laboratory have been reported to contain in excess of 100 ppm.

 EPA Staff Paper: Gasoline Sulfur                  3                              May 1, 1998

-------
the degree of emission control achievable by future vehicles may be limited, leading to higher
numerical Tier 2 emission standards. Selecting a gasoline sulfur level very different from
California's may also reduce or prevent further advanced technologies which are being
developed for use in California from being extended to the rest of the U.S.  These technologies
include further catalyst developments focused on California's proposed Phase n LEV standards
(LEV-n) and others which are focused on improved fuel efficiency.

       In addition to options regarding the level of the commercial fuel sulfur standard, there
may also be options regarding its geographical and temporal extent. In general, the costs of
sulfur control decrease as the region of sulfur control shrinks, but so do the environmental
benefits. For example, the potential Tier 2 vehicle standards are expected to apply nationally.  A
single sulfur standard applicable to all commercial fuel sold nationwide would allow vehicle
manufacturers to design their vehicles for a single fuel, instead of two or more.

       Many of the emission reductions associated with the potential Tier 2 standards and
gasoline sulfur reduction would also produce benefits  nationwide in addition to local benefits
such as ozone reductions. For example, VOC, NOx, and sulfur oxide (SOx) emissions all form
particulate matter (PM) in the atmosphere and are transported long distances. Thus, reductions in
these emissions, regardless of their location, are  likely to benefit areas currently not in attainment
of the PM national ambient air quality standards (NAAQS). Emissions of air toxics (compounds
which have been implicated in cancer and other health concerns) also affect human health,
regardless of the vehicles' locations.

       Automakers, state organizations, and others have recommended that EPA develop a
national maximum sulfur limit to ensure the benefits of sulfur control. A single, nationwide per-
gallon sulfur standard would tend to simplify enforcement of vehicle emission standards relative
to an averaging standard or to multiple standards applying to different regions of the country.
Both EPA and vehicle manufacturers could rely on the fact  that vehicles  tested in recall programs
were operated on fuels similar to those on which compliance was originally determined.
Furthermore, enforcement of the fuel standards would be much simplified if a single, per-gallon
cap were defined.

       On the other hand, representatives of the oil industry have proposed a regional, summer-
only program, focusing on the ozone-related benefits of the potential Tier 2 standards. The
majority of ozone nonattainment areas, and areas needing further emission reductions in order to
maintain their compliance status, are in the East and Midwest.  Also, high ozone is generally a
summertime problem. However, a sulfur control program designed to provide low sulfur fuel to
every area needing further emission reductions in order to attain or maintain their compliance
status would impact every region of the country, could significantly complicate fuel distribution,
and would not yield some of the other, non-ozone related benefits of gasoline sulfur reductions.

       The primary benefit of a regional, seasonal sulfur program is economic. The economics
of sulfur control in a refinery are such that the cost per gallon of control increases as the fraction
 EPA Staff Paper: Gasoline Sulfur                 4                              May 1, 1998

-------
of the refinery's gasoline production affected by the standard increases. Therefore, a refiner can
avoid the most costly controls if it can control less than 100% of its gasoline production.  Also,
individual refiners face different costs of sulfur control. With a regional program, some refiners
facing higher control costs could choose to produce fuel for sale outside of the affected area or in
areas with less stringent controls. While gasoline production costs would be less under a
regional control program, the costs of transporting and storing gasoline would be somewhat
higher, and the burdens to ensure compliance would be greater.

       The most important factor affecting any regional or seasonal sulfur control program is the
need for the sulfur effect to be temporary.  Vehicles are operated during all seasons and often
travel across state lines. Thus, under essentially any regional or seasonal program, many or all
vehicles in ozone non-attainment and maintenance areas would be occasionally or regularly
fueled with high sulfur gasoline. In order for sulfur control to reduce emissions over the life of
the vehicle, the emission impacts of sulfur must only last as long as the vehicle is being operated
on the higher sulfur fuel. This phenomenon  whereby the emissions decrease quickly when low
sulfur fuel is used after having increased as a result of high sulfur fuel use is termed reversibility.
Reversibility is not a certainty, either for vehicles meeting California LEV or potential federal
Tier 2 standards, or for those employing even more advanced designs.  Some vehicles designed
to meet the LEV standards have shown reversibility and other have not. Any significant  degree
of irreversibility would essentially eliminate consideration of regional and/or seasonal sulfur
control strategies.  Partial irreversibility would increase the stringency of the emission standards
for manufacturers, since they need to design based in part on the fuels a car will  see in-use, and
vehicles tested in an in-use recall program could have been operated on higher sulfur fuel at some
point in their lives.  Manufacturers  must compensate for this when they set their emission design
targets under the standards.  Thus, while reversibility does not in itself determine that a
regional/seasonal program is to be preferred  over a national program, it is a necessary component
of any regional/seasonal program.

       As the remainder of this paper discusses sulfur-related issues affecting vehicles and
refineries, constant mention  is made of how  the various factors could affect the degree of
commercial fuel sulfur control needed, as well as the nature of the sulfur control program. The
reader is encouraged to keep these factors in mind as the details of vehicle and refinery operation
are discussed.
 EPA Staff Paper: Gasoline Sulfur                  5                               May 1, 1998

-------
II.     Gasoline Sulfur's Impact on Vehicle Technologies

       Tailpipe emissions from modern motor vehicles depend on both the emissions coming
from the engine and the efficiency of the catalytic exhaust aftertreatment system in converting the
HC, CO, and NOx to carbon dioxide, water, nitrogen and oxygen.5 Fuel properties can impact
both engine-out emissions and catalyst efficiency.  For example, EPA has known that a wide
range of fuel variables (Reid vapor pressure, oxygen, benzene, sulfur, olefins, aromatics,
distillation properties) can affect emissions; the reformulated gasoline program is based on
control of many of these fuel properties. Lately, limited evidence has indicated that the
emissions impact of some of these fuel properties is diminishing as emission control technologies
have advanced.  However, recent research indicates that the emission impact of fuel sulfur may
be increasing with advances in emission control technology. The purpose of this section is to
summarize this research and estimate sulfur's emission impact on current and advanced vehicles.

       The sulfur in gasoline increases exhaust emissions of HC, CO, and NOx by decreasing
the efficiency of the three-way catalyst used in current and advanced emission control systems.
This effect has been demonstrated through numerous laboratory and vehicle fleet studies.6 The
degree of sulfur inhibition has been shown to be variable and to depend upon both catalyst
formulation and vehicle operating conditions, including:

       •      the type of precious metal, oxygen storage, support materials, and washcoat used,
       •      the catalyst surface area, cell structure, and layering of the washcoat,
       •      the oxygen content of the exhaust, and
       •      the operational temperature of the catalyst.

       The precious metal type and content are selected based on their ability to oxidize CO and
HC, reduce NO, achieve light-off temperature quickly, and be resistant to thermal aging. Only
three catalytic materials are currently available  which meet these criteria: platinum (Pt), rhodium
(Rh), and palladium (Pd).  These precious metals and their combinations vary in their sulfur
sensitivity. Similarly, metal oxides such as ceria and others within the washcoat are selected for
their ability to manage the oxygen concentration within the exhaust, their ability to support other
beneficial reactions within the catalyst, and their ability to improve thermal stability of both the
washcoat components and the precious metals.  Improvements in the oxygen storage and release
capabilities of catalysts have played a key role in the tremendous improvements in catalyst
efficiency in recent years.  Oxygen storage under lean conditions (excess oxygen) with
subsequent release under rich conditions (lack of oxygen) reduces the sensitivity of three-way
catalysts to perturbations in the air-to-fuel ratio encountered during operation of the vehicle.
This has particularly improved the catalyst's ability to reduce NOx, as the catalyst's NOx
       5  Unless otherwise noted, the term emissions will refer to tailpipe emissions and motor vehicle will refe
to light-duty vehicles and trucks.

       6  See, for example, Auto/Oil Air Quality Improvement Research Program, Technical Bulletin No. 2,
"Effects of Fuel Sulfur on Mass Exhaust Emissions," February, 1991.

EPA Staff Paper:  Gasoline Sulfur                 6                              May 1, 1998

-------
efficiency inherently decreases quickly as soon as excess oxygen is present on the catalytic
surface. The improved washcoats preferentially store excess oxygen until the exhaust returns to
an oxygen-poor (i.e., rich) condition. Other improvements which have also enhanced the
performance of recent catalysts include increased surface area per volume (small cells), thinner
coatings of catalytic material, and layering of the washcoat and catalytic material.

       Of the three precious metals currently used, palladium appears to be the most sensitive to
sulfur poisoning, followed by platinum and rhodium.  Laboratory studies have shown that
palladium's ability to reduce automotive pollutants decreases more dramatically than platinum's.
While sulfur's poisoning effect is primarily related to its presence on the surface of the catalyst,
sulfur appears to penetrate into palladium, but not platinum or rhodium.  This penetration slows
the processes whereby sulfur can be removed from the surface of palladium relative to platinum
or rhodium.

       Of the various metal oxides used for oxygen storage and other purposes, ceria appears to
be most sensitive and least reversible to sulfur poisoning. Sulfate penetrates  the crystalline
lattice of ceria and prevents ceria from taking up oxygen when the engine is operating lean and
giving off oxygen when the engine is operating rich.  The mechanisms of sulfur's interference in
both catalytic activity and oxygen storage are discussed further in Appendix A.

       While sulfur is a catalyst poison, it is not necessarily a permanent poison Jike lead.
Stored sulfur compounds can be removed from the catalyst components under certain conditions.
The ease of this removal depends on the chemical forms of the sulfur compounds, which are, in
turn, dependent on the amount of oxygen present in the exhaust, the exhaust temperature, and the
composition of the precious metals and other components of the catalyst. Sulfur from the fuel is
oxidized during combustion to sulfur dioxide (SO2).  Under lean conditions, the SO2 can react
with metal oxides within the washcoat to form sulfates.  This can be a particular problem with
washcoats having a high ceria content. Sulfation of ceria reduces the ability to store oxygen
within the catalyst, and inhibits other important chemical reactions that are helpful in promoting
high catalyst efficiency. Similarly, under rich conditions, SO2 is dissociated  and the resultant
sulfide strongly adsorbs onto precious metal surfaces. The adsorbed sulfide reduces the ability of
HC, CO, and NOx to adsorb onto the surfaces and undergo the reactions necessary to oxidize or
reduce them. Laboratory studies also suggest that sulfur increases the temperature required to
"light-off the catalyst after a cold engine start, as well as decreasing catalyst efficiency after
warm-up.7  Depending on these factors, high temperatures (650 to 700°C) and either rich
operation, lean  operation, or both, may be necessary to remove the sulfur compounds  from the
catalyst and reverse their adverse effects on catalyst performance after the vehicle has been
refueled on lower sulfur gasoline.
       7 Light-off temperature refers to the temperature (i.e., 350 to 400° C) at which the catalyst conversion
efficiency reaches a specified, minimally acceptable level (e.g., 50%).

EPA Staff Paper:  Gasoline Sulfur                 7                               May 1, 1998

-------
       Laboratory studies provide excellent insight into the chemical mechanisms involved in
the interaction between sulfur and the catalytic converter. However, these studies have
limitations in predicting the impact of fuel sulfur on vehicle emissions.  Their primary limitation
involves the conditions of the experiments. Laboratory studies are usually conducted under
conditions where the active surface of the catalyst is the limiting factor in determining catalyst
efficiency. This is purposely done in order to observe the impact of sulfur most clearly.
However, on a vehicle, operating conditions vary dramatically. Depending on the design of both
the engine and the catalytic converter, other factors not affected by sulfur may limit overall
catalyst performance. In these cases, the impact of sulfur on vehicle emissions will be less than
observed in the laboratory.

       Several vehicle fleet studies have been performed over recent years. These studies have
demonstrated that significant reductions in HC, CO, and in particular, NOx emissions can be
realized by reducing fuel sulfur levels. Although the results of these studies are in general
agreement with laboratory results, there are some noticeable differences. Laboratory studies
consistently show a significant degree of catalyst inhibition due to sulfur, whereas results from
vehicle studies show a high degree of variability. This suggests that other factors can have a
significant impact upon the degree of sulfur inhibition. Vehicle studies also indicate that sulfur
has the greatest effect on emissions during warmed-up operating  conditions. This suggests that
the sulfur effect on catalyst light-off times reported in recent laboratory studies may not be as
significant as shown in earlier laboratory studies.

       This chapter examines the data which are available to estimate the impact sulfur has on
today's cars and light trucks. Information is also presented about the reversibility of this  sulfur
impact (how quickly and completely the sulfur effect is reversed upon fueling  with lower sulfur
fuel). The chapter presents projections about how sensitive future technologies may be to sulfur,
using test data and information gathered from recent test programs and from discussions with
automotive and catalyst  manufacturers.  Finally, this chapter presents information on how sulfur
impacts the operation of the catalyst monitoring systems known as on-board diagnostics (OBD).
A.     Tier 0 Vehicle Emissions

       A number of vehicle test programs were conducted in the early 1990's which evaluated
the effect of fuel sulfur on emissions from Tier 0 vehicles (vehicles manufactured through 1993).
These included test programs by the Auto/Oil Air Quality Research Program (Auto/Oil), EPA,
and others. In 1993, EPA statistically evaluated all of the available data on vehicles which were
technologically equivalent to  1990 model year vehicles and used the results in the Complex
Model, which is used to evaluate compliance with federal reformulated gasoline requirements.
The Complex Model predicts the impact of sulfur on tailpipe emissions for both normal- and
high-emitting vehicles.  Based on this model, the impact of reducing sulfur from the current
national average of 330 ppm to 150 and 40 ppm is summarized in Table 1.
 EPA Staff Paper:  Gasoline Sulfur                 8                              May 1, 1998

-------
Table 1. Decrease in Emissions with Fuel Sulfur Decreasing from Average
In-Use Level (330 ppm) for Tier 0 Vehicles [Source: EPA Complex Model]

Normal Emitters
High Emitters
NMHC
150 ppm
4.6%
-0.4%
40 ppm
13.0%
-0.6%
CO
150 ppm
4.4%
3.5%
40 ppm
12.6%
10.4%
NOx
150 ppm
5.0%
1.7%
40 ppm
11.1%
6.1%
       The equations in the Complex Model are exponential in form.8 Since developing the
Complex Model, a number of investigations into the effect of sulfur on emissions from Tier 0
vehicles have concluded that other equation forms, such as logarithmic or square root, may better
represent the sulfur-emission relationship.  Such alternative equation forms would likely show
only small deviations from the Complex Model in the range of 330 - 150 ppm, but may  show
large deviations from the Complex Model in the range 150 - 40 ppm. The Agency is currently
evaluating the most appropriate form of the sulfur-emission relationship for Tier 0 vehicles in the
context of revisions to the MOBILE9 model.

       In performing the analyses to support the Draft Tier 2 Study, EPA made a number of
modifications to MOBILESb to reflect more recent emission data which has become available
since the model was developed in 1993. The result of these modifications is that the great
majority of in-use Tier 0 vehicles are now projected to be normal emitters. Therefore, the sulfur
impacts shown above for normal emitters now appear to be the most relevant to the Tier 0
vehicle fleet as a whole.
B.     Tier 1 Vehicle Emissions

       Only one test program has been performed evaluating the impact of sulfur on emissions
from Tier 1  vehicles (vehicles manufactured to meet the emission standards which took effect in
1994). Auto/Oil tested six early Tier 1 light-duty vehicles (LDVs) on two sets of otherwise
identical fuels containing nominally 40 ppm and 330 ppm sulfur. The results of this test program
are shown below in Table 2.  All of the vehicles tested met the Tier 1 emission standards, so they
would all be considered normal emitters.  As can be seen, the effect of sulfur on Tier 1 emissions
       8 The equations are in the form of an exponential of a polynomial, though the behavior of the model very
closely follows polynomial curvature.

       ' MOBILE is a model developed by EPA for predicting in-use emissions for an entire fleet of motor
vehicles (vehicles of different types and ages) in a specified calendar year. The most recent version of the model i
MOBlLESb. EPA is currently working on the next version of the model, which will be called MOB1LE6.
 EPA Staff Paper: Gasoline Sulfur
May 1, 1998

-------
of NMHC and CO are somewhat greater than those observed for normal emitting Tier 0 vehicles,
while that for NOx emissions is essentially the same.
Table 2. Decrease in Emissions with Fuel Sulfur Decreasing from Average
In-Use Level (330 ppm) for Tier 1 LDVs [Source: Auto/Oil AQIRP]

Normal Emitters
NMHC
150 ppm
—
40 ppm
16.3%
CO
150 ppm
—
40 ppm
16.4%
NOx
150 ppm
—
40 ppm
11.0%
C.     LEV and ULEV Emissions

       Two test programs evaluating the impact of sulfur on emissions from LEVs and ULEVs
were recently completed by the Coordinating Research Council (CRC) and the auto industry.
The CRC program consisted of twelve 1997 LEV passenger cars, representing six different
models from five different vehicle manufacturers.  The vehicles were tested with fuel sulfur
levels of 40 (the baseline level to represent California certification and in-use fuel sulfur levels),
100, 150, 330, and 600 ppm.  The remaining properties of the fuel represented national averages.
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 auto  industry testing was performed by members of the American Automobile
Manufacturers Association (AAMA) and the Association of International Automobile
Manufacturers (AIAM).  The AAMA/AIAM program consisted of 13 production and
production-intent LEV and ULEV LDVs and eight LEV and ULEV light-duty trucks (LDTs). A
total of ten vehicle manufacturers participated in the program. The vehicles were tested at the
same sulfur levels as the CRC program.  The other fuel properties were those of California Phase
II certification fuel.  All vehicles were equipped with aged components to simulate 100,000
miles.

       The results of the CRC and AAMA/AIAM programs have been combined.10 Table 3
shows the percent increase in emissions  associated with increasing the fuel sulfur level from 40
ppm to 150 ppm and 330 ppm, respectively, for both LDVs and LDTs designed to meet the LEV
and ULEV standards. Only the 100,000-mile data are presented. Analogous impacts for the Tier
0 and Tier 1 vehicles discussed above are shown for comparison.
       111 The test results from each pair of LEVs in the CRC test program were averaged and assumed to
 represent a single vehicle. The results for vehicles from the same model line and certified set of emission standard
 which were tested in both the CRC and AAMA/AIAM test programs were also averaged and assumed to represent a
 single vehicle.
 EPA Staff Paper: Gasoline Sulfur
10
May 1,1998

-------
Table 3. Increase in Emissions with Fuel Sulfur Increases from Baseline (40 ppm) for
LEVs and ULEVs (LDVs and LDTs) and Older Vehicles
[Source: CRC and AAMA/AIAM Test Programs]
Pollutant
Sulfur, ppm
NMHC
150 ppm
330 ppm
CO
150 ppm
330 ppm
NOx
150 ppm
330 ppm
LEV and ULEV, LDVs and LDTs
All LDV/LDT1 a
All LDT2/LDT3 b
26.7%
23.0%
43.0%
26.4%
58.0%
12.5%
75.8%
31.2%
65.7%
33.7%
136%
65.5%
LDV/LDTls Broken Down by NMHC+NOx Emissions with 40 ppm Sulfur Fuel
9 Lowest Emitters
9 Highest Emitters
26.8%
26.6%
44.0%
42.1%
—
—
—
...
127%
41.2%
264%
84.2%
LDV/LDTls Broken Down by the Sensitivity of NMHC+NOx Emissions to Sulfur
9 Least Sensitive
9 Most Sensitive
32.2%
25.1%
49.5%
42.2%
—
—
—
—
31.0%
122%
61.0%
251%
Tier 1 LDVs
Normal Emitters
—
20.9%
—
21.1%
—
13.6%
Tier 0 LDVs and LDTs
Normal Emitters
High Emitters
5.9%
-0.6%
16.3%
-1.6%
5.7%
4.7%
15.8%
12.9%
6.4%
2.8%
13.8%
7,6%
      These results indicate that LEV and ULEV emission control technologies being utilized
on LDVs are, on average, much more sensitive to sulfur than Tier 0 or Tier 1 technology. For
example, the percentage increases in NOx emissions for LEV and ULEV LDVs are roughly 10
times greater than those for Tier 0 and 1 vehicles.  Emissions from the LEV and ULEV LDTs are
also more sensitive than the Tier 0 and Tier 1 vehicles tested earlier, but to a much lesser extent.
The LDTs had a higher level of base emissions on 40 ppm sulfur fuel, which may indicate that
their technology differs less dramatically from the Tier 1 LDVs tested earlier.

      With respect to both LDVs and LDTs, the sensitivity of individual LEV and ULEV
models to sulfur varies dramatically. The impact of sulfur on four subsets of the 18 LDVs and
LDTls tested in the two test programs are also summarized in Table 3.

      The first breakdown of these vehicles was performed according to these vehicles'
baseline NMHC+NOx emission levels on 40 ppm sulfur fuel. The impact of sulfur on NMHC
emissions from the nine vehicles with the lowest NMHC+NOx emissions on California-like 40
ppm fuel is essentially the same as the impact of sulfur on the remaining nine vehicles.
However, the impact of sulfur on NOx emissions from the nine vehicles with the lowest
NMHC+NOx emissions on California-like 40 ppm fuel is much greater than the impact of sulfur
 EPA Staff Paper: Gasoline Sulfur
11
May 1, 1998

-------
on the remaining nine vehicles. This is consistent with the overall trend observed that sulfur's
impact increases as the overall stringency of emission control increases.

       The second breakdown of these vehicles was performed according to their sensitivity to
sulfur. This sensitivity was defined as the percentage change in NMHC+NOx emissions per
change in sulfur concentration." NOx emissions from the nine vehicles showing the least sulfur
sensitivity (in terms of NMHC+NOx emissions) increase less than half as much due to high
sulfur levels than the original 18 vehicles.  However, the effect of higher sulfur levels on NMHC
emissions from these nine least sensitive vehicles is roughly the same as the original 18. This is
due to the fact that the change in NOx emissions with higher sulfur is much greater on average
than that for NMHC emissions. Thus, the selection of the least sensitive vehicles in terms of
NMHC+NOx emissions was roughly the selection of the least sensitive vehicles in terms of NOx
emissions. As it turned out, those vehicles with the least sulfur sensitivity towards NOx
emissions had slightly greater than average sensitivity towards NMHC emissions.

       The sulfur sensitivity of the 9 least sensitive vehicles suggests that there are aspects of the
emission control system that can be modified to reduce the average sulfur sensitivity of the LDV
fleet in the future.  It is likely that auto manufacturers should, in general, be able to reduce the
sulfur sensitivity of future vehicles to that of the nine least sensitive LEVs and ULEVs.
However, one factor not reflected in these data is the need to meet stringent emission standards
applicable during periods of aggressive driving and use of the air conditioner starting in 2001.
These "off-cycle" standards will place additional limits on the fuel management strategy and
catalyst designs used by manufacturers. In particular, more precise air-fuel  mixture  control will
likely be required.  This could eliminate one method which appears  to reduce sulfur's emission
impact, wide air-fuel ratio swings. Future Tier 2 vehicles may also have to  meet more stringent
emission standards than today's LEVs, which could further limit the manufacturers' choices for
emission control system designs.  EPA plans to further analyze the emission data produced in
these test programs to determine the degree that off-cycle standards  would affect manufacturers
ability to design emission control systems with lower sulfur sensitivity.
D.     Sulfur Reversibility

        Sulfur has an almost immediate effect 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.
       1' Mathematically, this sensitivity is the sum of the percentage change in NMHC emissions multiplied by
 the base NMHC emissions plus the percentage change in NOx emissions multiplied by the base NOx emissions, all
 divided by the sum of base NMHC emissions plus base NOx emissions.

 EPA Staff Paper:  Gasoline Sulfur                 12                              May 1,1998

-------
       The potential reversibility of the sulfur effect could have substantial consequences for the
design of a commercial sulfur control program. If sulfur's effect on catalyst performance is not
substantially reversible, sulfur controls must be uniform nationwide and year-round, as the
benefits of a regional sulfur control program would be permanently compromised whenever
vehicles traveled between low and high sulfur areas. Similarly, under these circumstances a
seasonal sulfur control program would compromise essentially all vehicles' emission
performance since higher sulfur fuel would be used part of the year. On the other hand, if the
sulfur effect is quickly and completely reversible, this may enable control of commercial fuel
sulfur regionally and/or seasonally. However, the other benefits of nationwide sulfur standards,
enablement of future engine and emission control technology, and reductions in ambient PM,
etc., would still need to be considered before a regional control program could be determined to
be appropriate.

       In an effort to better understand and quantify sulfur reversibility, EPA reviewed the
available literature, including a literature review submitted to EPA by the American Petroleum
Institute (API).12 EPA found that a number of laboratory studies have been performed to
understand the nature of sulfur's impact on catalyst performance and the reversibility of this
impact. In contrast, relatively few vehicle test programs have been performed to investigate
sulfur reversibility.  The vehicle test programs discussed above utilized special procedures when
vehicles switched from one fuel to another to ensure that no sulfur-related effect was carried over
from the previous test fuel. These procedures generally involved vehicle operation which is not
typical of in-use driving.  Thus, these programs provide little insight into the reversibility of the
sulfur impact under typical in-use driving conditions.

       In addition to this information from the technical literature, EPA has obtained data from
individual companies that suggest sulfur is easily reversible for some vehicles and not easily
reversible for others during normal driving. For example,  Chrysler provided sulfur reversibility
data for a 1997 LEV-certified Dodge Neon. Emissions were measured with 40 ppm fuel, with
800 ppm fuel after 300 miles of in-use operation with 800 ppm fuel, and then again on 40 ppm
sulfur fuel after about 100 miles of light urban in-use driving with 40 ppm fuel.  Figures la, Ib,
and Ic show that CO and NOx emissions returned to the baseline 40 ppm sulfur levels upon the
return to 40 ppm fuel, while NMHC did not. This suggests that CO and NOx performance was
reversible after this type of driving, while NMHC performance was irreversible. It should be
noted that even though NMHC did not recover, the increase in NMHC emissions at 800 ppm was
only 13% over the emissions on 40 ppm sulfur fuel, which is much smaller than the average
sulfur impact on LEV emissions summarized in Table 3 above.  To return the NMHC emissions
to their baseline, 40 ppm sulfur levels, Chrysler operated the vehicle through a number of severe
accelerations to increase catalyst temperature and obtain a rich exhaust mixture.13  This procedure
returned the NMHC emissions to the original level.
       12 Truex, Timothy J., "Potential for Improved Sulfur Tolerance in Three-Way Automotive Catalysts,"
Center for Environmental Research and Technology, University of California at Riverside, November 26, 1997.

       13 Five to 10 wide-open throttle accelerations from 30 to 70 miles per hour.

EPA Staff Paper:  Gasoline Sulfur                 13                              May 1, 1998

-------
                              Su^ttle^tt at ffMBCEmltsiatt
                 Figure la: Sulfur Reversibility Study on a 1997 LEV
                  Dodge Neon Effect of Sulfur on NMHC Emissions
                 Figure Ib: Sulfur Reversibility Study on a 1997 LEV
                    Dodge Neon Effect of Sulfur on CO Emissions
EPA Staff Paper: Gasoline Sulfur
14
May 1,199

-------
                                   Sulfur Level* 
-------
                                  Effett efVariota Gasetote
                                Sulfur Ur* on NMHCEntelont
                   Figure 2a: Sulfur Reversibility Study on a ULEV
                   Ford Escort Effect of Sulfur on NMHC Emissions
                                   Effect tfyarlmt Gmn&te
                   Figure 2b: Sulfur Reversibility Study on a ULEV
                     Ford Escort Effect of Sulfur on CO Emissions
EPA Staff Paper: Gasoline Sulfur
16
May 1,1998

-------
                                    Effett of Yfcriott* Ga*>8n«
                                                              40
                                                            Rkfc&F
                    Figure 2c: Sulfur Reversibility Study on a ULEV
                     Ford Escort Effect of Sulfur on NOx Emissions
       Finally, API provided EPA with data for a 1995 Mazda Protege" transitional low emission
vehicle (TLEV) equipped with an enhanced catalyst that allowed the vehicle to meet LEV
standards with 40 ppm sulfur fuel. The vehicle was tested twice at 40 ppm, followed by a single
test at 100 ppm, a single test at 600 ppm, a single test at 300 ppm, and then two more tests at 4
ppm. There was no special vehicle operation in between refuelings. Figures 3 a and 3b show
that, although the Protege" did not achieve complete reversibility, it came very close, experiencin
about 85% recovery for NMHC and CO. The data in Figure 3c does indicate full reversibility for
NOx, but this result is tempered by the fact that the NOx emissions were erratic over the differen
sulfur levels.
 EPA Staff Paper: Gasoline Sulfur
                                           17
May 1,1998

-------
                                   Effect of Various Gasolttut
                 Figure 3a: Sulfur Reversibility Study on a 1995 Mazda
                  Protege TLEV Effect of Sulfur on NMHC Emissions
                                 Sulfur Ifteh on CQEmmisioar
                                           'TttT
                 Figure 3b: Sulfur Reversibility Study on a 1995 Mazda
                    Protege TLEV Effect of Sulfur on CO Emissions
EPA Staff Paper: Gasoline Sulfur
18
May 1,1998

-------
                                      Effect a/ Various Gtualinf
                                    SulforLertti On M* Bmitslffnt
                                                                 40
                  Figure 3c: Sulfur Reversibility Study on a 1995 Mazda
                    Protege TLEV Effect of Sulfur on NOx Emissions
       In addition to these individual vehicle test results, CRC is currently performing a test
program to explicitly evaluate sulfur reversibility with LEVs. The program involves the six LEV
models tested in the CRC LEV-sulfur emissions test program described earlier and is scheduled
to be completed by the end of May 1998.  This program will assess reversibility after light urban
operation and after more severe high speed, high load operation (i.e., the EPA US06 cycle). This
program will further our knowledge of the degree of reversibility of current LEVs and facilitate a
more accurate projection of the degree of reversibility achievable with potential Tier 2 vehicles.
E.     Sulfur Tolerant Automotive Technologies

        One of the clearest observations from the CRC and auto industry sulfur test programs is
that there is a great deal of variability in vehicle sulfur sensitivity. While none of the vehicles
were immune to sulfur's inhibition of catalyst performance, some of the vehicles showed that
they were far less sensitive than others. In order to determine if it is possible to develop
automotive technology that is substantially less sulfur-sensitive, EPA is in the process of
evaluating  the detailed emission and operating data obtained in the two LEV test programs. The
goal of this research is to identify those technologies which are more and less sensitive to sulfur
and whether the less sensitive technology can be applied to all manufacturers' vehicles (large and
small, car or truck) in a cost-effective manner. In addition to analyzing the data from both
industry test programs, EPA is also reviewing the latest technical literature, and meeting with
vehicle and catalyst manufacturers and other researchers in this field to ensure that all available
data are incorporated into this assessment.

       At this time, it is clear that there are no catalyst designs currently available that are fully
sulfur tolerant. The results of virtually every laboratory, engine dynamometer, and vehicle fleet
study has shown that all automotive catalyst designs, whether they use Pd, Pt, Rh or any
 EPA Staff Paper:  Gasoline Sulfur
19
May 1,1998

-------
combination of the three, have some inhibition in performance resulting from sulfur. However,
some catalyst designs are more tolerant of sulfur than others.

       There is a broad range of opinions regarding the feasibility of developing sulfur resistant
catalyst technology. C.H. Bartholomew of the Department of Chemical Engineering at Brigham
Young University stated at a recent CRC Symposium15 that sulfur tolerant catalysts could be
developed with a significant investment in research money and time, but that it is unlikely that a
catalyst could be developed that could tolerate more than 100 ppm to 200 ppm sulfur with no
emissions degradation.  A representative of Degussa/ICT, a catalyst manufacturer,  stated that it
was possible to develop a Pd/Rh catalyst system that was more resistant to sulfur, but that there
would still be some deterioration in HC and CO performance due to sulfur.  Christopher Bennett
of Johnson-Matthey, another catalyst manufacturer, stated that results from some of their studies
indicate that Pd-only catalyst performance may be improved through washcoat formulation, and
that these performance improvements were observed both in the presence and the absence of
sulfur.  Studies by Johnson-Matthey also showed that similar washcoat formulations, when
applied to Pd/Rh catalysts, could also result in more active catalyst.  However, laboratory and
engine dynamometer testing indicated that these advanced washcoat formulations still exhibit a
substantial decrease in catalytic performance when exposed to an increase in sulfur levels.

       Heinz Robota of Allied Signal Emission Components (ASEC)  Manufacturing, a
manufacturer of catalysts and other emission control components, stated at the symposium that
there are trade-offs in sulfur tolerant catalyst designs. He claimed that good hydrocarbon sulfur
resistance may result in poor NOx conversion resistance to sulfur and vice versa, while the
Automobile Emissions Control by Catalyst (AECC), an association of European emission control
manufacturers, claimed that sulfur resistant catalysts are not an option  because achieving this
capability would necessitate trading off catalyst performance for the removal of some pollutants.

       Recent discussions between EPA and several catalyst manufacturers have suggested that
there may be some techniques available to make catalysts more tolerant of sulfur.  However, in
addition to an economic cost of doing so, there will also likely be an overall loss in catalytic
activity.  This means that the catalyst may perform better with high sulfur levels than today's
highly active catalysts, but that the sulfur-tolerant catalyst will not perform as  well (at either low
or high sulfur levels) as today's catalysts perform with low sulfur fuel. Thus, higher sulfur levels
still imply higher emission levels,  though possibly not as high as is indicated by some of the
today's LEVs which are very sensitive to sulfur.  According to these catalyst suppliers, sulfur
will still be technology constraining. Although these companies believe they can make catalysts
that achieve low emissions even with moderate levels of sulfur, they can make considerable
improvements in catalyst performance and emission reductions with even lower sulfur levels.
Therefore, catalyst designs that might appear in Japan, Europe, and California, where sulfur
levels are very low, may not be available in the rest of the United States.
       13 CRC Auto/Oil Symposium, "The Search for Sulfur-Tolerant Emission Control Systems," Dearborn, MI,
September 11-12, 1997.

EPA Staff Paper: Gasoline Sulfur                 20                              May 1,1998

-------
       In addition to catalyst design, the calibration of the fuel control system can have a
significant effect on a catalyst's ability to tolerate sulfur.  Systems and components such as air-
fuel ratio controls and exhaust gas recirculation can have an effect on how sulfur reacts on the
catalyst surface.  This fact was supported at the CRC symposium by Johnson Matthey and also
by Ray Paggi of Texaco Fuels and Lubricants.  Paggi stated that evidence for this was the fact
that some Tier 0 and Tier 1 production vehicles appear to be somewhat tolerant to sulfur and that
fuel system calibrations appeared to be one of the contributing factors.

       A discussion of several potential catalyst design features which may reduce sulfur
sensitivity is contained  in Appendix A. API is currently performing a research program to
investigate the potential for developing sulfur tolerant catalyst designs. They hope to have a final
report discussing their findings in the near future.
F.     Projected Impacts on Advanced Engine Technologies

       Based on information in the technical literature and information received from
automotive manufacturers, it appears that gasoline sulfur may also have a detrimental impact on
advanced engine technologies. An example is the gasoline direct-injection (GDI) engine. There
are a growing number of vehicles currently sold in Japan which utilize direct fuel injection for
gasoline engines. The advantages of GDI include up to a 30% improvement in fuel economy,
improved engine response under variable operating conditions, and more rapid starting with less
cold enrichment. As a result, GDI engines should produce lower levels of NMHC and CO2
emissions.  The very lean (>19:1 air-fuel ratio) operation typical of GDI engines results in low
engine-out NOx emissions in comparison to more conventional, stoichiometric spark ignition
engines. However, this lean operation poses a considerable barrier to NOx control using current
three-way catalyst technologies, and very low tailpipe NOx emission standards would require
reductions beyond the low engine-out levels.

       Conventional spark ignition engines maintain an air-fuel ratio near stoichiometry (i.e.,
neither lean nor rich) to allow operation within the narrow window of operating conditions
necessary for efficient removal of CO, hydrocarbons, and NOx. Under lean conditions, the
ability of the catalyst to reduce NOx decreases  substantially. Some GDI engine designs
compensate by using a NOx trap to store NOx emissions while the engine is running lean. This
stored NOx is then released when the engine is operating near stoichiometry, permitting the
three-way catalyst to effectively reduce the NOx. Selective catalytic reduction (SCR) of the NOx
is another approach being investigated.  With SCR, a reducing agent is introduced into the
exhaust which facilitates the  reduction of NOx, even under lean conditions. Both the NOx
trap/three-way catalyst and SCR are often collectively referred to as "lean-NOx" catalysts.

       Some lean NOx catalyst strategies appear to be more sensitive to fuel sulfur than LEV
catalysts or the catalysts expected to be used on post-LEV vehicles.  Information supplied by
Toyota at the CRC Symposium indicates that NOx emissions increase dramatically with


EPA Staff Paper: Gasoline Sulfur                21                             May 1,1998

-------
increased fuel sulfur for one lean-NOx catalyst system. NOx emissions doubled when fuel sulfur
increased from 30 to 70 ppm, and quadrupled with sulfur levels of 500 ppm. Systems using SCR
without a NOx trap may or may not be as sensitive to sulfur.

       An important question regarding the use of GDI engines is their capability to achieve low
NOx levels even with low sulfur gasoline. While low sulfur levels may enable the lean NOx
catalyst systems mentioned above, it is not clear what level of NOx emissions vehicles with GDI
engines would be able to meet,  even with these catalysts. LEV-like NOx emission levels appear
to be feasible, at least with some GDI designs. However, the ability of this technology to reduce
NOx emissions below these levels is more uncertain.  Thus, it is important to characterize the
emission control potential of these engines with both low and high sulfur levels. This technology
is described in further detail in Appendix A.

       The fuel cell is another promising propulsion system that is being developed for possible
introduction to consumers early in the next century. A fuel cell is an electrochemical device,
similar to a battery, that generates electricity from a chemical reaction between hydrogen and
oxygen. The  necessary hydrogen can either be provided by a compressed gas (e.g., hydrogen,
methane) or extracted from a fuel, such as gasoline or methanol, carried on the vehicle. The
electricity produced from a fuel cell drives a traction motor that in turn drives the wheels.  In
comparison to conventional electric vehicles, fuel cell use gives a vehicle long range, good
performance,  rapid refueling, and similar (low or even zero) emission levels.

       If gasoline is used to provide the hydrogen for fuel cells, the sulfur content of the gasoline
can be very important.  Although fuel cell technology is still in its infancy, it has become clear
that the extraction of hydrogen  from gasoline will simultaneously produce hydrogen sulfide
and/or sulfur  dioxide from the sulfur contained in the gasoline. These sulfurous compounds can
reduce the efficiency of the fuel cell either through clogging of membranes or adsorption and
subsequent inhibition of hydrogen dissociation. Current fuel cell research makes use of gasoline
containing less than 40 ppm sulfur; it is not clear that all fuel cells can tolerate even this level o
sulfur and nor whether future developments will result in fuel cells that can tolerate even higher
levels of sulfur.
G.     Sulfur's Effect on OBD Catalyst Monitors and Implications for I/M Programs

     •  EPA's OBD regulations require that the OBD catalyst monitor identify those catalysts
with efficiencies reduced to the extent that tailpipe emissions would have emissions increases
exceeding 1.5 times.the applicable HC standard.  Current catalyst monitor designs do not actually
measure catalyst efficiency, but measure the oxygen storage capacity of the catalyst and infer HC
conversion efficiency from these readings. Sulfur can interfere with the catalyst's oxygen
storage. In some cases, sulfur can decrease oxygen storage and cause the OBD system to
determine that the catalyst has insufficient efficiency. In other cases, the chemical form of the
 EPA Staff Paper Gasoline Sulfur                 22                             May 1,1998

-------
sulfur adsorbed on the catalyst can mimic stored oxygen and cause the OBD system to determine
that the catalyst has sufficient efficiency, when in fact it does not.

       EPA evaluated the potential of sulfur's interference with the proper operation of OBD
catalyst monitors in 1997.16 EPA concluded that there were not sufficient data to indicate a
broad-based sulfur-related problem with OBD catalyst monitors. However, EPA also found that
there might be a need, in a limited number of cases, for manufacturers to adjust their
California-certified OBD systems for operation on higher sulfur levels outside of California.
These adjustments would reduce the possibility that sulfur would cause the OBD system to
prematurely illuminate the malfunction indicator light (MIL). EPA indicated that it would
consider allowing such adjustments on a case-by-case basis.

       EPA also acknowledged that it was not aware of any adjustments which could be made in
the short term to eliminate the potential for high sulfur levels to cause "false-passes," where
emissions exceed the monitor threshold but do not illuminate the MIL. This issue does not
represent a customer inconvenience, as would a "false-failure." However, it could reduce the
effectiveness of the OBD catalyst monitor, particularly in an inspection and maintenance (I/M)
program, since the monitor may not detect a poorly performing catalyst.

       The LEV and ULEV emission data presented above in Table 3  indicate that higher sulfur
levels have the potential to increase emissions in the range that the OBD system is designed to
detect (i.e., 1.5 times  the applicable emission standards). However, the performance of the OBD
systems on these test  vehicles was not reported in the test programs. EPA is not aware of any
other new data which has become available since the OBD-sulfur study was completed last year.
Thus, the possibility of an interference of the OBD catalyst monitor by higher sulfur levels still
exists.  EPA has previously indicated to automobile manufacturers that it would consider any
impacts of sulfur on OBD catalyst monitors during vehicle certification on a model-by-model
basis. This approach still appears to be appropriate.

       A regional sulfur control program could also affect I/M programs located outside of the
sulfur control region. The emissions measured in these I/M programs would likely be higher
than those measured elsewhere, possibly necessitating the use of unique emission cutpoints for
LEVs and Tier 2 vehicles. As mentioned above, since sulfur may affect the operation of the
OBD system on selected vehicle models, these I/M programs located outside of the sulfur control
area would need to incorporate this information into their checking for the presence of OBD
failure codes.

       Also, one study of a TLEV showed that sulfur can have an inordinate affect on emissions
over a loaded I/M test after a prolonged idle, which can occur while a vehicle is waiting in line
       16 EPA, "OBD & Sulfur Status Report: Sulfur's Effect on the OBD Catalyst Monitor on Low Emission
Vehicles," March 1997 and updated September 1997.

EPA Staff Paper: Gasoline Sulfur                 23                             May 1,1998

-------
for an I/M test.17 Thus, higher sulfur levels in-use may necessitate the use of special pre-
conditioning procedures for the I/M testing of LEVs and Tier 2 vehicles.
       17  Beck, D.D., W.A. Short, T.E. Angelos and R.R. Oils, "1M240 Emission Tests with a 2.2 L Corsica
TLEV," SAE Paper No. 942001,1994.

EPA Staff Paper: Gasoline Sulfur                  24                                May 1,1998

-------
III.    Cost of Reducing Gasoline Sulfur

       The data presented in the previous chapter indicate that control of sulfur in gasoline may
be necessary to achieve an appropriate level of emissions reduction from potential Tier 2 vehicles
and to enable the use of advanced technology vehicles in the U.S.  Control of gasoline sulfur
would also provide significant emissions reductions from NLEVs and older technology vehicles
and resolve concerns about the compatibility of current gasolines with existing emission  control
hardware.  The cost of reducing gasoline sulfur levels will be an important factor in evaluating
the level and geographical extent of any required sulfur reductions. This chapter summarizes
data regarding current sulfur levels in different grades and types of gasoline, and presents an
overview of how sulfur is removed from gasoline.  Following that is a discussion of refinery cost
modeling that EPA is currently performing jointly with the U.S. Department of Energy (DOE).
Preliminary cost estimates for gasoline sulfur reduction are presented, and additional aspects of a
sulfur control program which may affect the costs to the refining industry are discussed.
A.     Current Sulfur Levels in U.S. Gasoline

       Sulfur occurs naturally in crude oil. As crude oil is refined, some of the sulfur ends up in
gasoline. The amount of sulfur in the gasoline from any refinery depends on a number of factors:
the amount of sulfur in the crude oil, the fraction of gasoline produced from heavier petroleum
compounds,18 and the amount of desulfurization being applied to gasoline blendstocks and
gasoline precursor streams in the refinery.

       There are a number of standards or limits which affect the amount of sulfur in gasoline.
There is a 1000 ppm maximum sulfur specification for gasoline, developed by the American
Society for Testing and Materials (ASTM), which most refiners meet voluntarily and some states
have adopted by law. Federal reformulated gasoline (RFG) subject to the Complex Model is
currently limited to 500 ppm sulfur due to the valid range limits of the Complex Model.  Nearly
one-third of U.S. gasoline is covered by the RFG requirements.  Gasoline sold in California is
subject to the California Phase 2 RFG Program (also known as the California Cleaner Burning
Gasoline Program) requirements, which includes an annual average sulfur limit of 30 ppm sulfur,
with no individual gallon exceeding 80 ppm.19 Because refiners want to ensure that they will not
exceed the annual average limit and some refiners have elected to certify their gasoline to
       18 The conversion of high boiling hydrocarbons to gasoline-like material tends to put more sulfur into
gasoline, since the sulfur contained in crude oil is generally concentrated in the high boiling material.

       19 California provides an alternate approach permitting refiners to comply with a per-gallon "flat limit" of
40 ppm. Under either the 30 ppm average standard or 40 ppm flat limit standard, refiners can use California'
Predictive Model to identify and certify alternative fuel formulations which provide equivalent emission
performance. These alternative formulations can involve either more stringent or more lenient sulfur limits, but in
no case can these exceed the 80 ppm per gallon cap applicable to all gasoline sold in California.

EPA Staff Paper: Gasoline Sulfur                 25                              May 1,1998

-------
alternative requirements which may involve even more stringent sulfur control, California
gasoline sulfur levels currently average about 20 ppm.

       API and NPRA recently surveyed their member companies to determine the qualities of
gasoline and other petroleum products sold during the summer of 1996.20 This survey,
summarized in Table 4, showed that gasoline had the following average sulfur levels during that
period.
Table 4. Average Sulfur Levels in Gasoline Sold Outside California (ppm):
Summer, 1996 Production By Octane Grade and Gasoline Type
Gasoline Type
Conventional
Federal
Reformulated
All Gasoline
Regular Grade
400
366
394
Intermediate Grade
350
353
351
Premium Grade
142
200
159
All Grades
346
316
340
       Starting in the year 2000, refiners producing summertime gasoline for sale in RFG areas
must meet a NOx reduction standard which is expected to result in lower gasoline sulfur levels in
those areas, since sulfur control is believed to be the most cost-effective means to achieve the
NOx reduction. Refiners have projected that their summertime Phase n RFG sulfur levels will
average about 150 ppm. Sulfur levels could be higher in the winter, since winter RFG is subject
to less stringent NOx reduction requirements.

       Because of the wide range of gasoline sulfur levels seen in the U.S., it is likely that any
sulfur control program will require sulfur reductions in at least some of the gasoline currently
produced.  An evaluation of the costs of sulfur control requires an understanding of how sulfur
ends up in gasoline and what technologies are available to refiners to remove the sulfur. The
following section addresses these issues.
B.     Refinery Operations Affecting Sulfur

       Most sulfur in gasoline (up to 97%) comes from gasoline feedstocks produced in the
fluidized catalytic cracker (FCC), one of the key refinery processes for gasoline production.21
Without any treatment of the sulfur levels of the feed entering the FCC unit, or of the gasoline
       20 American Petroleum Institute and National Petroleum Refiners Association, "Survey of Refining
 Operations and Product Quality," Final Report, July 1997.

       21 MathPro, Inc., "Costs of Producing Gasolines With Low Sulfur Content," Final Report, April 28,1997
 EPA Staff Paper Gasoline Sulfur
26
May 1,1998

-------
streams exiting the unit, sulfur levels can be 1000-2000 ppm in the FCC gasoline blendstocks.  In
general, the heaviest portion of FCC gasoline contains the most sulfur, with the heaviest 10%
contributing about one-third of the total sulfur. A more complete explanation of these refinery
processes is presented in Appendix D.

       One means to reduce the sulfur content of gasoline, and other refined products, is to
switch to a lower sulfur crude oil.  However, it is not possible for all refiners to rely on this
strategy because the supply of low sulfur crude oil is limited and with increased demand its price
would likely rise. In addition, it may be difficult for refiners to maintain the same product mix if
they switch to low sulfur crude oils, since low sulfur crude oils tend to be lighter. The volume
and range of products that can be produced from lighter crude oils may or may not match the
refiner's market plans.  (Even with a change to lower sulfur crude oils, many refiners may still
need to add new equipment to further reduce their gasoline sulfur levels.)

       While the specific configuration and capabilities vary by refinery, there are generally two
main options for removing sulfur from gasoline.  Both of these options involve "hydrotreating," a
process which uses hydrogen gas at high pressures and temperature to force out the sulfur
compounds.22  When adding the capacity to hydrotreat a particular stream, refiners may have to
add facilities to produce more hydrogen as well as facilities to recover the sulfur removed from
gasoline. Refiners may also have to augment utility facilities to ensure sufficient support for
these new processing units.

       Of the two main options for reducing gasoline sulfur, the less costly and less capital
intensive approach is to hydrotreat the gasoline blendstocks produced by the FCC.  Depending on
the level of sulfur desired in the finished gasoline, part or all of the FCC gasoline blendstock can
be hydrotreated. It is least expensive to hydrotreat the heaviest portion of the FCC gasoline
because that portion contains the greatest amount of sulfur, and this stream usually undergoes
less octane loss when hydrotreated.23 This strategy requires a distillation tower to separate the
heavier FCC output streams from the lighter products. A downside to hydrotreating the FCC
gasoline is that it reduces the volume of gasoline produced, thereby adding slightly to the cost of
producing lower sulfur gasoline. To achieve very low sulfur levels in gasoline, the lighter
portion of the FCC gasoline blendstock may have to be hydrotreated as well, which can result in
a significant octane loss, depending on the specific hydrotreating process used. The octane value
lost with hydrotreating must be recouped by octane producing units in the refinery, or by the
addition of high octane oxygenates. A relatively new sulfur removal technology licensed by
Mobil, called OCTGAIN, does not cause as much octane loss for the same reduction in gasoline
       22 Another option for very limited sulfur reduction is achieved by extraction. This is not a commonly used
approach, because in most cases a substantial volume of sulfur reduction is required, or a substantial reduction in
sulfur levels is needed; extraction cannot meet these needs.

       23 Hydrotreating impacts not only sulfur-containing compounds but also compounds which contribute
octane, such as olefins. If olefins are hydrotreated, their octane value is reduced and the refiner must make up fo
the octane by blending other, higher octane, components.

EPA Staff Paper: Gasoline Sulfur                 27                              May 1,1998

-------
sulfur as older processes. An OCTGAIN unit has been built in one refinery in the U.S. and
another unit is under development in Europe.

       EPA has become aware of a new FCC gasoline hydrotreating technology which may
permit refiners to remove sulfur from the FCC gasoline while minimizing the drawbacks of
traditional hydrotreating (octane loss, lost volume) at a significant cost savings.  This process is
licensed by CDTECH and is termed catalytic distillation, indicating a process which performs
hydrotreating with a catalyst while distilling the gasoline into different temperature ranges. In
terms of gasoline sulfur reduction, the most important potential for this technology, hydrotreating
the heavy FCC gasoline, has not yet been tested in a commercial application. Thus, uncertainties
in the cost of this emerging technology will remain until a commercial application can verify the
efficacy and cost estimates. EPA has learned that at least one refiner plans to install this
technology in their refinery during 1998.

       An alternative to hydrotreating the gasoline from the FCC unit is to hydrotreat the FCC
feed stream. This approach avoids a decrease in gasoline octane and generally improves the
quality of other products produced in the FCC unit, such as distillates. It can also achieve very
low sulfur levels.  However, this tends to be a more expensive and capital intensive approach,
because a larger stream of material is processed. For some refiners this may be an attractive
option since this approach allows the processing of heavier, higher sulfur crude oils, reduces
sulfur levels in diesel fuel,24 and increases overall gasoline yield. Few refineries in the U.S.
currently have FCC feed hydrotreating capacity.

       The other major way to remove sulfur in gasoline blendstocks is with a hydrocracker.
Hydrocrackers are essentially FCC units with hydrotreating capacity combined in the  same unit;
this process both upgrades and hydrotreats the stream being treated.23 Hydrocrackers  are
generally used when the feedstock is too poor for upgrading with an FCC.  For example, some
crude oils are high in heavy metals, nitrogen, or sulfur, which would poison an FCC catalyst.
For this reason, some refineries have hydrocrackers for upgrading the heavy gas-oil streams
normally fed to the FCC. This is particularly common in California refineries which process
Alaskan crude which is high in sulfur and heavy metals. Hydrocrackers are capital-intensive
processes, similar to that of FCC feed hydrotreaters.  Because of the high capital expense and
because most refineries already have FCC units, many refiners outside of California would have
to incur substantial costs and process changes to adopt hydrocracking capabilities. Like feed
hydrotreaters, hydrocrackers would allow treatment of heavier, higher sulfur crude oils and may
be more suitable for environmental programs which involve changes to both diesel fuel and
gasoline.
       24 If diesel fuel sulfur levels were required to be reduced as well, FCC feed hydrotreating becomes a more
attractive option as the sulfur levels of both streams are reduced simultaneously. See Chapter IV: Other Fuel Issue
for more discussion of this issue.

       25 Berger, Bill D., Anderson, Kenneth E., Modem Petroleum, "A Basic Primer of the Industry," Pennwell
Books, 1981.

EPA Staff Paper:  Gasoline Sulfur                 28                             May 1,1998

-------
       The modeling of gasoline sulfur reduction in refineries has shown that the most economic
approach to removing sulfur from gasoline is to hydrotreat the gasoline-like material leaving the
FCC unit. When the OCTGAIN process has been included in a refinery model, including the
model used by EPA, it has been selected (based on the economic and performance data provided
to the model) over other sulfur reduction processes. The models also typically purchase
oxygenates to help make up the octane shortfall caused by hydrotreating, and this oxygenate
addition also helps to reduce sulfur in gasoline through dilution. The refinery models tend to
show that hydrotreating the FCC feed and the installation of hydrocracking capacity are less
economically attractive to refiners. The following section provides a brief overview of the
modelling EPA has performed to estimate gasoline sulfur control costs.
C.     Cost Estimation Methodology

       EPA has been working with DOE to develop cost estimates for gasoline sulfur control,
using Oak Ridge National Laboratory's refinery yield model (ORNL-RYM). (Appendix B
presents additional information about how the modeling work to be done was defined.)  In
planning the refinery modeling runs to be performed, EPA considered a range of geographic
areas that could be covered by a low sulfur gasoline program, based on the range of vehicle
technology issues being addressed and the emissions reductions which could be achieved. In
consideration of these issues, the geographic areas identified for consideration included: 1) the
entire U.S., excluding California; 2) the 37-state Ozone Transport Assessment Group (OTAG)
region; and 3) the 22-state OTAG region covered by the proposed NOx State Implementation
Plan (SIP) call. Table 5 shows the percentage of national gasoline affected for each of these
potential areas of control.
Table 5. Gasoline Production Impacted by Potential Control Areas
Sulfur Control Area
22-state OTAG Region
37-state OTAG Region
Entire U.S., excluding California
Gasoline Volume (percent of
U.S. total, outside California)
59%
89%
100%
       A national control strategy is one of the options being evaluated because of its
consistency with the expected national applicability of potential Tier 2 standards, the possibility
that the sulfur impact may not be fully reversible, and the other emissions benefits, such as
reductions in air toxics and primary and secondary PM, that could be achieved from current
vehicles. In addition, as discussed in Chapter n, a national program may enable the use of certain
advanced technology vehicles and would remove any remaining concerns that onboard diagnostic
systems may malfunction due to interference by sulfur. A national program would also create


EPA Staff Paper: Gasoline Sulfur                 29                             May 1,1998

-------
nationally uniform enforcement standards and compliance procedures for industry to abide by,
possibly reducing compliance costs and making it easier to ensure that gasoline is meeting the
new standards relative to a smaller program.

       Figure 4 shows the area covered by the 37-state OTAG region. This region was identified
as a potential sulfur control area because this was the region that was considered during the
OTAG process to identify strategies to control the transport of ozone. However, it excludes 11
projected future ozone nonattainment and maintenance areas in seven other states, affecting a
1990 population of 13.4 million people. As Figure 4 shows, the 37-state OTAG region almost
exactly corresponds to the fuel distribution area from the refineries in PADDs 1,2, and 3.26
Covering the entire volume of gasoline sold in these three PADDs may tend to simplify
enforcement of a sulfur control program compared to a smaller program affecting only a fraction
of the gasoline sold in these regions.

      The 22-state OTAG region (also shown in Figure 4), although smaller than the 37-state
region, encompasses many of the expected future ozone nonattainment and maintenance areas as
well as many of the areas which contribute to the transport of ozone. However, it would exclude
even more such areas than the 37-state OTAG region discussed above. The 22-state OTAG
region is sufficiently limited in gasoline demand to exclude production by some refineries
located in  PADDs 1-3 for which sulfur reduction would be very expensive, which would result in
lower costs for a given sulfur level compared to a program affecting a larger volume of gasoline.
However,  the complexities in enforcing a smaller program, particularly one that covers only a
fraction of the gasoline produced in a given PADD, would probably increase.  Because of the
gasoline distribution system for the refineries that would produce gasoline  for this region, a
program may have to be implemented to prevent "dumping" of high sulfur gasolines into
uncontrolled areas.
       26 U.S. refineries are grouped regionally into five Petroleum Administration Districts for Defense
 (PADDs) to represent the production and distribution systems. See Figure 8 for the division of the U.S. into the
 five PADDs.

 EPA Staff Paper:  Gasoline Sulfur                 30                             May 1,1998

-------
I
03
Petroleum Administration Districts for Defense (PAODs) and OTAG Areas
o
00
                                     37 State OTAG area •



                                     22 State OTAG area-
                         HAWAH
                  Figure 4
VO
VO
oo

-------
D.     Gasoline Sulfur Reduction Cost Estimates

       EPA has developed preliminary cost estimates for sulfur control in PADDs 1 and 3,
assuming that all of the gasoline produced in each PADD is controlled.  The costs of reducing
gasoline sulfur levels from the current average sulfur level of 340 ppm down to average levels of
150 ppm, 100 ppm, and 40 ppm have been estimated. These sulfur control levels were chosen
because they matched the sulfur levels of the AAMA/AIAM and CRC test programs described in
Chapter n.  EPA used the results from the  ORNL-RYM model to estimate the net costs to the
industry using a pre-established cost estimation methodology, with a few modifications, as
described in Appendix B.27  The estimated costs, listed in Table 6, include the cost of any
projected decreases in fuel economy associated with the use of oxygenates to reduce sulfur
levels. As explained in Appendix B, the costs were derived based upon estimates of summer
production costs, as this season represents  the most severe sulfur control season for a refiner.
However, as sulfur controls were assumed to be in effect year-round, the resultant capital costs
were amortized over the entire year.
Table 6. Preliminary Estimates of Sulfur Reduction Costs
(Volume- Weighted Average for PADDs 1 and 3, 8% Return on Investment)
Sulfur Control Level
150 ppm
l.l-1.8c/gal
100 ppm
1.9 - 3.0 c/gal
40 ppm
5.1-8.0c/gal
       The cost estimates in Table 6 should be considered to be only roughly indicative of the
potential cost of commercial gasoline sulfur control. These cost estimates only represent those
for East Coast and Gulf Coast refiners (PADDs 1 and 3), and thus do not cover all of the major
gasoline producing regions in the country which would be affected by a national sulfur control
program.  Also, these costs were developed without inclusion of some of the latest technologies
for sulfur removal, including the process for catalytic distillation developed by CDTECH
mentioned previously. At a recent conference on sulfur removal technologies,28 CDTECH
representatives claimed that sulfur control from current average levels down to about 40 ppm (on
average) could be achieved with the CDTECH catalytic distillation process at a cost of 1-2
cents/gallon. Other designers of new sulfur removal technologies made similar claims. As
indicated earlier, EPA plans to gather additional information about new technologies, and if
appropriate, will include these technologies in future evaluations of sulfur removal costs. EPA
wi]J also be estimating the costs of gasoline sulfur control to refiners in the rest of the country.
       27 Lester Wybomy, MEMORANDUM to Susan Willis, "Review of the Cost-Effectiveness of Phase II
 RFG NOx Control," February 27,1997.

       28 International Fuel Quality Information Center, "Fuel Quality and Technology Briefing," Detroit, MI,
 March 9,1998.
 EPA Staff Paper:  Gasoline Sulfur
32
May 1,1998

-------
This will facilitate more precise estimates of the costs of both regional and national control
programs, as well as the impact of using the most advanced sulfur removal technology.
E.     Other Sulfur Control Cost Studies

       In addition to performing the analysis discussed above, EPA has reviewed several other
recent estimates of the costs of gasoline sulfur control.  During the period when the Ozone
Transport Assessment Group (OTAG) was developing costs for NOx control strategies, the
American Automobile Manufacturers Association contracted with MathPro to estimate the cost
of gasoline sulfur reduction.29 MathPro estimated the cost of gasoline sulfur reduction in all the
gasoline produced by PADDs 1, 2, and 3 from the current average sulfur level down to 40 ppm.
A range of prices for the oxygenate methyl tertiary-butyl ether (MTBE) was assumed in the
modeling. If an MTBE price of 84 cents/gallon is assumed (to be consistent with the EPA/DOE
work described in the previous section), the cost of sulfur control estimated by the MathPro
analysis is 5.5 cents per gallon of summertime gasoline.

       The MathPro/AAMA cost estimate falls near the lower end of the range of preliminary
costs estimated by EPA/DOE. However, caution must be exercised in comparing different
refinery modeling studies without understanding the differences in the premises of the studies,
which may result in differences in cost. For example, the MathPro study included the cost of
sulfur control in PADD 2 refineries. It is expected that PADD 2 costs would be higher than the
average of PADD 1 and 3 costs. Thus, adding PADD 2 costs to EPA's estimates would be
expected to increase the average costs, since PADD 2 refineries currently produce higher sulfur
gasoline (on average) than PADD 1 and 3 refiners. By contrast, resolution of some of the other
modeling differences between the EPA/DOE work and  the MathPro/AAMA work could tend to
push the MathPro estimates higher.

       Recently, API also contracted with MathPro to estimate the cost of gasoline sulfur
reductions.30  That study estimated the cost of sulfur reduction from current average sulfur levels
down to average levels of 150 ppm, 100 ppm, and 40 ppm.31 To estimate these costs, MathPro
used a refinery modeling methodology similar to that used for AAMA, modeling the combination
of PADDs 1,2, and 3. However, the refinery model used for API was re-calibrated to represent
the actual 1996 summertime gasoline qualities determined through the API survey of gasoline
qualities referenced above in section A. The MathPro/API-estimated costs are summarized in
       29 MathPro, Inc., "Costs of Producing Gasolines with Low Sulfur Content," work performed for AAMA,
April 28,1997.

       30 MathPro Inc., "Costs of Alternative Sulfur Content Standards for Gasoline in PADDS 1-3," Study for
the American Petroleum Institute, February 5,1998.

       31 The MathPro analysis also considered a low sulfur gasoline scenario in which reformulated gasoline had
an average sulfur level of 100 ppm and conventional gasoline had an average sulfur level of 246 ppm. The cost for
this scenario, assuming an MTBE price of 90 c/gal, was estimated to be 1.90 c/gal.

EPA Staff Paper. Gasoline Sulfur                33                            May 1,1998

-------
Table 7 below.  Like the AAMA analysis, these costs are similar to those estimated by ORNL for
EPA/DOE, even though there were significant differences in the premises on which these
modeling exercises were based. However, the MathPro/API estimates for control to 150 ppm or
100 ppm are at or above the upper end of the EPA/DOE ranges, while the MathPro/API estimate
for 40 ppm sulfur is closer to the lower end of the EPA/DOE range. EPA will continue to
analyze the MathPro/API analysis to better understand the causes of these differences.
Table 7. Estimated Cost of Reducing Gasoline Sulfur, cents/gallon *
[Source: MathPro for API]
Average Sulfur Control Level (ppm):
Cost (c/gal):
150
2.7
100
3.4
40
5.1
' average of summer and winter costs based on an MTBE price of 90 c/gal
F.     Cost-Effectiveness of Sulfur Control

       EPA typically considers cost-effectiveness (the cost of reducing a ton of emissions) when
evaluating emission control programs. Calculating cost-effectiveness allows comparisons
between various emission control strategies that are aimed at achieving reductions in the same
pollutant (e.g., the cost of a motor vehicle or fuel  control strategy compared to the cost of various
stationary source control strategies for reducing emissions of NOx). Cost-effectiveness is a useful
tool for comparison, although other criteria are also used to determine which regulatory programs
to pursue.

       In determining the cost-effectiveness of a  regulatory option, both the costs and the
resulting emissions reductions must be quantified. The estimation of the cost is the most
straightforward of the two factors. The costs of vehicle standards can be derived from estimated
per-vehicle costs, multiplied by total affected sales. Similarly, the costs of a sulfur control
program can be derived from an estimated cost per gallon, multiplied by affected sales.

       It is the estimate of the emissions benefits to be achieved that can be more difficult,
particularly when the control program being evaluated involves more than one significant
benefit. Cost-effectiveness is most easily determined when a single pollutant is being controlled,
or at least reductions in the emissions of one pollutant are the predominant benefit. Cost-
effectiveness becomes more difficult to calculate when multiple benefits (e.g., reductions in
NMHC, CO, NOx, toxics, and SOx, enablement of future technology, etc.) are significant, as is
expected under potential Tier 2 and gasoline sulfur standards. Not only is it difficult to
determine how to weight, on a relative basis, the various emissions reduced (e.g., is a ton of NOx
reductions as beneficial as a ton of SOx), but it can also be difficult to compare the resulting
cost-effectiveness estimates with estimates of the cost-effectiveness of other control strategies
that may not have valued these emissions reductions in the same way.
 EPA Staff Paper: Gasoline Sulfur                 34                             May 1,1998

-------
       Recently, representatives of the API Economics Committee (including a representative of
NPRA) presented a proposed cost-effectiveness methodology to EPA.32  This methodology
combined the evaluation of Tier 2 standards and gasoline sulfur standards and focused on the
ozone-related benefits from both the current and future vehicle fleets.  API and NPRA
recommended that this methodology be used to determine the optimum mix of lower sulfur fuel
requirements and improved vehicle technology. They identified the specific  types of data needed
for this approach, as well as gaps in the available data.
G.     Other Issues Affecting Sulfur Control Costs

       In addition to the production cost for the average refinery, there may be other issues
which could affect the costs to the refining industry of a sulfur control program. For example,
one consideration is whether the program should be defined based on an average sulfur level, or
based on a per-gallon limit on sulfur.  The preliminary cost estimates by EPA, as well as the
other cost estimates cited previously, were for the costs for the average refinery controlling sulfur
to an average level.  If such a sulfur reduction standard  were promulgated, a refiner would have
to control the volume-weighted average sulfur level of its gasoline batches at or below the
averaging standard.  While an averaging standard leads to reduced sulfur levels overall, over the
year some batches of gasoline would be higher and others lower than the standard. These batch-
to-batch variations are due to the variances in the day-to-day operations of the refineries,
including variations in the sulfur content of the crude oils processed and both inadvertent and
required maintenance shutdowns of certain units.

       As explained in Chapter n, batches of gasoline  with sulfur levels above a certain (as yet
unidentified) level may cause serious complications for some emission control hardware.
Furthermore, the compliance costs associated with an averaging program are  likely to be higher
than the costs for a program which provides an absolute cap on sulfur content, because of the
complexities of enforcing an averaging standard  (as discussed in the next section). For these
reasons, it may be more desirable to establish a maximum per-gallon limit on gasoline sulfur
levels. With such a cap, refiners could not exceed the specified level in any batch.

       A per-gallon limit affects the refining industry much differently than an averaging
standard. This is because a cap forces an individual refiner to install capital equipment or take
other actions to reduce high sulfur batches to a greater degree than would be necessary under an
averaging program. Even if some of the historic variations in gasoline sulfur level were reduced,
under a per-gallon limit a refiner would have to produce gasoline which, on average, would be
substantially lower than the cap to ensure that they comply.

       Since the cost estimates derived from refinery models are all based on compliance on
average, it is difficult to evaluate the costs of a per-gallon limit. EPA has performed an analysis,
       32 "APMNPRA/EPA Meeting - 3/10/98 - Tier 2 Sulfur Control Cost-Effectiveness"

 EPA Staff Paper. Gasoline Sulfur                 35                             May 1,1998

-------
presented in Appendix C, to estimate a range of average sulfur levels that could arise if a
per-gallon cap were implemented. Using this approach, EPA may be able to estimate the costs of
a sulfur cap using costs derived from refinery modeling which yields costs for control on average.

       A second issue to consider in defining a sulfur control program is whether all refiners will
be able to make the investment needed to control sulfur. As with any regulatory program, a
gasoline sulfur control program - particularly one that affects the majority of gasoline produced
in the country - may result in particular hardship for some individual refineries. If a refinery
supplies a limited geographic area and if few other refiners can easily supply the same area, the
costs of the sulfur control program in that area may be substantially higher than the average costs
experienced elsewhere. In defining a sulfur control program, EPA plans to evaluate whether the
opportunity exists to include program design strategies which may result in reduced costs and
increased flexibility for the regulated industry, particularly those refiners most adversely affected
by the program. Examples of these strategies include trading and banking of credits for early or
excess compliance with standards, delayed implementation for small refiners, and phase-in of the
requirements (by fraction of production or by increasing stringency) over several years, to spread
out costs and investment. No estimates of the cost savings, other benefits, or emissions
implications that would result from pursuing such strategies have yet been made, and the
introduction of additional compliance options will further complicate the enforcement of the
program requirements.
H.     Enforcement of a Gasoline Sulfur Standard

       There are a variety of forms that a gasoline sulfur control program can take, and the
ultimate decision regarding which form is most appropriate will depend heavily on the costs
incurred by the regulated  industries. However, there are also enforcement implications for each
of the various choices, and the costs to industry of these different enforcement and compliance
burdens cannot be easily quantified, although they must be taken into consideration. The
enforcement implications are described here briefly; a more detailed discussion and evaluation of
these issues will appear in any regulatory documents EPA develops in the future.

       A nationwide, per-gallon gasoline sulfur standard would be the most straightforward
program from an enforcement perspective.  In this case, the Agency may be able to focus
primarily on the sulfur level of individual batches at the refinery gate and/or collect and analyze
gasoline samples for sulfur content at all levels of distribution. It is possible that the current
reporting requirements for all gasoline under the anti-dumping regulations could help serve this
purpose, avoiding the need to implement new reporting procedures for gasoline sulfur.

        If, however, any alternative form of a gasoline sulfur control program were implemented,
the compliance burden would increase. For instance, allowing averaging to meet a sulfur
standard might require: 1) a more complex set of compliance procedures for refiners to prove that
their volume-weighted, annual average sulfur levels meet the standard, 2) sampling surveys to


EPA Staff Paper: Gasoline Sulfur                 36                             May 1,1998

-------
establish appropriate compliance within a given area, and/or 3) imposition of downstream sulfur
standards which require distributors to establish compliance with the standard to ensure that the
gasoline sold at the retail level meets the standard, just as the gasoline that left the refinery did.
Furthermore, an averaging standard could require an enforcement program involving retail-level
fuel quality surveys and the threat of tightening standards (ratchets) to ensure that the average
sulfur standard applicable to a given area is actually being met in-use. A regional program could
lead to the use of segregated storage and distribution systems, to ensure that low-sulfur gasoline
is being delivered to the correct area(s).  Finally, the inclusion of phased-in requirements, credits
for early compliance, and delayed implementation might require additional enforcement
protocols having precedent in such fuels control programs as the phase-out of lead in gasoline.
The Agency will consider these and other enforcement implications in designing the most
appropriate sulfur control program.
 EPA Staff Paper: Gasoline Sulfur                 37                              May 1,1998

-------
IV.    Other Fuel Issues

       Few refinery operations are totally independent, and variations in the volume or quality of
one refinery product can impact the volume and quality of another product. Sulfur removal
technologies vary, and depending on which technology is selected, gasoline alone can be
impacted, or other refinery products, such as the distillates used for diesel fuel and other
products, can also be affected. Similarly, controlling fuel properties other than gasoline sulfur
can impact the cost of gasoline sulfur control.

       The impact of gasoline sulfur control on other refinery products is addressed directly in
the refinery model, as non-gasoline sulfur specifications must be met both before and after
gasoline sulfur control. The primary issue in this section is  the potential impact of changes to
non-sulfur properties of gasoline or other fuels on the cost of gasoline sulfur control.  Also, these
other fuel controls could increase the overall economic impact on a given refinery. Details on the
refinery interactions between gasoline sulfur control and the control of diesel fuel properties can
be found in Appendix D.

       This chapter summarizes several other fuel properties which may have significant impacts
on emissions and that vehicle and engine manufacturers may request improvements in these
properties to facilitate compliance with emission standards in the future. The first section
discusses the distillation properties of gasoline, which collectively represent the ease with which
liquid gasoline vaporizes and thus has implications for combustion and other phenomenon. A
number of automobile manufacturers have indicated that current levels of gasoline distillation
properties are limiting their ability to meet stringent emission standards. The second section
highlights several diesel fuel properties that may be implicated in elevated emissions.  Additional
controls in either of these areas could impact a given refiner's decisions regarding the
technological means through which sulfur is removed from  gasoline, and therefore the cost of
gasoline sulfur control.
A.     Distillation Properties of Gasoline

       It is widely understood that there is a relationship between the distillation properties of
gasoline and the emissions which result upon combustion of the fuel. Gasoline distillation
characteristics are typically described in terms of the temperature at which a given percent of the
fuel (e.g., 10%) evaporates during a standardized distillation test. Thus, T10 is defined as the
temperature at which 10% of the fuel has evaporated, T50 is the temperature at which 50% of the
fuel has evaporated,, and so on.
 EPA Staff Paper: Gasoline Sulfur                 38                             May 1,1998

-------
       The auto industry has defined the "driveability index" of a fuel, calculated from the
equation [DI = 1.5(T10) + 3.0 (T50) + T90],33 as a measure of the potential for a vehicle to
experience "driveability" problems - hesitations, stalls, etc. - as a result of the fuel. According to
the auto industry, a driveability index in excess of 1200 indicates that the fuel may result in
driveability problems.34 Auto industry representatives have also shared with EPA data which
suggests an historical trend of increasing values of the driveability index in gasolines sold
nationwide.

       Previous studies of the effect of fuel properties on emissions have demonstrated that
hydrocarbon emissions, in particular, are correlated with high values of individual distillation
temperatures, such as T50 and T90.  High hydrocarbon emissions can also be assumed to occur
during instances of poor driveability,35 such as hesitation during acceleration or engine stalls,
since such situations indicate that a significant amount of engine misfire is occurring. Engine
misfire typically indicates that most or all of the fuel inducted into the cylinder leaves the engine
unburned.  The fuel does not ignite because an insufficient amount of the fuel injected into the
intake manifold vaporizes to produce a flammable mixture in the engine cylinder.  (Combustion
of gasoline requires a specific amount of gasoline vapor, combined with  oxygen; if too much of
the gasoline remains liquid, combustion will not occur.)

       When misfire occurs, it also causes  a large amount of fuel and oxygen to be sent to the
catalyst.  If misfire occurs over a sufficient  length of time, this excess fuel and oxygen can
increase catalyst temperatures to the point where the catalyst is permanently damaged. Small
amounts of misfire are unlikely to be noticed by the driver, but can still significantly increase
tailpipe emissions of hydrocarbons. Whenever noticeable driveability problems are encountered,
emissions are likely to have  increased substantially.

       If the distillation properties of all gasoline were similar, automotive  engineers could
account for the situations where insufficient vaporization may occur when calibrating the amount
of fuel to be injected during  accelerations.  Fuels with high distillation temperatures vaporize
less, leading to lean air-fuel  mixtures at the beginning of accelerations. Fuels with low
distillation temperatures vaporize more, leading to rich air-fuel mixtures at the beginning of
accelerations.  However, when fuels with different distillation characteristics are encountered,
there is currently no  practical way to measure or sense the distillation properties of a fuel on the
vehicle and adjust the calibrations based on this information.
       33 Where the T10, T50, and T90 represent the temperatures at which 10%, 50% and 90% of the gasoline
evaporates, measured in degrees Fahrenheit.

       34 Although oxygenates decrease the T50 of the gasoline (and thus reduces the driveability index), it has
been suggested that oxygenated gasolines may result in driveability problems with a driveability index over 1150.

       33 Increased emissions resulting from driveability problems may not be reported on tests conducted under
the federal test procedure (FTP), since tests in which hesitations or stalls occur may have been invalidated.

EPA Staff Paper Gasoline Sulfur                  39                              May 1,1998

-------
the quality of diesel fuel used in nonroad diesel engines.  However, during 1998 and 1999, EPA
will review the appropriateness of the 2.4 gram per brake horsepower-hour (g/BHP-hr) HC plus
NOx standard applicable to new onroad heavy-duty diesel engines sold in 2004 and beyond.
This technology review will include consideration of fuel quality improvements. Sulfur in diesel
fuel is known to be a problem widi lean-NOx catalyst technology, which was addressed in
Chapter n above.  While control of nonroad diesel fuel quality appears less likely, it is possible
that a sulfur limit similar to that applicable to onroad diesel fuel could be implemented to
facilitate the use of oxidation catalysts or to reduce sulfate paniculate.  In 2001,  EPA will be
conducting a similar review of the feasibility of the future standards applicable to new nonroad
heavy-duty diesel engines. EPA (and California) are also conducting evaluations of the health
effects of exposure to diesel exhaust, including paniculate emissions from diesel engines. The
results of these evaluations could have implications for future decisions about diesel fuel
composition.

       Although any new controls applied to diesel fuel can potentially impact a given refiner's
decisions regarding the means through which sulfur is removed from gasoline, it appears that
diesel sulfur control has  the potential for the greatest refinery interactions with gasoline sulfur
control. As indicated above, a gasoline sulfur standard would most likely lead refiners to remove
sulfur from gasoline blendstocks. More stringent diesel sulfur standards would  likely lead
refiners to remove sulfur from diesel fuel blendstocks, as opposed to other strategies, such as
switching to lower sulfur crude oils.  However, if both changes were made at the same time,
refiners might remove sulfur from feedstocks going to both gasoline and diesel fuel (e.g., the
feedstream to the FCC unit) and reduce the total cost of sulfur control. Also, a single large
hydrogen plant could be built, instead of two smaller units, again at a cost  savings.

       The Agency's modeling of gasoline sulfur reductions costs completed to date has not
modeled the cost of the simultaneous imposition of any diesel fuel program with gasoline sulfur
control. However, the Agency is planning to perform such a study to understand the costs and
other implications of implementing programs affecting both gasoline and diesel fuels at the same
time.

       Although diesel sulfur control may have the most significant refinery interactions with
gasoline sulfur control, controls applicable to other diesel fuel properties also have the potential
to impact the way a refinery operates, and thus may also  impact gasoline sulfur control. The
need for changes to diesel fuel properties odier than sulfur will  also be evaluated in the context of
the technology review for the model year 2004 standard of 2.4 g/BHP-hr for onroad diesel
engines and the technology review for standards  applicable to nonroad diesel engines proposed
for 2001.
 EPA Staff Paper: Gasoline Sulfur                41                             May 1,1998

-------
V.     Gasoline Sulfur Reduction Programs in Other Countries

       This paper focuses primarily on the concerns with gasoline sulfur levels in the United
States. However, other countries are also taking steps to limit the sulfur content of gasoline as
well as diesel fuel.  The characteristics and operating circumstances of national and regional
gasoline refining industries differ to some degree from country to country.  While most refiners
have access to the same crude oils, some utilize local crude oils with unique compositions.  Also,
the mix of refined products differs from country to country. For example, the market in some
countries is more oriented to distillates (e.g., diesel fuel) and others to gasoline. However, the
primary benefits of sulfur control in other countries are the same as those here - to enable
advanced emission control technologies and to reduce emissions from current or future vehicles.
Harmonizing emission standards and test procedures across national boundaries which meet the
levels of control EPA determines to be appropriate is also a long-term goal of EPA.  While
considerable effort is still required to achieve this goal, it cannot be achieved without
harmonizing those aspects of fuel quality which affect vehicle technology and emissions. Thus,
international regulatory activity on vehicle standards and fuel quality should be considered as
EPA proceeds to develop regulatory programs affecting both vehicles and gasoline sulfur levels.
This chapter presents an overview of the current and proposed gasoline sulfur standards in other
countries as well as international requirements and activities for reducing gasoline sulfur levels.
A.     Canada

       In Canada, the Canadian General Standards Board (CGSB) has adopted voluntary
standards for gasoline sulfur following a consensus process involving the producers (refineries
and importers), users (e.g., automobile manufacturers), and other parties (e.g., governments).
This process is very similar to the process followed by ASTM in the U.S. The CGSB's standard
for gasoline sulfur is 1000 ppm, which is the same as the level established by ASTM in the U.S.
Although CGSB's standard for unleaded gasoline specifies a maximum sulfur content of 1000
ppm, there are no federal requirements that limit gasoline sulfur levels.  However, the Canadian
government is currently considering reductions in gasoline sulfur levels nationwide.36
Recommendations to reduce sulfur to levels as low as 30 ppm on average/80 ppm maximum
have been made, although no final decisions have been reached. The Canadian government is
collecting comments on the range of recommendations, and will hold a public meeting in May
1998 to gather additional comments.

       Individual Canadian provinces have adopted the CGSB standard, or some other standard,
for gasoline produced in that province.  For example, Ontario has adopted the CGSB limit, while
Quebec's "Reglement sur les Produits P6troliers" limits the maximum level of sulphur in gasoline
       36 See Reports referenced in: Glenn Allard, Environment Canada, Memorandum to Steering Committee
members, Sulphur in Gasoline and Diesel Studies, Reports Concerning Sulphur in Gasoline and Diesel Fuels, 29
August 1997.

EPA Staff Paper: Gasoline Sulfur                 42                             May 1,1998

-------
to 1500 ppm. British Columbia has cleaner burning gasoline regulations which specify average
sulfur reductions, but also permit compliance by demonstrating that reductions in NOx emissions
have been achieved, as determined using the U.S. EPA's Complex Model.  Hence, while the
sulfur levels specified in the regulations are expected to be seen on average, the potential exists
for some gasoline to have higher sulfur levels (while still meeting the NOx emissions
performance standards).  British Columbia's regulations specify that beginning in 1999, the
annual average sulfur level of any primary distributor must not exceed 150 ppm in Southwest
B.C. and Vancouver Island. Beginning in 2000, the annual average must not be greater than 200
ppm for the rest of the province.
B.     European Union and Countries within Europe

       Recently, the European Parliament adopted fuel quality specifications for gasoline and
diesel fuel. These fuel specifications include a 150 ppm cap on gasoline sulfur beginning in the
year 2000.  Beginning in the year 2005, this sulfur cap is lowered to 30 ppm. The Parliament
must now reach a "common position" with the European Council of Ministers. A common
position is anticipated and a final rule for the European Union is expected between May 15 and
June 1 of 1998.37  In addition, Sweden and Finland have had tax incentive programs since 1994
that encourage gasolines with sulfur levels below 100 ppm.3*
C.     Other Countries

       In addition to the United States, Canada, and Europe, other countries around the world are
also implementing programs to limit gasoline sulfur levels. A summary of the current and
expected future gasoline sulfur standards for various countries around the world is presented
below in Table 8.39  In Mexico, cleaner-burning fuels are required in Mexico City during the
winter.  In October 1996, the state-owned oil company PEMEX began offering and promoting a
gasoline called magna reformulada. This gasoline has been described as "comparable to CARB
Phase 2 RFG." However, the sulfur level of this gasoline can be as high as 500 ppm. The magna
reformulada fuel may be expanded to Monterrey and Guadalajara in the future. In addition, the
government plans to subsidize the retail price of the fuel.40
       37 Memorandum from Fred L. Potter, Information Resources, Inc., to Margo Oge, U.S. EPA, European
Parliament Passes Tight Fuel Specifications, 18 February 1998.

       38 Tushingham, Dr. Mark, "International Activities Directed at Reducing Sulphur in Gasoline and Diesel,"
Environment Canada, 25 August 1997.

       39 Alfred, Kenneth R., "Why Fuel Sulfur Varies Worldwide," Hart's Fuel Technology & Management
January/February 1998. [Refer to this document for sulfur levels in the following countries:  Germany, Austria,
Switzerland, France, United Kingdom, Central/So. Europe, Latin America, Asia (except Japan), and Australia.

       40 Tushingham, 25 August 1997.

EPA Staff Paper: Gasoline Sulfur                 43                             May 1,1998

-------
       Japan has established a maximum per-gallon limit for sulfur in gasoline of 100 ppm.
However, in 1996, Japanese regular gasoline had an average sulfur level of 27 ppm whereas the
average sulfur level of premium gasoline was 7 ppm. Although in the future there is a possibility
that Japan may set similar gasoline standards to those of the U.S. or the European Union (through
the new Japan Clean Air Program), there will most likely be no additional improvements in
gasoline sulfur levels in the near term due to the very low sulfur levels already seen in the
country.41 Elsewhere in Asia, Thailand regulations require a cap of 1000 ppm sulfur in unleaded
gasoline.

       In Australia, the government of the state of New South Wales (NSW) proposed to reduce
sulfur in gasoline to 100 ppm initially with a further reduction to 40 ppm. The current sulfur
standard of NSW gasoline is approximately 200 ppm. Australian gasoline overall has an average
sulfur level of 175 ppm in regular unleaded gasoline and 151 ppm in premium unleaded.
       41 Idemitsu Kosan Co., Ltd., "Clean Air Program in Japan," Presentation to the U.S. EPA, December 1997.

 EPA Staff Paper:  Gasoline Sulfur                44                             May 1,1998

-------
Table 8. Summary of Gasoline Sulfur Levels in Other Countries
Country
^^^^^^^^^^^^^^^^^^^^^^^^E
European Union
(Parliament)
Sweden
Finland
Germany, Austria,
Switzerland
France
United Kingdom
Central/So. Europe
Canada
Atlantic
Quebec
Ontario
Prairies
British Columbia*
Mexico
Latin America
Japan
Thailand
Asia (except Japan)
Australia
New South Wales
Current Avg.Sulfurfppm)
Regular
300


176
466
348
257
343
276
364
470
228
273

623
<100

245
175

Premium


115
249
287
124







503
<100

159
151

|
Current
Sulfur Standard
500 ppm
Recommended
100 ppm
(tax incentive)
Recommended
100 ppm
(tax incentive)




1000 ppm





500 ppm max
(Mexico City)

100 ppm max
1000 ppm


200 ppm
1
Future/Proposed
Sulfur Standard
^^^^^^^^^^^^^^^^^^^^^
150 ppm
30pprn






50-360 ppm




250 avg or 1993 baseline
150 ppm avg in SW BC
200 ppm avg, rest of BC
500 ppm max
(Monterey & Guadalajara)





100 ppm; 40 ppm later
Effective
Date
2000
2005






varies




1996
1999
2000







'British Columbia's regulations also permit compliance with NOx emissions reductions, determined using EPA's
complex model, which may result in higher sulfur levels than specified here.
EPA Staff Paper: Gasoline Sulfur
45
May 1,1998

-------
VI.    Stakeholder Positions on Sulfur

       A number of parties have taken positions regarding the need for and appropriateness of
gasoline sulfur control for the benefit of potential Tier 2 vehicles and the fleet as a whole.  The
following sections summarize the positions taken by various interested parties on this issue.
A.     Automotive Industry

       AAMA and AIAM have jointly petitioned EPA to limit gasoline sulfur level for all
gasoline (outside of California) "as soon as possible" to the level that is "as low as practicable
but in no case greater than 40 ppm per gallon by weight...or,... no greater than 30 ppm annual
average by weight with a per gallon cap of 80 ppm."42 The petition claims that sulfur control is
necessary for the following reasons:

       •      Higher sulfur levels will reduce the benefits of current and advanced technology
              vehicles by impairing the emission control systems.

       •      Fuel sulfur contributes to air quality problems such as paniculate, sulfur oxides,
              and air toxics emissions.

       •      The sulfur impact on emission control systems is irreversible under normal
              operating conditions.

       •      If sulfur is not controlled, more fuel-efficient technologies will not be introduced
              in the U.S. because of their great sensitivity to sulfur.

The petition also states that sulfur reduction technology is available to refiners, is a cost-effective
means to reduce emissions, and is already used in California and in other countries. EPA is
currently in the process of reviewing the petition.

       Several individual automakers have also released positions on the need for gasoline sulfur
control. Chrysler Corporation has concluded that high fuel sulfur levels reduce catalyst
efficiency, increase tailpipe emissions, and are a barrier to lean burn technology.43  To address
these issues, they have recommended California sulfur levels (30 ppm on average, 80 ppm max.)
       41 American Automobile Manufacturers Association and Association of International Automobile
Manufacturers, "Petition to Regulate Sulfur in Gasoline Under Section 211(c) of the Clean Air Act," Before the
United States Environmental Protection Agency, March 19,1998.

       43 Boltz, Ronald R., "What's Coming - Emission Control Standards and Technology," presentation to the
CRC Auto/Oil Symposium: The Search for Sulfur-Tolerant Emission Control Systems, Dearborn, Ml, September
11,1997.


EPA Staff Paper: Gasoline Sulfur                  46                              May 1,1998

-------
Similarly, Ford Motor Company has concluded that the catalyst technologies presently being
developed for 2002 model year and later vehicles will not be capable of meeting LEV/ULEV
standards with fuel sulfur levels greater than 300-500 ppm.44
B.     Oil Industry

       API and NPRA recently recommended a sulfur control program which they believe will
meet environmental needs and will be less costly than the approach advocated by the auto
industry.45 Acknowledging that Tier 2 vehicles will benefit from sulfur control, API and NPRA
advocate a regional fuels program, under which all gasoline sold in ozone nonattainment areas
and the 22 OTAG states would be limited to 150 ppm sulfur, on average, during the summer
ozone season.  The gasoline sold in the rest of the country would be limited to 300 ppm, on
average, during the same period.  The industry recommends that this program begin January 1,
2004, or when the Tier 2 standards take effect, whichever is sooner. These recommendations are
based on claims of significant NOx emissions reductions and a cost to the industry of about $3
billion, equivalent to a production cost increase of about two cents per gallon of gasoline sold in
the recommended covered areas.

       Tosco,  an independent refiner, has recommended that all gasoline produced in the U.S.
outside of California meet a per gallon cap of 80 ppm sulfur.
C.     Catalyst Manufacturers

       Catalyst manufacturers have expressed views which support many of the conclusions
reached by the automotive manufacturers. For example, the Manufacturers of Emission Controls
Association (MECA) has recommended setting a near-term cap on sulfur in the 180-220 ppm
range and a longer term cap of 80 ppm.4* MECA cited several factors which influence the impact
sulfur has on catalyst performance and the degree to which the impact is reversible, and
suggested that catalysts designed for LEVs may see a greater impact and less reversibility than
earlier models. An international association of European catalyst manufacturers and designers
has also concluded that sulfur's impact on catalyst conversion efficiency increases at very low
       44 Gandhi, Haren S., "Effect of Fuel Sulfur on Automotive Three Way Catalysts," presentation to the CRC
Auto/Oil Symposium: The Search for Sulfur-Tolerant Emission Control Systems, Dearborn, Ml, September 11,
1997.
       45 "AP1/NPRA Sulfur Recommendations," March 20,1998.

       46 Manufacturers of Emission Controls Association, "The Impact of Gasoline Fuel Sulfur on Catalytic
Emission Control Systems,"  October 1997.

EPA Staff Paper. Gasoline Sulfur                 47                              May 1,1998

-------
emission levels.47  This organization claims that the development of a sulfur resistant catalyst
would require a trade-off between catalyst performance and removal of other pollutants.
Furthermore, they state that sulfur has an immediate effect with just one tankful, although most
of the catalyst performance will be restored when low sulfur fuel is used. They also raise
questions about the long term durability of emission control systems operated on high sulfur fuel.
D.     State and Local Air Quality Managers

       Recently, the State and Territorial Air Pollution program Administrators (STAPPA) and
the Association of Local Air Pollution Control Officials (ALAPCO) jointly resolved that EPA
should adopt a two-phase national cap on sulfur in gasoline, specifying levels of approximately
200 ppm by 2001 and no higher than 80 ppm by 2005,48 This resolution was based on a number
of findings of the organizations, including the belief that reducing sulfur in gasoline can decrease
emissions from vehicles, the fact that EPA recently promulgated new ambient air quality
standards for ozone which will require additional emissions reductions, and their desire that
states and localities get maximum emission reductions from EPA's motor vehicle control
programs.
E.     OTAG recommendation

       OTAG, which is comprised of representatives from each of the 37 states in the Eastern
half of the U.S., was formed to identify options for controlling the transport of ozone and ozone
precursors. In addition to supporting the use of Federal reformulated gasoline, OTAG
recommended that EPA adopt an "appropriate" sulfur standard to "further reduce emissions and
assist the vehicle technology/fuel system to achieve maximum long term performance."49
       47 Searles, R.A., Automobile Emissions Control by Catalyst (AECQ, "The Effect of Sulfur on Durability
and Performance of Catalyst Based Emission Controls," 18 June 1997.

       48 STAPPA/ALAPCO RESOLUTION ON A NATIONAL GASOLINE SULFUR CAP, Adopted October
28,1997.

       49 Ozone Transport Assessment Group, "Recommendation: Gasoline," Approved by the Policy Group,
May 13,1997.

EPA Staff Paper:  Gasoline Sulfur                 48                             May 1,1998

-------
VII.   Conclusions

       Based on the data and information obtained to date, EPA staff believe that some level of
gasoline sulfur control is appropriate. While it is probable that vehicle emission control
technology will continue to develop and that catalysts may be able to be designed to be less
sensitive to sulfur, complete sulfur tolerance appears highly unlikely. The ability to design
emission control systems to achieve complete reversibility of the sulfur effect appears possible.
However, the need for manufacturers to meet stringent off-cycle emission standards beginning in
2001 will increase the difficulty of both of these tasks, particularly that of achieving easy
reversibility under normal driving.  Therefore, it appears appropriate to evaluate various levels of
sulfur reduction, applicable in both regional areas and across the entire nation, while the Agency
is developing Tier 2 standards for 2004 and later LDVs and LDTs.

       The appropriate sulfur level for commercial gasoline remains to be determined. EPA is
continuing to conduct refinery modeling in order to better estimate the cost of sulfur control,
including the use of advanced sulfur removal techniques. EPA will also be analyzing the
detailed emission data contained in the two industry LEV-sulfur test program in order to develop
improved estimates of the sulfur sensitivity and reversibility of Tier 2 vehicles.  When
determining the appropriate sulfur level for commercial gasoline, EPA will also address the
current mismatch of sulfur between commercial gasoline and EPA test fuel.

       To identify the appropriate sulfur control strategy and Tier 2 standards, EPA will continue
to work closely with interested parties, gathering their input along the way. A public workshop
will be held in May, 1998, to discuss the  technical issues raised in this Staff Paper. Throughout
the coming months, EPA will continue to interact with individual companies, trade
organizations, the states, and environmental organizations to determine the best approach to
dealing with this issue.  Staff are recommending that the Agency develop proposed gasoline
sulfur standards to be implemented coincidentally with the implementation of the planned
proposed Tier 2 emission standards for LDVs and LDTs.
 EPA Staff Paper. Gasoline Sulfur                 49                             May 1,1998

-------
              EPA Staff Paper on Gasoline Sulfur
                           Appendices
                  (Transmitted as a Separate File, app-fin.wpd)
EPA Staff Paper: Gasoline Sulfur             50                       May 1,1998

-------
                                     Appendix A

                   Additional Vehicle Technology Information
       This appendix is intended to supplement the information provided in Chapter n of the
EPA Staff Paper on Gasoline Sulfur Issues. It presents more detailed, technical information on
how sulfur affects catalysts and the potential impact of sulfur on advanced emission control
technologies.

1.     Effects of Sulfur on Three-way Catalyst Efficiency

       In spark ignition engines, combustion converts sulfur present in the gasoline primarily to
sulfur dioxide (SO2). The nature of the interaction of sulfur with three-way catalysts, and the
tenacity of this interaction, depends on a number of factors, including:

•      Precious metal, support, and washcoat composition
•      Distribution of precious metals and metal  oxides
•      Catalyst surface area, cell structure, and washcoat layering
•      Exhaust oxygen content
•      Exhaust temperature

       Increasing the sulfur content of gasoline has an almost immediate impact upon the
performance of a three-way catalytic converter. SO2 in the exhaust can react with the support
materials, metal oxides within the washcoat, and the precious metals of the catalyst.

a.     Precious Metal Sulfur Tolerance and Reversibility

       SO2 is adsorbed and dissociates on the precious metal surfaces, forming adsorbed oxygen
and adsorbed sulfide. Adsorbed sulfide effectively blocks pollutant adsorption to precious metal
surfaces at the surface site it is adsorbed to, and at nearby sites due to the electrical charge it
imparts. The effect appears to be more severe for palladium (Pd) and palladium-rhodium (Pd-
Rh) catalysts than for other precious metal combinations, such as those containing platinum (Pt).
In general, sulfur sensitivity of the precious metal catalysts can be described, in order of
decreasing sensitivity, as:

                                 Pd>Pd-Rh>Pt-Rh»Rh

       Catalysts containing Pd have been an important component in the development of LEV
and ULEV vehicles for use in California. Pd catalysts have demonstrated improved low
temperature oxidation of hydrocarbons  when compared to Pt-Rh catalysts. Pd-containing
catalysts have also demonstrated improved resistance to high-temperature sintering and thermal
aging, allowing their use in close-coupled applications and further reducing cold-start

 EPA Staff Paper: Gasoline Sulfur - Appendices      A-l                             May 1,  1998

-------
hydrocarbon emissions. Laboratory testing of a Pd three-way catalyst demonstrated a substantial
reduction in performance after 4 hours of operation upon switching from the equivalent of a 15
ppm S fuel to a 450 ppm S fuel.1

       Adsorbed sulfide can be oxidized and removed from precious metal surfaces at elevated
temperatures (>600°C). It is also possible under such oxidizing conditions that some of the
released SO2 would react with cerium oxide within the washcoat to form cerium oxide and
degrade the tolerance of the catalyst to perturbations in air-to-fuel ratio (see below). Sulfide can
also be removed from precious  metal surfaces under reducing conditions at even higher
temperatures (>700°C). Reversal of sulfur inhibition of palladium-containing catalysts appears
to be more difficult and slower  than for Pt-Rh catalysts.  A study of a Pd-only three-way catalyst
required a minimum of 10 minutes at temperatures exceeding 650 °C for complete reversibility to
occur under laboratory conditions.1  Another laboratory study2 showed that a Pt/Pd/Rh tri-metal
pelleted catalyst required a rich air-to-fuel ratio and a catalyst temperature of 700°C for full
sulfur reversibility.  The study also demonstrated that catalysts containing higher cerium oxide
contents were more susceptible to sulfur irreversibility. A similar study with a Pd-only catalyst,
found that sulfur effects were reversible at 700 °C treatment under oxidizing (lean) and then
reducing (rich) air-to-fuel ratio  mixtures.3 Such cycling of the air-to-fuel ratio has typically been
reduced in LEV-technology vehicles in order to improve simultaneous removal of CO,
hydrocarbons, and NOx by the three-way catalyst.  Substantial rich excursions at high
temperature conditions for purposes of sulfur reversibility could adversely affect hydrocarbon
and CO emissions levels under  the revised (supplemental) federal test procedure (SFTP).

       One possible explanation for the decrease in reversibility found with Pd and Pd-Rh three-
way catalysts is that there is evidence that sulfur migrates from the surface of the Pd into the
crystal lattice of the metal. Under such circumstances, the Pd would act as a sulfur reservoir. As
sulfur is removed from the surface, sulfur would eventually migrate back to the surface to re-
establish equilibrium, again blocking the precious metal surface and electrically interfering with
nearby surfaces.

b.     Effects of Sulfur on the Management of Oxygen Within  the Catalyst

       Cerium oxide within the catalyst washcoat can be used to regulate the oxygen content of
the exhaust:
       'Beck, D. D., Spanners, J. W., "Impact of Sulfur on the Performance of Vehicle-Aged Palladium
Monoliths," Applied Catalysis B, Vol. 6, p. 185-200,1995.

       2Beck, D. D., Krueger, M. H., Monroe, D. R. "The Impact of Sulfur on Three-way Catalysts: Storage and
Removal," SAE Technical Paper Series, No. 910844,1991.

       3Beck, D. D., Sommers, J. W., "Impact of Sulfur on Three-Way Catalysts: Comparison of Pd and Pt-Rh
Commercially Produced Monoliths," 3rd International Congress on Catalysis and Automotive Pollution Control,
Brussels, Belgium, April 20-22,1994.

 EPA Staff Paper: Gasoline Sulfur - Appendices      A-2                              May 1,1998

-------
                             oxidizing agent + C^Oj => Ce02
                             reducing agent + Ce02 =>
       Oxygen regulation within the catalyst by cerium oxide under lean and rich air-to- fuel ratio
conditions reduces the sensitivity of three-way catalysts to air-to-fuel ratio fluctuations. Catalytic
reactions which are most sensitive to air-to-fuel ratio fluctuations near the stoichiometric air-to-
fuel ratio are thus strongly promoted by cerium oxide forcing the ratio nearer to stoichiometric
along the catalyst bed (figures Al, A2). Cerium oxide also promotes reactions that directly
oxidize CO under rich conditions and (to a lesser extent) reduce NO under lean conditions
(figure A3).

       Reactions of  metal oxides,  such as cerium oxide, with SO2 under lean conditions to form
sulfates, such as Ce2O2(SO4), impedes catalyst function primarily by reducing the ability of the
catalyst to manage the oxygen content of the exhaust. Reversal of the sulfation of cerium oxide
can occur either through direct thermal decomposition at high temperature (>700 °C):
                        Ce2O2(SO4) ==> CejOj + SO3 (==> SO2+

or by reduction at moderate temperatures (> 500 °C), which results in emission of hydrogen
sulfide (H2S)4'5, known for its "rotten egg" odor.
       4Bazin, P., et al., "Influence of Platinum on Cerium Oxide Sulfation," Applied Catalysis B, Vol. 13, p. 265-
274,1997.
       3Waqif, M., et al., "Study of Cerium oxide Sulfation," Applied Catalysis B, Vol. 11, p. 193-205,1997.

EPA Staff Paper:  Gasoline Sulfur - Appendices      A-3                            May 1, 1998

-------
                       /\
A
   Figure Al: Exhaust oxygen content at the catalyst inlet (top) and after regulation by
                      cerium oxide over the catalyst bed (bottom).
EPA Staff Paper: Gasoline Sulfur - Appendices     A-4
                 May 1, 1998

-------
                 Lean
                 t"
  Oa » CejOj, NOx will break
 through the converter
               A/F
                                           If CO»CeOa. CO will break through
                                           the converter
                 Rich
  Figure A2: Air-to-fuel ratio perturbations that overwhelm cerium oxide capabilities.
                 Cerium Oxide Chemistry In Exhaust
     LEAN:
O2 + 2Ce2O3 ==> 4CeO2  (fast)
2NO + Ce2O3 = = > 2CeO2 +  N2 (slow)
    A/F
                                    A
                          CeO
                                             Ce2O3
     RICH:
CO + 2Ce02 ==> Ce203 + C02  (fast)
H2 +2CeO2 ==> Ce203 + H2O   (fast)
  Figure A3: A graphical representation of cerium oxide reactions that manage exhaust

             oxygen content and directly oxidize CO and reduce NO.
EPA Staff Paper: Gasoline Sulfur - Appendices     A-5
                                         May 1.1998

-------
2.     Reversibility of Sulfur Inhibition of Three-way Catalysts

       The return of catalyst efficiency upon lowering the sulfur content of the gasoline, or
sulfur reversibility, may be possible depending on engine operating conditions and the
composition and construction of the catalyst.

       Although a significant amount of information exists in the literature on sulfur effects and
sulfur reversibility using aged three-way catalysts under controlled laboratory conditions, the
laboratory studies were typically conducted with pre-mixed gases cycled between slightly
reducing and slightly oxidizing. Considering the degree of variability of test conditions and the
extremely transient nature of actual vehicle operation, observations and conclusions drawn from
data generated in closely controlled laboratory studies may not necessarily apply to actual in-use
vehicle performance. The few vehicle studies that have been conducted have focused primarily
on the effects of fuel sulfur content on emissions, not on sulfur reversibility.

       The vehicle test programs discussed above in Section C of Chapter n utilized special
procedures when vehicles switched from one fuel to another to ensure that no sulfur-related
effect was carried over from the previous test fuel.  These procedures utilized vehicle operation
which is not generally representative of typical in-use driving and/or required fuel enrichment
beyond the amount available using standard calibrations of the engine management systems.
Thus, these programs provided little insight into the reversibility of the sulfur impact.

       Thoss,  et al.6 found CO and NOx only partially recovered with standard vehicle operation
using low-sulfur fuel after high sulfur aging and evaluation. A subsequent treatment at 700 °C
with low sulfur fuel lead to some further, but incomplete, recovery. Hydrocarbon emission
results were inconclusive due to prolonged rich operation during cold-start. Haren S. Gandhi of
Ford Motor Company stated at the CRC Sulfur Symposium that a prototype ULEV Lincoln
Towncar exposed to 928 ppm sulfur did not experience full recovery until it was operated under
both high temperature and oxidizing (lean) or reducing (rich) conditions.

       As discussed in Chapter n, limited test data on sulfur reversibility has been provided by
Chrysler, Ford, and API.  The Ford data indicated sulfur was irreversible for a prototype ULEV
Escort without a special "purge cycle" that introduced high temperature and extremely rich air-
to-fuel ratio operation. The LEV Dodge Neon had the same results for hydrocarbons, but sulfur
inhibition was reversible  for CO and NOx without any special purge cycle.

       A TLEV Mazda ProtegS (with advanced catalyst) demonstrated almost complete sulfur
recovery for HC, CO, and NOx also without any special purge cycle. The catalyst temperatures
experienced during the FTP, the air-to-fuel ratio, or, in the case of the Protege\ any specifics
about the catalyst composition were not provided with the preliminary release of this data, but
       Thoss, J. E., Rieck, J. S., Bennet, C. J., "The Impact of Fuel Sulfur Level on FTP Emissions - Effect of
 PGM Catalyst Type," SAE Technical Paper Series, No. 970737,1997.

 EPA Staff Paper:  Gasoline Sulfur - Appendices      A-6                             May 1,1998

-------
will eventually be provided. Once this information has been obtained, it may be possible to
determine why the Escort is only reversible when operated over a rather extreme purge cycle,
while the Proteg6 appears reversible over the FTP.  It is possible that the Protege' experiences
catalyst temperatures during the FTP cycle in excess of 650 °C and also has sufficient amounts of
rich and/or lean operation, while the Escort may have lower catalyst temperatures and maintain
very tight air-to-fuel ratio at nearly stoichiometric conditions. It may also be possible that the
Escort catalyst uses a catalyst design that is less sulfur tolerant and/or that is more difficult to
reverse the effects of sulfur inhibition.

       Finally, both the laboratory and the vehicle studies are typically lacking in information as
to whether high catalyst temperature and rich and/or lean operation need to be continuous or
cumulative.  For example, how long does the catalyst have to be exposed to temperatures of 700
°C or greater? Can several seconds of these temperatures spread throughout the course of the
FTP cycle be sufficient to reverse sulfur inhibition? Or is it necessary to have some length of
continuous operation? What conditions will be necessary in actual vehicle usage? Catalyst
temperature  and air-to-fuel ratio are some of the parameters being measured in the ongoing CRC
reversibility study. This information should help illuminate the conditions under which sulfur
inhibition of advanced three-way catalyst designs is reversible.

       A number of design arid operating factors appear to directly or indirectly influence sulfur
tolerance and/or sulfur reversibility. In addition to the type of precious metals, higher precious
metal loading appears to increase tolerance to sulfur. Higher temperature operation, such as in
close coupled applications, appears beneficial for both sulfur tolerance and reversibility of
cerium oxide and precious metal components of the catalyst, although such operation can
increase thermal aging of the catalyst. Both rich and lean excursions in air-to-fuel ratio may be
beneficial from both a standpoint of sulfur tolerance and reversibility.  Some vehicles using fuel
shut-off during deceleration may provide the necessary oxidizing, high temperature conditions to
remove sulfur from precious metal surfaces. Rich operation appears to aid in reducing sulfation
of cerium oxide, and when combined with high temperatures, may allow simultaneous removal
of sulfur from precious metal surfaces. However, rich excursions may be an impediment to
passing SFTP requirements. Conversely, vehicles which use a higher degree of control of the air-
to-fuel ratio  through the engine management system require less cerium oxide in their catalyst
formulations, and thus may be less prone to the effects  of cerium oxide sulfation on management
of the oxygen content of the exhaust.

3.     Effects of Sulfur  on Lean NOx Catalysts

       Recently developed gasoline-direct-injection (GDI) engines offer fuel consumption
benefits  approaching those of turbocharged, direct-injection diesel engines, with PM emissions
nearer to those of conventional spark ignition gasoline engines. Although the very lean (20:1 to
40:1 air-to-fuel) operation made possible by GDI reduces engine out NOx emissions,
conventional three-way catalyst designs are ineffective for further NOx control under these
conditions.  More conventional spark ignition engines use closed loop control systems to
maintain a fuel-to-air ratio near stoichiometric to allow operation within the narrow window of

EPA Staff Paper:  Gasoline Sulfur - Appendices      A-7                             May 1,1998

-------
operating conditions necessary for efficient removal of CO, hydrocarbons, and NOx with
conventional three-way catalyst designs. Under conditions where the partial pressure of 02 in the
exhaust significantly exceeds the partial pressure of NO, as with GDI engines during much of
their operation, the rate of NO reduction with conventional three-way catalysts decreases
substantially. Thus, for NOx control to levels for vehicles equipped with lean GDI engines to be
at least equivalent to those of stoichiometric/three-way catalyst-equipped vehicles, the use of NO
adsorption or "NOx storage traps" with cycling of the air-to-fuel ratio, and/or the use of selective
catalytic reduction (SCR) will be necessary.  These types of lean-operation catalyst systems are
collectively referred to as "lean-NOx" catalysts.

       Effective SCR reductants include urea and fuel hydrocarbons. SCR systems typically use
a base metal catalyst that is capable of reducing NO in the presence of a suitable reductant. Use
of a hydrocarbon reductant eliminates the need for on-board storage of a separate reductant, and
is typically accomplished by direct injection of fuel during the expansion stroke. SCR lean-NOx
catalyst strategies appear to be relatively insensitive to gasoline sulfur content up to at least 300
ppm of sulfur, but have lower NOx removal efficiency (<60%) and are effective over a narrower
range of exhaust temperatures than lean-NOx catalysts using NOx storage7.  Based on
information provided by manufacturers, it does not appear feasible for GDI engines to achieve
NLEV LDV NOx emissions levels using SCR-only lean-NOx strategies.

       A NOx trap can be used either upstream of a catalyst substrate,  or can be incorporated in
the washcoat in a portion of the catalytic converter. The trap material is typically an alkaline or
alkaline earth material, with barium and calcium being two commonly  used materials. First, NO
is oxidized to NO2 over a platinum catalyst during lean operation. The NO2 adsorbs to the NOx
trap as, for example, barium or calcium nitrate. Under reducing (rich) conditions, the nitrate is
reduced (either directly or indirectly) and N2 is released. Only slightly  rich (0.9540:1 air-to-fuel in some cases) to rich operation.

       NOx trapping and reduction has a much higher NOx removal efficiency (>90% in some
instances) and is effective over a broader range of exhaust temperatures. It also appears to be
significantly less tolerant of sulfur. The trap materials adsorb SO2 continuously to form barium
or calcium sulfates, and eventually depleting the barium or calcium available for nitrate
       'Mitsubishi Motors, "4G93 GDI Engine: Technical Information," Available from Mitsubishi Motors, or
 on the World-Wide-Web: , 1998.

 EPA Staff Paper: Gasoline Sulfur - Appendices      A-8                             May 1,1998

-------
formation and severely degrading the performance of the catalyst.  Tests by Mitsubishi of a
prototype lean-NOx catalyst using a NOx storage trap showed significant degradation after 8000
km of aging using gasoline with 100 ppm of sulfur.  The catalyst originally performed at greater
than 90% removal efficiency for NOx, but degraded to 40 to 60% NOx removal efficiency after
aging on the higher sulfur gasoline. The effect will occur at any level of fuel sulfur in the
gasoline, and is essentially an integrative effect, occurring more rapidly at higher fuel sulfur
levels. Purging of the sulfate is similar in principle to nitrate purging, requiring slightly rich
operation, but the reaction rates are much slower and would require longer periods of rich
operation. Automotive and catalyst manufacturers have stated that using a  100 to 300 ppm sulfur
fuel would necessitate enough rich operation for sulfate purging that the fuel efficiency benefits
of using a lean GDI combustion strategy would be essentially negated. Fuel efficiency using a
20 to 50 ppm sulfur fuel, including rich operation for nitrate and sulfate purging, approaches that
of a direct injection turbo-charged diesel engine at near NLEV NOx levels and with PM levels
comparable to those conventional spark-ignition, stoichiometric engines. Considering the
coupling of fuel efficiency and NOx emissions using the NOx trap technology with GDI, it
appears that GDI equipped vehicles will only be feasible at gasoline sulfur levels in the range of
those in Japan and California (20 - 40 ppm). It does not appear likely that such vehicles could
meet even Tier 1 NOx levels if required to rely on SCR for NOx reduction due to higher fuel
sulfur levels.

4.     Effects of Sulfur on Gasoline Fuel Cells

       Fuel cells are electrochemical devices that can be used to generate electricity. They have
potential for automotive applications as a relatively high efficiency (>50%) source of electrical
power for electric cars. Fuel cells offer the possibility of weight and driving range advantages
over more conventional battery-equipped electric cars.  Thin-membrane, proton-exchange fuel
cell technologies can be provided with the necessary hydrogen fuel from a number of sources,
including reforming of methane, methanol, and gasoline. The Office of Advanced Automotive
Technologies within the U.S. Department of Energy is currently participating in the development
of a process for reforming gasoline to provide hydrogen for automotive fuel cells. They are
currently testing a system using California low sulfur gasoline (approximately 40 ppm sulfur).
The system uses a zinc oxide bed for trapping sulfur to prevent contamination of the fuel cell.
The technology is feasible at higher gasoline sulfur levels, but would require a disposable zinc
oxide bed rather than the current  system, which currently is approximately sized to last the useful
life of the vehicle.  The effect of the use of disposable traps on operational costs is not known.
 EPA Staff Paper: Gasoline Sulfur - Appendices      A-9                             May 1,1998

-------
                                    Appendix B

                            Fuel Sulfur Control Costs
       The purpose of this appendix is to provide the reader with additional information
regarding the analyses performed by EPA, working with DOE, to estimate the costs of gasoline
sulfur control.
1.    Refinery Modeling Plan and Modeling Premises

      EPA and DOE developed a modeling plan for studying the cost of reducing gasoline
sulfur levels. Table B-l lists the modelling runs, to be performed by Oak Ridge National
Laboratory under contract to DOE, that were planned.

      While the refinery modeling plan outlined in Table B-l describes the refinery modeling
cases to be run, certain premises must be established as well to set up the refinery model for
developing the most representative costs.  These premises include: the price of refinery raw
materials and products, refinery capacities, and the volumetric demand and production of
petroleum products. The approach for establishing these premises, and some of the premises that
were derived, are summarized below.

      •      Only summer cases would be run and the investments made would be amortized
             over the entire year.

      •      The potential sulfur reduction program would be implemented in the year 2004
             with the Phase n RFG program capital investments presumed to have already
             been made.

      •      The current production shares of RFG in PADDs 1  and 3 would apply in the year
             2004; these are taken from the Motor Gasoline Watch (DOE, 1997).  The result is
             that RFG represents 62 percent of gasoline production in PADD 1, and 18 percent
             in PADD 3.

      •      A summertime base case for year 2004 (including Phase n RFG) would be
             established using 1997 refinery capacity and/or construction reported in the
             Petroleum Supply Annual, the Oil and Gas Journal, National Petroleum Council
             1993 petroleum industry report and statistics by NPRA.  The refinery model
             would be permitted to add process capacity for the base case, and this investment
             would be sunk for the study cases which follow.
 EPA Staff Paper. Gasoline Sulfur - Appendices     A-10                           May 1,1998

-------
Table B-l. List of EPA's Planned Refinery Modeling Case Runs
Model Version1
LSG Share"
LSG Sulfur Content"
Status
Case Runs
PADD 3 RF
PADD3R
PADD 1 RF
PADD1R
PADD 3 RF
PADD 3 RF
PADD 3 RF
PADD 1 RF
PADD 2 RF
PADD 2 RF
PADD 2 R
PADD 4 & PADD
50CRF
100%
100%
100%
100%
55%
30%
55%
100%
100%
70%
100%
100%
Basecase, 150, 100 & 40 ppm
Basecase & 100 ppm
Basecase, 150, 100 & 40 ppm
Basecase & 100 ppm
100 & 40 ppm
40 ppm
40 ppm in N/A areas, and 100 ppm
in rest of the 22 States
40 ppm in N/A areas, and 100 ppm
in rest of the 22 States
Basecase & 100 ppm
100 ppm
Basecase & 100 ppm
Basecase, 150, 100 & 40 ppm
Completed
Completed
Completed
Completed
Completed
Completed
.Completed
Completed




Sensitivity Cases
PADD 3 RF
PADD 3 RF
PADD 3 RF
PADD 3 RF
100%, Investment
Adverse0
100%, Grade Split"
100%, Enhancedc
55%, Most Capable
Refineries'
100 ppm
100 ppm
100 ppm
Basecase & 100 ppm •




a. RF represents version of refinery model without ratio controls; R is the version with such controls, as explained
in the text.  PADD=Petroleum Administration District for Defense; PADD 5 OC = PADD 5, outside of California.
b. LSG means low sulfur gasoline.
c. Approximates an investment adverse climate by modeling with a higher hurdle rate of return for capital
investments.
d. Evaluates the cost of sulfur reduction with both premium and regular grades of gasoline represented in the
refinery model.
e. Provides opportunity for the refinery model to select SynSat and CDTECH sulfur removal technologies.
f. Estimates the sulfur reduction cost assuming only low cost sulfur reduction refineries participate.
 EPA Staff Paper: Gasoline Sulfur - Appendices      A-11
May 1,1998

-------
       •      Crude oil costs would be based on Petroleum Marketing Monthly (DOE, 1995 and
             later), and on the National Petroleum Council (NPC, 1993).  Historical price
             differentials and price ratios would be taken from NPC.

       •      Refinery inputs of raw materials would be estimated using the Petroleum Supply
             Annual 1996 (DOE, 1997), with extrapolations to the year 2004 per the Annual
             Energy Outlook (DOE, 1997). The crude mix would be based on the Petroleum
             Supply Annual and NPC estimates. Available swing crudes1 would include Saudi
             Arabian Light, Saudi Arabian Heavy, United Kingdom Brent and United
             Kingdom North Sea crudes.

       •      Refinery product output would be based on Petroleum Supply Annual (DOE,
             1996), extrapolated to the year 2004 with growth rates reported in the Annual
             Energy Outlook (DOE, 1997).

       •      Refinery investment in the refinery model would be based on a 15 percent return
             on investment. A 20 percent expansion of existing capacity would also be
             allowed at typical investment costs.  Other investment would be subject to
             exponential economies of scale.

       •      Methyl tertiary butyl ether (MTBE) would be assigned a purchase price of 84
             cents per gallon.  This value is an average of a range of costs established by a
             recent study of the cost of variations of RFG by OTAG, conducted with the
             participation of states, and the oil and automobile industries.

       •      All prices and costs would be expressed in 1997 dollars.

       The next section describes the results of the refinery modeling performed to date using
these premises.
2.     Summary of the Cost Calculation Methodology and the Cost of Reducing Gasoline
       Sulfur

       In the analyses performed to date by EPA/DOE, two different versions of the ORNL-
RYM refinery model were run to develop a range in costs for the program being investigated.
The two versions of the refinery model are referred to as the ratio and ratio-free versions of the
ORNL refinery model, because of the presence or absence of "ratio controls." Ratio controls
constrain the model from moving certain streams between refining processes based on the
recommendation of industry experts who believe that they are required to ensure  that the model
       1 Swing crudes are those crude oils which the refinery model can choose from to meet the incremental
 increases in energy demand caused by reducing gasoline sulfur.

 EPA Staff Paper: Gasoline Sulfur - Appendices     A-12                            May 1,1998

-------
accurately represents real refinery operations.  Ratio controls tend to increase the projected costs.
These two versions of the refinery model are explained further in a recent Oak Ridge National
Laboratory's draft technical report which summarized an analysis of the cost of the Phase n RFG
NOx standard.2

       The cost estimates developed in the first phase of work performed by EPA/DOE assumed
a program covering all gasoline produced in PADDs 1 and 3, which represents 75%  of national
gasoline production, California excluded. PADDs 1 and 3 provide a good starting point for
estimating costs because of the large gasoline volume produced in these two PADDs, because the
Northeast receives most of its gasoline from these two PADDs, and because of the range of
technologies used in the refineries in these PADDs. While an estimate of the cost for sulfur
control in either the 22- or 37-state OTAG scenarios could be made from the PADD 1 and 3
costs, a more detailed estimate would include an estimate of PADD 2 costs to supplement the
costs obtained for PADDs 1  and 3.  EPA and DOE plan to develop cost estimates for PADD 2 in
the future.

       Likewise, the cost of a nationwide sulfur control program could be estimated with costs
from PADDs 1, 2, and 3, as these three PADDs produce the great majority of gasoline produced
outside of California. EPA intends to perform additional refinery modeling runs to estimate the
costs to refineries in PADD 4 (the Rocky Mountain area) and in PADD 5 outside of California
("PADD 50C," which is the West coast, Hawaii and Alaska, without California). The average
cost of sulfur reduction in these two PADDs is expected to be somewhat higher than the average
cost in PADDs 1, 2, and 3  because of the small size of the refineries, and because the refineries
are generally less complex  which makes sulfur reduction more difficult and costly. However,
because the gasoline production volume is much lower for PADDs 4 and 5OC, including the cost
of sulfur reduction in these PADDs with the cost of sulfur reduction in PADDs 1,2 and 3 would
likely not significantly change the overall, national average cost estimate.

       The series of calculations used to determine cost are summarized below in Tables B-2 to
B-15. The cost in cents per gallon is calculated by dividing the change in objective function,
which can be simply described as the change in net income of the group of refineries being
modeled, by the volume of the gasoline pool experiencing the sulfur reduction. The next step in
the procedure is to adjust the portion of the cost related to new capital investment to represent a
range of rates of return on investment (ROI).3  The refinery model is run using a 15 percent ROI,
to represent the decision-making process of refiners.  The projected costs of sulfur control are
       1 Madder, G.R., Oak Ridge National Laboratory, "Analysis of Costs and Refinery Impacts of Further
Control of Gasoline Emissions - Draft, Incomplete Report," Draft 2, June 1,1996.

       3 The ROI is part of a mathematical relationship to determine the capital amortization factor, or the cost of
capital, on a yearly basis. If the ROI is 15 percent, 10 percent or 8 percent, then the capital amortization factor
would be (based on a depreciation and economic life of 13 years and a federal income tax rate of 34 percent) 0.232,
0.174 and 0.152, respectively.  This means that with a 10 percent ROI, the yearly capital cost for a $1 million
investment would be $174,000 per year ($1MM x 0.174). Based on an 8 percent ROI, the capital investment cost
would be $152,000 per year.

EPA Staff Paper: Gasoline Sulfur - Appendices     A-13                            May 1,1998

-------
determined using 8 and 10 percent ROIs, which more closely represent the actual return on
capital invested.

       To make these adjustments, the portion of the total cost of sulfur control which is related
to new capital equipment must be determined. This is accomplished by simulating the
accounting of capital costs performed by the model. The total capital cost investment (in dollars)
is multiplied by the annual capital recovery factor for a ROI of 15 percent, or 0.242. This annual
cost is then divided by the volume of low sulfur gasoline produced to estimate the per-gallon cost
of the new capital equipment as determined by the refinery model. This process is then repeated
for ROIs of 8 and 10 percent using annual capital cost recovery factors of 0.152 and 0.171,
respectively.  The difference between the per-gallon cost of new capital equipment with ROIs of
8 and 10 percent and that with an ROI of 15 percent is then subtracted from the total cost per
gallon as estimated by the refinery model.

       Other adjustments to the methodology were made to account for an apparent anomaly in
how octane requirements in the gasoline pool were maintained, and to account for the fuel
economy effect.  An examination of the ORNL-RYM data output revealed a trend of increasing
gasoline pool octane number (R+M)/2 in sulfur reduced gasoline pools compared to the base
case. Since desulfurization generally results in octane loss, and it is unlikely that in making up
octane refiners would exceed the required octane levels at an excess cost, these results were
judged to be erroneous. To correct for this error, it is estimated that the oil industry spends on
average about 4 cents per gallon to produce the 6 octane number spread between premium (93
(R+M)/2) and regular (87 (R+M)/2) grades.4  This cost rate per octane value difference was
applied to the change in octane value from any particular refinery modeling case relative to the
previous case, and the cost of sulfur reduction adjusted accordingly.  The estimated cost of sulfur
control must also be corrected for an expected reduction in fuel economy. The refinery model
generally projects that refiners will increase their use of oxygenates to meet a tighter sulfur
specification. Oxygenates contain less energy per gallon than non-oxygenated gasoline (81%
percent of gasoline heat content for MTBE and 65% percent for ethanol). It is assumed that fuel
economy is proportional to the energy content of gasoline. The decrease in total energy content
of the finished gasoline is estimated from the increased oxygenate content of the affected
gasoline pool relative to the previous refinery modeling case and multiplied by the retail cost of
gasoline minus taxes. This cost of decreased fuel economy is then added to the sulfur control
cost estimated by the refinery model.

       Concerning the handling of costs, one challenge which had to be overcome was to
determine to which gallons of gasoline the sulfur reduction costs would be allocated. This is
important because as the cost of the gasoline sulfur reduction is allocated to a smaller or larger
size of the gasoline pool, the per-gallon cost gasoline sulfur reduction is higher or lower,
respectively. When a sulfur specification is imposed over the entire pool, including reformulated
       4 U.S. DOE, "Table #32. Conventional Motor Gasoline Prices by Grade, Sales Type, PAD District and
 State"' Petroleum Marketing Annual 1995. Energy Information Administration.

 EPA Staff Paper: Gasoline Sulfur - Appendices     A-14                            May 1,1998

-------
and conventional gasoline, the initial sulfur reduction occurs in the conventional pool only as
RFG would already have lower sulfur levels as a result of the Phase n RFG NOx standard.
However, at some point, both pools experience sulfur reduction due to the sulfur standard. That
point is difficult to judge as it generally occurs somewhere in between refinery modeling case
runs.  For the costs presented below, the cost of sulfur reduction is allocated to the conventional
gasoline pool only down to 150 ppm, and is  allocated to both the conventional and reformulated
pool down to 100 and 40 ppm. While the sulfur reduction level at which both conventional and
reformulated pools experience sulfur reduction differs for both PADD 1 and 3 models (for
PADD 3  the convergence point is higher than 150 ppm, and for PADD 1 it is  lower than 150
ppm), assigning the convergence point at the 150 ppm sulfur reduction point simplifies the cost
calculations. For the refinery model runs with ratio controls, the gasoline pool size is determined
by a weighted fraction of the cost for sulfur reduction determined by the model without ratio
controls.  The weighting determines what fraction of the range of gasoline pool size, which
ranges from the conventional pool only to conventional and reformulated pool, that should be
used, and is based on the costs for sulfur  reduction from the basecase to 150 ppm relative to the
cost of sulfur reduction from 150 to 100 ppm, for the same PADD model.
 EPA Staff Paper. Gasoline Sulfur - Appendices     A-15                            May 1,1998

-------
Table B-2. Calculation and Adjustment of Cost - PADD 1 Refinery Model Without Ratio Controls
Change
in Sulfur
Level
(ppm)
450 to
150
150 to
100
100 to
40
Obj.
Funct
Change
($M/day)
275.5
245.5
1522.4
Volume of
Fuel
Affected
(Mbbl/day)
290.5
755.1
755.1
Unadjust.
cost
(c/gal)
2.26
0.78
4.80
Change in
Invest.
($MM)
262.1
110.4
1204
15% ROI
Annual
Amort.
(c/gal)
1.42
0.232
2.52
10%ROI
Annual
Amort.
(c/gal)
1.01
0.164
1.78
Adj. Cost
(c/gal)
1.84
0.711
4.06
8% ROI
Annual
Amort.
(c/gal)
0.90
0.15
1.58
Adj. Cost
(c/gal)
1.73
0.69
3.86
Table B-3. Fuel Economy Adjustment of Cost - PADD 1 Refinery Model without Ratio Controls
Change in Sulfur Level
450 - 150
150 - 100
100-40
Fuel Economy Adjustment
(c/gal)
0.49
-0.02
0.58
Fuel Economy Adjust Cost
10% ROI (c/gal)
2.34
0.68
4.56
Fuel Economy Adjust Cost
8% ROI (c/gal)
2.23
0.68
4.36
EPA Staff Paper:  Gasoline Sulfur - Appendices
A-16
May 1,1998

-------
Table B-4. Calculation and Adjustment of Cost - PADD 3 Refinery Model without Ratio Controls
Change in
Sulfur
Level
(ppm)
322 to 150
150 to 100
100 to 40
Obj. Funct
Change
($M/day)
1108
1678
5815
Volume of
Fuel
Affected
(Mbbl/day)
2957
3677
3678
Unadjust.
Cost
(c/gal)
0.892
1.09
3.77
Change in
Invest.
($MM)
671.6
1182
4845
15% ROI
Annual
Amort.
(c/gal)
0.359
0.507
2.08
10%ROI
Annual
Amort.
(c/gal)
0.253
0.358
1.47
Adj. Cost
(c/gal)
0.787
0.938
3.15
8% ROI
Annual
Amort.
(c/gal)
0.225
0.319
1.31
Adj. Cost
(c/gal)
0.759
0.898
2.99
Table B-5. Fuel Economy Adjustment of Cost
PADD 3 Refinery Model without Ratio Controls
Change in
Sulfur Level
322 - 150
150 - 100
100-40
Fuel Economy
Adjustment (c/gal)
0.33
0
0.01
Fuel Economy Adjust.
Cost, 10% ROI (c/gal)
1.07
0.83
3.14
Fuel Economy Adjust Cost
8% ROI (c/gal)
1.04
0.79
2.97
EPA Staff Paper:  Gasoline Sulfur - Appendices
A-17
May 1,1998

-------
Table B-6. Calculation and Adjustment of Cost - PADD 1 Refinery Model with Ratio Controls
Change in
Sulfur
Level
(ppm)
420 to 100
Obj Funct
Change
($M/day)
717
Volume of
Fuel
Affected
(Mbbl/day)
409
Unadjust.
Cost
(c/Gal)
465
Change in
Invest.
($MM)
690.5
15% ROI
Annual
Amort.
(c/gal)
2.67
10%ROI
Annual
Amort.
(c/gal)
1.89
Adj. Cost
(c/gal)
3.40
8% ROI
Annual
Amort.
(c/gal)
1.66
Adj. Cost
(c/gal)
3.17
Table B-7. Fuel Economy Adjustment of Cost
PADD 1 Refinery Model with Ratio Controls
Change in
Sulfur Level
420 - 100
Fuel Economy
Adjustment (c/gal)
0.34
Fuel Economy Adjust.
Cost, 10% ROI (c/gal)
3.73
Fuel Economy Adjust Cost
8% ROI (c/gal)
3.51
EPA Staff Paper Gasoline Sulfur - Appendices
A-18
May 1,1998

-------
Table B-8. Calculation and Adjustment of Cost - PADD 3 Refinery Model with Ratio Controls
Change in
Sulfur Level
(ppm)
450 to 100
Obj Funct
Change
($M/day)
4808
Volume of
Fuel
Affected
(Mbbl/day)
3405
Unadjust.
Cost
(c/gal)
3.41
Change in
Invest.
($MM)
3787
15% ROI
Annual
Amort.
(c/gal)
1.78
10%ROI
Annual
Amort.
(c/gal)
1.26
Adj. Cost
(c/gal)
2.88
8% ROI
Annual
Amort.
(c/gal)
1.11
Adj. Cost
(c/gal)
2.74
Table B-9. Fuel Economy Adjustment of Cost
PADD 3 Refinery Model with Ratio Controls
Change in
Sulfur Level
367 - 100
Fuel Economy
Adjustment (c/gal)
0.35
Fuel Economy Adjust.
Cost 10% ROI (c/gal)
3.18
Fuel Economy Adjust
Cost8% ROI (c/gal)
3.04
EPA Staff Paper Gasoline Sulfur - Appendices
A-19
May 1, 1998

-------
Table B-10. Volume-weighted PADD 1 and 3 Cost of Sulfur Reduction, Based on the Refinery
Model Without Ratio Controls, 10% ROI
Sulfur Reduction
to
150 ppm
150 to
100 ppm
100 to
40 ppm

Cost (c/gal)
Volume affected
(Mbbl/day)
Cost (c/gal)
Volume affected
(Mbbl/day)
Cost (c/gal)
Volume affected
(Mbbl/day)
PADD1
2.34
290.5
0.68
755.1
4.56
755.1
PADD 3
1.07
2957
0.829
3677
3.14
3678
Volume Weighted
PADDs 1 and 3
1.18
3248
0.80
4433
3.38
4433
Table B-l 1. Volume- weighted PADD 1 and 3 cost of Sulfur Reduction, Based on the Refinery
Model with Ratio Controls, 10% ROI
Sulfur Reduction
to
100 ppm

Cost (c/gal)
Volume affected
(Mbbl/day)
PADD1
3.73
409
PADD 3
3.19
3405
Volume Weighted
PADDs 1 and 3
3.19
3814
EPA Staff Paper. Gasoline Sulfur - Appendices      A-20
May 1,1998

-------
Table B-12. Volume-weighted PADD 1 and 3 Cost of Sulfur Reduction, Based on the Refinery
Model Without Ratio Controls, 8% ROI
Sulfur Reduction
to
150 ppm
150 to
100 ppm
100 to
40 ppm

Cost (c/gal)
Volume affected
(Mbbl/day)
Cost (c/gal)
Volume affected
(Mbbl/day)
Cost (c/gal)
Volume affected
(Mbbl/day)
PADD1
2.23
290.5
0.678
755.1
4.36
755.1
PADD 3
1.04
2957
0.789
3677
2.97
3678
Volume
Weighted
PADDs 1 and 3
1.15
3248
0.776
4433
3.21
4433
Table B-13. Volume-weighted PADD 1 and 3 cost of Sulfur Reduction, Based on the Refinery
Model With Ratio Controls, 8% ROI
Sulfur Reduction
to
100 ppm

Cost (c/gal)
Volume affected
(Mbbl/day)
PADD1
3.51
409
PADD 3
3.04
3405
Volume Weighted
PADDs 1 and 3
3.14
3814
Table B-14. Summary of Volume-Weighted, PADD 1 and 3
Sulfur Reduction Costs (cents/gallon)
Sulfur Level
10 Percent ROI
8 Percent ROI
Base to 150 ppm
1.2-1.9
1.1 -1.8
Base to 100 ppm
2.0 -3.2
1.9-3.0
Base to 40 ppm
5.4 -8.5
5.1 -8.0
EPA Staff Paper: Gasoline Sulfur - Appendices      A-21
May 1,1998

-------
3.     Estimating the Costs for Individual Refiners

       While most gasoline sulfur control cost estimates, including the preliminary estimates
presented above, represent costs to the average refiner, it is important to consider the range of
costs expected to be experienced by individual refiners. Several years ago, EPA and DOE
prepared a draft report which assessed the costs to the refining industry of various gasoline
quality changes, and in the process developed a cost differential between "sulfur challenged"
refineries and the refineries in the rest of PADD 3.5 The term "sulfur challenged" refinery was
defined based on three criteria: relatively large capacity (specifically, 100-350 thousand barrels
per day, which demonstrated that high costs and difficulties in complying with new requirements
are not limited to small refineries), no FCC hydrotreating capability, and 1990 average gasoline
sulfur levels above 450 ppm.  Based on this analysis, these challenged refineries in PADD 3 were
estimated to expend 1 cent per gallon just to bring their sulfur levels down to the PADD average
gasoline sulfur level, which was about 320 ppm.  From there, the challenged refineries would
incur 0.7 c/gal more than the rest of the PADD to control gasoline sulfur levels down to an
average of 150 ppm.  Thus, some refiners are expected to incur substantially higher costs  than
other refiners for gasoline sulfur reduction.

       Another study, performed by Wright-Killen, has also explored the potential impact on the
refining industry of a gasoline sulfur program.  This study estimated that a number of refineries
nationwide could be in danger of closing if a low sulfur (about 40 ppm) gasoline program were
pursued nationally.6 These refineries at risk produce about 10 percent of U.S. gasoline
production.  The reasons given for forecasting this financial risk is that these refineries currently
lack gasoline sulfur removal equipment, have difficulty in obtaining capital due to a relatively
poor financial standing,  and must compete with other refiners which are better prepared to meet
such requirements. EPA staff have reviewed the conclusions in this study and believe that this
estimate of potential refinery closures represents the high end of a likely range.  EPA's position is
based in part on the fact that some experts in the industry have concluded that various,
compelling reasons exist which keep refiners in business, even when faced with financial
hardship.7'8-9
       5Hadder, G.R., Oak Ridge National Laboratory, "Analysis of Costs and Refinery Impacts of Further
Control of Gasoline Emissions - Draft, Incomplete Report," Draft 2, June 1,1996.

       6 Hobbs, Horace, Ernst and Young-Wright Killen, Low Sulfur Gasoline Implementation Issues,
Presentation at the CRC Symposium on Conventional and Emerging Technologies for Sulfur Removal from
Gasoline, August 1996.

       7 Comments by Jeff Utley, Refinery Manager, Flying J. Inc., at a Hart-lRl conference entitled 'Telling it
like it is," summarized in Fuel Technology and Management, July/August, 1996.

       8 Stetzer, C. Martin, Price Waterhouse L.L.C., Redefining the Context of Refinery Pacesetter Performance,
1997 NPRA Annual Meeting, March 16 -18,1997.

       9 Wright-Killen, Clean Air Rules Could Wipe Out 1.5 million b/d in U.S. Refining Capacity, Oil and Gas
Journal, December 28,1992.

EPA Staff Paper: Gasoline Sulfur - Appendices     A-22                             May 1,1998

-------
                                     Appendix C

          Relationship Between Per-Gallon and Averaging Standards
       The purpose of this analysis is to estimate the average sulfur levels which would result
from various per-gallon caps on gasoline sulfur content. The refinery cost estimates available to
EPA are based on average costs to refiners in a given region of the country. Existing refinery
models are not capable of estimating the average costs to the industry of a cap. However, as
discussed in the main body of this paper, there are several factors that would likely lead to a cap
being a more effective means of sulfur control than a standard based on the average sulfur level.
The analysis presented here is an initial attempt to estimate the average sulfur levels which would
likely result under various cap standards so that the costs presented in this report (applicable to
average sulfur levels) could be associated with various cap standards.

       API recently suggested a possible methodology for relating caps and averages based on an
analysis of sulfur levels in batches of gasoline produced by U.S. refineries.1 API assumed that
the 90th percentile sulfur level from a refiner's batches of fuel would represent the cap
achievable by that refiner. API presumed that the 10 percent of the batches with the highest
sulfur levels could be reduced to below the 90 percentile level at an insignificant cost. The
average sulfur level of all of the batches produced by the refinery would represent the average
level that would result from meeting the per-gallon cap.

       Based on this premise, API estimated the average sulfur levels associated with two
different caps using data from the 1996 gasoline survey performed by API and NPRA.2 API
found that a cap on gasoline sulfur levels of 300 ppm for regular grade gasoline could result in an
average sulfur content of about 200 ppm.  Similarly, a cap of 200 ppm could result in an average
sulfur content of 145 ppm, although the estimates for individual refiners ranged from 80-180
ppm.

       After viewing the API analysis, EPA performed a similar analysis utilizing more data and
additional percentiles. Specifically, EPA used conventional gasoline compositional data
submitted in 1996 by the refining industry to meet the reporting requirements of the RFG  and
anti-dumping programs. In EPA's analysis, refiners with similar average gasoline sulfur levels
were first grouped together. The distribution of gasoline sulfur levels in the regular grade
gasoline batches of each group of refineries was then analyzed. (Only the regular grade batches
need to be evaluated since premium gasoline, which is already relatively low sulfur, is not
expected to be directly affected by a cap (unless if the per-gallon sulfur standard is very
       'Letter from Tim Hogan, API, to Charles Freed, U.S. EPA, October 10,1997.

       'American Petroleum Institute and National Petroleum Refiners Association, "Survey of Refining
 Operations and Product Quality," Final Report, July 1997.

 EPA Staff Paper: Gasoline Sulfur - Appendices     A-23                            May 1,1998

-------
stringent)). In addition to determining the 90th percentile of the batch gasoline sulfur levels as
suggested by API, the 75th, 80th, and 95th percentiles were determined as well.  The results of
this analysis are summarized in Table C-l below.

       Prior to reviewing the results of this analysis, a couple of limitations should be noted in
the potential applicability of this methodology. Refiners with the lower average sulfur levels are
likely producing these low sulfur levels by processing low-sulfur crude oils, not because of sulfur
reduction processing equipment implemented to meet a sulfur reduction standard. Thus, the
relationship between potential sulfur caps and averages at the lower sulfur levels could be very
different from that estimated here. Refiners currently have no incentive to minimize the
maximum sulfur levels of individual batches produced through out the year. Their actions could
be quite different under a cap. Also, the data analyzed here are not volume-weighted. Thus, if
gasoline sulfur level is somehow associated  with refinery size or if the batches produced by a
particular refinery vary in volume, this analysis could have a bias.

       There is also the issue of which (if any) of the percentiles actually represents the cap
associated with the average of the batches of fuel.  API recommended the 90th percentile.
However, API used the 90th percentile of the whole gasoline pool while EPA only evaluated
batches of regular grade gasoline. API's 90th percentile is roughly equivalent to the 85th
percentile of regular grade gasoline (i.e., halfway in between the 80th and 90th percentiles shown
in Table C-l.) (This equivalency presumes that none of the premium batches is above the 90th
percentile.) As mentioned above, API estimated that sulfur caps of 300 and 200 ppm would
result in average sulfur levels of 200 and 145 ppm. The results in Table C-l  indicate that these
average levels more closely resemble the 75th percentiles as opposed to the 85th percentiles.

       There are few actual examples of sulfur caps  in the U.S. and elsewhere which could also
be used to assess the relationship between a cap standard and the resulting average levels. One
example is the 300 ppm gasoline sulfur cap which existed  in California prior to California Phase
n RFG, which resulted in an average sulfur level of 150 ppm. That cap/average relationship
corresponds to the 95th percentile, based on EPA's analysis. That example, however, may not be
representative of the rest of the country since California refiners generally have substantial
hydrocracking capability to process the dirty feedstocks being processed and thus would tend to
produce lower sulfur gasoline even in the absence of a cap.3  While not gasoline, another
example of a cap is the federal 500 ppm limit on the  sulfur content of highway diesel. That cap
results in an average sulfur level of 340 ppm in the highway diesel sulfur pool.4  Based on the
results in Table C-l, this cap/average relationship roughly  corresponds with the 80th percentile.
Again, the applicability of this relationship is limited because it is based on the average diesel
sulfur level, which is controlled by a slightly different desulfurization technology than is
gasoline.  In Japan, refiners must meet a 100 ppm sulfur cap, and average gasoline pool sulfur
       3See the discussion in Chapter III, Section B: Refinery Operations Affecting Sulfur.

       4David Korotney, MEMORANDUM to Susan Willis, "A review of current and historical nonroad diesel
sulfur levels," March 3,1998.

EPA Staff Paper:  Gasoline Sulfur - Appendices      A-24                            May 1,1998

-------
levels there are about 30 ppm. Although not necessarily representative of U.S. refiners, this
relationship between the cap and average sulfur level is more consistent with the 95th percentile.
It is important to assess how the relationship in Japan between the average gasoline sulfur level
and the gasoline sulfur cap would apply to such a relationship here in the U.S., taking into
account the differences in refinery capability, crude oil slate, and other fuel sulfur or emission
control standards which may be affecting gasoline sulfur levels.

       Based on this assessment of the likely average sulfur level which would result from a cap
on gasoline sulfur levels, a preliminary range of average sulfur levels can be estimated from
suggested cap values. Since significant uncertainty exists in this methodology, the range will
encompass all  the cap/average relationships derived above. These relationships are summarized
in the figure below where the two curves mark the upper and lower boundaries of the range of the
above estimates. One observation which can be made from applying this relationship, presuming
it does indeed properly estimate the difference between a per-gallon cap and average pool
gasoline sulfur level, is that as the cap is increased in stringency, the spread between the average
and cap sulfur levels decreases in absolute terms and increases in relative terms. This is evident
in the increased slope at lower sulfur levels.
EPA Staff Paper: Gasoline Sulfur - Appendices      A-25                             May 1,1998

-------
Table C-l: Relationship of the Percentiles of Sulfur Levels in Regular Summertime Gasoline Batches
to the Mean Sulfur Level of the Gasoline Pool for U.S. Refineries
Mean Sulfur
Level of
Refinery
All Data
Analyzed
S<50 ppm
50
-------
600
500
400
|
a 300
?
£
a.
5
200
100
0
Estimated Relationship Between a Cap Sulfur Standard and Gasoline Pool Average Sulfur Level


t





























—Q.

	 H













•*-















-7^













••































&—
























~^"~



















'-"
























©


















..-J ' '














-'•*





























































— ~\















r*1-















..<:.































~^















-^















U—















^-«-














































. *
7B-
































































































































) 50 100 150 200 250 300 350
Estimated Average Sulfur Level in Gasoline Pool (ppm)
EPA Staff Paper:  Gasoline Sulfur - Appendices
A-27
May 1,1998

-------
                                     Appendix D

               Diesel and Gasoline Desulfurization in the Refinery
       In addition to examining the need for gasoline sulfur reductions in the context of future
light-duty vehicle/truck emission standards, the upcoming emission standards for diesel engines
provides compelling reason to examine the need for changes to diesel fuel. In particular, the
potential need for diesel sulfur reductions in the near future would be important to study because
there are associations between how a refiner might desulfurize both gasoline and diesel fuel.
These associations would make it advantageous for many refiners to address sulfur reduction
requirements of both gasoline and diesel at the same time. Even if refiners are not required to
control both fuels at the same time, if they are at least aware of impending diesel sulfur
regulations, then they could invest in gasoline sulfur reduction, or respond in other ways, to
facilitate a lower capital or operational cost  for meeting a diesel sulfur requirement at a later date.
There are several strategies which refiners could apply to address gasoline and diesel
concurrently which would reduce the cost of addressing each individually. These include
investments in larger refinery processing units which could cover the needs of both requirements,
investments in different types of processing units that could achieve both fuel changes at a lower
overall investment, or perhaps changes in the raw material and product slate.

       Before investigating these options in more detail, it would be useful to examine the
configuration of two typical refineries which produce motor vehicle fuels. One refinery is rather
complex, which makes it capable of using a heavier, higher sulfur crude oil to make a wide range
of refined products.  The complex refinery is also capable of producing a larger fraction of its
products as gasoline, and could more easily produce reformulated gasoline. This refinery is
shown in Figure Dla.  The second typical refinery is a simpler refinery with less processing
capability, and is shown in Figure D2a. The various units within these representative refineries
are summarized below in Table D-1.
EPA Staff Paper: Gasoline Sulfur - Appendices     A-28                            May 1,1998

-------
I
I
o
CA
s
I

>
       Hydrogen Plant
Natural

  Gas
•ss
Fuel Gas


 LPG
                                                                           Gasoline
                                                                          Aromatics



                                                                           Kerosine


                                                                           Jet Fuel




                                                                           Low Sulfur

                                                                             Diesel




                                                                      ••Off-road Diesel

                                                                       *•  Fuel Oil



                                                                             Resid



                                                                             Coke
                                                                Coker
                         Figure Dla:  Diagram of a Typical Complex Refinery

-------
I
C/J
O
g
C/l


I
>

O
         Crude Tower
Fuel Gas
                                                                              LPG
Gasoline
                                                                           Kerosine


                                                                            Jet Fuel
                                                                           Low Sulfur

                                                                             Diesel
                                                                         Off-road Diesel

                                                                            Fuel Oil
                                                                             Resid
                                                                             Coke
                                                                 Coker
                            Figure D2a: Diagram of a Typical Refinery

-------
Table D-l. Summary of Principal Refinery Processing Units and their Role in Gasoline and Diesel Refining
Unit Name
Crude Tower
Vacuum Tower
Naphtha Splitter
Straight Run
Splitter
Gas Plant
Fluidized Cat
Cracker
Hydrocracker
Distillate
Hydrotreater
Reformer
Hydrotreater
Reformer
Isomerization
Alkylation,
Polymerization
Hydrogen Plant
Affect Gasoline?
Y
Y
Y
Y
Y
Y
Y
N
Y
Y
Y
Y
Y
Affect Diesel?
Y
Y
N
N
N
Y
Y
Y
N
N
N
N
Y
Role in the Refinery
Distill crude oil into various streams for refining, distillate
and gasoline cuts may go directly to blending
Distill crude tower bottoms into streams for refining
Distill naphtha into streams for refining
Distill volatile naphtha into streams for gasoline refining
Distill very volatile streams for refinery fuel, LPG sales and
for further refinery processing to produce gasoline
Breaks large molecules into smaller ones for making more
marketable refined products such as gasoline and diesel
Removes impurities while breaking large molecules into
smaller ones, makes low sulfur refined products
Reduces sulfur in distillate blendstocks for meeting low
sulfur needs
Reduces sulfur in feed to reformer which uses a sulfur
sensitive catalyst
Rearranges molecules to make aromatics which improves
gasoline octane; makes hydrogen for desulfurization
Makes branched chain molecules out of straight chain
molecules to improve octane value of naphtha
Joins together very small molecules to make gasoline
blendstocks with high octane value
Makes hydrogen from natural gas for desulfurization
       A number of observations can be made from these figures concerning the production of
gasoline and diesel. First, gasoline is often produced from a number of different units, all of
which are designed to either convert other boiling stocks into gasoline blendstocks, or to improve
the octane quality of gasoline. Conversely, diesel is mostly made up of straight run streams; the
rest is a result of minimal processing (streams from the fluidized catalytic cracker (FCC) being
the critical exception).  Outside of the initial distillation, gasoline and diesel blendstocks are
created from the same units which are designed to crack the heavier part of the crude oil into
more marketable products.  These units are the FCC, the hydrocracker and the coker. Even
though these refining units do not comprise a large volumetric overlap between the production of
gasoline and diesel, when considering the total amount of sulfur which ends up in gasoline and
diesel, the FCC and coker refinery processing units provide a substantial contribution. -These two
 EPA Staff Paper  Gasoline Sulfur - Appendices     A-31
May 1,1998

-------
figures also show that refineries in the U.S. are already equipped with some facilities for
hydrotreating distillate to make lower sulfur diesel and other products.  However, the
hydrotreating for these streams is moderate, as they are only designed to reduce diesel sulfur
levels to meet the 500 ppm federal diesel sulfur standard.1  Some sulfur reduction units will
likely be added in those refineries producing gasoline to meet the year 2000 Phase IIRFG
standards.

       A number of possible processing changes could be made in the refinery to reduce the
sulfur content of gasoline, diesel fuel, or both fuels.  For gasoline, these are summarized in
Chapter in of this paper. These processes include hydrotreating of FCC gasoline blendstock
outputs, hydrotreating the feed to the FCC unit, and using hydrocrackers. Diesel desulfurization
involves similar processes. These include hydrotreating of diesel blendstocks (from the FCC or
other units), hydrotreating the feed to  the FCC unit, and using hydrocrackers.

       Hydrotreating the FCC gasoline and hydrotreating diesel blendstocks are processes with
no overlap in the hydroprocessing units themselves, since these are essentially "end of the pipe"
treatment processes for these two refined products (shown as units "B" and "C", respectively, in
the figures above).  However, there still may be economic advantages to implementing these two
processes at the same time based on the need to provide a more economical incremental supply
of hydrogen or additional utilities. Refiners could install a larger hydrogen production facility
which would provide a lower cost supply of hydrogen for the hydrodesulfurization.

       Hydrotreating the FCC feed affects the sulfur levels of both gasoline and diesel (shown as
unit "A" in the figures above). This is because the feed hydrotreater reduces the sulfur level in
the entire feed to the FCC unit, and a large portion of the FCC unit product stream ends up in
gasoline and diesel. Since the heaviest part of crude oil contains the most sulfur, these streams
will contain much more sulfur than other gasoline and diesel blendstocks, except for coker
streams. FCC feed hydrotreating could be a more cost-effective option if both gasoline and
diesel sulfur levels are controlled, since it reduces the sulfur levels from both fuels, and since the
hydrogen supply and other utility needs could be met for both fuels at the same time.  In addition,
FCC feed hydrotreating improves the  performance of the FCC, which would improve the product
yields of that unit.

       Another alternative as part of a strategy to reduce both gasoline and diesel sulfur levels is
to install a hydrocracker. Hydrocrackers (shown as an existing unit in Figure Dla and as unit
"D" in Figure D2a) convert the heavier parts of crude oil into compounds which can be used in
gasoline and diesel fuel. However, unlike an FCC unit, a hydrocracker removes much of the
sulfur present at the same time, unlike the FCC unit, which leaves the sulfur in the product
streams.  As stated earlier in this report, some refiners use hydrocrackers to process less desirable
feedstocks which would be troublesome to treat with methods other than with hydrocracking.
Alternatively, by investing in hydrocracking technology for their refinery, a refiner could change
       'In-use, highway diesel fuel sulfur average levels are about 340 ppm.

 EPA Staff Paper:  Gasoline Sulfur - Appendices     A-32                           May 1,1998

-------
its crude slate to cheaper, higher sulfur crude oils or other similar feedstocks at a raw material
cost savings.

       Finally, to provide more flexibility and to achieve low levels of sulfur, a refinery may
install several sulfur reduction units. For example, refiners without hydrocrackers could install
both a hydrocracker and FCC gasoline and distillate hydrotreaters. This may allow the refinery
to meet stringent sulfur reduction requirements and offset some of their costs with purchasing
higher sulfur feedstocks.  Another alternative would be for refiners to install a hydrocracker and
FCC feed hydrotreating. However, many refiners may not be in the position financially to make
such large capital investments, even with anticipated improvements in refinery profit margins.

       EPA and DOE plan to investigate the impact of low sulfur diesel fuel requirements on the
cost of gasoline sulfur control using the ORNL-RYM refinery model. The model will be used to
estimate the cost and required capital equipment needed to simultaneously control diesel and
gasoline sulfur. These processing changes and resulting costs will be compared to those resulting
from the control of gasoline sulfur alone to assess whether there is much potential for synergies.
However, the results will be limited in part by the current configuration of the refinery model. N
The ORNL-RYM model has never been used to simulate diesel sulfur control to such low levels
and there is some question about whether it contains sufficient refinery modeling capability to
generate realistic cost estimates. The model's capability will be assessed after performing
several initial modeling runs.
 EPA Staff Paper.  Gasoline Sulfur - Appendices     A-33                            May 1,1998

-------