Control of Hazardous Air Pollutants From Mobile Sources

[Federal Register: March 29, 2006 (Volume 71, Number 60)]
[Proposed Rules]
[Page 15853-15902]
From the Federal Register Online via GPO Access [wais.access.gpo.gov]
[DOCID:fr29mr06-34]

[[pp. 15853-15902]]
Control of Hazardous Air Pollutants From Mobile Sources
[[Continued from page 15852]]
[[Page 15853]]

Table VI.B-4.--Schedule for In-Use Standards for HLDVs/MDPVs
----------------------------------------------------------------------------------------------------------------
Model year of introduction 2010 2011 2012 2013 2014 2015
----------------------------------------------------------------------------------------------------------------
Models years that the in-use standard is 2010 2011 2012 2013 2014 2015
available for carry-over test groups......... 2011 2012 2013 2014 2015 2016
2012 2013 2014 2015 2016
2013 2014 2015
----------------------------------------------------------------------------------------------------------------

7. Monitoring and Enforcement
Under the proposed programs, manufacturers could either report that
they met the relevant corporate average standard in their annual
reports to the Agency, or they could show via the use of credits that
they have offset any exceedance of the corporate average standard.
Manufacturers would also report their credit balances or deficits. EPA
would monitor the program.
As in Tier 2, the averaging, banking and trading program would be
enforced through the certificate of conformity that manufacturers must
obtain in order to introduce any regulated vehicles into commerce.\198\
The certificate for each test group would require all vehicles to meet
the emissions level to which the vehicles were certified, and would be
conditioned upon the manufacturer meeting the corporate average
standard within the required time frame. If a manufacturer failed to
meet this condition, the vehicles causing the corporate average
exceedance would be considered to be not covered by the certificate of
conformity for that engine family. A manufacturer would be subject to
penalties on an individual vehicle basis for sale of vehicles not
covered by a certificate.
---------------------------------------------------------------------------

\198\ ``Control of Air Pollution from New Motor Vehicles: Tier 2
Motor Vehicle Emissions Standards and Gasoline Sulfur Control
Requirements'', Final Rule, 65 FR 6797, February 10, 2000.
---------------------------------------------------------------------------

EPA would review the manufacturer's sales to designate the vehicles
that caused the exceedance of the corporate average standard. We would
designate as nonconforming those vehicles in those test groups with the
highest certification emission values first, continuing until a number
of vehicles equal to the calculated number of noncomplying vehicles as
determined above is reached. In a test group where only a portion of
vehicles would be deemed nonconforming, we would determine the actual
nonconforming vehicles by counting backwards from the last vehicle
produced in that test group. Manufacturers would be liable for
penalties for each vehicle sold that is not covered by a certificate.
We are proposing to condition certificates to enforce the
requirements that manufacturers not sell credits that they have not
generated. A manufacturer that transferred credits it did not have
would create an equivalent number of debits that it would be required
to offset by the reporting deadline for the same model year. Failure to
cover these debits with credits by the reporting deadline would be a
violation of the conditions under which EPA issued the certificate of
conformity, and nonconforming vehicles would not be covered by the
certificate. EPA would identify the nonconforming vehicles in the same
manner described above.
In the case of a trade that resulted in a negative credit balance
that a manufacturer could not cover by the reporting deadline for the
model year in which the trade occurred, we propose to hold both the
buyer and the seller liable. We believe that holding both parties
liable will induce the buyer to exercise diligence in assuring that the
seller has or will be able to generate appropriate credits and will
help to ensure that inappropriate trades do not occur.
We are not proposing any new compliance monitoring activities or
programs for vehicles. These vehicles would be subject to the
certification testing provisions of the CAP2000 rule. We are not
proposing to require manufacturer in-use testing to verify compliance.
There is no cold CO manufacturer in-use testing requirement today
(similarly, we do not require manufacturer in-use testing for SCO3
standards under the SFTP program). As noted earlier, manufacturers have
limited cold temperature testing capabilities and we believe these
facilities will be needed for product development and certification
testing. However, we have the authority to conduct our own in-use
testing program for exhaust emissions to ensure that vehicles meet
standards over their full useful life. We will pursue remedial actions
when substantial numbers of properly maintained and used vehicles fail
any standard in-use. We also retain the right to conduct Selective
Enforcement Auditing of new vehicles at manufacturers' facilities.
The use of credits would not be permitted to address Selective
Enforcement Auditing or in-use testing failures. The enforcement of the
averaging standard would occur through the vehicle's certificate of
conformity. A manufacturer's certificate of conformity would be
conditioned upon compliance with the averaging provisions. The
certificate would be void ab initio if a manufacturer failed to meet
the corporate average standard and did not obtain appropriate credits
to cover their shortfalls in that model year or in the subsequent model
year (see proposed deficit carryforward provision in section
VI.B.5.e.). Manufacturers would need to track their certification
levels and sales unless they produced only vehicles certified to NMHC
levels below the standard and did not plan to bank credits.
We request comments on the above approach for compliance monitoring
and enforcement.

C. What Evaporative Emissions Standards Are We Proposing?

We are proposing to adopt a set of numerically more stringent
evaporative emission standards for all light-duty vehicles, light-
trucks, and medium-duty passenger vehicles. The proposed standards are
equivalent to California's LEV II standards, and these proposed
standards are shown in Table VI.C-1. The proposed standards would
represent about a 20 to 50 percent reduction (depending on vehicle
weight class and type of test) in diurnal plus hot soak standards from
the Tier 2 standards that will be in effect in the years immediately
preceding the implementation of today's proposed standards.\199\ As
with the current Tier 2 evaporative emission standards, the proposed
standards vary by vehicle weight class. The increasingly higher standards
for heavier weight class vehicles account for larger vehicle sizes

[[Page 15854]]

and fuel tanks (non-fuel and fuel emissions).\200\
---------------------------------------------------------------------------

\199\ Diurnal emissions (or diurnal breathing losses) means
evaporative emissions as a result of daily temperature cycles or
fluctuations for successive days of parking in hot weather. Hot soak
emissions (or hot soak losses) are the evaporative emissions from a
parked vehicle immediately after turning off the hot engine. For the
evaporative emissions test procedure, diurnal and hot soak emissions
are measured in an enclosure commonly called the SHED (Sealed
Housing for Evaporative Determination).
\200\ Larger vehicles may have greater non-fuel evaporative
emissions, probably due to an increased amount of interior trim,
vehicle body surface area, and larger tires.

Table VI.C-1.--Proposed Evaporative Emission Standards
[Grams of hydrocarbons per test]
------------------------------------------------------------------------
Supplemental 2-
Vehicle class 3-day diurnal day diurnal
plus hot soak plus hot soak
------------------------------------------------------------------------
LDVs.................................. 0.50 0.65
LLDTs................................. 0.65 0.85
HLDTs................................. 0.90 1.15
MDPVs................................. 1.00 1.25
------------------------------------------------------------------------

1. Current Controls and Feasibility of the Proposed Standards
Evaporative emissions from light-duty vehicles and trucks will
represent about 35 percent of the light-duty VOC inventory and about 4
percent of the benzene inventory in 2020. As described earlier, we are
proposing to reduce the level of the evaporative emission standards
applicable to diurnal and hot soak emissions from these vehicles by
about 20 to 50 percent. These proposed standards are meant to be
effectively the same as the evaporative emission standards in the
California LEV II program. Although the California program contains
evaporative emissions standards that appear more stringent than EPA
Tier 2 standards if one looks only at the level of the standard, we
believe they are essentially equivalent because of differences in
testing requirements. For these same reasons, some manufacturers
likewise view the programs as similar in stringency. (See section
VI.C.5 below for further discussion of such test differences, e.g.,
test temperatures and fuel volatilities.) Thus, some manufacturers have
indicated that they will produce 50-state evaporative systems that meet
both sets of standards (manufacturers sent letters indicating this to
EPA in 2000).201 202 203 In addition, a review of recent
model year certification results indicates that essentially all
manufacturers certify 50-state systems, except for a few limited cases
where manufacturers have not yet needed to certify a LEVII vehicle in
California due to the phase-in schedule. Also, in recent discussions,
manufacturers have restated that they plan to continue producing 50-
state evaporative systems in the future. Based on this understanding,
we do not project additional VOC or air toxics reductions from the
evaporative standards we are proposing today.\204\ Also, we do not
expect additional costs since we expect that manufacturers will
continue to produce 50-state evaporative systems. Therefore,
harmonizing with California's LEV-II evaporative emission standards
would be an ``anti-backsliding'' measure--that is, it would prevent
potential future backsliding as manufacturers pursue cost
reductions.\205\ It would thus codify (i.e., lock in) the approach
manufacturers have already indicated they are taking for 50-state
evaporative systems.
---------------------------------------------------------------------------

\201\ DaimlerChrysler, Letter from Reginald R. Modlin to Margo
Oge of U.S. EPA, May 30, 2000. A copy of this letter can be found in
Docket No. EPA-HQ-OAR-2005-0036.
\202\ Ford, Letter from Kelly M. Brown to Margo Oge of U.S. EPA,
May 26, 2000. A copy of this letter can be found in Docket No. EPA-
HQ-OAR-2005-0036.
\203\ General Motors, Letter from Samuel A. Leonard to Margo Oge
of U.S. EPA, May 30, 2000. A copy of this letter can be found in
Docket No. EPA-HQ-OAR-2005-0036.
\204\ U.S. EPA, Office of Air and Radiation, Update to the
Accounting for the Tier 2 and Heavy-Duty 2005/2007 Requirements in
MOBILE6, EPA420-R-03-012, September 2003.
\205\ Anti-backsliding provisions can satisfy the requirement in
section 202 (l) (2) that emission reductions of hazardous air pollutants
be the greatest achievable. Sierra Club v. EPA, 325 F. 3d at 477.
---------------------------------------------------------------------------

We believe this proposed action would be an important step to
ensure that the federal standards reflect the lowest possible
evaporative emissions, and it also would provide states with certainty
that the emissions reductions we project to occur due to 50-state
compliance strategies will in fact occur. In addition, the proposed
standards will assure that manufacturers continue to capture the abilities
of available fuel system materials to minimize evaporative emissions.
We also considered the possibility of whether it is feasible to
achieve further evaporative emission reductions from motor vehicles. In
this regard, it is important to note that California's LEV II program
includes partial zero-emission vehicle (ZEV) credits for vehicles that
achieve near zero emissions (e.g., LDV evaporative emission standards
for both the 2-day and 3-day diurnal plus hot soak tests are 0.35
grams/test, which are more stringent than proposed standards).\206\ The
credits would include full ZEV credit for a stored hydrogen fuel cell
vehicle and 0.2 credits for (among other categories for partial credit)
a partial zero emission vehicle (PZEV).\207\ Currently, only a fraction
of California's certified vehicles (gasoline powered, hybrid, and
compressed natural gas vehicles) meet California's optional PZEV
standards, but this number is expected to increase in coming
years.208 209 These limited PZEV vehicles require additional
evaporative emissions technology or hardware (e.g., modifications to
fuel tank and secondary canister) than we expect to be needed for
vehicles meeting the proposed standards. At this time, we need to
better understand the evaporative system modifications (i.e.,
technology, costs, lead time, etc.) potentially needed for other
vehicles in the fleet to meet PZEV-level standards before we can
rationally evaluate whether to adopt more stringent standards. For
example, at this point we cannot even determine whether the PZEV
technologies could be used fleetwide or on only a limited set of
vehicles. Thus, in the near term, we lack any of the information
necessary to determine if further reductions are feasible, and if they
could be achievable considering cost, energy and safety issues.
However, we intend to consider

[[Page 15855]]

more stringent evaporative emission standards in the future, and
revisiting this issue in a future rulemaking will allow us time to
obtain the important necessary additional information for such standards.
---------------------------------------------------------------------------

\206\ California Air Resources Board, Fact Sheet, LEV-II
Amendments to California's Low-Emission Vehicle Regulations, February 1999
\207\ PZEV meets California super ultra low emission vehicle
exhaust emission standards and have near zero evaporative emissions.
California Air Resources Board, News Release, ARB Modifies Zero
Emission Vehicle Regulation, April 24, 2003.
\208\ California Air Resources Board, Fact Sheet, California
Vehicle Emissions, April 8, 2004.
\209\ California Air Resources Board, Consumer Information: 2006
California Certified Vehicles, November 7, 2005.
---------------------------------------------------------------------------

2. Evaporative Standards Timing
We are proposing to implement today's evaporative emission
standards in model year 2009 for LDVs/LLDTs and model year 2010 for
HLDTs/MDPVs. Today's proposed rule is not expected to be finalized
until February 2007, at which time many manufacturers already will have
begun or completed model year 2008 certification. Thus, model year 2009
is the earliest practical start date of new standards for LDVs/LLDTs.
For HLDTs/MDPVs, the phase-in of the existing Tier 2 evaporative
emission standards ends in model year 2009. Thus, the model year 2010
is the earliest start date possible for HLDTs/MDPVs. Since the proposed
standards are an anti-backsliding measure and we believe that
manufacturers already meet these standards, there is no need for
additional lead time beyond the implementation dates proposed. We
request comment on this proposed schedule.
3. Timing for Multi-Fueled Vehicles
As discussed earlier in this section, manufacturers appear to view
the Tier 2 and LEV II evaporative emission programs as similar in
stringency, and thus, they have indicated that they will produce 50-
state evaporative systems that meet both sets of standards. For multi-
fueled vehicles capable of operating on alternative fuel (e.g., E85
vehicles--fuel is 85% ethanol and 15% gasoline) and conventional fuel
(e.g., gasoline),\210\ this commitment for 50-state systems would still
apply. However, a few multi-fueled vehicles were certified only on the
conventional fuel (gasoline) for the California LEV II program even
though they had 50-state evaporative emission systems. For such cases,
manufacturers did not intend to sell these vehicles for operation on
the alternative fuel (e.g. E85) in California (only for operation on
conventional fuel in California), but they did certify and plan to sell
these vehicles in the federal Tier 2 program for operation on the
alternative and conventional fuels.\211\ For these few types of multi-
fueled vehicles, manufacturers are potentially at risk of not complying
with the proposed new evaporative emission certification standards
(which are equivalent to California LEV II certification standards)
when operating on the alternative fuel.
---------------------------------------------------------------------------

\210\ 40 CFR 86.1803-01 defines multi-fuel as capable of operating on
two or more different fuel types, either separately or simultaneously.
\211\ For the Tier 2 program, multi-tier vehicles must meet the
same standards on conventional and alternative fuel.
---------------------------------------------------------------------------

For such multi-fueled vehicles or evaporative emission systems,
manufacturers would need a few additional years of lead time to adjust
their evaporative systems to comply with the proposed evaporative
emission certification standards when operating on the alternative
fuel. Thus, to reduce the compliance risk for these types of multi-
fueled vehicles (or evaporative families) when they first certify to
the more stringent evaporative standards, the proposed evaporative
emission certification standards would apply to the non-gasoline
portion of multi-fueled vehicles beginning in the fourth year of the
program--2012 for LDVs/LLDTs and 2013 for HLDTs/MDPVs. The proposed
evaporative emission certification standards would be implemented in
2009 for LDVs/LLDTs and 2010 for HLDTs/MDPVs for the gasoline portion
of multi-fueled vehicles and vehicles that are not multi-fueled. We
believe this additional three years of lead time would provide
sufficient time for manufacturers to make adjustments to their new
evaporative systems for multi-fueled vehicles, which are limited
product lines.
The provisions for in-use evaporative emission standards described
below in section VI.C.4 would not change for multi-fueled vehicles. We
believe that three additional years to prepare vehicles (or evaporative
families) to meet the certification standards, and to simultaneously
make vehicle adjustments from the federal in-use experience of other
vehicles (other vehicles that are not multi-fueled) is sufficient to
resolve any issues for multi-fueled vehicles. Therefore, the proposed
evaporative emission standards would apply both for certification and
in-use beginning in 2012 for LDVs/LLDTs and 2013 for HLDTs/MDPVs.
4. In-Use Evaporative Emission Standards
As described earlier in this section, we are proposing to adopt
evaporative emission standards that are equivalent to California's LEV
II standards for all light duty vehicles, light trucks, and medium duty
passenger vehicles. Currently, the Tier 2 evaporative emission
standards are the same for certification and in-use vehicles. However,
the California LEV II program permits manufacturers to meet less
stringent standards in-use for a short time period in order to account
for potential variability in-use during the initial years of the
program when technical issues are most likely to arise.\212\ The LEV II
program specifies that in-use evaporative emission standards of 1.75
times the certification standards will apply for the first three model
years after an evaporative family is first certified to the LEV II
standards (only for vehicles introduced prior to model year 2007, the
year after 100 percent phase-in).213 214 An interim three-
year period was considered sufficient to accommodate any technical
issues that may arise.
---------------------------------------------------------------------------

\212\ California Air Resources Board, ``LEV II'' and ``CAP
2000'' Amendments to the California Exhaust and Evaporative Emission
Standards and Test Procedures for Passenger Cars, Light-Duty Trucks
and Medium-Duty Vehicles, and to the Evaporative Emission Requirements for
Heavy-Duty Vehicles, Final Statement of Reasons, September 1999.
\213\ 1.75 times the 3-day diurnal plus hot soak and 2-day
diurnal plus hot soak standards.
\214\ For example, evaporative families first certified to LEV
II standards in the 2005 model year shall meet in-use standards of
1.75 times the evaporative certification standards for 2005, 2006,
and 2007 model year vehicles.
---------------------------------------------------------------------------

Federal in-use conditions may raise unique issues (e.g., salt/ice
exposure) for evaporative systems certified to the new proposed
standards (which are equivalent to the LEV II standards), and thus, we
propose to adopt a similar, interim in-use compliance provision for
federal vehicles. As with the LEV II program, this provision would
enable manufacturers to make adjustments for unforeseen problems that
may occur in-use during the first three years of a new evaporative
family. Like California, we believe that a three-year period is enough
time to resolve these problems, because it allows manufacturers to gain
real world experience and make adjustments to a vehicle within a
typical product cycle.
Depending on the vehicle weight class and type of test, the Tier 2
certification standards are 1.3 to 1.9 times the LEV II certification
standards. On average the Tier 2 standards are 1.51 times the LEV II
certification standards. Thus, to maintain the same level of stringency
for the in-use evaporative emission standards provided by the Tier 2
program, we propose to apply the Tier 2 standards in-use for only the
first three model years after an evaporative family is first certified
under today's proposed standards instead of the 1.75 multiplier
implemented in the California LEV II program. Since the proposed
evaporative emission certification standards (equivalent to LEV II
standards) would be implemented in model year 2009 for LDVs/LLDTs and
model year 2010 for HLDTs/MDPVs, these same certification

[[Page 15856]]

standards would apply in-use beginning in model year 2012 for LDVs/
LLDTs and model year 2013 for HLDTs/MDPVs.\215\
---------------------------------------------------------------------------

\215\ For example, evaporative families first certified to the
proposed LDV/LLDT evaporative emission standards in the 2011 model
year would be required to meet the Tier 2 LDV/LLDT evaporative
emission standards in-use for 2011, 2012, and 2013 model year
vehicles (applying Tier 2 standards in-use would be limited to the
first three years after introduction of a vehicle), and 2014 and
later model year vehicles of such evaporative families would be required
to meet the proposed LDV/LLDT evaporative emission standards in-use.
---------------------------------------------------------------------------

5. Existing Differences Between California and Federal Evaporative
Emission Test Procedures
As described above, the California LEV II evaporative emission
standards are numerically more stringent than EPA's Tier 2 standards,
but due to differences in California and EPA evaporative test
requirements, EPA and most manufacturers view the programs as similar
in stringency. The Tier 2 evaporative program requires manufacturers to
certify the durability of their evaporative emission systems using a
fuel containing the maximum allowable concentration of alcohols
(highest alcohol level allowed by EPA in the fuel on which the vehicle
is intended to operate, i.e., a ``worst case'' test fuel). Under
current requirements, this fuel would be about 10 percent ethanol by
volume.\216\ (We are retaining these Tier 2 durability requirements for
the proposed evaporative emissions program.) California does not
require this provision. To compensate for the increased vulnerability
of system components to alcohol fuel, manufacturers have indicated that
they will produce a more durable evaporative emission system than the
Tier 2 numerical standards would imply, using the same low permeability
hoses and low loss connections and seals planned for California LEV II
vehicles.
---------------------------------------------------------------------------

\216\ Manufacturers are required to develop deterioration
factors using a fuel that contains the highest legal quantity of
ethanol available in the U.S.
---------------------------------------------------------------------------

As shown in Table VI.C-2, combined with the maximum alcohol fuel
content for durability testing, the other key differences between the
federal and California test requirements are fuel volatilities, diurnal
temperature cycles, and running loss test temperatures.\217\ The EPA
fuel volatility requirement is 2 psi greater than that of California.
The high end of EPA's diurnal temperature range, is 9[deg] F lower than
that of California. Also, EPA's running loss temperature is 10[deg] F
lower than California's.
---------------------------------------------------------------------------

\217\ Running loss emissions means evaporative emissions as a
result of sustained vehicle operation (average trip in an urban
area) on a hot day. The running loss test requirement is part of the
3-day diurnal plus hot soak test sequence.

Table VI.C-2.--Differences in Tier 2 and LEV II Evaporative Emission
Test Requirements
------------------------------------------------------------------------
Test requirement EPA tier 2 California LEV II
------------------------------------------------------------------------
Fuel volatility (Reid Vapor 9................... 7.
Pressure in psi).
Diurnal temperature cycle 72 to 96............ 65 to 105.
(degrees F).
Running loss test 95.................. 105.
temperature (degrees F).
------------------------------------------------------------------------

Currently, California accepts evaporative emission results
generated on the federal test procedure (using federal test fuel),
because available data indicates the federal procedure to be a ``worst
case'' procedure. In addition, manufacturers can obtain federal
evaporative certification based upon California results (meeting LEV II
standards under California fuels and test conditions), if they obtain
advance approval from EPA.\218\
---------------------------------------------------------------------------

\218\ EPA may require comparative data from both federal and
California tests.
---------------------------------------------------------------------------

D. Opportunities for Additional Exhaust Control Under Normal Conditions

In addition to the cold temperature NMHC and evaporative emission
standards we are proposing, we evaluated an additional option for
reducing hydrocarbons from light-duty vehicles. This option would
further align the federal light-duty exhaust emissions control program
with that of California. We are not proposing this option today for the
reasons described below. It is possible that a future evaluation could
result in EPA reconsidering the option of harmonizing the Tier 2
program with California's LEV-II program or otherwise seeking emission
reductions beyond those of the Tier 2 program and those being proposed
today.\219\
---------------------------------------------------------------------------

\219\ See Sierra Club v. EPA, 325 F.3d at 480 (EPA can
reasonably determine that no further reductions in MSATs are
presently achievable due to uncertainties created by other recently
promulgated regulatory provisions applicable to the same vehicles).
---------------------------------------------------------------------------

As explained earlier, section 202(l)(2) requires EPA to adopt
regulations that contain standards which reflect the greatest degree of
emissions reductions achievable through the application of technology
that will be available, taking into consideration existing motor
vehicle standards, the availability and costs of the technology, and
noise, energy and safety factors. The cold temperature NMHC program
proposed today is appropriate under section 202(l)(2) as a near-term
control: That is, a control that can be implemented relatively soon and
without disruption to other existing vehicle emissions control program.
We are not proposing long-term (i.e., controls that require longer lead
time to implement) at this time because we lack the information
necessary to assess appropriate long-term controls. We believe it will
be important to address the appropriateness of further MSAT controls in
the context of compliance with other significant vehicle emissions
regulations (discussed below).
In the late 1990's both the EPA and the California Air Resources
Board finalized new and technologically challenging light-duty vehicle/
truck emission control programs. The EPA program, known as Tier 2,
focused on reducing NO_X emissions from the light-duty fleet.
The California program, which is the second generation of their low
emission vehicle (LEV) program and is known as LEV-II, focuses
primarily on reducing hydrocarbons by tightening the light-duty NMOG
standards. Both programs are expected to present the manufacturers with
significant challenges, and will require the use of hardware and
emission control strategies not used in the fleet under previously
existing programs. Both programs will achieve significant reductions in
emissions. Taken as a whole, the Tier 2 program presents the
manufacturers with significant challenges in the coming years. Bringing
essentially all passenger vehicles under the same emission control
program regardless of their size, weight, and application is a major
engineering challenge. The Tier 2 program represents a comprehensive,
integrated package of exhaust, evaporative, and fuel quality standards
which will achieve significant reductions in

[[Page 15857]]

NMHC, NO_X, and PM emissions from all light-duty vehicles in
the program. These reductions will include significant reductions in
MSATs. Emission control in the Tier 2 program will be based on the
widespread implementation of advanced catalyst and related control
system technology. The standards are very stringent and will require
manufacturers to make full use of nearly all available emission control
technologies.
Today the Tier 2 program remains early in its phase-in. Cars and
lighter trucks will be fully phased into the program with the 2007
model year, and the heavier trucks won't be fully entered into the
program until the 2009 model year. Even though the lighter vehicles
will be fully phased in by 2007, we expect the characteristics of this
segment of the fleet to remain in a state of transition at least
through 2009, because manufacturers will be making adjustments to their
fleets as the larger trucks phase in. The Tier 2 program is designed to
enable vehicles certified to the LEV-II program to cross over to the
federal Tier 2 program. At this point in time, however, it is difficult
to predict the degree to which this will occur. The fleetwide NMOG
levels of the Tier 2 program will ultimately be affected by the manner
in which LEV-II vehicles are certified within the Tier 2 bin structure,
and vice versa. We intend to carefully assess these two programs as
they evolve and periodically evaluate the relative emission reductions
and the integration of the two programs.
Today's proposal addresses toxics emissions from vehicles operating
at cold temperatures. The technology to achieve this is already
available and we project that compliance will not be costly. However,
we do not believe that we could reasonably propose further controls at
this time. There is enough uncertainty regarding the interaction of the
Tier 2 and LEV-II programs to make it difficult to evaluate today what
might be achievable in the future. Depending on the assumptions one
makes, the LEV-II and Tier 2 programs may or may not achieve very
similar NMOG emission levels. Therefore, the eventual Tier 2 baseline
technologies and emissions upon which new standards would necessarily
be based are not known today. Additionally, we believe it is important
for manufacturers to focus in the near term on developing and
implementing robust technological responses to the Tier 2 program
without the distraction or disruption that could result from changing
the program in the midst of its phase-in. We believe that it may be
feasible in the longer term to seek additional emission reductions from
the base Tier 2 program, and the next several years will allow an
evaluation based on facts rather than assumptions. For these reasons,
we are deferring a decision on seeking additional NMOG reductions from
the base Tier 2 program.

E. Vehicle Provisions for Small Volume Manufacturers

Prior to issuing a proposal for this proposed rulemaking, we
analyzed the potential impacts of these regulations on small entities.
As a part of this analysis, we convened a Small Business Advocacy
Review Panel (SBAR Panel, or the Panel). During the Panel process, we
gathered information and recommendations from Small Entity
Representatives (SERs) on how to reduce the impact of the rule on small
entities, and those comments are detailed in the Final Panel Report
which is located in the public record for this rulemaking (Docket EPA-
HQ-OAR-2005-0036). Based upon these comments, we propose to include
lead time transition and hardship provisions that would be applicable
to small volume manufacturers as described below in section VI.E.1 and
VI.E.2. For further discussion of the Panel process, see section XII.C
of this proposed rule and/or the Final Panel Report.
As discussed in more detail in section XII.C in addition to the
major vehicle manufacturers, three distinct categories of businesses
relating to highway light-duty vehicles would be covered by the new
vehicle standards: Small volume manufacturers (SVMs), independent
commercial importers (ICIs),\220\ and alternative fuel vehicle
converters.\221\ We define small volume manufacturers as those with
total U.S. sales less than 15,000 vehicles per year, and this status
allows vehicle models to be certified under a slightly simpler
certification process. For certification purposes, SVMs include ICIs
and alternative fuel vehicle converters since they sell less than
15,000 vehicles per year.
---------------------------------------------------------------------------

\220\ ICIs are companies that hold a Certificate (or
certificates) of Conformity permitting them to import nonconforming
vehicles and to modify these vehicles to meet U.S. emission standards.
\221\ Alternative fuel vehicle converters are businesses that
convert gasoline or diesel vehicles to operate on alternative fuel
(e.g., compressed natural gas), and converters must seek a
certificate for all of their vehicle models.
---------------------------------------------------------------------------

About 34 out of 50 entities that certify vehicles are SVMs, and the
Panel identified 21 of these 34 SVMs that are small businesses as
defined by the Small Business Administration criteria (5 manufacturers,
10 ICIs, and 6 converters). Since a majority of the SVMs are small
businesses and all SVMs have similar characteristics as described below
in section VI.E.1, the Panel recommended that we apply the lead time
transition and hardship provisions to all SVMs. These manufacturers
represent just a fraction of one percent of the light-duty vehicle and
light-duty truck sales. Our proposal today is consistent with the
Panel's recommendation.
1. Lead Time Transition Provisions
In these types of vehicle businesses, predicting sales is difficult
and it is often necessary to rely on other entities for technology (see
earlier discussions in section VI on technology needed to meet the
proposed standards).222 223 Moreover, percentage phase-in
requirements pose a dilemma for an entity such as a SVM that has a
limited product line. For example, it is challenging for a SVM to
address percentage phase-in requirements if the manufacturer makes
vehicles in only one or two test groups. Because of its very limited
product lines, a SVM could be required to certify all their vehicles to
the new standards in the first year of the phase-in period, whereas a
full-line manufacturer (or major manufacturer) could utilize all four
years of the phase-in. Thus, similar to the flexibility provisions
implemented in the Tier 2 rule, the Panel recommended that we allow
SVMs, manufacturers with sales less than 15,000 vehicles per year
(includes all vehicle small entities that would be affected by this
rule, which are the majority of SVMs) the following flexibility options
for meeting cold temperature NMHC standards and evaporative emission
standards as an element of determining appropriate lead time for these
entities to comply with the standards.
---------------------------------------------------------------------------

\222\ For example, as described later in section VI.E.3, ICIs
may not be able to predict their sales because they are dependent
upon vehicles brought to them by individuals attempting to import
uncertified vehicles.
\223\ SMVs (those with sales less than 15,000 vehicles per year)
include ICIs, alternative fuel vehicle converters, companies that
produce specialty vehicles by modifying vehicles produced by others,
and companies that produce small quantities of their own vehicles,
but rely on major manufacturers for engines and other vital emission
related components.
---------------------------------------------------------------------------

For cold NMHC standards, the Panel recommended that SVMs simply
comply with the standards with 100 percent of their vehicles during the
last year of the 4 year phase-in period. Since these entities could
need additional lead time flexibility and proposed standards for light-
duty vehicles and light light-duty trucks would begin in model year
2010 and would end in model year 2013 (25%, 50%, 75%, 100% phase-in over 4

[[Page 15858]]

years), we propose that the SVM provision would be 100 percent in model
year 2013. Also, since the proposed standard for heavy light-duty
trucks and medium-duty passenger vehicles would start in 2012 (25%,
50%, 75%, 100% phase-in over 4 years), we propose that the SVM
provision would be 100 percent in model year 2015.
In regard to evaporative emission standards, the Panel recommended
that since the proposed evaporative emissions standards would not have
phase-in years, we allow SVMs to simply comply with standards during
the third year of the program (we have implemented similar provisions
in past rulemakings). Given the additional challenges that SVMs face,
as noted above, we believe that this recommendation is reasonable.
Therefore, for a 2009 model year start date for light-duty vehicles and
light light-duty trucks, we propose that SVMs meet the evaporative
emission standards in model year 2011. For a model year 2010
implementation date for heavy light-duty trucks and medium-duty
passenger vehicles, we propose that SVMs comply in model year 2012.
2. Hardship Provisions
In addition, the Panel recommended that hardship provisions be
extended to SVMs for the cold temperature NMHC and evaporative emission
standards as an aspect of determining the greatest emission reductions
feasible. These entities could, on a case-by-case basis, face hardship
more than major manufacturers (manufacturers with sales of 15,000
vehicles or more per year), and we are proposing this provision to
provide what could prove to be a needed safety valve for these
entities. SVMs would be allowed to apply for up to an additional 2
years to meet the 100 percent phase-in requirements for cold NMHC and
the delayed requirement for evaporative emissions. As with hardship
provisions for the Tier 2 rule, we propose that appeals for such
hardship relief must be made in writing, must be submitted before the
earliest date of noncompliance, must include evidence that the
noncompliance will occur despite the manufacturer's best efforts to
comply, and must include evidence that severe economic hardship will be
faced by the company if the relief is not granted.
We would work with the applicant to ensure that all other remedies
available under this rule are exhausted before granting additional
relief. To avoid the very existence of the hardship provision prompting
SVMs to delay development, acquisition and application of new
technology, we want to make clear that we would expect this provision
to be rarely used. Our proposed rule contains numerous flexibilities
for all manufacturers and it delays implementation dates for SVMs,
which effectively provides them more time. We would expect small volume
manufacturers to prepare for the applicable implementation dates in
today's proposed rule.
3. Special Provisions for Independent Commercial Importers (ICIs)
Although the SBAR panel did not specifically recommend it, we are
proposing to allow ICIs to participate in the averaging, banking, and
trading program for cold temperature NMHC fleet average standards (as
described in Table IV.B.-1), but with appropriate constraints to ensure
that fleet averages will be met. The existing regulations for ICIs
specifically bar ICIs from participating in emission related averaging,
banking, and trading programs unless specific exceptions are provided
(see 40 CFR 85.1515(d)). The concern is that they may not be able to
predict their sales and control their fleet average emissions because
they are dependent upon vehicles brought to them by individuals
attempting to import uncertified vehicles. However, an exception for
ICIs to participate in an averaging, banking, and trading program was
made for the Tier 2 NO_X fleet average standards, and today
we propose to apply a similar exception for the cold temperature NMHC
fleet average standards.
If an ICI is able to purchase credits or to certify a test group to
a family emission level (FEL) below the applicable cold temperature
NMHC fleet average standard, we would permit the ICI to bank credits
for future use. Where an ICI desires to certify a test group to a FEL
above the applicable fleet average standard, we would permit them to do
so if they have adequate and appropriate credits. Where an ICI desires
to certify to an FEL above the fleet average standard and does not have
adequate or appropriate credits to offset the vehicles, we would permit
the manufacturer to obtain a certificate for vehicles using such a FEL,
but would condition the certificate such that the manufacturer can only
produce vehicles if it first obtains credits from other manufacturers
or from other vehicles certified to a FEL lower than the fleet average
standard during that model year.
Our experience over the years through certification indicates that
the nature of the ICI business is such that these companies cannot
predict or estimate their sales of various vehicles well. Therefore, we
do not have confidence in their ability to certify compliance under a
program that would allow them leeway to produce some vehicles to a
higher FEL now but sell vehicles with lower FELs later, such that they
were able to comply with the fleet average standard. We also cannot
reasonably assume that an ICI that certifies and produces vehicles one
year, would certify or even be in business the next. Consequently, we
propose that ICIs not be allowed to utilize the deficit carryforward
provisions of the proposed ABT program.

VII. Proposed Gasoline Benzene Control Program

A. Overview of Today's Proposed Fuel Control Program

As discussed in sections I, IV, and V above, people experience
elevated risk of cancer and other health effects as a result of
inhalation of air toxics. Mobile sources are responsible for a
significant portion of this risk. As required by section 202(l) of the
Clean Air Act, EPA has evaluated options to reduce MSAT emissions by
setting standards for motor vehicle fuel. We have determined that there
are fuel-related technologies available to feasibly reduce MSAT
emissions and that these reductions are achievable, considering cost,
energy, and other factors. These feasible reductions would be in
addition to those resulting from actions taken by the industry in
response to the earlier fuel-related MSAT programs described in section
V above. Accordingly, we believe a fuel control program is necessary
and appropriate to reduce air toxics emissions from motor vehicles to
the greatest extent achievable (in addition to the programs proposed
elsewhere in this notice to reduce MSAT emissions by changes to
gasoline-powered motor vehicles and gas cans). This section of the
preamble describes our proposed fuel control program.
The section begins with a detailed description of today's proposed
program. In summary, we propose that beginning January 1, 2011,
refiners would meet an average gasoline benzene content standard of
0.62% by volume on all their gasoline (reformulated and conventional)
nationwide.\224\ We also propose that refiners could generate benzene
credits and use or sell them as a part of a nationwide averaging,
banking, and trading (ABT) program.

[[Page 15859]]

We believe that the proposed benzene standard, combined with the
proposed ABT program, would result in the largest feasible overall
reductions in benzene emissions of any potential fuel-based MSAT
control program. Finally, as an aspect of achieving the greatest
emission reductions, we also propose special compliance flexibility for
approved small refiners.
---------------------------------------------------------------------------

\224\ The State of California has a similar benzene standard and
gasoline sold there is not covered by this proposal. For more
information, see California Code of Regulations, Title 13 Section 2262.
---------------------------------------------------------------------------

This section then describes in detail how we arrived at the
proposed program. We discuss a range of potential approaches to
reducing MSATs through changes in fuel, concluding that benzene
emissions would be significantly more responsive to fuel changes than
emissions of any other fuel-related MSAT. This is followed by
discussion of alternate methods of reducing benzene emissions,
resulting in the proposed approach of directly controlling benzene
content. We also discuss how we arrived at the proposed level of 0.62
volume percent (vol%) for the benzene standard. We discuss why we
believe that incorporating the proposed ABT program would be crucial
for the effectiveness of the overall benzene control program and
describe how the system would work. Finally, we review the
recommendations of the special panel that was convened to assess the
potential for disproportionate impacts of the proposed program on small
refiners, and present our reasoning for the special small refiner
provisions we are proposing today.
Today's proposed action would fulfill several statutory and
regulatory goals for gasoline-related MSAT emissions, which are
discussed in more detail in this section. The program would meet our
commitment in the MSAT1 program to consider further MSAT control. The
program would also allow EPA to streamline the regulatory provisions
for the air toxics performance requirements of the reformulated
gasoline (RFG) and Anti-dumping programs. The expected levels of
benzene control by individual refiners under this proposal, combined
with other gasoline controls such as sulfur, RVP, and VOC controls,
mean that compliance with these provisions is expected to lead to
compliance with the annual average requirements for benzene and toxics
performance for RFG and the annual average Anti-dumping toxics
performance for conventional gasoline. EPA is therefore proposing that
upon full implementation in 2011, the regulatory provisions for the
benzene control program would become the single regulatory mechanism
used to implement these RFG and Anti-dumping annual average toxics
requirements, replacing the current RFG and Anti-dumping annual average
provisions (although the 1.3 vol% benzene cap would still apply for
RFG). The proposed benzene control program would also replace the MSAT1
requirements. In addition, the program would satisfy certain fuel MSAT
conditions of the Energy Policy Act of 2005. By consciously designing
this proposed program to address these separate but related goals, we
would significantly consolidate and simplify the existing national
fuel-related MSAT regulatory program.
Finally, this section concludes with a detailed summary of our
assessment of the technological feasibility for different types of
refineries, and the refining industry as a whole, to meet the program
as proposed. We request general and specific comment on all aspects of
the proposed program, and we request that comments include supporting
data whenever possible.

B. Description of the Proposed Fuel Control Program

Today's proposed program has three main components, the development
of each of which is further described later in this section:

--A gasoline benzene content standard. We propose that an annual
average gasoline benzene standard of 0.62 vol% be implemented beginning
January 1, 2011. This single standard would apply to all gasoline, both
reformulated (RFG) and conventional (CG) nationwide (except for gasoline
sold in California, which is already covered by a similar state program).
--An averaging, banking, and trading (ABT) program. From 2007-2010
refiners could generate benzene credits by taking early steps to reduce
gasoline benzene levels. Beginning in 2011 and continuing indefinitely,
refiners could generate credits by producing gasoline with benzene
levels below the 0.62% average standard. Refiners could apply the
credits towards company compliance, ``bank'' the credits for later use,
or transfer (``trade'') them to other refiners nationwide (outside of
California) under the proposed program. Under this program, refiners
could use credits to achieve compliance with the benzene content
standard, regardless of their actual gasoline benzene levels.\225\
---------------------------------------------------------------------------

\225\ However, the per-gallon benzene cap (1.3 vol%) in the RFG
program would continue to apply separately.
---------------------------------------------------------------------------

--Hardship provisions. Refiners approved as ``small refiners'' would
have access to special temporary relief provisions. In addition, any
refiner facing extreme unforeseen circumstances or extreme hardship
circumstances could apply for similar temporary relief.

C. Development of the Proposed Gasoline Benzene Standard

EPA believes that benzene control is by far the most effective
fuel-based means of achieving MSAT emissions control, as described in
this section. There are other options that can target individual MSATs
or reduce overall VOCs and thereby reduce MSATs as well. We have
evaluated these other options, as discussed below, and our analysis
indicates that the potential MSAT reductions would be considerably
smaller and more expensive.
1. Why Are We Focusing on Controlling Benzene Emissions?
We considered controlling emissions of several MSATs through
changes to fuel parameters. There are only a limited number of MSATs
that are affected through fuel changes, each of which we discuss below.
For several reasons, we have concluded that the most effective and
appropriate means of reducing fuel-related MSATs is to reduce the
benzene emissions attributable to gasoline.
Benzene emissions can be reduced much more significantly through
fuel changes than can emissions of other MSATs. Relatively small
changes in gasoline can result in very significant reductions in
benzene emissions. This relative responsiveness of benzene emissions to
fuel controls (specifically to control of gasoline benzene content, as
discussed in the next section) is coupled with little negative impact
on other important characteristics of gasoline or refining processes. A
related and critical advantage of fuel control of benzene emissions, as
compared to fuel control of emissions of other MSATs as discussed
below, is that controlling benzene emissions does not significantly
increase emissions of other MSATs.\226\
---------------------------------------------------------------------------

\226\ A key tool in evaluating the potential for fuel changes to
affect MSAT emissions is EPA's Complex Model. This model relates
changes in gasoline parameters with emissions of specific MSATs and
was developed for refiners and EPA to assess compliance with the
RFG, Anti-dumping, and MSAT1 programs. (See section V.D.1 above.)
Given a set of gasoline parameters, it estimates the emissions of an
average vehicle based on a large set of fuel effects data. We
further discuss the Complex Model, as well as other sources of
information the relationships between fuel changes and MSAT
emissions, in chapter 6 of the RIA.
---------------------------------------------------------------------------

In determining an appropriate approach to fuel-related MSAT
control, a key consideration was octane value.

[[Page 15860]]

Among potential approaches to fuel-related MSAT emission reduction,
only benzene emission reduction can avoid major losses in octane value
and the negative cost and environmental consequences discussed below of
replacing that lost octane value. Finished gasoline must meet minimum
specifications for octane value; these specifications are tied to the
operational needs of motor vehicles. Thus, refiners must be keenly
aware of how any changes in gasoline production might reduce the octane
value of their fuel, what approaches to restore the octane value might
be available, and the costs in material and operational changes of any
selected approach.
There are a limited number of approaches refiners have at their
disposal to restore gasoline octane value lost through control of MSAT
emissions. These approaches vary in their economics and effectiveness,
and their availability may be limited by the specific configuration of
a given refinery. However, all methods of replacing octane value have
cost implications, and as shown in the next paragraph, air toxics
implications as well.
In the case of changes in gasoline production that are intended to
reduce MSAT emissions, it is also important to consider whether
restoring any lost octane might itself significantly increase other
MSAT emissions. Some methods of replacing octane value can increase
other MSATs. For example, increasing aromatics would increase benzene
emissions; adding MTBE would increase formaldehyde emissions; and
adding ethanol would increase acetaldehyde emissions. Given the very
large MSAT emission reduction associated with benzene control, these
impacts on other MSATs are relatively insignificant. However, in the
case of changes in other fuel qualities (e.g., aromatics control), the
relative impacts on other MSATs would be greater.
We encourage comment on our decision to propose a program that
directly controls gasoline benzene content, including comments on each
of the alternate approaches to MSAT control discussed in the following
paragraphs.
a. Other MSAT Emissions
As alternatives to the proposed program focusing on benzene
emission reductions, we considered other MSATs that are responsive to
fuel-based emission control. Each of these is discussed next.
Polycyclic Organic Matter, or POM, is composed of a number of
combustion products of gasoline. According to the Complex Model, POM
emissions are a function of exhaust VOC. Several fuel parameters
including volatility and sulfur content affect VOC emissions. As
discussed below, little data exists about the potential impacts of
changes in gasoline volatility and sulfur content on VOC, and thus POM,
emissions from new Tier 2-compliant vehicles. In any event, because POM
is only a tiny fraction of vehicle VOC emissions, we expect that
further changes in these fuel parameters would have only small effects
on POM. As a result, we are not proposing fuel controls to address POM
emissions in today's action.
Emissions of the compound 1,3-butadiene can be reduced by reducing
the olefin content of gasoline. However, olefin reduction yields
relatively small reductions in 1,3-butadiene and can increase VOC
emissions. In addition, olefin reduction significantly affects octane,
with the negative cost and MSAT emissions consequences of octane
replacement. We are thus not proposing to address 1,3-butadiene
emissions through fuel changes.
Emissions of the compound formaldehyde can only be effectively
reduced by reducing use of the octane enhancer methyl tertiary butyl
ether (MTBE). This is because formaldehyde increases significantly as a
combustion product when MTBE is added to gasoline. Formaldehyde also
increases to a lesser extent when ethanol is added to gasoline, as
described below. For a number of years, MTBE has been used as a cost-
effective way to meet mandated fuel oxygenate requirements and to boost
octane. In recent years, many states have banned the use of MTBE
because it has leaked from storage tanks and caused significant
groundwater contamination. More recently, in the wake of the removal of
the oxygenate requirement in the Energy Policy Act of 2005, many
refiners are taking action to remove MTBE from their gasoline as soon
as possible. As a result, MTBE use and the resulting formaldehyde
emissions are expected to continue to decline, and no additional
federal action appears warranted at this time.
The compound acetaldehyde is a combustion product of gasoline when
ethanol is added. Controlling acetaldehyde would require reductions in
the use of ethanol as a gasoline additive. However, the Energy Policy
Act of 2005 (section 1501) includes a renewable fuels program that will
increase use of ethanol in gasoline nationwide. That Act requires a
study of the Act's impacts on public health, air quality, and water
resources. We accordingly intend to defer further evaluation of
acetaldehyde emissions to the analyses associated with the Energy
Policy Act.
b. MSAT Emission Reductions Through Lowering Gasoline Volatility or
Sulfur Content
We also considered two approaches to fuel-related MSAT control that
would involve increasing the stringency of two existing emission
control programs. Both were originally promulgated primarily to address
ozone but also have the effect of reducing some MSAT emissions by
virtue of their control of VOC emissions. As explained in section V,
the Tier 2 program included the pairing of lower vehicle emissions
standards with large reductions in gasoline sulfur levels. The low
sulfur fuel helped enable development of more advanced catalytic
aftertreatment systems needed to meet the stringent tailpipe standards.
These actions will result in large reductions of VOC, NO_X,
and air toxics emissions. In development of today's proposal, we
considered whether further reductions in fuel sulfur would bring
significant additional reductions in MSAT emissions.
The second program considered for additional stringency was the
gasoline volatility program, which was implemented in 1989 to address
evaporative VOC emissions from gasoline vehicles. Reducing the
volatility of gasoline can reduce evaporative VOC emissions as well as
exhaust emissions. Evaporative VOC emissions include benzene. As a
result, in developing this proposal we have considered whether further
reductions in gasoline volatility may be effective in further reducing
MSAT emissions.
In the cases of both further reductions in RVP and sulfur
reductions below the current 30 ppm standard, the available data is not
sufficient to conclude that additional control of either would be a
valuable MSAT emission reduction strategy. Historic data suggest that
reducing both RVP and sulfur content would reduce overall VOC emissions
from vehicles, in turn reducing both MSATs and ozone formation.
However, vehicles complying with the stringent new Tier 2 emission
standards have dramatically lower VOC emissions than earlier vehicles.
Furthermore, it is likely that VOC emissions for these vehicles would
react differently to RVP and sulfur control than older vehicles, as new
catalysts and control systems may have more or less sensitivity to
these variables. Since the dominant effect on MSAT emissions of
changing these fuel parameters is through their impact on total VOC
mass, it is not possible to

[[Page 15861]]

properly assess the impact of changes in these fuel parameters on MSAT
emissions without additional data. We have begun collecting data on
some of these new vehicles, but more work will be required before we
can draw conclusions about the effectiveness of these fuel controls in
reducing MSAT emissions. Therefore, we are not proposing additional
control of gasoline volatility or sulfur at this time, but will
continue to evaluate them for possible future action. We request
comments on these potential fuel controls as emission reduction
strategies, in particular for MSAT emissions, including any data that
does or does not support the effectiveness of such controls.
i. Gasoline Sulfur Content
In general, reducing gasoline sulfur levels increases the
effectiveness of the catalytic converter at destroying unburned fuel
and other VOCs in vehicle exhaust. Catalytic converters contain a
variety of physical and chemical structures that act as reaction sites
for conversion of raw exhaust gases into less harmful ones before they
are emitted into the atmosphere. Over time, sulfur compounds in the
exhaust gases interfere with these processes, making the catalyst less
effective under normal driving conditions.\227\ Since many air toxics
are part of the exhaust VOCs, reduction of fuel sulfur would be
expected to reduce air toxics emissions. As with the Tier 2 program,
however, desulfurizing gasoline further would reduce gasoline octane.
Most options for recovering this lost octane (e.g., increasing
aromatics) would result in some offsetting MSAT emissions increases.
---------------------------------------------------------------------------

\227\ For further discussion on sulfur effects on emissions, see
the Tier 2 Regulatory Impact Analysis, EPA 420-R-99-023.
---------------------------------------------------------------------------

EPA primarily uses two computer models for examining emissions
impacts when considering changes in fuel properties: the Complex Model
and the MOBILE model. The Complex Model (CM) was developed as a
compliance tool that refiners use to ensure their gasoline meets its
baseline requirements under the RFG, Anti-dumping, and MSAT1 programs.
Given a set of fuel parameters, it estimates the emissions of an
average vehicle using regression relationships drawn from a large set
of fuel effects data. The CM contains data on test fuels with sulfur
levels as low as 5 ppm, but is based on the Auto/Oil research programs
of the early 1990s, and reflects performance of vehicles on the road
during that time period. With a sulfur reduction from 30 ppm to 10 ppm
applied to average 2003 conventional gasoline, the CM projects a
decrease of approximately 1% for exhaust benzene, NO_X and CO.
MOBILE was developed to estimate aggregate emissions on a county,
state, or national scale. It uses a fuel effects dataset that includes
the CM dataset with some updates, along with driving data, to predict
emissions inventories of pollutants for a specified time period and
area of the country. MOBILE6.2 contains updates from a small number of
LEV and ULEV vehicles in addition to the CM dataset, but applies a
lower limit of 30 ppm to fuel sulfur content being modeled to avoid
extrapolation beyond the range of available emissions data.
Based primarily on the above models, the analyses done for the Tier
2 rulemaking suggested benzene emission reductions on the order of 9%
could be expected in 2020 as a result of the fuel sulfur reduction
expected from that program alone (the final Tier 2 program included low
sulfur gasoline as well as tightened vehicle standards).\228\ A recent
study done on vehicles meeting LEV, TLEV, and ULEV standards indicates
that sulfur reductions from 30 to 5 ppm may reduce NMHC by more than
10%, bringing similar reductions in air toxics.\229\ Additional
analyses done by EPA on sulfur reductions in this range suggest VOC
emission reductions on the order of 5% may be expected, with refining
costs estimated at about a half cent per gallon. Given these analyses
using available data, using sulfur reductions as air toxics control
alone would not be as cost-effective as other options in this proposal.
Further discussion of the feasibility and costs are available in
Chapters 6 and 9, respectively, of the RIA.
---------------------------------------------------------------------------

\228\ Tier 2 Regulatory Impact Analysis, EPA 420-R-99-023
\229\ AAM-Honda fuel effects study, 2000
---------------------------------------------------------------------------

Since our models do not reflect the significant improvements in
emissions control technology over the past decade, more fuel effects
studies are necessary on newest-technology vehicles before going
forward with sulfur control. A small cooperative test program is
currently underway between EPA and the Alliance of Automobile
Manufacturers to evaluate the effects of reducing sulfur below 10 ppm
on Tier 2 Bin 5 compliant vehicles.
In addition to potential air toxics reductions from adjustment of
gasoline sulfur to 10 ppm, reducing sulfur may also provide significant
VOC and NO_X emission reductions. These emission reductions
may be important for states in complying with the National Ambient Air
Quality Standards (NAAQS) for ozone. Since the implementation of the
RFG program, several states and localities have made their own unique
fuel property requirements in an effort to further improve air
quality.\230\ As a result, by summer 2004 the gasoline distribution and
marketing system in the U.S. had to differentiate between more than 12
different fuel specifications, when storing and shipping fuels between
refineries, pipelines, terminals, and retail locations. These unique
fuels decrease nationwide fungibility of gasoline, which can lead to
local supply problems and amplify price
fluctuations.231, 232 In addition to the existing state fuel
programs, we are aware of a number of other states considering new
programs (although in the context of the recently enacted Energy Policy
Act it is unclear what will occur). While the timeline for state action
on new fuel formulations could be prior to any nationwide ultra-low
sulfur standard, implementation of such a standard could help diminish
issues related to small-market fuel programs in the long term.
---------------------------------------------------------------------------

\230\ These changes have focused almost exclusively on
additional RVP control, with just one program also controlling
sulfur to 30 ppm earlier than required by EPA.
\231\ EPA, Study of Unique Gasoline Fuel Blends (``Boutique
Fuels''), Effects on Fuel Supply and Distribution and Potential
Improvements, EPA420-P-01-004
\232\ GAO, Special Gasoline Blends Reduce Emissions and Improve
Air Quality, but Complicate Supply and Contribute to Higher Prices,
GAO-05-421
---------------------------------------------------------------------------

From the perspective of gasoline production, reducing sulfur to
ultra-low levels does not happen completely independently of other fuel
parameters. The emissions benefits of further sulfur reduction gained
in vehicle aftertreatment may be offset by unintended changes in other
gasoline properties. The refining process modifications required to
bring sulfur to ultra-low levels begin to have a stronger effect on
other components of gasoline, such as olefins (the effect of which is
discussed in the previous section). These impacts must be further
evaluated before moving forward with a proposal of additional sulfur
reductions for the purpose of air toxics reduction. These issues are
also discussed in more detail in Chapter 6 of the RIA.
Refiners with whom we have met have generally expressed disapproval
of further sulfur control. The Tier 2 gasoline sulfur program requires
refiners to meet an average standard of 30 ppm. In response many have
invested in and brought online desulfurization units, which would not
have the capacity to

[[Page 15862]]

reach a new, lower standard of 10 ppm in many cases. Modifications
would have to be made to units that have recently been installed to
comply with the current gasoline sulfur requirements. In some cases
these units might have to be replaced with new units. EPA requests
comments on the magnitude of the impact of a new, lower sulfur
standard, including the potential effect on refiners that have recently
installed desulfurization units.
On the automotive side, sulfur reduction may encourage further
development of lean-burn or direct-injection gasoline technology.
Leaner combustion of gasoline results in greater fuel economy and less
VOC and carbon dioxide emissions, but generally produces more engine-
out nitrogen oxides. Reducing fuel sulfur to 10 ppm would improve
feasibility and reduce cost of next-generation aftertreatment designed
to control these higher levels of nitrogen oxides. EPA will continue to
evaluate further gasoline sulfur reductions, and seeks comment on it,
especially with data supporting or opposing such action.
ii. Gasoline Vapor Pressure
According to the Complex Model and the MOBILE model, reducing fuel
vapor pressure reduces evaporative as well as exhaust VOC emissions.
Reducing VOC emissions in turn reduces MSAT emissions. A portion of
this MSAT emission decrease through VOC control would likely be offset
through an increase in the relative concentration of MSAT emissions. As
volatility is decreased, non-aromatic compounds are removed from the
gasoline, increasing the concentration of aromatics. Furthermore, these
non-aromatic compounds are higher in octane, which would have to be
offset--perhaps with still further increases in aromatics. Such
increases in aromatics would lead to an increase in the relative
concentration of benzene in VOC emissions. However, since changing
vapor pressure has an effect on evaporative emissions, reducing vapor
pressure can also reduce evaporative benzene from stationary sources
related to gasoline distribution and marketing. Moreover, reducing
overall VOC emissions reduces ground level ozone in urban areas, which
itself has a significant impact on health and welfare.
Currently, in reformulated gasoline (RFG) areas, fuel is limited to
roughly 7.0 psi Reid vapor pressure (RVP) in the summer season in order
to meet the VOC performance standard. Additional vapor pressure
controls considered for this proposal would regulate RVP levels to 7.0
or 7.8 in some conventional gasoline (CG) ozone nonattainment areas,
resulting in an impacted volume of gasoline equal to about 50% of that
of current federal RFG. Further details of these analyses are covered
in Chapter 6 of the RIA.
As with the sulfur analyses above, EPA also uses the Complex Model
and MOBILE to estimate emissions impacts of changes in gasoline vapor
pressure. In terms of the fuel parameter itself, this process is
somewhat simpler than modeling sulfur effects since the range of vapor
pressures useful in conventional vehicles has been well-defined for a
number of years and is not expected to change. However, parallel to the
arguments made above for sulfur, data on the effects of RVP changes on
air toxics in these models is dated and does not represent newest
technology. Since our models do not reflect improvements in emissions
control technology for the Tier 2 program, more fuel effects studies
must be carried out before making decisions on further gasoline vapor
pressure controls. The cooperative test program between EPA and the
Alliance of Automobile Manufacturers described above is also examining
some of the effects of changes in RVP.
Looking beyond emissions benefits, more stringent national vapor
pressure standards could also help avoid additional small market
(``boutique'') fuels. Several states and localities have adopted their
own seasonal requirements for vapor pressure in an effort to improve
air quality, contributing to constraints on gasoline supply and
potential for price volatility.233 234
---------------------------------------------------------------------------

\233\ EPA, Study of Unique Gasoline Fuel Blends (``Boutique
Fuels''), Effects on Fuel Supply and Distribution and Potential
Improvement, EPA420-P-01-004.
\234\ GAO, Special Gasoline Blends Reduce Emissions and Improve
Air Quality, but Complicate Supply and Contribute to Higher Prices,
GAO-05-421.
---------------------------------------------------------------------------

Feedback from refiners on further volatility control has
highlighted concerns with the summer-winter butane balance and
resulting potentially adverse supply implications. Currently, refiners
who produce large quantities of RFG must remove a significant amount of
the light-end components from their fuel in the summer to meet the
vapor pressure specifications. These light components, primarily
butanes, are often stored and then blended back into gasoline in the
winter when higher fuel vapor pressures are needed for drivability
reasons. Several refiners have indicated that a new rule adding a
number of reduced RVP areas would cause the amount of butanes removed
in summer to exceed what is useable in winter, resulting in a net loss
of volume from the annual pool and a need to make up supply at
additional expense. EPA will continue to evaluate further gasoline
volatility reductions, and seeks comment on it, especially with data
supporting or opposing such action.
c. Toxics Performance Standard
While we are not proposing it, we considered and are seeking
comment on the merits of expressing the standard as an air toxics
performance standard rather than as a benzene content standard. Such a
standard would be analogous to the current MSAT1 standard, but more
stringent and with an ABT component. In theory, a toxics performance
standard could provide broader environmental benefits by addressing
other toxics in addition to benzene. However, because controlling
benzene is more cost-effective than controlling emissions of other
MSATs, refiners are unlikely to reduce emissions of other MSATs whether
or not the standard is in the form of a toxics performance standard or
a benzene content standard. Setting a toxics performance standard at an
appropriate level also requires us to predict future changes in fuel
properties in addition to benzene, and to be able to establish as
precisely as possible the effects of those fuel properties on emissions
of several MSATs. In addition, a toxics emission performance standard
is more complex to implement and enforce than a benzene content
standard. For all of these reasons, as discussed more fully below, we
believe a benzene content standard offers more certain environmental
results and less complexity. However, we seek comment on the overall
merits of an air toxics performance standard, including comments
specifically on the tradeoff between the complexity of complying with a
performance standard and the additional environmental benefits it could
provide.
Based on our analysis for this proposal, fuel benzene control is by
far the most effective and cost-effective means of achieving MSAT
emission reductions. This is consistent with our experience with the
MSAT1 and other air toxics control programs, which have shown that even
when refiners have the flexibility to choose among different fuel
changes to achieve MSAT control, reduction in benzene content is the
predominant choice. Only when other fuel changes that impact MSAT
emission performance are mandated (e.g., sulfur control, oxygenate use)
have refiners made fuel changes other than benzene content to control MSAT

[[Page 15863]]

emissions. As a result, even if we were to express the proposed
standard as an air toxics performance standard rather than a benzene
content standard, we would expect the outcome to be the same--benzene
content control with corresponding benzene emission reductions and no
changes in other MSAT emissions. Our analysis of the feasibility and
cost of the program would be identical as well. If future fuel
parameters are significantly different than we have projected in this
analysis such that emissions of other MSATs decrease, then a toxic
performance standard would result in less benzene control than would be
achieved by the benzene content standard we propose today, with a
corresponding overall reduction in cost. If future fuel parameters are
significantly different such that emissions of other MSATs increase,
then refiners would need to reduce benzene content to levels that are
not feasible considering cost, but overall toxics performance would be
maintained.
If we were to set an air toxics performance standard, the accuracy
of the model used in estimating the real world effects of the many
different fuel parameters on MSAT emissions also becomes of critical
importance. To the extent fuel changes are projected to result in air
toxics emission reductions that are not in fact borne out in-use, then
the standard will have less benefit. There was a great deal of work
done in the early 1990's to develop the Complex Model for the
reformulated gasoline program. It estimates VOC, NO_X, and
certain MSAT emissions (benzene, 1,3-butadiene, formaldehyde,
acetaldehyde, and POM) as a function of eight fuel properties (RVP,
oxygen, aromatics, benzene, olefins, sulfur, E200, and E300) for 1990
technology vehicles. However, a similar set of comprehensive data does
not yet exist for new Tier 2 vehicles. Some of the fuel effects that
were found to be statistically significant in the Complex Model may not
be significant for Tier 2 vehicles (e.g., distillation properties).
Others that impacted MSAT emissions primarily through their impact on
VOC emissions may be of much less importance, due to the much lower VOC
emissions of Tier 2 vehicles.\235\ To the extent that the Complex Model
gives air toxics credit for fuel changes that are later found to be
much smaller or not valid at all, a toxics performance standard could
result in less fuel benzene control and less in-use MSAT control. Of
all the fuel changes from past modeling, we would have the greatest
confidence that the benzene relationships are unlikely to change
significantly. This is due to the direct relationship between benzene
fuel content and benzene evaporative and exhaust emissions, and due to
the magnitude of these impacts. Thus, we would have the greatest
confidence that the MSAT emission reductions projected from a fuel
benzene content standard will be realized in-use.
---------------------------------------------------------------------------

\235\ This is one reason why the Energy Policy Act of 2005
requires EPA to create an updated gasoline emissions model by 2009.
---------------------------------------------------------------------------

In addition, if we were to set an air toxics performance standard,
it would be important to have a clear understanding of the changes in
fuel properties anticipated in the future independent of today's
proposal. Significant changes in the composition of gasoline are
anticipated over the next several years as a result of the Energy
Policy Act of 2005 (EPAct). MTBE is being removed from gasoline,
ethanol use is increasing dramatically, and the oxygenate mandate for
RFG is being eliminated. To the extent that these changes would result
in reductions in modeled MSAT emission performance automatically, then
refiners could comply with an air toxics performance standard with less
benzene control than would be achieved under today's proposed benzene
standard, and with lower overall costs. Conversely, to the extent that
these changes would result in increases in modeled MSAT emission
performance, an air toxics performance standard would require refiners
to take additional measures to maintain overall MSAT performance, but
these measures may not be cost-effective.
Although a toxics performance standard could theoretically give
refiners more flexibility than a program focusing only on benzene
emissions, we do not believe that such flexibility would be meaningful
in actual practice. As discussed above, in order to comply with a new
total MSAT standard, we expect that refiners would rely almost
exclusively on benzene control. However, if their emission performance
for other MSATs changed in the future (due to such factors as changes
in oxygenate use, octane needs, or crude oil quality), refiners could
find themselves unable to maintain overall MSAT performance using cost-
effective controls.
For all these reasons, we are not proposing to address fuel-related
MSAT emissions with a toxics performance standard, but we seek comment
on this option.\236\ We also seek comment on the merits of applying an
air toxics performance standard in addition to a fuel benzene content
standard, and how such a dual standard could be implemented. From a
theoretical standpoint, this dual standard might serve as a backstop to
ensure overall toxics performance is maintained. However, it is not
clear how such an approach could be realistically implemented,
especially in the context of ABT programs that apply to both.
---------------------------------------------------------------------------

\236\ As explained further in section VII.C.5 below, based on
the use of the currently available models, the proposed rule would
result in greater overall reduction of air toxics from all gasoline
than the current MSAT 1 program, and (consistent with section
1504(b)(2) of the EPact) greater overall reductions of air toxics
from reformulated gasoline than would be obtained under amended
section 211(k)(1)(B) as well.
---------------------------------------------------------------------------

d. Diesel Fuel Changes
We are also not proposing today to reduce MSATs by changing diesel
fuel. The existing major diesel fuel sulfur programs being implemented
in the next few years for highway and nonroad diesel fuel will have a
very large impact on reducing MSAT emissions `` specifically diesel
particulate matter and exhaust organic gases. We have found in the on-
highway diesel engine rulemaking that these are the greatest reductions
achievable and reiterate that finding here. (See also section V.D.1.f
above.) We are not aware of other changes to diesel fuel that could
have a significant effect on emissions of any other MSATs. We welcome
comment on our decision to focus this proposed program exclusively on
changes to gasoline.
2. Why Are We Proposing To Control Benzene Emissions By Controlling
Gasoline Benzene Content?
In the previous section, we describe how we decided to focus
today's proposed fuel program on gasoline benzene emissions. This
section describes our decision to propose to reduce benzene emissions
through a gasoline benzene content standard. We also describe our
consideration of two other potential approaches to reducing benzene
emissions, both of which would indirectly reduce gasoline benzene
content: a standard to control the gasoline content of all aromatic
compounds; and a standard to control benzene emissions.
a. Benzene Content Standard
For several reasons we have decided that a benzene content standard
would be the most cost-effective and most certain way to reduce
gasoline benzene emissions (and thereby MSAT emissions in general).
First, a small change in gasoline benzene content results in large
reductions in benzene emissions `` benzene typically

[[Page 15864]]

represents around 1 percent of gasoline, but this contributes about 25
percent of benzene exhaust and evaporative emissions.\237\ Second, we
have high confidence in the benzene emission reductions that would
result from fuel benzene control. Historical data across a range of
vehicles and engine types continues to support the relationship between
fuel benzene content and benzene emissions. Even if Tier 2 vehicles
react differently, the relationship is unlikely to change
significantly. Third, because a relatively small change in gasoline
properties is needed to achieve the desired result, reducing benzene
content does not have a large impact on octane value. Benzene itself
does contribute to the octane value of gasoline, but the small loss of
octane from reducing benzene content is much less than the octane loss
from reducing other aromatics for the same benzene emission effect, as
discussed below, and the consequences of refiners having to replace
that octane value are also much less. (This is why, as noted earlier,
we anticipate that refiners would seek to comply with any toxics
standard by reducing benzene levels in any case.) Fourth, we believe
that a direct benzene content standard would best ensure real benzene
emission reductions, including both exhaust and evaporative benzene
emissions. We discuss this conclusion below, in the context of the
potential alternative of a benzene emission standard.
---------------------------------------------------------------------------

\237\ Based on the Complex Model.
---------------------------------------------------------------------------

b. Gasoline Aromatics Content Standard
Because benzene emissions are formed from benzene and other
aromatics that are present in gasoline, we considered a standard that
would limit the aromatics content of gasoline. However, we believe that
reducing benzene emissions through a more general reduction in gasoline
aromatics content would be much less cost-effective than direct benzene
reduction. Non-benzene aromatics account for on average about 30
percent of gasoline (typically ranging between about 20 percent and 40
percent), and this fraction contributes about 30 percent of benzene
emissions. In contrast, benzene only makes up about 1 percent of
gasoline but is responsible for about 25 percent of benzene emissions.
The remaining benzene emissions are formed from other compounds. Based
on the Complex Model, it would require about a 20 percent reduction in
non-benzene aromatics to achieve the same benzene emission reductions
as the proposed benzene content standard. As we discussed earlier, a
major consequence of removing a significant amount of the aromatics in
gasoline is the need to replace the large loss in octane value. As a
result, it is much more costly for refiners to reduce benzene emissions
through aromatics control than through benzene control. We have not
evaluated the cost of aromatics control recently, but when we did so
for the RFG rule in the early 1990s, the cost was about 5 times more to
achieve the same benzene reduction through aromatics control than
through benzene control.\238\ In recent years a variety of factors have
reduced the use of MTBE as an octane booster; we expect that this trend
will raise the relative cost of aromatics control even further.
---------------------------------------------------------------------------

\238\ Final Regulatory Impact Analysis for Reformulated
Gasoline, AEPA420-R-93-017, December 1993.
---------------------------------------------------------------------------

In addition, aromatics reductions would have to be offset with
other high-octane compounds, such as ethanol and ethers (e.g., ETBE and
MTBE). Increasing other high-octane compounds tends to significantly
increase other air toxics emissions (like acetaldehyde or
formaldehyde). Consequently, the benzene emission reductions would be
substantially offset by increases in other toxics. For these reasons,
aromatics control has historically only been cost-effective for
refiners when other requirements are placed on them, such as state or
federal oxygenate mandates that also serve to boost octane value. For
this same reason, we anticipate that further aromatics reductions will
occur as a result of the near doubling of the use of ethanol in
gasoline due to the renewable fuels standard contained in the EPAct.
Given a mandate for ethanol use and the cost associated with it,
refiners can reduce their refining costs by further reducing aromatics.
Aromatics control would also affect other recent fuel control
programs. For example, many refineries depend on the reforming process
that produces aromatics to also supply much or all of the hydrogen
needed for gasoline and diesel desulfurization processes. Reducing
aromatics thus would indirectly reduce hydrogen supply, which would
then likely require refiners to either purchase hydrogen or build
hydrogen production facilities.
At the same time, although it would not be constrained, we do not
believe that in the absence of aromatics control, refiners would be
likely to increase gasoline aromatics content in the future. Aromatics
are a relatively valuable gasoline component, and refiners are
generally careful not to make changes that would increase aromatics
content more than is needed for octane purposes. In addition, as
mentioned previously, the Renewable Fuel Standard that will be
promulgated under the new Energy Policy Act will, by boosting ethanol
use, increase the octane of the gasoline pool. We expect that this, in
turn, will prompt refiners to reduce their use of aromatics for octane
enhancement. Also, higher gasoline prices recently have reduced the
demand for premium grade gasoline, which generally has higher aromatics
levels. To the extent that this trend continues, we expect that it will
tend to further reduce the levels of aromatics in the overall gasoline pool.
For all of these reasons, we believe that reducing benzene
emissions through a benzene content standard would be much superior to
doing so through an aromatics content standard. However, there may be
other benefits associated with aromatics control in addition to benzene
emissions. EPA is working to improve its understanding of the effect of
mobile source emissions on ambient PM, especially secondary PM. For
example, there is limited data that suggest that aromatic compounds
(toluene, xylene, and benzene) react photochemically in the atmosphere
to form secondary particulate matter (in the form of secondary organic
aerosol (SOA)), although our current modeling tools do not fully
reflect this. One caveat regarding this work is that a large number of
gaseous hydrocarbons emitted into the atmosphere having the potential
to form SOA have not yet been studied in this way. It is possible that
hydrocarbons which have not yet been studied produce some of the SOA
species which are being used as tracers for other gaseous hydrocarbons.
This means that the current interpretation of the available studies may
over-estimate the amount of SOA formation in the atmosphere. We seek
comment on the potential benefits, costs, and other implications of
aromatics control for consideration in the future.
c. Benzene Emission Standard
In addition to the benzene or aromatics fuel content standards
discussed above, we have considered reducing benzene emissions through
a benzene emission standard. The primary argument for such an approach
is that it would focus on the environmental outcome we are interested
in `` reduced benzene emissions `` while providing refiners some
flexibility in how that goal was met.
In order to fully discuss this option, it is useful to clarify how such a

[[Page 15865]]

benzene emission standard would be implemented. Instead of directly
measuring gasoline content to determine compliance, as would be the
case with a benzene (or aromatics) content standard, compliance would
be determined using EPA's Complex Model or an updated version of it.
Several parameters of a refiner's gasoline (including benzene and
aromatics content) would be used as inputs into the model. Based on
these and other assumed properties of the gasoline, the model would
estimate the expected level of benzene emissions from that gasoline
formulation.
As compared to a program based on the direct measurement of benzene
content in gasoline, we believe that one relying on modeled estimates
of benzene emissions would be difficult to set today. As with the
toxics performance standard we considered above, gasoline parameters
and their effects on MSAT emissions will be changing in the future due
to the Energy Policy Act, changes in crude oil supplies, and perhaps
other unknown factors. In addition, the effects of fuel changes on MSAT
emissions from the new Tier 2 vehicles now entering the light-duty
fleet are poorly represented in our modeling. Thus, it would be
difficult to accurately predict future gasoline parameters and set an
appropriate benzene emission standard that ensured the greatest
emission reduction achievable, especially a standard that could remain
stable for a number of years. As benzene content has been and is sure
to remain by far the most important fuel parameter in estimating
benzene emissions, a benzene content standard provides greater
assurance of actual benzene emission reduction in-use.
Even if it were practical to set a long-term benzene emission
standard, such an approach would be problematic for other reasons. As
we have stated, the only significant option for reducing benzene
emissions other than reducing benzene content is reducing aromatics
content. Since we do not believe that requiring control of gasoline
aromatics is appropriate at this time, a benzene emission standard
would not result in appreciably different emission reductions than
would result from a benzene content standard. However, given that
aromatics control is a less effective means of reducing benzene
emissions and has a more disruptive effect on octane values (as just
discussed), requiring more aromatics control could dramatically
increase the cost of compliance. Finally, although a benzene emission
standard might be assumed to offer additional flexibility to refiners,
we do not believe that such flexibility would actually exist. Faced
with a dependence on aromatics to meet octane requirements, and in some
cases to provide hydrogen supply for desulfurization of gasoline and
diesel fuel, we believe that refiners would choose benzene content
reduction over aromatics reductions even when they theoretically had
the choice to do otherwise. Experience with the MSAT1 emissions
performance standard has confirmed this. However, as mentioned
previously, gasoline parameters do change, octane requirements can
decrease, ethanol will supply additional octane, and therefore aromatic
reductions may occur in the future regardless. Were this to occur, a
benzene emission standard set today could allow benzene content to
increase in the future. Given the additional complexity and uncertainty
associated with a benzene emission standard, we have therefore elected
to propose a benzene content standard exclusively. We request comment
on this approach and on a benzene emission standard.
3. How Did We Select the Level of the Proposed Gasoline Benzene Content
Standard?
a. Current Gasoline Benzene Levels
In selecting an appropriate level for the proposed benzene content
standard, we began by evaluating the current status of the industry
regarding gasoline benzene. Benzene content varies widely among
refineries, depending on such factors as refinery configuration and
proximity to benzene markets. The national average benzene level was
1.6 vol% in 1990. Due to the 0.95 vol% requirement of the 1995 RFG
program, the introduction of gasoline oxygenate requirements, and other
factors, benzene levels have since declined. By 2003, RFG averaged 0.62
vol% benzene. (See section V.D.1 above.)
Benzene levels have also declined for CG over the same period, to
an average of 1.14 vol%. This is in part because when faced with
investing in new processes to comply with the RFG benzene standard,
some refiners found it economical to install more benzene extraction
capacity than was needed to meet the standard. As a result, in many
cases, these refiners have also controlled benzene from CG.
b. The Need for an Average Benzene Standard
Even before considering the level of the benzene content standard,
we first needed to consider the standard's potential form. A standard
for this purpose could be expressed as a per-gallon benzene limit,
which would ensure that no gasoline exceeded a specified benzene level.
In contrast, a benzene content standard could be expressed as a
flexible average level, allowing some of the existing variability in
current benzene levels to remain while reducing overall benzene levels.
For several reasons, it became clear that an average standard was the
most appropriate for this program.
As mentioned above, there is a great diversity in the benzene
content of gasoline currently produced at refineries across the
country. In 2003, the annual average benzene content of refineries
ranged nationally from under 0.5 vol% to above 3.5 vol%. This variation
among refineries is also reflected in large regional differences in
average gasoline benzene content, as illustrated below (Tables VII.C-2
and VII.F-1).
In addition to average benzene levels varying widely across
refineries and regions, per-gallon benzene levels for individual
batches produced by a refinery also vary dramatically depending on the
crude oil supply and the refinery streams used to produce a particular
batch. This variation occurs as a result of a wide range of day-to-day
decisions necessary in producing marketable gasoline within a refinery
on a continuous basis. We reviewed actual batch data for a typical
refinery producing both RFG and CG with an average benzene content of
1.6 vol% for all its gasoline, and batch benzene levels ranged from
under 0.1 to 3.0 vol% for CG. The range for RFG is typically narrower
due to the existing 1.3 vol% per gallon cap, but still shows
significant batch to batch fluctuations. Batches that refiners produce
with benzene higher than 1.3 vol% are marketed as CG.
We considered controlling benzene emissions with a fixed, per-
gallon benzene content standard to be met at all refineries. By capping
gasoline benzene content in this way, the program would ensure that all
gasoline nationwide would have benzene levels below the selected upper
limit. However, as we developed the rule, it became clear that with the
large variation in benzene levels among refineries and regions
(reflecting the variation in the economics of reducing benzene), a per-
gallon standard would have to be so high (to account for maximum,
legitimate potential variability) as to leave most refineries with
little or no need to reduce benzene. Moreover, the burden of the
national control program would fall almost entirely on the refineries
where the challenges of control would be greatest, and where the most
lead time would be

[[Page 15866]]

required for compliance. With many refineries able to comply without
making any changes, we do not believe such a program would represent
the greatest reduction feasible, as the Clean Air Act requires.
The typical fluctuations in benzene content among batches at
individual refineries, as discussed above, also indicate the need for
refiners to have a degree of flexibility in producing gasoline, as
would be provided by an average benzene standard. Restrictions on day-
to-day fluctuations would not significantly affect average benzene
levels, but would certainly increase costs as refiners invested in
avoiding occasionally higher benzene batches. We believe that allowing
refiners to average batches with fluctuating benzene over a year's
time, as we propose, would result in a more cost-effective program.
Most importantly, it is clear that with the incorporation of a
carefully-designed benzene credit averaging, banking, and trading (ABT)
program, a more stringent benzene standard would be feasible, and
implementation could occur earlier. Thus, we are proposing a 0.62 vol%
annual average standard to begin in 2011. Under the proposed ABT
program, refiners could generate early credits by making early
reduction efforts prior to 2011. Refiners would have an incentive to do
so, because the credits generated could be used to postpone more
expensive final investments in benzene control technology. In this way,
the ABT program would allow the economic burden of the benzene standard
to be more efficiently distributed among refiners and over time. The
proposed ABT program would result in lower benzene levels in all areas
of the country compared to today's levels, as described in more detail
below in section VII.D.
c. Potential Levels for the Average Benzene Standard
We evaluated a range of potential standards on a national refinery
annual average basis from 0.52 to 0.95 vol% benzene.\239\ Our refinery-
by-refinery model incorporates data on individual refineries whenever
possible and estimates the likely technological approaches that
refiners would choose for each refinery to comply with each potential
standard at the least cost. The model chooses among several
technological options that are the most common and effective methods
available to refiners to reduce gasoline benzene content. (Section
VII.F below and Chapter 6 of the RIA have more detailed discussions of
benzene reduction technologies).
---------------------------------------------------------------------------

\239\ For this evaluation we used both refinery linear
programming (LP) models and a refinery-by-refinery model developed
specifically for this rule.
---------------------------------------------------------------------------

All of the methods that we considered focus on reducing benzene
content in the reformate stream, which is the product of the reformer
unit. The role of the reformer unit is to increase gasoline octane,
which it does by generating aromatic compounds from simpler
hydrocarbons. Benzene is one of the aromatic compounds produced by the
reformer. Reformate accounts for 30-40% of gasoline volume and can
contain as much as 12% benzene. As a result, reformate contributes the
majority of the total benzene content of gasoline. For these reasons,
treatment of reformate is usually the most effective and economical
means of reducing benzene content. Several proven and commercially
available technologies exist for reducing benzene creation in the
reformer and removing it from the reformate product.
The least stringent standard we evaluated, a national average of
0.95 vol% benzene, would not require any changes at most refineries.
For the refineries where action would be needed, we project that most
could be brought into compliance by reducing creation of benzene in the
reformer using the simplest and least costly of the technology options
evaluated. We do not believe that a standard at this level would meet
the statutory requirements of section 202(l) of the Clean Air Act to
achieve the greatest reductions achievable considering cost and other
factors since, as discussed below, greater reductions are feasible at
reasonable cost, and without adverse energy or safety implications.
As the most stringent case, we evaluated a national average benzene
content standard of 0.52 vol%. Our analysis indicates that a standard
at this level would require all refiners to invest in the most
effective technologies used today that remove the benzene from their
reformate product streams (benzene saturation and benzene extraction,
as discussed below). If the ABT program were fully utilized (all
credits generated were used), we believe all refiners might comply with
this average standard. Because of the almost universal need for
refineries to use the most expensive reformate-based benzene control
technologies, we believe a standard of 0.52 vol% would be very
challenging economically for many refineries, and we believe that such
a standard would not be achievable taking costs into consideration, as
we are required to do under section 202(l). In addition, if, as appears
likely, ``perfect'' credit trading did not occur, some refiners would
have to use additional, more extreme approaches that would be even more
costly and would require more difficult compromises in the operation of
the refineries. (We discuss these technological and operational
approaches to benzene reduction in more detail in section VII.F below
and in Chapter 6 of the RIA.)
In 2003, the average benzene level in RFG was 0.62 vol%.\240\ We
believe an annual average benzene standard of 0.62 vol% applied to all
gasoline (both CG and RFG) would be feasible considering cost and other
factors. Furthermore, implementing an average benzene standard of 0.62
vol% would achieve several other important program goals. At this
level, the same benzene standard could be applied to both RFG and CG
nationwide, and our analysis shows that the RFG benzene reductions
already achieved by the industry to date would not be lost. We expect
that refiners currently producing RFG with benzene levels below 0.62
vol% would continue to be committed to producing low-benzene gasoline
based on prior investment in benzene extraction equipment or ABT credit
incentives. Additionally, as discussed below in VII.C.5, a gasoline
benzene standard of 0.62 vol% would achieve sufficient mobile source
air toxic reductions allowing this program to supersede the additional
MSAT requirements under EPAct. Finally, an average benzene standard
applied to both CG and RFG, would allow for a uniform nationwide ABT
program providing additional flexibility and reduced compliance costs
to refiners, resulting in the greatest achievable reductions within the
meaning of section 202(l).
---------------------------------------------------------------------------

\240\ Volume-weighted average benzene level based on January 1,
2003 to December 31, 2004 RFG batch reports.
---------------------------------------------------------------------------

At a national average standard of 0.62 vol%, we estimate that a
number of refiners would produce gasoline with significantly lower fuel
benzene levels, creating enough benzene credits to allow refiners in
less economically favorable positions to purchase these credits on an
on-going basis and use them for compliance purposes. We project that
further reductions would occur not only in CG, but also in RFG, despite
the fact that RFG is already averaging 0.62 vol%. As discussed in
section IX below and in Chapter 9 of the RIA, as the stringency is
pushed below 0.62 vol%, the overall program costs would begin to rise
more steeply. This is because in meeting a lower average standard,
there would be fewer

[[Page 15867]]

refineries able to comply at low cost, resulting in fewer credits being
generated. This in turn would require more investment among refiners
with higher costs of compliance.
We also considered a program that would apply separate benzene
content standards to RFG and CG. In the context of any nationwide ABT
program that allowed trading across both RFG and CG, separate standards
for these two gasoline pools would not be fundamentally different from
the proposed unified standard. The only impact would be to somewhat
change which refiners generated credits and which used credits, and to
what degree. For separate RFG and CG standards to have a meaningful
impact in comparison to today's proposed program, separate trading
programs for each of the two gasoline pools would be required. Our
modeling shows that without the credits generated by RFG producers in a
nationwide trading program, it would not be possible to set as
stringent a standard for CG. The higher-benzene refineries that would
most need credits to meet a stringent average standard are a subset of
refineries that produce CG. As a result, in a program with separate RFG
and CG pools, we would expect to set a slightly more stringent standard
for RFG alone, but we would need to set a substantially relaxed
standard for CG. The net result would be, at best, the same nationwide
average benzene reductions in the RFG and CG pools that would be
expected under a unified standard. However, there would be a clear risk
that the reduced generation of credits by lower-cost refineries would
lead to either a significant increase in the cost of the program
(because higher-cost refineries would need to make refinery changes
earlier) or the potential for fewer reductions through the process of
setting the levels for the separate CG and RFG standards. Conversely,
with a unified standard and nationwide ABT, we believe that the program
would achieve the maximum economical reduction in all areas and greater
overall benzene reduction over the CG and RFG pools.
In addition, we considered a somewhat less stringent national
average standard than the proposed 0.62 vol% (e.g., 0.65 or 0.70 vol%).
Such standards would still achieve significant benzene emission
reductions. However, we are concerned that a less stringent standard
would not satisfy our statutory obligation for the most stringent
standard feasible considering cost and other factors. Furthermore, such
standards would not allow us to accomplish several important
programmatic objectives. Given that the average benzene content of RFG
in 2003 was already 0.62 vol%, such higher standards would not provide
the certainty that the air toxics performance of RFG would decline in
the future. This would then trigger the provisions in the 2005 EPAct to
adjust the MSAT1 baseline for RFG. The only way of avoiding this
situation would be to maintain separate standards for RFG and CG where
the RFG standard was still more stringent than 0.62 vol% and credits
could not be used from CG to comply. As discussed above, having
separate standards with separate ABT programs raises additional cost
and feasibility issues.
For all of the above reasons, we believe that a refinery annual
average benzene content standard of 0.62 vol% applying to all gasoline
nationwide (excluding California), in conjunction with an
appropriately-designed ABT system, would maximize benzene emission
reductions considering cost and other factors.
Section 202(l)(2) also requires that we consider lead time in
determining the greatest reductions achievable. We are proposing that
the standard of 0.62 vol% become effective on January 1, 2011. Because
the final rule will be completed in early 2007, this would allow about
4 years for refiners to plan and execute the necessary capital projects
and operational changes needed to meet the program requirements. We
discuss our assessment of necessary lead time in section VII.F below.
We believe that this proposed level for the standard, the proposed ABT
program, and the proposed implementation date together meet the
statutory requirement that the program results in the greatest emission
reduction achievable considering costs and other factors.
We encourage comment on our selection of this level for the
standard, especially with data and analysis that support the comments.
d. Comparison of Other Benzene Regulatory Programs
In addition to the benzene content standard of the RFG program,
California and several countries have regulatory limits on the benzene
content of gasoline. Table VII.C-1 shows the basic provisions of each
of these programs.
Canada has limits similar to those covering U.S. RFG. In Canada,
producers may either comply with a 1.0 vol% flat limit or an averaging
standard of 0.95 vol%, with a per-gallon cap of 1.5 vol%. The European
Union regulates fuel to the same level in all its member countries,
currently a per-gallon cap of 1.0 vol%. Japan has the same limit as the
E.U., while South Korea will be moving from a cap of 1.5 to 1.0 vol% in
2006.
California is the only state that has implemented a benzene
standard, and it is similar to the standard we are proposing today.
California's average standard is 0.7 vol%, with a per-gallon cap of 1.1
vol%. Together, these standards result in an average 0.62 vol% in-use
gasoline benzene level.

Table VII.C-1.--Other Gasoline Benzene Control Programs
--------------------------------------------------------------------------------------------------------------------------------------------------------
California
Federal RFG phase 3 RFG Canada South Korea Japan European Union
--------------------------------------------------------------------------------------------------------------------------------------------------------
Average Std (vol%)...................................... 0.95 a 0.7 0.95 .............. .............. ..............
Per-gallon Cap (vol%)................................... 1.3 1.1 1.5 1.5 b 1.0 1.0
--------------------------------------------------------------------------------------------------------------------------------------------------------
a Producers may also comply with a per-gallon cap of 1.0.
b Limit to be lowered to 1.0 in 2006.

4. How Do We Address Variations in Refinery Benzene Levels?
a. Overall Reduction in Benzene Level and Variation
As explained above, there is currently a wide variation in gasoline
benzene levels across the country. According to summer 2003 batch data
(proposed baseline \241\), average benzene content ranged from 0.41 to
3.81 vol%, including both RFG and CG. The current

[[Page 15868]]

variation in benzene levels is primarily attributable to differences in
crude oil quality, different refinery configurations, and differences
in refinery operations. Our analysis of the proposed program,
summarized below, concludes that average benzene levels would be
reduced in all areas of the country (PADDs \242\) and variation among
refineries would also be reduced. We believe that under the proposed
rule, virtually all refineries would reduce their benzene levels and
that no refineries would increase their benzene levels.
---------------------------------------------------------------------------

\241\ For the purpose of our analyses, we selected 2003 to
represent current (baseline) conditions because it reflected the
most recent batch data available. The refinery-by-refinery model
used to predict refinery behavior (discussed later in section IX) is
based on inputs from the linear programming (LP) model, which is set
up to only model the summer season. As a result, we have used summer
2003 as our baseline period.
\242\ The Department of Energy divides the United States into
five Petroleum Administration for Defense Districts, or PADDs. The
states included in each PADD are defined at 40 CFR 80.41.
---------------------------------------------------------------------------

Upon implementation of the proposed 0.62 vol% benzene standard in
2011, we believe that some refiners would reduce benzene levels to
below the standard while others would reduce benzene levels but would
need to rely partially or largely on credits generated and traded under
the proposed ABT program, as described below. Refiners' compliance
strategies would ultimately be driven by economics. For many it would
be economical to reduce gasoline benzene levels to 0.62 vol% or below.
For others it would be economical to make some reduction in gasoline
benzene levels and rely partially upon credits. For some refineries
already below the standard, no benzene reduction efforts would be
necessary. For the limited number of remaining technologically-
challenged refineries it would be most economical to rely wholly upon
credits. Regardless of the compliance strategies selected, under the
proposed program, benzene levels and variation would be reduced nationwide.

Table VII.C-2.--Benzene Levels in Gasoline Produced Currently and Under the Proposed Program
--------------------------------------------------------------------------------------------------------------------------------------------------------
Number of refineries by gasoline benzene level (vol%) Benezene level (vol%) *
-------------------------------------------------------------------------------------------------------------
<0.5 0.5-<1.0 1.0-<1.5 1.5-<2.0 2.0-<2.5 >=2.5 Min Max Range ** Avg ***
--------------------------------------------------------------------------------------------------------------------------------------------------------
Starting Gasoline Benzene Levels***
--------------------------------------------------------------------------------------------------------------------------------------------------------
PADD 1.................................... 4 3 3 0 2 0 0.41 2.19 1.77 0.62
PADD 2.................................... 0 5 8 11 1 1 0.60 2.85 2.25 1.32
PADD 3.................................... 4 18 10 7 0 2 0.41 3.10 2.69 0.86
PADD 4.................................... 0 1 4 6 3 2 0.60 3.56 2.96 1.60
PADD 5 ****............................... 0 0 1 3 2 2 1.36 3.81 2.44 2.06
-------------------------------------------------------------------------------------------------------------
Total................................. 8 27 26 27 8 7 0.41 3.81 3.39 0.97
--------------------------------------------------------------------------------------------------------------------------------------------------------
Benzene Levels After Program Implementation
--------------------------------------------------------------------------------------------------------------------------------------------------------
PADD 1.................................... 4 5 1 2 0 0 0.41 1.96 1.54 0.51
PADD 2.................................... 1 22 1 2 0 0 0.49 1.95 1.46 0.73
PADD 3.................................... 10 27 3 0 1 0 0.36 2.07 1.71 0.55
PADD 4.................................... 0 8 7 1 0 0 0.53 1.94 1.40 0.95
PADD 5 ***................................ 0 4 2 2 0 0 0.54 1.84 1.30 1.04
-------------------------------------------------------------------------------------------------------------
Total................................. 15 66 14 7 1 0 0.36 2.07 1.71 0.62
--------------------------------------------------------------------------------------------------------------------------------------------------------
* Starting benzene levels based on summer 2003 batch data.
** Range in benzene level (MIN-MAX).
*** Average volume-weighted benzene level.
**** PADD 5 excluding California.

As shown in Table VII.C-2, average benzene levels would be reduced
by 36%, from 0.97 vol% (baseline) to 0.62 vol% once the program is
fully implemented. Variation in benzene level, measured in terms of
range, would be reduced by 50% (from 3.39 vol% to 1.71 vol%). In
addition the areas with the highest starting benzene levels and
variation (PADDs 2, 3, 4 and 5) would experience the greatest reductions.
In conclusion, we project that under the proposed program all areas
of the country would see reductions in average benzene level and
variation among refineries would also be reduced. Refiners would have
several motivations for making the benzene reductions projected by our
analysis. First, reducing actual benzene levels could be the most
economically-favorable compliance strategy. Secondly, reducing benzene
levels would help reduce or eliminate the uncertainty associated with
relying on credits. Finally, reducing benzene levels could generate
credits that would be valuable to the refining industry.
b. Consideration of an Upper Limit Standard
We believe that the proposed program would provide significant
benefits in all areas of the nation. Nevertheless, we recognize that
some commenters are likely to be concerned that under a flexible ABT
program it is possible that some refiners could maintain their current
benzene levels or even increase them and comply through the use of
credits. If such a refinery dominated a particular market, then even
though nationally there would be significant benzene reductions, they
might not occur in that market. While our analysis does not lead us to
believe that such an outcome would happen, we have nevertheless
considered whether an upper limit on benzene (in addition to the
average standard) would be valuable to prevent that outcome from
happening.\243\ We considered two different forms of an upper benzene
limit to complement the average standard: a per-gallon cap standard and
a maximum average standard.
---------------------------------------------------------------------------

\243\ Upper limits on benzene are a part of comparable programs
in California and in other countries.
---------------------------------------------------------------------------

i. Per-Gallon Cap Standard
A cap would require that each gallon (or batch) of gasoline
produced or imported not contain more than a specified concentration of
benzene. Such a standard would force those refineries with the highest
benzene levels to make physical changes to their gasoline instead of
having the option of relying exclusively on credits. In addition to
formally limiting the maximum benzene content sold anywhere in the
country, such a cap would also be straightforward to enforce

[[Page 15869]]

at any point in the distribution system. Note that we are proposing
that the existing per-gallon cap of 1.3 vol% benzene would remain in
effect for RFG under this rule. EPA invites comment on whether the RFG
benzene cap should be retained.
The primary disadvantage of adding a rigid cap is that it would not
allow for occasional, short-term fluctuations in benzene content.
Refiners are faced with a range of unexpected or planned circumstances
that could cause temporary spikes in benzene content, including
equipment malfunctions and periodic maintenance. Although the 1.3 vol%
cap would remain for RFG, to apply a cap in this range to CG would
eliminate a necessary market for higher benzene batches.\244\ With no
ability to market the gasoline, the refiner would be forced to suspend
gasoline production. This could in turn force the shutdown of the
entire refinery, sacrificing supply of all products. To attempt to
avoid this situation, refiners would need to invest more heavily in
benzene control than needed to meet the average standard, simply to
provide back-up control to protect against short-term fluctuations. For
some higher-benzene refineries, a cap could make complying with the
program prohibitively expensive.
---------------------------------------------------------------------------

\244\ As explained in section VII.C.5 below, CG provides a
limited safety valve for occasional batches of high-benzene RFG due
to the Anti-dumping provisions.
---------------------------------------------------------------------------

Consequently, we concluded that if we were to impose a per-gallon
cap, it would have to be high enough to allow most refineries to
continue to operate even in such upset situations (in order to account
for legitimate maximum potential daily variability), thereby providing
little overall benefit.\245\ Alternatively, we would have to allow
exceptions to the per-gallon cap for such upset situations, which would
be burdensome to implement and also result in little overall benefit.
---------------------------------------------------------------------------

\245\ In California and other countries with benzene control
programs, the refining industry tends to be more homogeneous than in
the U.S. as a whole and face different market situations, resulting
in different considerations regarding upper limits.
---------------------------------------------------------------------------

If refiners with higher-benzene refineries need to invest in
greater benzene control in order to protect against unpredictable
upsets, their costs would be even higher relative to those of lower-
benzene refineries. As in the case of a program with no ABT at all, the
statutory requirement to balance the degree of feasible emission
reduction with cost (and other factors) would have the
counterproductive effect of requiring a less stringent overall program.
At the same time, the per-gallon cap would appear to provide no
overall additional reduction in benzene levels. Despite the increased
costs, particularly for higher-benzene refiners, our analysis indicates
that little additional emission reduction would result (primarily
because the higher-benzene refineries represent a relatively small
fraction of nationwide gasoline production). Instead, as discussed
below, emission reductions are expected to simply shift from one region
of the country to another, with no change in the overall emission
reductions. Because of this, and due to the potential deleterious cost
impacts, we are not proposing a per-gallon cap benzene standard.
ii. Maximum Average Standard
Another means of ensuring some reduction by those refiners with the
highest benzene concentrations would be to impose a maximum average
standard. An annual maximum average standard for each refinery would
limit the average benzene content of its actual production over the
course of the year, regardless of the extent to which credits may have
been used for compliance. While slightly less restrictive than a per-
gallon cap standard in that some shorter-term fluctuations in benzene
levels could occur, a maximum average standard would still limit the
flexibility otherwise available through the ABT program. Our modeling
shows that a number of refiners would need to invest substantially more
to ensure compliance with both the average and maximum average
standards. With the addition of a maximum average standard, we expect
emission reductions to simply shift from one region of the country to
another with no net change in overall emission reductions. For example,
when analyzing a 1.3 vol% maximum average standard, benzene levels were
lowered in two PADDs and raised in three PADDs compared to our proposed
program yet the overall emission reductions remained the same.\246\
Since we believe that a maximum average standard would increase costs
but not achieve any greater emission reduction, we are not proposing
such a standard.
---------------------------------------------------------------------------

\246\ This program comparison is discussed further in Chapter 9
of the RIA (Table 9.6-7).
---------------------------------------------------------------------------

We believe that the proposed ABT program, in combination with the
proposed 0.62 vol% benzene standard without a cap or maximum average
limit, would result in the maximum feasible reduction in benzene
emissions, considering costs, energy, and safety issues. The proposed
ABT program would provide refiners with compliance flexibility while
ensuring that the national program achieves significant overall benzene
emission reductions.
We invite comment on our conclusions about having an upper limit in
addition to an average standard.
5. How Would the Proposed Program Meet or Exceed Related Statutory and
Regulatory Requirements?
Three fuels programs (RFG, Anti-dumping and MSAT1) currently
contain direct controls on the toxics performance of gasoline.\247\
Based on our analyses of the proposed program, including the proposed
ABT program, we expect that meeting the proposed fuel benzene content
standard combined with other fuel controls would also lead to
compliance with the toxics requirements of all these programs.
---------------------------------------------------------------------------

\247\ Other gasoline fuel controls, such as sulfur, RVP or VOC
performance standards, indirectly control toxics performance by
reducing overall emissions of VOCs.
---------------------------------------------------------------------------

The RFG program, implemented in 1995, contains a fuel benzene
standard that requires a refinery's or importer's RFG to average no
greater than 0.95 vol% benzene annually.\248\ In addition, RFG has a
per-gallon benzene cap of 1.3 vol%. Each refinery's or importer's RFG
must also achieve at least a 21.5% annual average reduction in total
toxics emissions compared to 1990 baseline gasoline.\249\ The Anti-
dumping regulations require that a refinery's or importer's CG produce
no more exhaust toxics emissions on an annual average basis than its
1990 gasoline.\250\ This program keeps refiners from shifting fuel
components responsible for elevated toxic emissions into CG as a way to
comply with the RFG standards. Section V.D.1 above describes these
programs in more detail.
---------------------------------------------------------------------------

\248\ 40 CFR 80 Subpart D. Refiners also have the option of
meeting a per gallon limit of 1.0 vol%.
\249\ Emissions determined using the Complex Model, as defined
in 40 CFR 80.45.
\250\ CFR 80 Subpart E, emissions determined using the Complex Model.
---------------------------------------------------------------------------

The MSAT1 program, implemented in 2002, was overlaid on the RFG and
Anti-dumping programs.\251\ As explained in section V.D above, it was
not designed to further reduce MSAT emissions, but to lock in
overcompliance on toxics performance that was being achieved in RFG and
CG under the RFG and Anti-dumping programs. The MSAT1 rule requires the
annual average toxics performance of a refinery's or importer's
gasoline to be at least as clean as the average performance of its
gasoline during the three-year baseline period 1998-

[[Page 15870]]

2000.\252\ Compliance with MSAT1 is determined separately for each
refinery's or importer's RFG and CG.
---------------------------------------------------------------------------

\251\ 40 CFR 80 Subpart J.
\252\ Emissions determined using the Complex Model, as defined
in 40 CFR 80.45.
---------------------------------------------------------------------------

Today's proposed 0.62 vol% benzene content standard would apply to
all of a refinery's or importer's gasoline `` that is, the total of its
RFG and CG production or imports. This level of benzene control would
far surpass the RFG standard of 0.95 vol%, and would put in place a
benzene content standard for CG for the first time.\253\ As described
further in Chapter 6 of the RIA, we analyzed the expected overall
toxics performance under today's proposed program of benzene and
vehicle standards using currently-available models and compared it to
toxics performance under the pre-existing standards.\254\ When RFG and
CG toxics emissions are evaluated at this new level of benzene control,
it is clear that the benzene standard proposed today would result in
the MSAT1 toxics emissions performance requirements being surpassed
(i.e., bettered) not only on average nationwide, but for every PADD.\255\
---------------------------------------------------------------------------

\253\ Proposed program retains the 1.3 vol% maximum benzene cap
for RFG required by 40 CFR 80.41.
\254\ As discussed previously, the existing models contain
limited data on the impacts of fuel changes on 2004 and later
technology vehicles, making such projections difficult. However, we
do not believe the conclusions would change for these reasons: (1)
The fuel effect changes modeled here related to benzene, for which
we expect data for new technology vehicles to show similar trends as
those for older vehicles; (2) much of the projected change in future
emissions are due to changes in vehicles technology, not fuel
changes; and (3) for this analysis we need only look at the relative
changes, and given the magnitude of the projected effects we do not
expect that the direction of the result would change even if
significantly different values for absolute emissions were submitted.
\255\ The analysis shows an even greater benefit in overall
toxics reductions when the combined effect of the benzene standard
and the vehicle standards are considered.
---------------------------------------------------------------------------

To address compliance with statutory requirements currently in
effect through the RFG and Anti-dumping programs, we carried out a
refinery-by-refinery analysis of toxics emissions performance using the
Complex Model (the same model used for determining compliance with
these programs). We used 2003 exhaust toxics performance for CG and
2003 total toxics performance for RFG as benchmarks, which are at least
as stringent as the relevant toxics performance baselines. We applied
changes to each refiner's fuel parameters for today's proposed
standards and the gasoline sulfur standard phased in this year (30 ppm
average, 80 ppm max). The results indicate that all refineries
maintained or reduced their emissions of toxics over 2003. We expect
large reductions in sulfur for almost all refineries under the gasoline
sulfur program, and large reductions in CG benzene levels along with
modest reductions in RFG benzene levels. We do not expect backsliding
in sulfur levels by the few refiners previously below 30 ppm because
they had been producing ultra-low sulfur gasoline for reasons related
to refinery configuration. Furthermore, because of its petrochemical
value and the credit market, we do not expect any refiners to increase
benzene content in their gasoline.
In addition, we expect significant changes in oxygenate blending
over the next several years, but these are very difficult predict on a
refinery-by-refinery basis. Regardless of how individual refineries
choose to blend oxygenates in the future, we believe their gasoline
will continue to comply with baseline requirements. This is because all
RFG is currently overcomplying with the statutory requirement of 21.5%
annual average toxics reductions by a significant margin. Similarly,
most CG is overcomplying with its 1990 baselines by a significant
margin. Furthermore, we believe most refiners currently blending oxygenates
will continue to do so at the same or greater level into the future.
EPA is thus proposing that upon full implementation in 2011 the
regulatory provisions for the benzene control program would become the
single regulatory mechanism used to implement these RFG and Anti-
dumping annual average toxics requirements, replacing the current RFG
and Anti-dumping annual average provisions. However, the 1.3 vol%
maximum benzene cap would remain in place for RFG under 40 CFR 80.41;
we are requesting comment on the need to retain this requirement for
RFG. The proposed benzene control program would also replace the MSAT1
requirements.
Section 1504(b) of the Energy Policy Act of 2005 (EPAct) requires
that the MSAT1 toxics emissions baselines for RFG be adjusted to
reflect 2001-2002 fuel qualities, which would make them slightly more
stringent than the 1998-2000 baselines originally used in the MSAT1
program. However, as provided for in the Act, this action becomes
unnecessary and can be avoided if today's proposed program achieves
greater overall reductions of toxics emissions from RFG (i.e., PADDs 1
and 3) than would be achieved by this baseline year adjustment.
Therefore, in addition to comparing the proposed standard to the
current MSAT1 program, we also compared it to the program as the
standards would be modified by the EPAct.
We performed an analysis of aggregate toxics emissions for the
relevant baseline periods as well as for future years with and without
the proposed program. This analysis was carried out using MOBILE6.2
because that model accounts for changes in the vehicle fleet, which is
important when modeling future years. Results are shown in Table VII.C-
3. Since this modeling approach was intended to compare emissions from
different fuels and fleet year mixes, the emissions figures generated
here are different from those used for gasoline compliance determination.
The first row shows mg/mi air toxics emissions in 2002 under the
MSAT1 refinery-specific baseline requirements. The second row shows how
these would change by updating the RFG baselines to 2001-02 as
specified in EPAct. Since significant changes are expected in the
gasoline pool between 2002 and the proposed implementation time of the
fuel standard, such as gasoline sulfur reductions and oxygenate
changes, we decided to model a ``future baseline'' to allow comparison
with the proposed standard at the time it would become effective in
2011. As a result, the third row shows the projected mg/mi emissions in
2011 under the EPAct baseline adjustments, but without today's proposed
program. The large reductions in air toxics emissions between the EPAct
baseline and this 2011 baseline are primarily due to nationwide
reduction in gasoline sulfur content to 30 ppm average and significant
phase-in of Tier 2 vehicles into the national fleet.
An important comparison is made between rows three and four, where
the estimated toxics emissions under the proposed fuel standard only
are compared to the projected emissions without the proposed standard.
The fourth row shows small reductions for RFG and more significant
reductions for CG with the introduction of the proposed benzene
standard in 2011. We also evaluated the effects of the vehicle standard
also proposed today on toxics emissions at two points in time, shown in
the last two rows of the table.

[[Page 15871]]

Table VII.C-3.--Estimated Annual Average Total Toxics Performance of Light Duty Vehicles in mg/mi Under Current and Proposed Programs a
--------------------------------------------------------------------------------------------------------------------------------------------------------
Fleet RFG by PADD CG by PADD
Regulatory scenario --------------------------------------------------------------------------------------------------
Year I II III I II III IV V
--------------------------------------------------------------------------------------------------------------------------------------------------------
MSAT1 Baseline b (1998-2000)......................... 2002 108 124 89 104 135 96 137 152
EPAct Baseline b (RFG: 2001-2002).................... 2002 103 121 85 104 135 96 137 152
EPAct Baseline, 2011 c............................... 2011 67 79 51 62 79 54 77 96
Proposed program, 2011 c (Fuel standard only)........ 2011 66 78 50 59 74 51 71 85
Proposed program, 2011 c (Fuel + vehicle standards).. 2011 63 76 47 55 72 47 67 81
Proposed program, 2025 c (Fuel + vehicle standards).. 2025 39 46 30 35 44 31 42 50
--------------------------------------------------------------------------------------------------------------------------------------------------------
a Total toxics performance for this analysis includes overall emissions of 1,3-butadiene, acetaldehyde, acrolein, benzene and formaldehyde as calculated
by MOBILE6.2. Although POM appears in the Complex Model, it is not included here. However, it contributes a small and relatively constant mass to the
total toxics figure (4%), and therefore doesn't make a significant difference in the comparisons.
b Baseline figures generated in this analysis were calculated differently from the regulatory baselines determined as part of the MSAT1 program, and are
only intended to be a point of comparison for future year cases.
c Future year scenarios include (in addition to the controls proposed today, where stated) effects of the Tier 2 vehicle and gasoline sulfur standards
and vehicle fleet turnover with time, as well as rough estimates of the renewable fuels standard and the phase-out of ether blending.

Based on these analyses, we believe the fuel program proposed in
this notice, as well as the combined fuel and vehicle program, would
also achieve greater overall toxics reductions than would be achieved
under the EPAct were the RFG baseline period updated to 2001-2002.
In summary, today's proposed action for fuels would fulfill several
statutory and regulatory goals related to control of gasoline mobile
source air toxics emissions. The proposed program (in conjunction with
the proposed vehicle standards) would meet our commitment in the MSAT1
rulemaking to consider further MSAT control. It would also result in
air toxics emission reductions greater than required under all pre-
existing gasoline toxics programs, as well as under the baseline
adjustments specified by the Energy Policy Act. By designing this
program to address these separate but related goals, we would be able
to achieve a benefit in addition to the emissions reductions: A significant
consolidation and simplification of regulation of gasoline MSATs.
As part of today's action, in addition to the streamlining of
toxics requirements, we propose that the gasoline sulfur program become
the sole regulatory mechanism used to implement gasoline NO_X
requirements. Gasoline producers are required to show reductions from
their RFG relative to the 1990 Clean Air Act baseline gasoline
NO_X emissions, as determined using the Complex Model.
Conventional gasoline must comply with Anti-dumping individual
NO_X baselines for each refinery, similar to the Anti-dumping
toxics standards. A refinery-by-refinery NO_X analysis
parallel to that described above indicated that with the final
implementation of the gasoline sulfur program (January 1, 2006), all
gasoline will continue to meet or exceed the NO_X
requirements of the RFG and Anti-dumping programs.
As discussed elsewhere in this preamble, we believe that today's
proposed nationwide program would achieve significant reductions in
gasoline-related benzene emissions. The program would also have the
effect of preempting states from regulating gasoline benzene content.
The program is proposed under Clean Air Act section 211(c), which
includes preemption of state fuel programs in section 211(c)(4).\256\
The existing RFG benzene program, also authorized under section
211(c)(1), preempts states in RFG areas from regulating benzene.
Today's nationwide program expands this preemption to all states except
California, which is exempt from this preemption.
---------------------------------------------------------------------------

\256\ See discussion of statutory authority in section I.C. of
this preamble.
---------------------------------------------------------------------------

D. Description of the Proposed Averaging, Banking, and Trading (ABT) Program

1. Overview
As mentioned earlier, we are proposing a specially-designed ABT
program to allow EPA to set a more stringent nationwide gasoline
benzene standard than otherwise possible. The proposed ABT program
would allow refiners and importers to use benzene credits generated or
obtained under the provisions of the ABT program to comply with the
0.62 vol% refinery average standard in 2011 and indefinitely
thereafter. Benzene credits could be generated by refineries that make
qualifying early baseline reductions prior to 2011 and by refineries
and importers that overcomply with the 0.62 vol% standard in 2011 and
beyond. All credits generated could be used internally towards company
compliance (``averaged''), ``banked'' for future use, and/or
transferred (``traded'') to another refiner or importer.
The majority of the ABT credit provisions we are proposing are
similar to those offered in the gasoline sulfur program, with a few
exceptions. The major difference is that in the proposed program,
credit use would not be restricted by an upper limit (discussed in
VII.C.4.b above) and in fact would be encouraged by extended credit
life and nationwide credit trading provisions. We are able to propose a
flexible ABT program and a gradual phase-in of the 0.62 vol% benzene
because there is no corresponding vehicle standard being proposed that
is dependent on gasoline benzene content. A program with fewer
restrictions would help ensure that the overall proposed benzene
control program would result in the greatest achievable benzene
reductions, considering cost and other factors.
Because of the wide variation in current benzene levels among
refineries, we recognize that some refiners would be better situated
than others, technologically and financially, to respond to the
proposed benzene standard. As we discuss below, we believe that the
credit trading provisions of the ABT program would be well suited to
moderate the financial impacts that could otherwise occur with the
proposed benzene control program.
However, in other air quality programs, we have used other trading

[[Page 15872]]

mechanisms to address the varying impacts of such programs on different
regulated entities. For example, in EPA's Acid Rain program a limited
number of ``emissions allowances'' are allocated among entities, which
can then be banked and traded. We invite comment on this and other
alternative credit approaches that might be appropriate to gasoline
benzene control.
The following paragraphs provide more details on our proposed
benzene ABT program. We encourage comments on the design elements we
have proposed for the program. If you believe that alternative
approaches would make the program more effective, please share your
specific comments and recommendations with us.
2. Standard Credit Generation (2011 and Beyond)
We are proposing that standard benzene credits could be generated
by any refinery or importer that overcomplies with the 0.62 vol%
gasoline benzene standard on an annual volume-weighted basis in 2011
and beyond. For example, if in 2011 a refinery's annual average benzene
level was 0.52, its standard benzene credits would be determined based
on the margin of overcompliance with the standard (0.62-0.52 = 0.10
vol%) divided by 100 and multiplied by the gallons of gasoline produced
during the 2011 calendar year. The credits would be expressed as
gallons of benzene. Likewise, if in 2012 the same refinery produced the
same amount of gasoline with the same benzene content they would earn
the same amount of credits. The standard credit generation opportunities
for overcomplying with the standard would continue indefinitely.
The refinery cost model discussed further in section IX.A, predicts
which refineries would reduce benzene levels in an order of precedence
based on cost until the 0.62 vol% refinery average standard is
achieved. The model also predicts which refineries would overcomply
with the standard in 2011 and beyond and in turn generate standard
credits.\257\ Credits would be generated by two main sources.
---------------------------------------------------------------------------

\257\ The refinery cost model assumes that all credits generated
are used each year. To the extent that this does not occur, more
refiners would have to invest in technology to comply, increasing
the cost of the program.
---------------------------------------------------------------------------

First, standard credits would be generated by refineries whose
current gasoline benzene levels are already below the 0.62 vol%
standard. According to the model, 19 refineries are predicted to
maintain current gasoline benzene levels and overcomply with the
standard without making any additional process improvements. These
refineries would generate approximately 42 million gallons of benzene
credits per year without making any investment in technology.
Additionally, the model predicts that 5 other refineries would reduce
gasoline benzene levels even further below 0.62 vol% resulting in
deeper overcompliance and an additional 6 million gallons of benzene
credits per year.
Second, standard credits would be generated by refineries whose
current gasoline benzene levels are above 0.62 vol% but are predicted
by the model to overcomply with the standard based on existing refinery
technology, access to capital markets, and/or proximity to the benzene
chemical market. The model predicts that 34 refineries with gasoline
benzene levels above 0.62 vol% would make process improvements to
reduce benzene levels below the standard and in turn generate
approximately 40 million gallons of benzene credits per year.
For the refineries which the model predicts to make process changes
to overcomply with the standard, the incremental cost to overcomply is
relatively small or even profitable in some cases of benzene
extraction.\258\ As expected, refineries with the lowest compliance
costs would have the greatest incentive to overcomply based on the
value of the credits to the refining industry.
---------------------------------------------------------------------------

\258\ Despite the low costs of benzene extraction, without a
benzene control standard refiners are reluctant to invest in
capital-intensive processes such as extraction. This is because many
other projects involving capital investments that they may be
considering typically have a better or more certain payout (past
price volatility in the benzene chemical market can discourage
future investment). Thus, refiners tend to postpone capital projects
such as extraction even if they may appear to be profitable today.
---------------------------------------------------------------------------

3. Credit Use
We are proposing that refiners and importers could use benzene
credits generated or obtained under the provisions of the ABT program
to comply with the 0.62 vol% gasoline benzene standard in 2011 and
indefinitely thereafter. Refineries and importers could use credits to
comply on a one-for-one basis, applying each benzene gallon credit to
offset the same volume of benzene produced in gasoline above the
standard. For example, if in 2011 a refinery's annual average benzene
level was 0.72, the number of benzene credits needed to comply would be
determined based on the margin of under-compliance with the standard
(0.72-0.62 = 0.10 vol%) divided by 100 and multiplied by the gallons of
gasoline produced during the 2011 calendar year. The credits needed
would be expressed in gallons of benzene.
We believe that individual refineries would rely differently upon
credits, depending on their unique refinery situations. As mentioned
earlier, the current range in gasoline refinery technologies and
starting benzene levels would make it significantly more expensive for
some refineries to comply with the standard based on actual reduced
benzene levels than others. As such, some technologically-challenged
refiners may choose to rely largely or entirely upon credits because it
would be much more economical than making process improvements to
reduce benzene levels. Other refiners may choose to make incremental
process improvements to reduce refinery benzene levels and then rely
partially on credits to fully comply. Still others may choose to reduce
benzene levels to at or around 0.62 vol% and maintain an ``emergency
supply'' of credits to address short-term spikes in benzene levels due
to refinery malfunctions. Overall, the proposed credit trading program
would encourage low-cost refineries to comply or overcomply with the
standard while allowing high-cost refineries to rely upon credits to
comply. This would reduce the total economic burden to the refining
industry.
a. Credit Trading Area
We are proposing a nationwide credit trading program with no
geographic restrictions on trading. In other words, a refiner or
importer could obtain benzene credits and use them towards compliance
regardless of where the credits were generated. We believe that
restricting credit trading could reduce refiners' incentive to generate
credits and hinder trading essential to this program. As explained in
Chapter 6 of the RIA, if PADD restrictions were placed on credit trading,
there would be an imbalance between the supply and demand of credits.
In other fuel standard ABT programs (e.g., the highway diesel
sulfur program), credit trading restrictions were necessary to ensure
there was adequate low-sulfur fuel available in each geographic area to
meet the corresponding vehicle standard. Since there is no vehicle
emission standard being proposed that is dependent on gasoline benzene
content, we do not believe there is a need for geographic trading
restrictions. As mentioned above, we project that under the proposed
ABT program, all areas of the country (i.e., all PADDs) would

[[Page 15873]]

experience a large reduction in gasoline benzene levels as a result of
the standard.
As discussed earlier, California gasoline would not be subject to
the proposed benzene standards. However, California refiners that
produce gasoline that is used outside of California would be able to
generate credits on that gasoline (and use credits to achieve
compliance on their non-California gasoline if necessary). Likewise, as
proposed, refiners outside of California that produce gasoline that is
used in California would not be allowed to use that gasoline as the
basis for any credit generation, or compliance with the proposed
benzene standard. However, we request comment on whether and how credits
could be allowed to be generated on California gasoline benzene reductions
and applied to the benzene compliance for non-California gasoline.
EPA seeks comment on the proposed nationwide trading provision, its
effect on incentives for refiners to generate credits, and
environmental impacts.
b. Credit Life
We are proposing limited credit life to enable proper enforcement
of the program and to encourage trading of credits. Since the proposed
standard is a refinery gate standard (i.e., enforced as the fuel leaves
the refinery) with no enforceable downstream standard, it is critical
that EPA be able to conduct enforcement at the refinery. A reasonable
limitation on credit life would allow EPA to verify the validity of
credits through record retention. Credit information must be
independently verifiable such that, in the event of violations
involving credits, the liable party is identifiable and accountable.
EPA enforcement activities are limited by the five-year statute of
limitations in the Clean Air Act. As a consequence, credit life greater
than five years creates potentially serious enforcement difficulties.
This is particularly important given the ongoing changes in business
relationships, ownership, and merger practices that are characteristic
of the refining industry. In addition, since credit trading plays an
essential role in moderating program costs, it is important that
refiners have an incentive to trade credits rather than hoard them.
Instituting a credit expiration date would promote trading because
refiners would be forced to ``use it or lose it.'' In summary, we
believe the proposed credit life provisions, described in more detail
below, are limited enough to satisfy enforcement and trading concerns
yet sufficiently long to provide program flexibility.
We are proposing that standard credits generated in 2011 and beyond
would have to be used within five years of the year in which they were
generated. For example, credits generated based on 2011 gasoline
production would have to be used towards compliance with the 2016
calendar year or earlier, otherwise they would expire. Standard credits
traded to another party would still have to be used during the same
five-year period because credit life is tied to the date of generation,
not the date of transfer.
We are proposing that early credits generated prior to 2011
(discussed in the paragraphs to follow) would have a three-year credit
life from the start of the program. In other words, early credits would
have to be applied to the 2011, 2012, and/or 2013 compliance years or
they would expire.
These proposed credit life provisions are similar to those
finalized in the gasoline sulfur program, except the early credit life
is three years instead of two. We are proposing a three-year early
credit life because it corresponds with the number of early credits
projected to be generated according to our refinery cost model.\259\
Additionally, we predict that three years would be more than sufficient
time for all early credits generated to be utilized. We believe that
this certainty that all credits could be utilized would strengthen
refiners' incentive to generate early credits and subsequently
establish a more reliable credit market for trading.
---------------------------------------------------------------------------

\259\ Derivation of three-year early credit lag is found in
Chapter 6 of the RIA (section 6.5.3.1).
---------------------------------------------------------------------------

In addition to the above-mentioned provisions, we are proposing
that credit life may be extended by two years for early credits and/or
standard credits generated by or traded to approved small refiners. We
are offering this provision as a mechanism to encourage more credit
trading to small refiners. Small refiners often face special
technological challenges, so they would tend to have more of a need to
rely on credits. At the same time, they often have fewer business
affiliations than other refiners, so they could have difficulty
obtaining credits. We believe this provision would be equally
beneficial to refiners generating credits. This additional credit life
for credits traded to small refiners would give refiners generating
credits a greater opportunity to fully utilize the credits before they
expire. For example, a refiner who was holding on to credits for
emergency purposes or other reasons later found to be unnecessary,
could trade these credits at the end of their life to small refiners
who could utilize them for two more years. However, EPA is concerned
that extending credit life beyond the five-year statute of limitations
in the Clean Air Act (net 7-year credit life for standard credits
generated by or traded to small refiners) could create significant
enforceability problems. Consequently, EPA seeks comment on provisions
that could be included in the regulations that would address this
enforceability concern regarding the extended credit life for small
refiner standard credits.
As discussed in Section X.A, we are also seeking comment on
different ways of structuring the program that may be able to allow for
unlimited credit life since, unlike in the gasoline sulfur program,
there is no vehicle standard being proposed that is dependent on fuel
quality. We considered that unlimited credit life could further promote
credit generation and allow refiners to maintain an ongoing supply of
credits in the event of an emergency. However, for several reasons we
have elected to propose a limited credit life based on the context of
the rest of the proposed program. If unlimited credit life were to
discourage trading of credits, this could force refineries with more
expensive benzene control technologies to comply and thus increase the
total cost of the program. In addition, unlimited credit life would
make it more difficult to verify compliance with the standard. One way
of addressing this concern would be to require refiners to retain
credit records indefinitely. Even then, given the fluid nature of
refiner and importer ownership in recent years, in many cases it would
still be difficult to verify the validity of historical credit
generation and use. Since the proposed benzene standard would be
enforced solely at the refinery, it is critical that such enforcement
be as simple and straightforward as possible. Nonetheless, as discussed
in Section X.A, it may be possible to design the overall program in such a
way to address these concerns and still allow for infinite credit life.
In conclusion, we are proposing a reasonably limited credit life
for both early and standard benzene credits. We seek comment on
unlimited credit life. Please share with us any additional ideas you
may have on how unlimited credit life could be beneficial to this
program and/or how associated recordkeeping and enforcement issues
could be mitigated.

[[Page 15874]]

4. Early Credit Generation (2007-2010)
To encourage early application of and innovation in benzene control
technology, we are proposing that refiners could generate early benzene
credits from June 1, 2007 to December 31, 2010 by making qualifying
reductions from their pre-determined refinery baselines. A discussion
of how refinery baselines are established and what constitutes a
qualifying benzene reduction is found in the subsections to follow. The
early credits generated under this program would be interchangeable
with the standard credits generated in 2011 and beyond and would follow
the above-mentioned credit use provisions.
The early reductions we are projecting to occur would be the
initial steps of each refinery's ultimate benzene control strategy, but
completed earlier than required. We project that from mid-2007 to 2010,
refiners could implement operational changes and/or make small capital
investments to reduce gasoline benzene. These actions would create a
two-step phase down in gasoline benzene prior to 2011 as shown in
Figure VII.D-1.

BILLING CODE 6560-50-P
[GRAPHIC]
[TIFF OMITTED]
TP29MR06.006

BILLING CODE 6560-50-C
The credits generated under the early credit program could be used
to provide refiners with additional lead time to make their
investments. If properly implemented, we project that the delay could
be as much as three years as described in Chapter 6 of the RIA.
Accordingly, we are proposing a three-year early credit life, as
discussed earlier. The additional lead time would allow the refining
industry to spread out demand for design, engineering, construction and
other related services, reducing overall compliance costs.
Importers would not be permitted to generate early credits, for
several reasons.\260\ First, unlike refineries, importers would not
need additional lead time to comply with the standard, since they would
not be investing in benzene control technology. Additionally, because
importer operations are more variable than refinery operations,
importers could potentially redistribute the importation of foreign
gasoline based on benzene level to generate early credits without
making a net reduction in gasoline benzene. This type of scheme could
result in a large number of early credits being generated with no net
benzene emission reduction value. This is not expected to occur for
refineries because they are already operating at high capacity and do
not have the flexibility

[[Page 15875]]

to quickly increase, decrease, or shift production volumes.
Additionally, under the proposed program, refineries are prohibited
from moving benzene-rich blendstocks around to generate early credits
as described below.
---------------------------------------------------------------------------

\260\ As discussed in section VII.I.1 below, foreign refiners
may generate early credits under the proposed 40 CFR 80.1420 provisions.
---------------------------------------------------------------------------

We believe that refiners would have several motivations for making
early benzene reductions. For refiners who have a series of technology
improvements to make, early innovative improvements would help the
refiner get one step closer to compliance. Early reductions would also
generate credits which could be used to postpone subsequent
investments. For refiners capable of making early advancements to
reduce their benzene levels below 0.62 vol%, the early credits
generated would not be needed for their own future use. For these
refiners, trading early credits to other refiners may be a way to
offset the cost of their early capital investment(s).
a. Establishing Early Credit Baselines
We are proposing that any refiner planning on generating early
credits would have to obtain an individual refinery benzene baseline in
order to provide a starting point for calculating early credits.
Refinery benzene baselines would be defined as the annualized
volume-weighted benzene content of gasoline produced at a refinery from
January 1, 2004 to December 31, 2005. We are proposing a two-year
baseline period to account for normal operational fluctuations in
benzene level. We propose using the 2004 and 2005 calendar years
because we believe this would represent the most current batch gasoline
data available prior to today's proposal.
We would require refiners to submit individual baselines for each
refinery that is planning to generate early benzene credits. Refinery
benzene baselines would be calculated using the 2004-2005 batch data
submitted to us under the RFG and Anti-dumping requirements.\261\ We
propose that joint ventures, in which two or more refiners collectively
own and operate one or more refineries, be treated as separate refining
entities for early credit generation purposes.
---------------------------------------------------------------------------

\261\ RFG, 40 CFR 80.75; Anti-dumping, 40 CFR 80.105.
---------------------------------------------------------------------------

Refiners would be required to submit their refinery baselines in
writing to EPA. We propose that refiners could begin applying for 2004-
05 benzene baselines as early as March 1, 2007. There would be no
single cut-off date for applying for a baseline; however, a refiner
planning on generating early credits would need to submit a baseline
application at least 60 days prior to beginning credit generation. We
are proposing a shorter notification period for this rule (past rules
were 120 days) to accommodate our proposed early credit generation
start date of June 1, 2007. EPA would review all baseline applications
and notify the refiner of any discrepancies found with the data
submitted. If we did not respond within 60 days, the baseline would be
considered to be approved, subject to later review by EPA.
Under the proposed program, refiners would be prohibited from
moving gasoline and gasoline blendstock streams from one refinery to
another in order to generate early credits. This type of transaction
would result in artificial credits with no associated emission
reduction value. If traded and used towards compliance, these
artificial credits could negatively impact the benefits of the program.
We considered basing credit generation for multi-refinery refiners on
corporate benzene baselines instead of individual refinery baselines,
but determined that this could hinder credit generation. If a valid
reduction was made at one refinery and an unrelated expansion occurred
at another facility during this time, the credits earned based on a
corporate baseline could be reduced to zero. Instead, we propose to
validate early credits based on existing reporting requirements (e.g.,
batch reports and pre-compliance reporting data). We seek comment on
this approach.
b. Early Credit Reduction Criteria (Trigger Points)
We are proposing that to generate early credits, refiners would
first need to reduce gasoline benzene levels to 0.90 times their
refinery benzene baseline during a given averaging period. The purpose
of setting an early credit generation trigger point is to ensure that
changes in benzene level are representative of real process
improvements. Without a trigger point, refineries could generate
``windfall'' early credits based on normal year to year fluctuations in
benzene level associated with MSAT1. These artificial credits would
compromise the environmental benefits of an ABT program because they
would have no real associated benzene emission reduction value.
In designing the early credit generation program, we considered a
variety of different types of trigger points. We performed sensitivity
analyses around absolute level trigger points (refineries must reduce
gasoline benzene levels to a certain concentration), fixed reduction
trigger points (refineries must reduce gasoline benzene levels by a
certain concentration), and percent reduction trigger points
(refineries must reduce gasoline benzene by a percentage). Based on our
analysis found in Chapter 6 of the RIA, we found absolute level trigger
points to be too restrictive for high benzene level refineries that
could benefit from reductions the most. We also found fixed reduction
trigger points to be too restrictive to low benzene level refineries
which would be penalized for already being ``cleaner.'' Percent
reduction trigger points were found to be consistently limiting towards
all refineries, regardless of starting benzene level. As such, we
propose to conclude that a percent reduction trigger point would be the
most appropriate early credit validation tool to address the wide range
in starting benzene levels.
To determine an appropriate value for the percent reduction trigger
point, we considered a range of reductions from 5-40% and examined the
resulting early credit generation outcomes. We found that as the value
of the percent reduction trigger point increased, the potential for
windfall credit generation decreased, but unfortunately so did the
number of early credits generated from legitimate refinery
modifications. To address this competing relationship between windfall
and early credit generation, we are proposing a 10% reduction trigger
point. We believe that this trigger point is restrictive enough to
prevent most windfall credit generation, but not too restrictive to
discourage refineries from making early benzene reductions. The
proposed 10% reduction trigger point roughly coincides with the average
fluctuation in benzene level in 2004 as discussed in Chapter 6 of the
RIA. A 10% reduction trigger point for early credits was also finalized
in the gasoline sulfur rulemaking, which also affected the entire
gasoline pool and had to encompass a variety of unique refinery
situations.\262\ EPA requests comments on the proposed trigger point
and seeks alternate recommendations for validating early credits.
---------------------------------------------------------------------------

\262\ 40 CFR 80.305.
---------------------------------------------------------------------------

c. Calculating Early Credits
We are proposing that once the 10% reduction trigger point was met,
refineries could generate early credits based on the entire reduction.
In terms of benzene levels, a refinery would first have to reduce its
average benzene level to 0.90 times its original baseline benzene level
during a given averaging period in order to generate credits. For

[[Page 15876]]

example, if in 2008 a refinery reduced its annual benzene level from a
baseline of 2.00 vol% to 1.50 vol% (below the trigger of 0.90 x 2.00 =
1.80 vol%), its benzene credits would be determined based on the
difference in annual benzene content (2.00-1.50 = 0.50 vol%) divided by
100 and multiplied by the gallons of gasoline produced in 2008. The
credits would be expressed in gallons of benzene.
5. Additional Credit Provisions
a. Credit Trading
The potential exists for credits to be generated by one party,
subsequently transferred or used in good faith by another, and later
found to have been calculated or created improperly or otherwise
determined to be invalid. As in past programs, we propose that should
this occur both the seller and purchaser would have to adjust their
benzene calculations to reflect the proper credits and either party (or
both) could be determined to be in violation of the standards and other
requirements if the adjusted calculations demonstrate noncompliance
with the 0.62 vol% standard. This would allow the credit market to
properly allocate any such risk.
As with ABT programs in other rules, we are proposing that credits
should be transferred directly from the refiner or importer that
generated them to the party that would use them for compliance
purposes. This would ensure that the parties purchasing them would be
better able to assess the likelihood that the credits were valid, and
would aid in compliance monitoring. An exception would exist where a
credit generator transferred credits to a refiner or importer who could
not use all the credits, in which event that transferee could transfer
the credits to another refiner or importer. However, based on the
increased difficulty in assuring the validity of credits as the credits
change hands more than once, we are proposing that credits could only
be transferred a limited number of times. We are requesting comment on
the maximum number of allowable trades, in the range of 2 to 4 trades.
After the maximum number of trades, such credits would have be used or
terminated.
We propose no prohibitions against brokers facilitating the
transfer of credits from one party to another. Any person could act as
a credit broker, whether or not such person was a refiner or importer,
so long as the title to the credits was transferred directly from the
generator to the user. Further discussion of these credit trading
provisions and alternative options is found in section X.A below.
b. Pre-Compliance Reporting Requirements
In order to provide an early indication of the credit market for
refiners planning on relying upon benzene credits as a compliance
strategy in 2011 and beyond, we are requesting that refiners submit
pre-compliance reports to us in 2008, 2009, and 2010. EPA would then
summarize this information (in such a way as to protect confidential
business information) in a report available to the industry. This is
similar to the way pre-compliance reports are used for the ultra-low
sulfur diesel program. In addition, we are proposing that refiners
provide us with a final summary pre-compliance report in 2011, to allow
for a complete account of early credit generation.\263\ The reports
would be due annually by June 1st and would contain refiners' most up-
to-date implementation plans for complying with the 0.62 vol% benzene
standard. More specifically, we would require refiners to annually
submit to us engineering and construction plans and the following data:

\263\ Based on their proposed January 1, 2015 compliance date,
small refiners would be required to submit annual pre-compliance
reports to us in 2008 through 2014 with a final summary pre-
compliance report in 2015.
---------------------------------------------------------------------------

--Actual/projected gasoline production volume and average benzene level
for the June 1, 2007 through December 31, 2007 annual averaging period,
and for the 2008-2015 annual averaging periods.
--Actual/projected early credits generated during the June 1, 2007
through December 31, 2007 annual averaging period, and for the 2008-
2010 annual averaging periods (June 1 through December 31, 2007 and
2008-2014 for small refiners).
--Standard credits projected to be generated during the 2011-2015
annual averaging periods (2015 for small refiners).
--Credits projected to be needed for compliance during 2011-2015 annual
averaging periods (2015 for small refiners).

Pre-compliance reporting has proven to be an indispensable
mechanism in implementing the gasoline and diesel sulfur programs, and
we expect this to be the case in today's proposed program. A detailed
understanding of how individual refiners and the industry at large are
progressing toward final implementation of the proposed standards would
help identify early concerns and allow timely action if necessary to
prevent the development of major problems.
6. Special ABT Provisions for Small Refiners
Approved small refiners would follow all the above-mentioned ABT
provisions with the exception of special credit generation provisions
which accommodate their 2015 compliance start date. Early credits could
be generated by small refiners from June 1, 2007 to December 31, 2014
for refineries that reduce their average gasoline benzene level to 0.90
times their original 2004-2005 baseline level. Standard credits could
also be generated by small refiners beginning January 1, 2015 and
continuing indefinitely for refineries that overcomply with the
standard by producing gasoline with an annual average benzene content
below 0.62 vol%. Additionally, all credits generated by or traded to
approved small refiners would have an additional two-year credit life
as described above in VII.D.3.b.

E. Regulatory Flexibility Provisions for Qualifying Refiners

1. Hardship Provisions for Qualifying Small Refiners
In developing our proposed MSAT program, we evaluated the need and
the ability of refiners to meet the proposed benzene standards as
expeditiously as possible. We believe it is feasible and necessary for
the vast majority of the program to be implemented in the proposed time
frame to achieve the air quality benefits as soon as possible. However,
based on information available from small refiners, we believe that
refineries owned by small businesses generally face unique hardship
circumstances, compared to larger refiners. Thus, we are proposing
several special provisions for refiners that qualify as ``small
refiners'' to reduce the disproportionate burden that the proposed
standards would have on these refiners. These provisions are discussed
in detail below.
a. Qualifying Small Refiners
EPA is proposing several special provisions that would be available
to companies that are approved as small refiners. Small refiners
generally lack the resources available to larger companies that help
large companies, including those large companies that own small-
capacity refineries, to raise capital for investing in benzene control
equipment. These resources include shifting internal funds, securing
financing, or selling assets. Small refiners are also likely to have more

[[Page 15877]]

difficulty in competing for engineering resources and completing
construction of the needed benzene control equipment (and any necessary
octane recovery) equipment in time to meet the standards proposed
today. Therefore, we are proposing small refiner relief provisions in
today's action as an aspect of realizing the greatest emission
reductions achievable.
Since small refiners are more likely to face hardship circumstances
than larger refiners, we are proposing temporary provisions that would
provide additional time to meet the benzene standards for refineries
owned by small businesses. This approach would allow the overall
program to begin as early as possible, while still addressing the
ability of small refiners to comply.
i. Regulatory Flexibility for Small Refiners
As explained in the discussion of our compliance with the
Regulatory Flexibility Act below in section XII.C and in the Initial
Regulatory Flexibility Analysis in Chapter 14 of the RIA, we considered
the impacts of today's proposed regulations on small businesses. Most
of our analysis of small business impacts was performed as a part of
the work of the Small Business Advocacy Review (SBAR) Panel convened by
EPA, pursuant to the Regulatory Flexibility Act as amended by the Small
Business Regulatory Enforcement Fairness Act of 1996 (SBREFA). The final
report of the Panel is available in the docket for this proposed rule.
For the SBREFA process, EPA conducted outreach, fact-finding, and
analysis of the potential impacts of our regulations on small
businesses. Based on these discussions and analyses by all Panel
members, the Panel concluded that small refiners in general would
likely experience a significant and disproportionate financial hardship
in reaching the objectives of today's proposed program.
One indication of this disproportionate hardship for small refiners
is the higher per-gallon capital costs projected for the removal of
benzene from gasoline under the proposed program. Refinery modeling of
refineries owned by refiners likely to qualify as small refiners, and
of non-small refineries, indicates that small refiners could have
significantly higher costs to apply some technologies. For two of the
technologies that we believe that refiners would use to reduce their
benzene levels, routing the six carbon hydrocarbon compounds around the
reformer and isomerizing these compounds, we anticipate that small
refiners' costs would likely be similar to non-small refiners, as very
little capital investment would need to be made for these technologies.
However, for technologies such as benzene saturation and benzene
extraction, we anticipate that the costs to small refiners would be
higher. Due to the poorer economies of scale, benzene saturation is
expected to cost small refiners about 2.2 cents per gallon (while it is
projected that benzene saturation would cost a non-small refinery about
1.3 cents per gallon).\264\ Likewise, benzene extraction is estimated
to cost those refineries able to use this technology about 0.1 cents
per gallon; however, for small refiners benzene extraction is expected
to cost about 0.5 cents per gallon.
---------------------------------------------------------------------------

\264\ Smaller refineries are less likely to be able to take
advantage of economies of scale. For example, a portion of the
capital costs invested for a benzene control unit is fixed (i.e.,
engineering design costs) resulting in similar costs for each
investment project. However, when amortized over the volume of fuel
processed by a small versus large unit, the per-gallon capital costs
are higher for the smaller unit, resulting in poorer economies of scale.
---------------------------------------------------------------------------

The Panel also noted that the burden imposed on the small refiners
by the proposed benzene standard could vary from refiner to refiner.
Thus, the Panel recommended that more than one type of burden reduction
be offered so that most, if not all, small refiners could benefit. We
have continued to consider the issues that were raised during the
SBREFA process and have decided to propose the provisions recommended
by the Panel.
ii. Rationale for Small Refiner Provisions
Generally, we structured these proposed provisions to reduce the
burden on small refiners while still achieving the air quality benefits
that this program would provide. We believe that the proposed
regulatory flexibility provisions for small refiners are a necessary
aspect of standards reflecting the greatest achievable emission
reductions considering costs and lead time, because they would
appropriately adjust potential costs and lead time for the dissimilarly
situated small refiner industry segment, and at the same time allow EPA
to propose a uniform benzene standard for all refineries.
First, the proposed compliance schedule for this program, combined
with flexibility for small refiners, would achieve the air quality
benefits of the program as soon as possible, while still ensuring that
small refiners that choose to comply by raising capital for benzene
reduction technologies would have adequate time to do so. As noted
above, most small refiners have limited additional sources of income or
capital beyond refinery earnings for financing and typically do not
have the financial backing that larger and generally more integrated
companies have. Therefore, they could benefit from additional time to
accumulate capital internally or to secure capital financing from lenders.
Second, providing small refiners more time to comply would increase
the availability of engineering and construction resources to them.
Some refiners would need to install additional processing equipment to
meet the proposed benzene standard. We anticipate that there could be
increased competition for technology services, engineering resources,
and construction management and labor. In addition, vendors would be
more likely to contract with the larger refiners first, as their
projects would offer larger profits for the vendors. Temporarily
delaying compliance for small refiners would spread out the demand for
these resources and probably reduce any cost premiums caused by limited
supply.
Third, we are anticipating that many small refiners may choose to
comply with the proposed benzene standard by purchasing credits. Having
additional lead time (which could also result in additional time to
generate credits for some small refiners) could help to ensure that
there would be sufficient credits available and that there would be a
robust credit trading market. Furthermore, offering two years of
additional credit life for credits traded to small refiners, as
discussed in section VII.D.3.b, would improve credit availability.
Lastly, we recognize that while the proposed benzene standard may
be achieved using the four technologies suggested above, new
technologies may also be developed that may reduce the capital and/or
operational costs. Thus, we believe that allowing small refiners some
additional time for newer technologies to be proven out by other
refiners would have the added benefit of reducing the risks faced by
small refiners. The added time would likely allow for small refiners to
benefit from the lower costs of these technologies. This would help to
offset the potentially disproportionate financial burden facing small
refiners.
We discuss below the provisions that we are proposing to help
mitigate the effects on small refiners. Small refiners that chose to
make use of the small refiner delayed provision would also delay, to
some extent, the benzene emission reductions that would otherwise have
been achieved. However, the overall impact of these postponed
reductions would be

[[Page 15878]]

reasonable, for several reasons. Small refiners represent a relatively
small fraction of national gasoline production. Our current estimates
(of refiners that we expect would qualify as small refiners) indicate
that these refiners produce about 2.5 percent of the total gasoline
pool. In addition, these small refiners are generally dispersed
geographically across the country and the gasoline that they produce is
sometimes transported to other areas, so the limited loss in benzene
emissions reduction would also be dispersed. Finally, absent small
refiner flexibility, EPA would likely have to consider setting a less
stringent benzene standard or delaying the overall program (until the
burden of the program on many small refiners was diminished), which
would serve to reduce and delay the air quality benefits of the overall
program. By providing temporary relief to small refiners, we are able
to adopt a program that would reduce benzene emissions in a timely and
feasible manner for the industry as a whole.
The proposed small refiner provisions should be viewed as a subset
of the hardship provisions described in section VII.E.2.b. Rather than
dealing with many refineries on a case-by-case basis through the
general hardship provisions (described later), we limit the number by
proposing to provide predetermined types of relief to a subset of
refineries based on criteria designed to identify refineries most
likely to be in need of such automatic relief.
b. How Do We Propose To Define Small Refiners for the Purpose of the
Hardship Provisions?
The definition of small refiner for this proposed program is in
most ways the same as our small refiner definitions in the Gasoline
Sulfur and Highway and Nonroad Diesel rules. These definitions, in
turn, were based on the criteria use by the Small Business
Administration. However, we are proposing to clarify some ambiguities
about the definition that have existed in the past.
A small refiner would need to demonstrate that it met all of the
following criteria:
Produced gasoline from crude during calendar year 2005.
Small refiner provisions would be limited to refiners of gasoline
from crude because they would be the ones that bore the investment
burden and therefore the inherent economic hardship. Therefore,
blenders and importers would not be eligible, nor would be additive
component producers.
Small refiner status would be limited to refiners that owned and
operated the refinery during the period from January 1, 2005 through
December 31, 2005. New owners that purchased a refinery after that date
would do so with full knowledge of the proposed regulations, and should
have planned to comply along with their purchase decisions. As with the
earlier fuel rules, we are proposing that a refiner that restarts a
refinery in the future may be eligible for small refiner status. Thus,
a refiner restarting a refinery that was shut down or non-operational
between January 1, 2005 and January 1, 2006 could apply for small
refiner status. In such cases, we would judge eligibility under the
employment and crude oil capacity criteria based on the most recent 12
consecutive months prior to the application, unless we conclude from
data provided by the refiner that another period of time is more
appropriate. However, unlike past fuel rules, we propose to limit this
to a company that owned the refinery at the time that it was shut down.
New purchasers would not be eligible for small refiner status for the
same reasons described above. Companies with refineries built after
January 1, 2005 would also not be eligible for the small refiner
hardship provisions.

--Had no more than 1,500 employees, based on the average number of
employees for all pay periods from January 1, 2005 to January 1, 2006;
and,
--Had a crude oil capacity less than or equal to 155,000 barrels per
calendar day (bpcd) for 2005.

In determining its total number of employees and crude oil
capacity, a refiner would need to include the number of employees and
crude oil capacity of any subsidiary companies, any parent companies,
any subsidiaries of the parent companies, and any joint venture
partners. There has been some confusion in past rules regarding how
these provisions were interpreted, and as a result, we are proposing to
clarify (and, in some cases, modify) them here. For example, in
previous rules we defined a subsidiary to be a company in which the
refiner or its parent(s) has a 50 percent or greater interest. We
realize that it is possible for a parent to have controlling ownership
interest in a subsidiary despite having less than 50 percent ownership.
Similarly, we realize that it is also possible for multiple parents to
each have less than 50 percent ownership interest but still maintain a
controlling ownership interest. Therefore, in order to clarify our
rules, we are proposing to define a parent company as any company (or
companies) with controlling interest, and to define a subsidiary of a
company to mean any company in which the refiner or its parent(s) has a
controlling ownership interest. In many cases, there are likely to be
multiple layers of parent companies, with the ultimate parent being the
one for which no one else has controlling interest. The employees and
crude capacity of all parent companies, and all subsidiaries of all
parent companies, would thus be taken into consideration when
evaluating compliance with these criteria.
As with our earlier fuel sulfur regulations, we are also proposing
today that refiners owned and controlled by an Alaska Regional or
Village Corporation organized under the Alaska Native Claims Settlement
Act, would also be eligible for small refiner status, based only on the
refiner's employees and crude oil capacity.\265\
---------------------------------------------------------------------------

\265\ 43 U.S.C. 1626.
---------------------------------------------------------------------------

c. What Options Would Be Available For Small Refiners?
We are proposing several provisions today to help reduce the
burdens on small refiners, as discussed above. In addition, these
provisions would also allow for incentives for small refiners that make
reductions to their benzene levels.
i. Delay in Standards
We propose that small refiners be allowed to postpone compliance
with the proposed benzene standard until January 1, 2015, which is four
years after the general program would begin. While all refiners would
be allowed some lead time before the general proposed program began, we
believe that in general small refiners would still face
disproportionate challenges. The proposed four-year delay for small
refiners would help mitigate these challenges. Further, previous EPA
fuel programs have included two to four year delays in the start date
of the effective standards for small refiners, consistent with the lead
time we believe appropriate here.
Small refiners have indicated to us that an extension of available
lead time would allow them to more efficiently carry out necessary
capital projects with less direct competition with non-small refiners
for financing and for contractor to carry out capital improvements.
There appears to be merit in this position, and we propose that
approved small refiners have four years of additional lead time. This
would provide three years after the 2012 review of the program, which
we believe would be enough time for such

[[Page 15879]]

refiners to complete necessary capital projects if they chose to pursue
them.
ii. ABT Credit Generation Opportunities
While we have anticipated that many small refiners would likely
find it more economical to purchase credits for compliance, some have
indicated they would make reductions to their gasoline benzene levels
to meet the proposed benzene standard. Further, a few small refiners
indicated that they would likely do so earlier than would be required
by the January 1, 2015 proposed small refiner start date. Therefore, we
are proposing that early credit generation be allowed for small
refiners that take steps to meet the benzene requirement prior to their
effective date. Small refiner credit generation would be governed by
the same rules as the general program, described above in section
VII.D, the only difference being that small refiners would have an
extended early credit generation period of up to seven years. Early
credits could be generated by small refiners making qualifying
reductions from June 1, 2007 to December 31, 2014, after which credits
could be generated indefinitely for those that overcomplied with the
standard.
iii. Extended Credit Life
As discussed previously, in order to encourage the trading of
credits to small refiners, we are proposing that the useful life of
credits be extended by 2 years if they are generated by or traded to
small refiners. This is meant to directly address concerns expressed by
small refiners that they would be unable to rely on the credit market
to avoid large capital costs for benzene control.
iv. ABT Program Review
As previously stated, we are anticipating that it may be more
economically sound for some refiners to purchase and use credits.
During discussions with small refiners, all of the small refiners
voiced their concerns about reliance on a credit market for compliance
with the benzene standard. Specifically, small refiners feared that:
(1) there could be a shortage of credits, (2) that larger refiners
would not trade credits with smaller refiners, and (3) that the cost of
credits could be so high that the option to purchase credits for
compliance would not be a viable option. Due to these concerns it was
suggested that EPA perform a review of the ABT program (and thus, the
small refiner flexibility options) by 2012, one year after the general
program begins.
Such a review would take into account the number of early credits
generated, as well as the number of credits generated and transferred
during the first year of the overall benzene control program. Further,
requiring the submission of pre-compliance reports from all refiners,
similar to the highway and nonroad diesel programs, would aid in
assessing the ABT program prior to performing the review. A small
refiner delay option of four years after the compliance date for other
refiners, coupled with a review after the first year of the overall
program, would still provide small refiners with roughly three years
that we believe would be needed to obtain financing and perform
engineering and construction. We are proposing to perform a review
within the first year of the overall program (i.e., by 2012). To aid
the review, we are also proposing the requirement that all refiners
submit refinery pre-compliance reports annually beginning June 1, 2008.
Refiners' 2011 annual compliance reports will be similar to the pre-
compliance reports, but the annual compliance reports will also contain
information such as credits generated, credits used, credits banked,
credit balance, cost of credits purchased. EPA would aggregate the data
(to protect individual refiners' confidentiality) and make the results
available to the industry. When combined with the four-year delay
option, this would provide small refiners (and others) with the
knowledge of the credit trading market's status before they would need
to make a decision to either purchase credits or to obtain financing to
invest in capital equipment.
Further, we are requesting comment on elements to be included in
the ABT program review, and suggested actions that could be taken
following such a review. Such elements could include:

--Revisiting the small refiner provisions if it is found that the
credit trading market did not exist to a sufficient degree to allow
them to purchase credits, or that credits were only available at a
cost-prohibitive price.
--Options to either help the credit market, or help small refiners gain
access to credits.

With respect to the first element, the SBAR Panel recommended that
EPA consider establishing an additional hardship provision to assist
any small refiners that were unable to comply with the benzene standard
even with a viable credit market. Such a hardship provision would
address the case of a small refiner for which compliance would be
feasible only through the purchase of credits, but it was not
economically feasible for the refiner to do so. This hardship would be
provided to a small refiner on a case-by-case basis following the
review and based on a summary, by the refiner, of technical or
financial infeasibility (or some other type of similar situation that
would render its compliance with the standard difficult). This hardship
provision might include further delays and/or a slightly relaxed
standard on an individual refinery basis for up to two years. Following
the two-year relief, a small refiner would be allowed to request
multiple extensions of the hardship until the refinery's material
situation changed. We are proposing the inclusion of such a hardship
provision which could be applied for following, and based on the
results of, the ABT program review.
With respect to the second element, the Panel recommended that EPA
develop options to help the credit market if it is found (following the
review) that there is not an ample supply of credits or that small
refiners are having difficulty obtaining credits. These options could
include the ``creation'' of credits by EPA that would be introduced
into the credit market to ensure that there are additional credits
available for small refiners. Another option the Panel discussed to
assist the credit market was to impose additional requirements to
encourage trading with small refiners. These could include a
requirement that a percentage of all credits sold be set aside and only
made available for small refiners. Similarly, we could require that
credits sold, or a certain percentage of credits sold, be made
available to small refiners before they are allowed to be sold to any
other refiners. Options such as these would help to ensure that small
refiners were able to purchase credits. One such recommendation by the
Panel, to extend credit life for small refiners, is included in today's
proposal and described above.
We welcome comment on additional measures that could be taken
following the review if it was found that there was a shortage of
credits or that credits were not available to small refiners.
d. How Would Refiners Apply for Small Refiner Status?
A refiner applying for status as a small refiner would be required
to apply and provide EPA with several types of information by December
31, 2007. (The detailed application requirements are summarized below.)
All refiners seeking small refiner status under this program would need
to apply for small refiner status, regardless of whether or not the
refiner had been approved for small refiner status under another fuel
program. As with applications for relief under other rules,
applications for small refiner status under this proposed rule

[[Page 15880]]

that were later found to contain false or inaccurate information would
be void ab initio.

Requirements for small refiner status applications:

--The total crude oil capacity as reported to the Energy Information
Administration (EIA) of the U.S. Department of Energy (DOE) for the
most recent 12 months of operation. This would include the capacity of
all refineries controlled by a refiner and by all subsidiaries and
parent companies and their subsidiaries. We would presume that the
information submitted to EIA is correct. (In cases where a company
disagreed with this information, the company could petition EPA with
appropriate data to correct the record when the company submitted its
application for small refiner status. EPA could accept such alternate
data at its discretion.)
--The name and address of each location where employees worked during
the 12 months preceding January 1, 2006; and the average number of
employees at each location during this time period. This would include
the employees of the refiner and all subsidiaries and parent companies
and their subsidiaries.
--In the case of a refiner who reactivated a refinery that was shutdown
or non-operational between January 1, 2005, and January 1, 2006, the
name and address of each location where employees worked since the
refiner reactivated the refinery and the average number of employees at
each location for each calendar year since the refiner reactivated the
refinery.
--The type of business activities carried out at each location.
--An indication of the small refiner option(s) the refiner intends to
use (for each refinery).
--Contact information for a corporate contact person, including: name,
mailing address, phone and fax numbers, e-mail address.
--A letter signed by the president, chief operating officer, or chief
executive officer of the company (or a designee) stating that the
information contained in the application was true to the best of his/her
knowledge and that the company owned the refinery as of January 1, 2007.
e. The Effect of Financial and Other Transactions on Small Refiner
Status and Small Refiner Relief Provisions
In situations where a small refiner loses its small refiner status
due to merger with a non-small refiner, acquisition of another refiner,
or acquisition by another refiner, we are proposing provisions which
are similar to those finalized in the nonroad diesel final rule to
allow for an additional 30 months of lead time. A complete discussion
of this provision is located in the preamble to the final nonroad
diesel rule.
2. General Hardship Provisions
Unlike previous fuel programs, today's program includes inherent
flexibility because there is a nationwide credit trading program.
Refiners would have the ability to avoid or minimize capital
investments indefinitely by purchasing credits, and we expect that many
refiners would utilize this option. We also expect that refiners and
importers who normally would produce or import gasoline that met the
proposed standard would periodically rely on credits in order to
achieve compliance. As discussed in section VII.D, we expect that
sufficient credits would be available on an annual basis to accommodate
the needs of the regulated industry, and we expect that these credits
would be available at prices that are comparable to the alternative
cost of making the capital investment necessary to produce compliant
gasoline. We are proposing to require that refiners submit pre-
compliance reports beginning in 2008. These reports would indicate how
the refinery plans to achieve compliance with the 0.62 vol% standard as
well as the amount of credits expected to be generated or expected to
be needed. The information provided in these reports would enable an
assessment of the robustness of the credit market and the ability of
refiners to rely on credits as the program began.
Although we expect credits to be available at competitive prices to
those who need them, we are proposing hardship provisions to
accommodate an inability to comply with the proposed standard at the
start of the program, and to deal with unforeseen circumstances. These
provisions would be available to all refiners, small and non-small,
though relief would be granted on a case-by-case basis following a
showing of certain requirements, primarily that compliance through the
use of credits was not feasible. We are proposing that any hardship
waiver would not be a total waiver of compliance. Rather, such a waiver
would allow the refiner to have an extended period of deficit
carryover. Under regular circumstances, our proposed deficit carryover
provision would allow an entity to be in deficit with the proposed
benzene standard for one year, provided that they made up the deficit
and were in compliance the next year. The proposed hardship provisions
would allow a deficit to be carried over for an extended, but limited,
time period. EPA would determine an appropriate extended deficit
carryover time period based on the nature and degree of the hardship,
as presented by the refiner in their hardship application, and on our
assessment of the credit market. Note that any waivers granted under
this proposed rule would be separate and apart from EPA's authority
under the Energy Policy Act to issue temporary waivers for extreme and
unusual supply circumstances, under section 211(c)(4).
a. Temporary Waivers Based on Unforeseen Circumstances
We are proposing a provision which, at our discretion, would permit
any refiner to seek a temporary waiver from the MSAT benzene standard
under certain rare circumstances. This waiver provision is similar to
provisions in prior fuel regulations. It is intended to provide
refiners relief in unanticipated circumstances--such as a refinery fire
or a natural disaster--that cannot be reasonably foreseen now or in the
near future.
Under this provision, a refiner could seek permission to extend the
deficit carryover provisions of the proposal for more than the one year
already allowed if it could demonstrate that the magnitude of the
impact was so severe as to require such an extension. We are proposing
that the refiner would be required to show that: (1) The waiver would
be in the public interest; (2) the refiner was not able to avoid the
nonconformity; (3) it would meet the proposed benzene standard as
expeditiously as possible; (4) it would make up the air quality
detriment associated with the nonconforming gasoline, where
practicable; and (5) it would pay to the U.S. Treasury an amount equal
to the economic benefit of the nonconformity less the amount expended
to make up the air quality detriment. These conditions are similar to
those in the RFG, Tier 2 gasoline sulfur, and the highway and nonroad
diesel regulations, and are necessary and appropriate to ensure that
any waivers that were granted would be limited in scope.
As discussed, such a request would be based on the refiner's
inability to produce compliant gasoline at the affected facility due to
extreme and unusual circumstances outside the refiner's control that
could not have been avoided through the exercise of due diligence. The
hardship request would also need to show that other avenues for
mitigating the problem,

[[Page 15881]]

such as the purchase of credits toward compliance under the proposed
credit provisions, had been pursued and yet were insufficient or
unavailable. Especially in light of the credit flexibilities built into
the proposed overall program, we expect that the need for additional
relief would be rare.
b. Temporary Waivers Based on Extreme Hardship Circumstances
In addition to the provision for short-term relief in extreme
unforeseen circumstances, we are also proposing a hardship provision
where a refiner could receive an extension of the deficit carryover
provisions based on extreme hardship circumstances. Such hardship could
exist based on severe economic or physical lead time limitations of the
refinery to comply with the benzene standard at the start of the
program, and if they were unable to procure sufficient credits. A
refiner seeking such hardship relief under this proposed rule would
have to demonstrate that these criteria were met. In addition to
showing that unusual circumstances exist that impose extreme hardship
in meeting the proposed standard, the refiner would have to show (1)
best efforts to comply, including through the purchase of credits, (2)
the relief granted under this provision would be in the public
interest, (3) that the environmental impact would be acceptable, and
(4) that it has active plans to meet the requirements as expeditiously
as possible. Because such a demonstration could not be made prior to
the development of the credit market, EPA would not begin to consider
such hardship requests until August 1, 2010, that is, until after the
final pre-compliance reports are submitted. Consequently, requests for
such hardship relief would have to be received prior to January 1, 2011.
If hardship relief under these circumstances was approved, we would
expect to impose appropriate conditions to ensure that the refiner was
making best efforts to achieve compliance offsetting any loss of
emission control from the program through the deficit carryforward
provisions. We believe that providing short-term relief to those
refiners that need additional time due to hardship circumstances would
help to facilitate the adoption of the overall MSAT program for the
majority of the industry. However, we do not intend for hardship waiver
provisions to encourage refiners to delay planning and investments they
would otherwise make. Again, because of the flexibilities of the
proposed overall program, we expect that the need for additional relief
would be rare.
c. Early Compliance With the Proposed Benzene Standard
We are also requesting comment on a means for allowing refineries,
under certain conditions, to meet the proposed benzene standard early
in lieu of MSAT1. In order to meet the proposed benzene standard early,
refiners would need to meet several criteria similar to those used in
the past when EPA has adjusted refinery baselines under the MSAT1
program. Specifically, the eligibility for such provisions would be
limited to refiners that have historically had better than average
toxics performance, lower than average benzene and sulfur levels, and a
significant volume of gasoline impacted by the phase-out of MTBE as an
oxygenate. The result of not allowing such early compliance could be
less supply of their cleaner fuel and more supply of fuel with higher
toxics emissions, with a worsening of overall environmental performance
under MSAT1. A refiner opting into such provisions would not be allowed
to generate benzene credits on the affected fuel prior to 2011, since
an ability to reduce benzene further would presumably negate the need
for an early compliance option.

F. Technological Feasibility of Gasoline Benzene Reduction

This section summarizes our assessment of the feasibility for the
refining industry to reduce benzene levels in gasoline to an average of
0.62 vol% starting January 1, 2011. Based on this assessment, we
believe that it is technologically feasible for refiners to meet the
benzene standard by the start date using technologies that are
currently available.
We begin this section by describing where benzene comes from and
the current levels found in gasoline. Next we discuss the benzene
reduction technologies available to refiners today and how they are
expected to be used to meet the proposed benzene standard. Then we
provide our analysis of the lead time necessary for complying with the
benzene standard. All of these issues are discussed in more detail in
Chapters 6 and 9 of the Regulatory Impact Analysis.
1. Benzene Levels in Gasoline
EPA receives information on gasoline quality, including benzene
levels, from each refinery and importer in the U.S. under the reporting
requirements of the RFG and CG programs. As discussed earlier in this
section, benzene levels averaged 0.94 vol% for gasoline produced in and
imported into the U.S. in 2003, which is the most recent year for which
complete data is available. However, for individual refineries, daily
batch gasoline benzene levels and annual average levels can vary
significantly from the national average. As indicated earlier in
describing our decision-making process for the type and level of
gasoline benzene standard, it is very important to understand how
current benzene levels vary by individual refinery, by region, as well
as day-to-day by batch.
The variability in 2003 average annual gasoline benzene levels by
individual refinery is shown in Figure VII.F-1. This figure contains a
summary of annual average gasoline benzene levels by individual
refinery for CG and RFG versus the cumulative volume of gasoline produced.

[[Page 15882]]
[GRAPHIC]
[TIFF OMITTED]
TP29MR06.007

Figure VII.F-1 shows that the annual average benzene levels of CG
as produced by individual refineries varies from 0.29 to 4.01 vol%.
Based on the data in the figure, the volume-weighted average benzene
content for U.S. CG is 1.10 vol%. As expected, the annual average
benzene levels of RFG as produced by individual refineries are lower,
ranging from 0.10 to 1.09 vol%. The volume-weighted average benzene
content for U.S. RFG (not including California) is 0.62 vol%.
The information presented for annual average gasoline benzene
levels does not illustrate the very large day-to-day variability in
gasoline batches produced by each refinery. We evaluated the batch-by-
batch gasoline benzene levels for several refineries that produce both
RFG and CG, using information submitted to EPA as part of the reporting
requirements for the RFG and CG Anti-dumping Programs. One refinery had
no particular trend for its CG benzene levels, with benzene levels that
varied from 0.1 to 3 vol%. That same refinery's RFG averaged around
0.95 vol% benzene, ranging from 0.05 to 1.1 vol%. The second refinery
had RFG benzene levels that averaged around 0.4 vol% ranging from 0.1
to 1.0 vol%. Its CG benzene levels averaged about 0.6 vol% with batches
that ranged from 0.1 to 1.2 vol%. The batches for both RFG and CG
varied on a day-to-day basis and, overall, by over an order of
magnitude. It is clear from our review of batch-by-batch data submitted
to EPA that benzene variability is typical of refineries nationwide.
There are several contributing factors to the variability in
refinery gasoline benzene levels across all the refineries. We will
review these factors and describe how each impacts batch-by-batch and
annual average gasoline benzene levels.
The first factor contributing to the variability in gasoline
benzene levels is crude oil quality. Each refinery processes a
particular crude oil slate, which tends to be fairly constant except
for seasonal changes that reflect changes in product demand. Crude oil
varies greatly in aromatics content. Since benzene is an aromatic
compound, its level tends to vary with the aromatics content of crude
oil. For example, Alaskan North Slope crude oil contains a high
percentage of aromatics. Refiners processing this crude oil in their
refineries shared with us that their straight run naphtha contains on
the order of 3 vol% benzene (the production of naphtha is discussed
further below). This is one reason why the gasoline in PADD 5 outside
of California is high in benzene. Conversely, refiners that process
very paraffinic crude oils (low in aromatics) usually have a low amount
of benzene in their straight run naphtha. Because crude oil supplies
tend to be constant over periods of months, crude oil quality is not a
major contributor to day-to-day variations in benzene among gasoline
batches. However, because crude oil supplies often vary from refinery
to refinery, differences in crude quality are an important factor in
the variability among refineries.
The second factor contributing to the variability in benzene levels
is differences in the types of processing units and gasoline
blendstocks among refineries. If a refinery is operated to rely on its
reformer for virtually all of

[[Page 15883]]

its octane needs--especially the type that operates at higher pressures
and temperatures and thus tends to produce more benzene--it will likely
have a high benzene level in its gasoline. Refineries with a reformer
and without a fluidized catalytic cracking (FCC) unit are particularly
prone to higher benzene levels, since they rely heavily on the product
of the reformer (reformate) to meet octane needs. However, refineries
that can rely on other means for boosting their gasoline octane can
usually rely less on the reformer and can run this unit at a lower
severity, resulting in less benzene in their gasoline pool. Examples of
such other octane-boosting refinery units include the alkylation unit,
the isomerization unit and units that produce oxygenates. Refiners may
have these units in their refineries, or in many cases, they can
purchase the gasoline blendstocks produced by these units from other
refineries or third-party producers. The blending of the products of
these processes--alkylate, isomerate, and oxygenates--into the gasoline
pool provides a significant octane contribution, which can allow
refiners to rely less on the octane from reformate. Since refiners make
individual decisions about producing or purchasing different
blendstocks for each refinery, this variation is another important
contributor to differences in gasoline benzene content among
refineries. In addition, the variation in gasoline blendstocks used to
produce different batches of gasoline is by far the most important
factor in the drastically differing benzene levels among batches of
gasoline at any given refinery.
This practice by refiners of producing or purchasing different
blendstocks and blending them in different ways to produce gasoline is
an integral and essential aspect of the refining business. Thus, in
designing an effective benzene control program, it is critical that
benzene levels be reduced while refiners retain the ability to change
blendstocks (and crude supplies) as needed from batch to batch and
refinery to refinery. We believe that the proposed program accomplishes
these goals.
A third important source of variability in existing benzene levels
in gasoline is the fact that many refiners are already operating their
refineries today to intentionally reduce benzene levels in their
gasoline, while others are not. For example, refiners that are
currently producing RFG must ensure their RFG averages 0.95 vol% or
less and is always under the 1.3 vol% cap (see discussion of the
current toxics program in section VII.C.5 above). Similarly, refiners
producing gasoline to comply the California RFG program need to produce
gasoline with reduced benzene. These refiners are generally using
benzene control technologies to actively produce gasoline with lower
benzene levels. If they are producing CG along with the RFG, their CG
is usually lower in benzene as well compared with the CG produced by
other refiners, since the benzene control technology often affects some
of the streams used to blend CG. In addition, some refiners add
specific refinery units such as benzene extraction to intentionally
produce chemical-grade benzene. Benzene commands a much higher price on
the chemical market compared to the price of gasoline. For these
refiners, the profit from the sale of benzene pays for the equipment
upgrades needed to greatly reduce the levels of benzene in their
gasoline. In most cases, refineries with extraction units are marketing
their low-benzene gasoline in the RFG areas.
The use of these benzene control technologies by some refiners
contributes to the variability in gasoline benzene levels among
refineries. The use of these technologies can also contribute to the
batch-to-batch variability in benzene levels. This is because, as with
different blendstocks, refiners need to be able to change the operating
characteristics of these technologies to meet varying needs in gasoline
quality. In addition, planned or unexpected shut-downs of benzene
control equipment may result in temporarily high batch benzene levels
relative to the normally low gasoline levels when the unit is operating.
The variations in gasoline benzene levels among refineries also
lead to variations in benzene levels among regions of the country.
Table VII.F-1 shows the average gasoline benzene levels for all
gasoline produced in (and imported into) the U.S. by PADD for 2003. The
information is presented for both CG and RFG.

Table VII.F-1.--Benzene Levels by Gasoline Type Produced in or Imported Into Each PADD in 2003
----------------------------------------------------------------------------------------------------------------
PADD 1 PADD 2 PADD 3 PADD 4 PADD 5 CA U.S.
----------------------------------------------------------------------------------------------------------------
Conventional Gasoline............................ 0.84 1.39 0.94 1.54 1.79 0.63 1.11
Reformulated Gasoline............................ 0.60 0.82 0.56 n/a n/a 0.62 0.62
Gasoline Average................................. 0.70 1.28 0.87 1.54 1.79 0.62 0.94
----------------------------------------------------------------------------------------------------------------

Table VII.F-1 shows that benzene levels vary fairly widely across
different regions of the country. PADD 1 and 3 benzene levels are lower
because the refineries in these regions produce a high percentage of
RFG for both the Northeast and Gulf Coast. Also, a number of refineries
in these two regions are extracting benzene for sale into the chemicals
market, contributing to the much lower benzene level in these PADDs. It
is interesting to note that, in addition to RFG, CG benzene levels are
low in PADDs 1 and 3. There are two reasons for this. First, some RFG
produced by refineries ends up being sold as CG. Second, as mentioned
above, refiners that are reducing the benzene levels in their RFG
generally also impact the benzene levels in their CG. In contrast,
other parts of the U.S. with little to no RFG production and little
extraction have much higher benzene levels.
2. Technologies for Reducing Gasoline Benzene Levels
a. Why Is Benzene Found in Gasoline?
To discuss benzene reduction technologies, it is helpful to first
review some of the basics of refinery operations. Refineries process
crude oil into usable products such as gasoline, diesel fuel and jet
fuel. For a typical crude oil, about 50 percent of the crude oil falls
within the boiling range of gasoline, jet fuel and diesel fuel. The
rest of crude oil boils at too high a temperature to be blended
directly into these products and therefore must be cracked into lighter
compounds. Material that boils within the gasoline boiling range is
called naphtha. There are two principal sources of naphtha. The first
is ``straight run'' naphtha, which comes directly off of the crude oil
atmospheric distillation column. Another principle source of naphtha is
that generated from the cracking reactions. Each type of naphtha
contributes to benzene in gasoline.
Typically, little of the benzene in gasoline comes from benzene
naturally

[[Page 15884]]

occurring in crude oil. Straight run naphtha, which comes directly from
the distillation of crude oil, thus tends to have a low benzene
content, although it can contain anywhere from 0.3 to 3 vol% benzene.
While straight run naphtha is in the correct distillation range to be
usable as gasoline, its octane value is too low for blending directly
into gasoline. Thus, the octane value of this material must be
increased to enable it to be used as a gasoline blendstock.
The primary means for increasing the octane value of naphtha
(whether straight run or from cracking processes) is reforming.
Reforming reacts the heavier portion of straight run naphtha (six-
carbon material and heavier) over a precious metal catalyst at a high
temperature. The reforming process converts many of the naphtha
compounds to aromatic compounds, which raises the octane of this
reformate stream to over 90 octane numbers. (``Octane number'' is the
unit of octane value.) Since benzene is an aromatic compound, it is
produced along with toluene and xylene, the other primary aromatic
compounds found in gasoline. The reforming process increases the
benzene content of the straight run naphtha stream from 0.3 to 3 vol%
to 3 to 11 vol%.
There are two ways that benzene levels increase in the reformer
above the benzene levels occurring naturally in crude oil--the
conversion of non-aromatic six-carbon hydrocarbons into benzene, and
the cracking of heavier aromatic hydrocarbon compounds into
benzene.\266\ In the discussion below about how benzene in the
reformate stream can be reduced, we elaborate further about the
opportunities that refiners have to manage both of these benzene-
producing processes.
---------------------------------------------------------------------------

\266\ In the process of converting the straight run naphtha into
aromatics, a significant amount of hydrogen is produced that is
critical for the various hydrotreating operations in refineries. As
discussed later, the impact on hydrogen production is an important
consideration in reducing benzene levels.
---------------------------------------------------------------------------

Three factors contribute to the wide range in benzene levels in the
reformate stream, and these factors are important in the decisions
refiners would make in response to the proposed benzene control
program. First, different feedstocks contain different amounts of
benzene and different levels of benzene precursors that are more or
less capable of being converted to benzene by the reformer. Second, the
type of reformer being used affects how much benzene is produced during
the reforming process. For example, refineries with the older, higher
pressure reformers tend to form more benzene by cracking heavier
aromatics than refineries with newer, lower pressure units. Third, the
severity with which the reformer is being operated also affects benzene
levels in reformate. The greater the severity at which the reformer is
operated, the greater the conversion of feedstocks to aromatics (and
the more hydrogen is produced). However, more severe operation shortens
the time between the catalyst regeneration events that the reformer
must periodically undergo. Greater severity also lowers the gasoline
yield from this unit. Because refiners balance these operation and
production factors individually at each refinery in deciding on how
severely to operate the reformer, these decisions contribute to the
range of benzene levels found in reformate from refinery to refinery.
In addition to benzene occurring in the reformate stream, another
source of benzene in gasoline is naphtha produced from cracking
processes. There are three primary cracking processes in the refinery--
the FCC unit, the hydrocracker, and the coker. The naphthas produced by
these cracking processes contain anywhere from 0.5 to 5 vol% benzene.
The benzene in these streams is typically formed from the cracking of
heavier aromatic compounds into lighter compounds that can then be
blended into gasoline. The benzene content of cracked streams is
therefore largely a function of the aromatics content of the crude oil
feedstocks and the need of a particular refinery to produce gasoline
from heavier feedstocks. As we discuss later, we do not expect that
benzene reductions from these cracked naphthas would be a major avenue
for compliance with the proposed benzene control program for most refiners.
Finally, there are other intermediate streams that contribute to
benzene in gasoline but that have such low benzene content or are found
in such low volumes in gasoline that they are of very limited
importance in reducing benzene levels. Examples of these are light
straight run naphtha and the oxygenates MTBE and ethanol.
Table VII.F-2 summarizes the typical ranges in benzene content and
average percentages of gasoline of the various intermediate streams
that are blended to produce gasoline.

Table VII.F-2.--Benzene Content and Typical Gasoline Fraction of Various
Gasoline Blendstocks
------------------------------------------------------------------------
Average volume
Process or blendstock name Typical benzene in gasoline
level (vol%) (percent)
------------------------------------------------------------------------
Reformate.......................... 3-11 30
FCC Naphtha........................ 0.5-2 36
Alkylate........................... 0 12
Isomerate.......................... 0 4
Hydrocrackate...................... 1-5 3
Butane............................. 0 4
Light Straight Run................. 0.3-3 4
MTBE/Ethanol...................... 0.05 3
Natural Gasoline................... 0.3-3 3
Coker Naphtha...................... 3 1
------------------------------------------------------------------------

Table VII.F-2 shows that the principal contributor of benzene to
gasoline is reformate. This is due both to its high benzene content and
the relatively large gasoline fraction that reformate comprises of the
gasoline pool. The product stream from the reformer, reformate,
accounts for between 15 and 50 percent of the content of gasoline,

[[Page 15885]]

depending on the refinery (typically about 35 percent.) For this reason
and as discussed below, reducing the benzene in reformate is the
primary focus of the various benzene reduction technologies available
to refiners. Control of benzene from the other streams quickly becomes
cost prohibitive due to either the low concentration of benzene in the
stream, the low volume of the stream, or both.
b. Benzene Control Technologies Related to the Reformer
There are several technologies that reduce gasoline benzene by
controlling the benzene in the feedstock to and the product stream from
the reformer.\267\ One approach is to route the intermediate refiner
streams that have the greatest tendency to form benzene in a way that
bypasses the reformer. This approach is very important in benzene
control, but it is limited in its effectiveness because it does not
address any of the naturally-occurring benzene and some of the benzene
formed in the reformer. For this reason, refiners often use a second
category of technologies that remove or destroy benzene, including both
the naturally occurring benzene as well as that formed in the reformer.
These technologies are isomerization, benzene saturation, and benzene
extraction. We discuss each of these approaches to benzene reduction
below. The effectiveness of these technologies in reducing the benzene
content of reformate varies from approximately 60% to 96%. The actual
impact on an individual refinery's finished gasoline benzene content,
however, will be a function of many different refinery-specific
factors, including the extent to which they are already utilizing one
of these technologies.
---------------------------------------------------------------------------

\267\ The benzene reduction technologies are discussed here in
the context of the feasibility for reducing the benzene levels of
gasoline to meet a gasoline benzene content standard. However, this
discussion applies equally to the feasibility of a total air toxics
standard, since we believe that benzene control would be the only
means that refiners would choose in order to comply with such a standard.
---------------------------------------------------------------------------

i. Routing Around the Reformer
The primary compounds that are converted to benzene by the
reforming unit are the six-carbon hydrocarbon compounds contained in
the straight run naphtha fed to the reformer. These compounds, along
with the naturally-occurring benzene in this straight run naphtha
stream, can be removed from the feedstock to the reforming unit using
the upstream distillation unit, bypassed around the reforming unit, and
then blended directly into gasoline. Routing these compounds around the
reformer prevents the formation of much of the benzene in the reformer,
though it does not reduce the naturally-occurring benzene.
For a typical refinery, the technology to route the six-carbon
material around the reformer would likely require only a small capital
investment. Compared with a scenario where all of this material goes to
the reformer, the combined rerouted and reformate streams would overall
have about 60 percent less benzene, and finished gasoline would have
about 31 percent less benzene. However, in most cases this would not be
sufficient to achieve a 0.62 vol% benzene standard, and some
combination of the technologies discussed next would also be needed.
ii. Routing to the Isomerization Unit
A variation of routing around the reformer involves the
isomerization of the re-routed benzene precursors. Rather than directly
blending the rerouted stream into gasoline, this stream can first be
processed in the isomerization unit. This has two main advantages.
First, it increases the effectiveness of benzene control, since the
isomerization process converts the naturally-occurring benzene in this
rerouted stream to another compound. Second, it recovers some of the
octane otherwise lost by the conversion of benzene.
The typical role of the isomerization unit is to convert five-
carbon hydrocarbons from straight-chain to branched-chain compounds,
thus increasing the octane value of this stream. If the isomerization
unit at a refinery has sufficient additional capacity to handle the
rerouted six-carbon hydrocarbons, that stream can also be sent to this
unit, where the benzene present in that stream would be saturated and
converted into another compound (cyclohexane). (This benzene saturation
process is similar to what occurs in a dedicated benzene saturation
unit, as described below.) Compared to a scenario where all this
material goes to the reformer, routing the six-carbon compounds to the
isomerization unit in this manner can reduce the benzene levels in the
combined rerouted and reformate streams by about 80 percent. The option
of isomerization is currently available to those refineries with
sufficient capacity in an existing isomerization unit to treat all of
the six-carbon material.
iii. Benzene Saturation
The function of a benzene saturation unit is to react hydrogen with
the benzene in the reformate (that is, to saturate the benzene) in a
dedicated reactor, converting the benzene to cyclohexane. Because
hydrogen is used in this process, refiners that choose this technology
need to ensure that they have a sufficient source of hydrogen. Refiners
cannot afford to saturate other aromatic compounds present in their
reformate as it would cause too great an octane loss. Thus, it is
necessary to separate a six-carbon stream, which contains the benzene,
from the rest of reformate, and only feed the six-carbon stream to the
benzene saturation unit. This separation is done with a distillation
unit called a reformate splitter placed just after the reformer.
There are two vendors that produce benzene saturation units. UOP
produces a technology named Bensat. There are at least six Bensat units
operating in the U.S. today and many more around the world. CDTech
licenses another, somewhat newer technology for this purpose called
CDHydro. There are six CDHydro units operating today, mostly outside of
the U.S. Benzene saturation can reduce benzene in the reformate by
about 96 percent.
iv. Benzene Extraction
Extraction is a technology that chemically removes benzene from
reformate. The removed benzene can be sold as a high-value product in
the chemicals market. To extract only benzene from the reformate, a
reformate splitter is installed just after the reformer to separate a
benzene-rich stream from the rest of the reformate. The benzene-rich
stream is sent to an extraction unit which separates the benzene from
the rest of the hydrocarbons. Since the benzene must be sufficiently
concentrated before it can be sold on the chemicals market, a very
thorough distillation step is incorporated with the extraction step to
concentrate the benzene to the necessary purity. Where it is economical
to use, benzene extraction can reduce benzene levels in the reformate
by 96 percent.
There are two important considerations refiners have with respect
to using benzene extraction. The first is the price of chemical grade
benzene. If the price of chemical grade benzene is sufficiently higher
than the price of gasoline, benzene extraction can realize an
attractive return on capital invested and is often chosen as a
technology for achieving benzene reduction. The difference in price
between benzene and gasoline has been significantly higher than its
historic levels during the last few years. While we expect that this
difference will return closer to the lower historic levels by the time
the proposed program

[[Page 15886]]

would be implemented, the difference in prices should still be
sufficient to make extraction a very cost-effective technology for
reducing gasoline benzene levels. A more detailed discussion about
benzene prices is contained later in this preamble (section IX) and in
Chapter 9 of the RIA.
The other consideration in using benzene extraction is the distance
that a refinery is from the markets where benzene is used as a chemical
feedstock. Transportation of chemical grade benzene requires special
hazardous-materials precautions, including protection against leaks.
Certain precautions are also necessary to preserve the purity of the
benzene during shipment. These special precautions are costly for
shipping benzene over long distances. Thus if a refinery were located
far from the chemical benzene markets, the economics for using
extraction would be much less attractive compared to that of refiners
located near benzene markets.
The result has been that chemical grade benzene production has been
limited to those refineries located near the benzene markets. This
includes refineries on the Gulf and on the East Coast and to a limited
extent, several refineries in the Midwest. This could change if the
very high benzene prices in 2004 and the beginning of 2005 were to
continue, instead of returning to lower historical levels. However,
even if benzene prices remain high by the time that a benzene control
standard would take effect, refineries located away from the benzene
markets may be concerned that the higher benzene prices may not be
certain enough for the long term to warrant investment in extraction.
Our analysis for today's proposal conservatively assumes that only
refineries on the Gulf and East coasts would choose to use benzene
extraction to lower their gasoline benzene levels. Despite some
existing extraction units in the Midwest, the benzene market there is
small and no additional benzene extraction is assumed to occur there.
c. Other Benzene Reduction Technologies
We are aware of other, less attractive technologies capable of
achieving benzene reductions in gasoline. These technologies tend to
have more serious impacts on other important refinery processes or on
fuel quality and are generally capable of only modest benzene
reductions. We do not currently have sufficient information about how
widely these approaches are or could be utilized or their potential
costs, and in our modeling we have not assumed that refiners would use
them. However, because they may be feasible in some unique situations,
we mention these potential gasoline benzene reduction approaches here.
One of these less attractive opportunities for additional benzene
reduction would be for refiners to capture more of the reformate
benzene in the reformate splitter and send this additional benzene to
the saturation unit. Refiners attempt to minimize both the capital and
operating costs when splitting a benzene-rich stream out of the
reformate stream for treating in a benzene saturation unit. To do this,
they optimize the distillation cut between benzene and toluene, thus
achieving a benzene reduction of about 96 percent in the reformate
while preserving all but about 1 percent of the high-octane toluene.
However, if a refiner were to be faced with a dire need for additional
benzene reductions, it could change its distillation cut to send the
last 4 percent of the benzene to the saturation unit. Since this cut
would also bring with it more toluene than the normal optimized
scenario, this toluene would also be saturated, resulting in a larger
loss in octane and greater hydrogen consumption.
Some refineries with hydrocracking units may have another means of
further reducing the gasoline benzene levels. They may be able to
reduce the benzene content of one of the products of the hydrocracker,
the light hydrocrackate stream. Today, light hydrocrackate is normally
blended directly into gasoline. Light hydrocrackate contains a moderate
level of benzene, although its contribution to the gasoline benzene
levels is significant only in those refineries with hydrocrackers.
Light hydrocrackate could be treated by routing this stream to an
isomerization unit, similar to how refiners isomerize the six-carbon
straight run naphtha as discussed above. Alternatively, refiners could
use additional distillation equipment to cut the light hydrocrackate
more finely. In this way, more of the benzene could be shifted to the
``medium'' hydrocrackate stream, which in most refineries is sent to
the reformer and thus would be treated along with the reformate.
Another way that we believe some refiners could further reduce
their benzene levels would be to treat the benzene in natural gasoline.
Many refiners, especially in PADDs 3 and 4, blend some light gasoline-
like material, which is a by-product of natural gas wells, into their
gasoline. In most cases, we believe that this material is blended
directly into gasoline. Because the benzene concentration in this
stream is not high, it would be costly to treat the stream to reduce
benzene. However, there could be other reasons that refiners might find
compelling for treating this stream. First, since its octane is fairly
low to begin with, it could be fed to the reformer and its benzene
would be treated in the reformate, along with the benefit of improving
the octane quality of this stream. Second, refiners producing low-
sulfur gasoline under the gasoline sulfur program may not be able to
easily tolerate the sulfur from this stream if it were blended directly
into gasoline. Thus, if they treat this stream in the reformer, it
would undergo the hydrotreating (desulfurization) that is necessary for
all streams fed to the reformer. Overall, we do not have sufficient
information to conclude whether treating natural gasoline might become
more attractive in the future.
Another approach to benzene reduction that we believe could be
attractive in certain unique circumstances relates to the benzene
content in naphtha from the fluidized catalytic cracker, or FCC unit.
As shown in Table VII.F-2, FCC naphtha contains less than 1 percent
benzene on average. Despite the very low concentration of benzene in
FCC naphtha, the large volumetric contribution of this stream to
gasoline results in this stream contributing a significant amount of
benzene to gasoline as well. There are no proven processes which treat
benzene in FCC naphtha. This is because its concentration is so low as
well as because FCC naphtha contains a high concentration of olefins.
Segregating a benzene-rich stream from FCC naphtha and sending it to a
benzene saturation unit would saturate the olefins in the same boiling
range, resulting in an unacceptable loss in octane value. Also, some
refiners operate their FCC units today more severely to improve octane,
an action that also increases benzene content. Conceivably, refiners
could redesign their FCC process (change the catalyst and operating
characteristics) to reduce the severity and produce slightly less
benzene. We do not have sufficient information to know whether many
refiners are already operating at high FCC severity and thus have the
potential to reduce benzene by reducing that severity.
We request comment on our assessment of benzene reduction
approaches, including data related to the current or potential usage
and potential effectiveness of each approach.

[[Page 15887]]

d. Impacts on Octane and Strategies for Recovering Octane Loss
All these benzene reduction technologies affect the octane of the
final gasoline. Regular grade gasoline must comply with a minimum 87
octane (R+M)/2 rating (or a sub-octane rating of 86 for driving in
altitude), while premium grade gasoline must comply with an octane
rating which ranges from 91 to 93 (R+M)/2. Gasoline must meet these
octane ratings to be sold as gasoline at retail. Routing the benzene
precursors around the reformer reduces the octane of the six-carbon
compound stream, which normally exits the reformer with the rest of the
reformate. Without these compounds in the reformate, a loss of octane
in the gasoline pool of about 0.14 octane numbers typically occurs. If
this rerouted stream can be sent to an isomerization unit, a portion of
this lost octane can be recovered, provided that sufficient capacity
remains in that unit to continue treating the five-carbon naphtha
compounds. Benzene saturation and benzene extraction both affect the
octane of reformate and therefore the gasoline pool. Benzene saturation
typically reduces the octane of gasoline by 0.24 octane numbers, and
benzene extraction typically reduces the octane by 0.14 octane numbers.
Refiners can recover the lost octane in a number of ways. First,
the reformer severity can be increased. However, if the refiner is
reducing benzene through precursor rerouting or saturation, this
strategy can be somewhat counterproductive. This is because increased
severity increases the amount of benzene in the reformate and thus
increases the cost of saturation and offsets some of the benzene
reduction of precursor rerouting. Increasing reformer severity would
also decrease the operating cycle life of the reformer, requiring more
frequent regeneration. However, where benzene extraction is used,
increased reformer severity can improve the economics of extraction
because not only is lost octane replaced but the amount of benzene
extracted is increased. Again, operating the reformer more severely
would have the negative impact of shortening the reformer's operating
cycle between regeneration events.
Lost octane can also be recovered by increasing the activity of
other octane-producing units at the refinery. As discussed above,
saturating benzene in the isomerization unit loses the octane value of
that benzene, but octane is increased by the simultaneous formation of
branch-chain compounds. Also, many refineries produce a high-octane
blendstock called alkylate. Alkylate is produced by reacting normal
butane and isobutane with isobutylene over an acid catalyst. Not only
is this stream high in octane, but it converts compounds that are too
volatile to be blended in large amounts into the gasoline pool into
heavier compounds that can be readily blended into gasoline. If the
refinery is short of feedstocks for alkylate, then the operations of
the FCC unit, which is the principal producer of these feedstocks, can
be adjusted to produce more of the feedstocks for the alkylate unit,
increasing the availability of this high octane blendstock.
Octane can also be increased by purchasing high-octane blendstocks
and blending them into the gasoline pool. For example, some refiners
with excess octane production capacity market high octane blendstocks
such as alkylate or aromatics such as toluene. Oxygenates, such as
ethanol, can also be blended into the gasoline pool. Other oxygenates
such as methyl tertiary butyl ether (MTBE), ethyl tertiary butyl ether
(ETBE), tertiary amyl methyl ether (TAME), and other ethers are
sometimes used. The availability and cost of oxygenates for octane
replacement vary according to material prices as well as state and
federal policies that either encourage or discourage their use. (For
example, the Energy Policy Act of 2005 requires an increase in the
volume of renewable fuels, including ethanol, which are blended into
gasoline).
e. Experience Using Benzene Control Technologies
All of the benzene reduction technologies and octane generating
technologies described above have been demonstrated in refineries in
the U.S. and abroad. All four of these technologies have been used for
compliance purposes for the federal RFG program, which has required
that benzene levels be reduced to an average of 0.95 vol% or lower
since 1995.
According to the Oil and Gas Journal's worldwide refining capacity
report for 2003, there were 27 refineries in the U.S. with extraction
units. Those refineries that chose extraction often reduced their
benzene to levels well below 0.95 vol% because of the value of benzene
as a chemical feedstock, as discussed above. Once a refiner invests in
extraction, they have a strong incentive to maximize benzene production
and thus the availability of benzene to sell to the chemical market,
often reducing gasoline benzene more than is required by regulation.
The RFG program also led to the installation of a small number of
benzene saturation units in the Midwest to produce RFG for the markets
there. California has its own RFG program which also put into place a
stringent benzene standard for the gasoline sold there. The Oil and Gas
Journal's Worldwide Refining Report shows that four California
refineries have benzene saturation units. If we assume that those RFG
and California refineries that do not have extraction or saturation
units are routing their precursors around their reformer, then there
are 28 refineries using benzene precursor rerouting as their means to
reduce benzene levels. Thus, these technologies have been demonstrated
in many refineries since the mid-1990s in the U.S. and are considered
by the refining community as commercially proven technologies.
Worldwide experience provides further evidence of the commercial
viability of these benzene control technologies. A vendor of benzene
control technology has shared with us how the refining companies in
other countries have controlled the benzene levels of their gasoline in
response to the benzene standards put in place there. In Europe,
benzene control is typically achieved by routing the benzene precursors
around the reformer and feeding that rerouted stream to an
isomerization unit. In Japan, much of the benzene is extracted from
gasoline and sold to the chemicals market. Finally, in Australia and
New Zealand, refiners tend to use benzene saturation to reduce the
benzene levels in their gasoline.
f. What Are the Potential Impacts of Benzene Control on Other Fuel
Properties?
With the complex nature of modern refinery operations, most changes
to fuel properties affect other fuel properties to some degree. In the
case of benzene control, the ``ripple effects'' on other fuel
properties tends to be limited. However, as discussed above, the
reduction in benzene content that we are proposing in this rule,
depending on how it is accomplished, would in most cases slightly
reduce the overall octane of the resulting gasoline. Refiners would
likely compensate by increasing the volume of reformate (other
aromatics) blended into the gasoline, requiring a small increase in
reformer severity and energy inputs. Some analysis of gasoline property
survey data suggests that as benzene is reduced in gasoline, other
aromatics may increase somewhat to help compensate.
Another option refiners might consider in response to the proposed
rule is match-blending ethanol to make up octane and increase supply volume.

[[Page 15888]]

This has been done for several years with MTBE as an economical way to
meet toxics performance requirements and octane targets for RFG. Like
MTBE, ethanol has a relatively high blending octane, and is already
added in many markets to take advantage of tax benefits or to support
local suppliers. Since the use of ethanol is being encouraged in the
recently-enacted energy legislation, refiners will likely seek to
capture the octane benefits as part of their process, which could help
offset the octane loss some refiners will see as a result of benzene
reduction processes. Furthermore, to the extent that current MTBE
production is shifted to production of isooctene, isooctane, and
alkylate, these compounds would be available as high-octane, low-
benzene gasoline blendstocks.
Finally, refiners may blend in isomerate or alkylate, which are
very ``clean'' gasoline blendstocks, thereby reducing the levels of
``dirtier'' gasoline blendstocks, and reducing overall sulfur, olefins,
and aromatics. We do not anticipate major changes in other fuel
properties due to reductions in benzene. Our modeling of the emissions
impacts of the proposed benzene standard does account for the modest
changes in other fuel properties. As discussed in section V of this
preamble and Chapter 2 of the RIA, this emissions modeling indicates
that the proposed benzene standard has negligible impacts on the
emissions of other mobile source air toxics.
3. Feasible Level of Benzene Control
A key aspect of our selection of the level of the proposed average
benzene standard of 0.62 vol% was our evaluation of the benzene levels
achievable by individual refineries. Our modeling analyses, which
combine our understanding of technological and economic factors, is
summarized in section IX below and discussed in detail in Chapter 9 of
the RIA. Later in this section we summarize our conclusions about the
overall feasibility of the program in terms of the requirements of the
Clean Air Act.
We assessed the benzene levels achievable for each refinery,
assuming that each refinery pursued the most stringent form of
reformate benzene control available to it--installing either benzene
saturation or extraction units. Based on this assessment, we project
that the most stringent benzene level achievable on average for all
U.S. gasoline would be 0.52 vol% benzene.\268\ As discussed above,
however, a standard at this level would require significant investment
at essentially all refineries--that is, near-universal installation of
either benzene saturation or benzene extraction capability. As
discussed in section IX below, this would be a very expensive result--
costing about three times more than the proposed program--that we do
not believe would be reasonable when costs are taken into account.
---------------------------------------------------------------------------

\268\ This analysis is within the constraints of our modeling
and the refinery-specific information available to us at the time of
this proposal.
---------------------------------------------------------------------------

Furthermore, the model projects that all refineries would use
optimal combinations of actual benzene reductions and/or credit
purchases and would meet the average standard without going beyond the
primary technologies of reformate benzene reduction discussed earlier
in this section. To reach this conclusion, our model assumes a fully
utilized credit trading program (that is, each refiner is assumed to
minimize its average costs and to freely trade credits among companies
so that all credits generated are used). Although the assumption of a
fully utilized credit trading program is appropriate for our modeling
purposes, it is very possible that this would not occur in practice.
For example, some refiners might choose to hold onto credits that they
generate, saving them for potential ``emergencies'' when unexpected
events would otherwise cause noncompliance with the benzene standard.
Given the high cost of control for some refineries and the
potential that credit trading would be less-than-fully utilized, we
have looked at standards less stringent than 0.52 vol% that might be
feasible, considering cost. Based on our modeling, we believe that with
the proposed ABT program all gasoline could be produced at the proposed
average level of 0.62 vol% without extreme economic consequences. We
believe that sufficient credits would be generated such that refineries
facing the highest costs of benzene control would have sufficient access
to credits and would not need to turn to cost prohibitive technologies.
From a strict feasibility standpoint, we have also assessed whether
all refineries could meet the proposed benzene level in cases where
sufficient credits were not available to every refinery that might want
them. We found that, despite the application of maximum reformate
benzene control in the refinery model to all refineries, the analysis
concluded that 13 refineries would still have benzene levels that
exceeded a 0.62 benzene level, with one refinery as high as 0.77 vol%
benzene. We have evaluated how these 13 refineries might use the other,
less attractive benzene control technologies discussed above (assuming
that an ABT option is not available to them).
The approach of capturing more of the reformate benzene in the
reformate splitter and sending this additional benzene to the
saturation unit would allow 7 of the 13 challenged refineries to reach
the 0.62 vol% level. Then, those refineries with a hydrocracker or a
coker could reduce the benzene content of the light hydrocrackate or
coker stream. This step would allow 5 more refineries to reach the
target level. Finally, the treatment of benzene in natural gasoline
would bring the remaining 1 refinery to the 0.62 vol% level or below.
(Because of our lack of information about the potential for reducing
the severity of the FCC unit, and because we do not believe that
reducing the benzene level of FCC naphtha is feasible, we did not
consider FCC options in this analysis.) Again, we expect that at the
proposed standard level of 0.62 vol% in the context of the proposed ABT
program, all refineries would be able to comply. This analysis
demonstrates that there are options, although extreme and costly, for
challenged refineries even if the ABT program does not fully function
as projected.
4. Lead Time
Our proposal for the gasoline benzene standard to begin on January
1, 2011 would allow about four years after we expect the rulemaking to
be finalized for refiners to comply with the program's requirements. As
discussed below, we believe that four years of lead time would allow
refiners sufficient time to install the capital equipment they would
need to lower their benzene levels, and would also allow this program
to avoid significant conflict with other fuel programs being
implemented around the same time. In addition, the ABT program would
allow the industry to phase in the program, through the early credit
provisions, so that significant benzene reductions would occur earlier
than the program start date. The credits earned could allow the
investment in higher capital cost and less cost-effective technologies
to be delayed relative to the program start date.
In recent years, the implementation of the gasoline sulfur and
highway diesel sulfur programs has provided an opportunity to observe
the response of the refining industry to major fuel control
requirements. Many refiners have demonstrated their ability to make
very large, expensive sulfur control modifications to their refineries
in less than four years, and in some cases significantly less. It is
helpful to

[[Page 15889]]

compare this sulfur control experience with the types of technologies
refiners would use to reduce benzene.
Refiners could implement approaches to benzene control that require
very little or no capital equipment, including routing of benzene
precursors around the reformer and the use of an existing isomerization
unit, with very little lead time requirements. We believe that
approaches using moderately complex capital equipment, including
improving the effectiveness of precursor rerouting and expanding
existing extraction capacity, would generally require one to two years
of lead time. Projects that involve the installation of new equipment,
including benzene saturation and extraction units, require more time,
generally two to three years. This includes time for the equipment
installation as well as related offsite equipment and any necessary
capital equipment for production of hydrogen or high-octane
blendstocks. Of all the benzene control approaches, benzene extraction
is closest in scope and complexity to the technologies the industry is
using for fuel sulfur control. In addition to the time needed for
planning and installing the extraction unit and related equipment,
extraction also requires time to install additional facilities for
storing extracted benzene and for loading it for transport. Thus, as
with the earlier programs, we believe the refiners choosing to add a
benzene extraction unit could in some cases need as much as four years
to complete the project. Overall, we believe that four years of lead
time would ensure that all refiners would have sufficient time to
comply, regardless of the benzene control technology they select.
Another factor in selecting an appropriate date to begin the
program is the timing of the implementation of other large fuel control
programs, especially the Nonroad Diesel rule.\269\ The 15 ppm sulfur
standard mandated by the Nonroad Diesel Fuel program applies to nonroad
diesel fuel in 2010 and to locomotive and marine diesel fuel in 2012.
Refiners modifying their refineries to produce either ultra low sulfur
nonroad or locomotive and marine diesel fuel will do so during the
several years prior to 2010 and 2012. For each of those start dates,
there is a progression of actions which includes planning, design,
construction and start-up all during the four year run-up toward the
start date of the program. For example, the engineering and
construction (E&C) industry will be busy designing and constructing
each of the units that will be installed. Different portions of the E&C
industry will be engaged at specific periods of time leading up to the
time that the unit is started up. For this reason, staggering the start
year of this benzene fuel standard with the start years for the Nonroad
Diesel program would help to avoid excessive demand on specific parts
of the E&C industry. The staggering of today's proposed program's start
date with those of the Nonroad Diesel program may also help refiners
that might be seeking to acquire capital through banks or other lending
institutions by spreading out the requests.
---------------------------------------------------------------------------

\269\ The months leading up to January 2010 will also be when
several small refiners and refiners that were granted hardship
relief will be implementing their gasoline sulfur programs. We
believe that any serious interference among implementation projects
that individual refiners might demonstrate during this time period
could be addressed under the small refiner or general hardship
provisions of the proposed rule.
---------------------------------------------------------------------------

We believe that the proposed implementation date of January 1, 2011
would minimize overlap and possible interference with the
implementation of the Nonroad Diesel rule. Implementation of the
proposed benzene standard one year earlier or one year later would
overlap directly with one of the two Nonroad Diesel implementation
dates. We also believe that the additional year of lead time, compared
to a 2010 start date, would make the program more effective. Because we
expect that the proposed ABT program would encourage many refiners to
reduce benzene levels early whenever possible, we believe that
significant benzene reductions would occur prior to 2011. We discuss
this expected early benzene reduction further as a part of the
description of the proposed ABT program in section VII.D above.
For these reasons, we are proposing that the gasoline benzene
standard be implemented beginning January 1, 2011. We request comment
on the issue of lead time, including data supporting four years or a
different length of time.
5. Issues
a. Small Refiners
Small refiners are technically capable of realizing a similar
benzene reduction from their gasoline as large refiners. Because of
economies of scale, however, some of the benzene control technologies
which would be more affordable for larger refineries would be much more
challenging and more expensive for small refiners. This is due to the
poorer economies of scale that the small refiners are faced with
installing capital into their refineries. Two of the benzene control
technologies discussed above would be particularly attractive to small
refiners for implementing into their refineries. These are benzene
precursor rerouting, and, if the refinery has an isomerization unit,
routing the benzene precursors to the isomerization unit. These
technologies would be attractive to small refiners because they would
require little or no capital investments to implement for reducing
their gasoline benzene levels. Therefore, the per-gallon cost of these
two technologies is about the same as that for large refineries.
Smaller refineries tend to have fewer process units and blending
streams, which generally also means that they will have fewer options
for recovering lost octane. For example, these refineries are less
likely to have an alkylation unit. An alkylation unit gives refiners
short on octane the option to change the operations of their FCC unit
to make more olefins and then send the appropriate olefins to their
alkylation unit to produce more of that high octane blendstock. This is
not an option for several of the small refiners that do not have an
alkylation unit. Also, small refineries are more likely to have a
higher pressure reforming unit. The higher pressure reformer units tend
to produce more benzene from the cracking of heavier aromatic compounds
and will tend to do this more as their severity is increased. A higher
pressure reformer also has a more difficult regeneration cycle and
shorter cycle lengths as it is operated more severely. Thus, while
other refiners with lower pressure units may be able to increase the
severity of their reformers to make more octane without producing much
more benzene and greatly reducing the cycle lengths of their reformers,
many of the small refiners may not have as much flexibility in this
area. In any event, these greater technological challenges can be
offset somewhat where it is economical to purchase high octane blendstocks
or oxygenates from other refiners or from the petrochemical industry.
b. Imported Gasoline
Although the majority of petroleum products in the U.S. are made
from imported crude oil, only about five percent of the gasoline
consumed in this country was imported as finished gasoline in 2003.
This imported fuel is approximately half RFG and half CG, and had an
average benzene content of 0.8% volume in 2003. No batches of imported
gasoline had a benzene level above 2.4%. Over 90% of the imported
gasoline was delivered into the East Coast and Florida, with about 5%
arriving on the West Coast, and the

[[Page 15890]]

remainder being brought into other regions of the country. The origin
of the majority of this gasoline was Canada (40%), Western Europe
(31%), and South America (17%).
Since imported finished gasoline is not processed in a domestic
refinery, where refiners would be taking steps to meet the proposed
benzene standard, importers would be affected in other ways. Importers
would most likely either begin to purchase gasoline that is low enough
in benzene to meet the standard, or they would continue to import
gasoline with benzene at current levels but purchase credits to cover
the fuel being above the standard. As shown above, over 70 percent of
imported gasoline comes from countries that have already set benzene
limits on their gasoline. As a result, we believe that gasoline with
some degree of benzene control will be easily available for importers
to market. In some cases, we also expect that some foreign refiners may
produce for export some fraction of their gasoline to meet our proposed
0.62 vol% average standard benzene. This would provide importers
further options in the U.S. gasoline market.

G. How Does the Proposed Fuel Control Program Satisfy the Statutory
Requirements?

As discussed earlier in this section, we have concluded that the
most effective and appropriate program for MSAT emission reduction from
gasoline is a benzene control program. Today's action proposes such a
program, with an average benzene content standard of 0.62 vol% and a
specially-designed averaging, banking, and trading program. In section
VII.F above, we summarize our evaluation of the feasibility of the
proposed program, and in section IX.A we summarize our evaluation of
the costs of the program. The analyses supporting our conclusions in
these sections are discussed in detail in Chapters 6 and 9 of the RIA.
Taking all of this information into account, we believe that a
program more stringent than the proposed program would not be feasible,
taking into consideration cost. As we have discussed, making the
standard more stringent would require more refiners to install the more
expensive benzene control equipment, with very little improvement in
benzene emissions. Also, we have shown that related costs increase very
rapidly as the level of the standard is made more stringent.
Conversely, while it would provide significant benzene emission
reductions, we are concerned that a somewhat less stringent national
average standard than the proposed 0.62 vol% (e.g., 0.65 or 0.70 vol%)
would not satisfy our statutory obligation for the most stringent
standard feasible considering cost and other factors. Furthermore, such
standards would not accomplish several important programmatic
objectives as discussed in section VII.C.
We have also considered energy implications of the proposed
program, as well as noise and safety, and we believe the proposed
program would have very little impact on any of these factors. Analyses
supporting these conclusions are also found in Chapter 9 of the RIA. We
carefully considered lead time in establishing the stringency and
timing of the proposed program (see section VII.F above).
Consequently, we believe that the proposed program would meet the
requirements of section 202(l) of the Clean Air Act, reflecting ``the
greatest degree of emission reduction achievable through the
application of technology which is available, taking into consideration
* * * the availability and costs of the technology, and noise, energy,
and safety factors, and lead time.''

H. Effect on Energy Supply, Distribution, or Use

This rule is not a ``significant energy action'' as defined in
Executive Order 13211, ``Actions Concerning Regulations That
Significantly Affect Energy Supply, Distribution, or Use'' (66 FR 28355
(May 22, 2001)) because it is not likely to have a significant adverse
effect on the supply, distribution, or use of energy. If promulgated,
the gasoline benzene provisions of the proposed rule would shift about
22,000 barrels per day of benzene from the gasoline market to the
petrochemical market. This volume represents about 0.2 percent of
nationwide gasoline production. The actual impact of the rule on the
gasoline market, however, is likely to be less due to offsetting
changes in the production of petrochemicals, as well as expected growth
in the petrochemical market absent this rule. The major sources of
benzene for the petrochemical market other than reformate from gasoline
production are also derived from gasoline components or gasoline
feedstocks. Consequently, the expected shift toward more benzene
production from reformate due to this proposed rule would be offset by
less benzene produced from other gasoline feedstocks.
The rule would require refiners to use a small additional amount of
energy in processing gasoline to reduce benzene levels, primarily due
to the increased energy used for benzene extraction. Our modeling of
increased energy use indicates that the process energy used by refiners
to produce gasoline would increase by about one percent. Overall, we
believe that the proposed rule would result in no significant adverse
energy impacts.
The proposed gasoline benzene provisions would not affect the
current gasoline distribution practices.
We discuss our analysis of the energy and supply effects of the
proposed gasoline benzene standard further in section IX of this
preamble and in Chapter 9 of the Regulatory Impact Analysis.
The fuel supply and energy effects described above would be offset
substantially by the positive effects on gasoline supply and energy use
of the proposed gas can standards also proposed in today's action.
These proposed provisions would greatly reduce the gasoline lost to
evaporation from gas cans. This would in turn reduce the demand for
gasoline, increasing the gasoline supply and reducing the energy used
in producing gasoline.

I. How Would the Proposed Gasoline Benzene Standard Be Implemented?

This section discusses the details associated with meeting the
proposed 0.62 vol% benzene standard.
1. General Provisions
a. What Are the Implementation Dates for the Proposed Program?
We are proposing that refiners and importers would achieve
compliance with the requirements of the proposed benzene program
beginning with the annual averaging period beginning January 1, 2011.
Refineries with approved benzene baselines could generate early credits
from June 1, 2007, through December 31, 2010. Refineries and importers
could generate standard credits beginning with the annual averaging
period beginning January 1, 2011, provided that the average benzene
content of the gasoline they produce or import during the year was less
than 0.62 vol% benzene.
Approved small refiners would be allowed to delay compliance with
the 0.62 vol% standard until the annual averaging period beginning
January 1, 2015. They could, however, generate early credits beginning
June 1, 2007 through December 31, 2014, provided that they had an
approved benzene baseline. They would be able to generate standard
credits beginning January 1, 2015.

[[Page 15891]]

b. Which Regulated Parties Would Be Subject to the Proposed Benzene
Standards?
Domestic refiners and importers would be subject to the proposed
standards. We are proposing that each refinery of a refiner must meet
the standard, and all associated requirements, individually. Refinery
grouping, or aggregation, as allowed in the Anti-dumping and MSAT1
program for CG, would not be permitted for purposes of complying with
the proposed benzene standard (although the ABT provisions provide
similar flexibility, and the credit generation and transfer provisions
would perform basically the same functions). For an importer, we are
proposing that the requirements apply to the entire volume imported
during the averaging period regardless of import locations or sources.
In addition, where a company has both refinery and import operations,
each operation would have to achieve its own compliance with the 0.62
vol% benzene standard. We are proposing that those who only added
oxygenate or butane to gasoline or gasoline blending stock would not be
subject to the proposed standards for that gasoline unless they also
added other blending components to the blend. This would be similar to
the current treatment of these entities and their gasoline under the
RFG, Anti-dumping and MSAT1 programs, where specialized accounting and
calculation procedures are specified. In these cases, the refinery (or
importer) that produces gasoline or gasoline blendstock includes the
oxygenate in its own compliance determination. We are proposing that
this practice would continue under today's program. Transmix processors
would not be subject to the proposed requirements for gasoline produced
from transmix, but gasoline produced from transmix to which other
blendstocks were added would be subject to the proposed benzene standard.
We are proposing that all gasoline produced by foreign refineries
for use in the United States would be included in the compliance and
credit calculation of the importer of record. Under the Anti-dumping
and MSAT1 rules, as well as the gasoline sulfur requirements,
additional requirements applicable to foreign refiners who chose to
comply with those regulations separately from any importer were
included to ensure that enforcement of the regulation at the foreign
refinery would not be compromised. We are proposing similar provisions
here. Specifically, we are proposing to allow foreign refiners to
generate early credits and to apply for temporary hardship relief and
small refiner status. See proposed 40 CFR 80.1420. However, under the
earlier rules, few foreign refiners have chosen to undertake these
additional requirements, and almost all gasoline produced at foreign
refineries is included in an importer's compliance determination for
the current EPA gasoline programs.\270\ We invite comment on the value
of extending these provisions to this proposed benzene program.
---------------------------------------------------------------------------

\270\ Often, the importer of record is the foreign reiner. In
these instances, the importer/foreign refiner has simply opted to
achieve compliance via the applicable importer provisions.
---------------------------------------------------------------------------

As mentioned, we are proposing to extend the small refiner
provisions to foreign refiners. Our experience in past rules is that
they are not taken advantage of for various reasons. Most foreign
refineries are state-owned or owned by large multinational companies,
and would exceed the employee-count criterion. Others have typically
not been interested in fulfilling the enforcement-related requirements
that apply to foreign refineries. We request comment on extending the
small refiner provisions to foreign refiners.
c. What Gasoline Would Be Subject to the Proposed Benzene Standards?
All finished gasoline produced by a refinery or imported by an
importer would be subject to the proposed benzene content standard. In
addition, gasoline blending stock which becomes finished gasoline
solely upon the addition of oxygenate would also be subject to the
proposed standard.\271\ Other gasoline blendstocks which are shifted
among refiners prior to turning them into finished gasoline would not
be subject to the benzene standard. They would be included at the point
they are converted or blended to produce finished gasoline.
---------------------------------------------------------------------------

\271\ As stated earlier, both blending stock and oxygenate would
be included in the refinery's or importer's compliance
determination. Conventional gasoline refiners are required to have
agreements with downstream oxygenate blenders to ensure that the
appropriate type and amount of oxygenate is added to the gasoline
blending stock, per 40 CFR 80.10(d). Absent such agreements, the
refinery may only include the gasoline blending stock in its
compliance determination and the oxygenate is not included in any
compliance determination.
---------------------------------------------------------------------------

We are proposing to exclude gasoline produced or imported for use
in California from this benzene requirement. Although California's
benzene averaging standard is greater than 0.62 vol%, California in-use
benzene levels are currently below the level of the proposed
standard.\272\ We expect this situation will continue. There would be
no additional benefit to consumers of California gasoline or to the
implementation and benefits of the proposed program by the inclusion of
gasoline used in California.
---------------------------------------------------------------------------

\272\ California Code of Regulations, Title 13 Section 2262.
---------------------------------------------------------------------------

This proposal also would exclude those specialized gasoline
applications that have been exempted from other EPA gasoline rules,
such as gasoline used to fuel aircraft, or for sanctioned racing
events, gasoline that is exported for sale and use outside of the U.S.,
and gasoline used for research, development or testing purposes, under
certain circumstances.
d. How Would Compliance With the Benzene Standard Be Determined?
Compliance with the proposed benzene standard would be on an
annual, calendar year basis, similar to almost all other current
gasoline controls. A refiner's or importer's compliance (or Compliance
Benzene Value, as used in the proposed regulation) would be determined
from the annual average benzene content of its gasoline (produced or
imported), any credits used for compliance purposes, and any deficit
carried over from the previous year, and would have to be 0.62 vol% or
lower, on a benzene volume basis. The Compliance Benzene Value would
differ from the refiner's or importer's actual annual average benzene
concentration because the latter would be solely a volume weighted
average of the benzene concentrations of the refinery's or importer's
actual gasoline batches.
Credits, in any amount, could be used to achieve compliance. As
mentioned, we are also proposing to allow a deficit to be carried
forward for one year. Under these circumstances, in the next compliance
period, the refinery or importer would have to be in compliance, that
is, the refinery or importer would have to, through production or
import practices, and/or the use of credits, make up the deficit from
the previous year and be in compliance with the proposed benzene
standard. This provision could be especially helpful to refiners in the
first year of the program, until the availability and need for credits
was established.
In the RFG and Anti-dumping programs, and MSAT1, by extension,
refiners and importers generally include oxygenate added downstream
from the refinery or the import facility in their compliance
calculations.\273\ Refiners

[[Page 15892]]

and importers of RBOB are required to account for the oxygenate in
their own compliance. As mentioned earlier, refiners and importers of
conventional gasoline can include the oxygenate if they have met the
Anti-dumping requirements for ensuring that the amount and type of
oxygenate was indeed added. We are not proposing any changes to these
provisions for the purposes of compliance with the proposed benzene
program. However, average pool benzene levels are expected to decrease
as a result of increased ethanol use due to requirements of the Energy
Policy Act of 2005, and this would affect both early and standard
credit generation, as will be discussed below. However, we request
comment on how, if at all, additional oxygenate use should be
considered, and perhaps limited, in compliance determinations for the
proposed program.
---------------------------------------------------------------------------

\273\ As a result, oxygenate blenders would not be subject to
the RFG, Anti-dumping or MSAT1 regulations except for gasoline to
which they add other blendstocks in addition to the oxygenate.
---------------------------------------------------------------------------

2. Averaging, Banking and Trading Program
a. Early Credit Generation
As discussed, early credit generation could occur as early as the
averaging period beginning June 1, 2007, through the averaging period
ending December 31, 2010, or ending December 31, 2014, for small
refiners. In order to generate early benzene credits, a refinery would
first establish a benzene baseline which is its average benzene
concentration over the period January 1, 2004, through December 31,
2005. A refinery would be eligible to generate early credits when it
reduced its annual average benzene concentration by at least 10%
compared to its benzene baseline. Credits would then be calculated
based on the entire reduction in benzene below the baseline. Generation
of early credits for the first averaging period, June 1, 2007 through
December 31, 2007, which is less than a calendar year, would be based
on the average benzene level of the gasoline produced only during this
period. Gasoline produced before June 1, 2007, would not be included in
the credit generation determination.
We are proposing to allow only refiners (and not importers) to
generate early benzene credits because it is at the refinery, or
production level, where real changes in the production of gasoline can
be made. Importers would simply seek out blending streams or gasoline
with lower benzene, but would not have to invest or take other action
involving the production of the lower benzene gasoline. Furthermore,
many importer operations grow in volume, shrink in volume, come into
existence and go out of existence on a continual basis, making it
difficult to assess the appropriateness of both the baseline and any
early credits. Thus, even though an importer may have had regular,
consistent import activity during the 2004-2005 baseline period, we are
proposing that only refiners would be allowed to apply for a benzene
baseline, and if approved, to generate early benzene credits based on
reductions in future averaging period gasoline benzene levels.
As discussed above, one of the purposes of allowing the early
generation of benzene credits would be to promote reductions in benzene
through refinery processing changes. We are concerned that benzene
reductions due to increased oxygenate use would result in reduced
benzene concentrations. Oxygenate use (in the form of ethanol) in CG is
expected to increase as a result of the Energy Policy Act
requirements.\274\ This additional oxygenate will dilute gasoline
benzene levels as well as extend the gasoline pool. As a result,
refinery average benzene levels would be likely to be lower during the
early credit generation period than during the benzene baseline period
(2004-2005) if there is an increase in the amount of CG refiners send
for downstream blending with ethanol (CBOB). We are concerned that
reductions in fuel benzene levels due to oxygenate addition
significantly beyond the average levels of recent years could result in
windfall early credit generation for some refineries. We request
comment on the likelihood of windfall early credit generation, and if
such a situation were to occur, whether it would warrant limiting early
benzene credits by consideration of the average oxygenate use during
the baseline period compared to the early credit generation period or
by adjusting the early credit trigger point. We believe this would be
less of an issue during the standard credit generation period beginning
in 2011 (2015 for small refiners) because of the more stringent
requirements for generating standard credits (getting below the 0.62
vol% standard) compared to the early credit generation requirements
(achieving a minimum 10% reduction in baseline benzene levels).
---------------------------------------------------------------------------

\274\ Even though the Energy Policy Act of 2005 eliminated the
oxygen mandate for RFG, oxygenate use (in the form of ethanol) in
RFG is expected to continue.
---------------------------------------------------------------------------

b. How Would Refinery Benzene Baselines Be Determined?
As mentioned above, a refiner would submit a benzene baseline
application to EPA for any of its refineries which planned to generate
early credits. The benzene baseline would be the volume-weighted
average of the benzene levels of the gasoline produced by the refinery
during 2004-2005. Note that the gasoline would be the combination of
the refinery's RFG and CG, if applicable, and would exclude California
gasoline and other fuels exempted from the proposed standard. The
benzene values used in the benzene baseline calculation should be the
same as used in the RFG, Anti-dumping and MSAT1 compliance
determinations. We are not proposing provisions for adjusting these
benzene baselines based on circumstances during the baseline years or
otherwise.
Though we expect that most refineries that apply for a benzene
baseline would have data for both 2004 and 2005, if a refinery was shut
down for part of the 2004-2005 period, it could still be able to
establish a benzene baseline. Under these circumstances, the refiner
would have to provide and justify, using refinery and engineering
analyses, an appropriate adjusted value that reflects the likely
average benzene concentration for the refinery, had it been fully
operational. A refinery that was non-operational for the entire period
January 1, 2004 through December 31, 2005 would not be able to
establish a benzene baseline and therefore not allowed to generate
early credits.
c. Credit Generation Beginning in 2011
Credits could be generated in any annual averaging period beginning
January 1, 2011, or for small refiners, beginning January 1, 2015.
These credits, also called standard benzene credits, could be generated
by a refinery or importer when the refinery's or importer's annual
average benzene concentration was less than the proposed standard of
0.62 vol%.
While the proposed benzene standard is a 49-state standard due to
the fact that California would maintain its existing benzene standard,
we request comment on the appropriateness of allowing California
refineries to generate credits that could be used to demonstrate
compliance outside of California.
d. How Would Credits Be Used?
We are proposing that all gasoline benzene credits that are
properly created may be used equally and interchangeably. That is, once
generated, there would be no difference

[[Page 15893]]

between early credits and standard credits, except for their credit
life, as discussed below. Under this proposal, credits could be
transferred to another refiner or importer, or they could be banked by
the refinery or importer that created them for use or transfer in a
later compliance period.
As in past credit programs, we are proposing some limits on credit
use. First, we are proposing to limit the number of times a credit
could be transferred. At the end of the allowable number of transfers,
the credit would have to be used by the last transferee before its
expiration date. Second, we are proposing that credits would have a
finite life whether or not transferred. We are proposing that early
credits, those generated prior to 2011, would have a three-year credit
life from the start of the program in 2011. These credits would have to
be used to achieve compliance with the proposed benzene standard in
2011, 2012, and/or 2013, or they would expire. In addition, we are
proposing that credits generated in 2011 and beyond (or early credits
generated by small refiners during this period) would have to be used
within five years of the year in which they were generated. We had
considered requiring credits be used in order of their generation date,
that is, credits generated earlier would have to be used before credits
generated later. However, the finite credit life is likely to ensure
this usage, and thus we are not proposing to regulate credit use in
this manner. We are also proposing that credit life could be extended
by two years for any credits that are generated by or traded to
approved small refiners.
Under the proposed regulations, a refiner or importer would have to
use all benzene credits in its possession before being allowed to have
deficit carryover, and would have to meet its own compliance
requirement before transferring any gasoline benzene credits. In the
case of invalid credits, or credits improperly created, all parties
would have to adjust their credit records, reports, and compliance
calculations to reflect proper credit use. The transferor would first
correct its own records and ensure its own compliance, and then apply
any remaining properly created credits to the transferee before trading
or banking those credits. See section X.A below for more discussion of
these issues.
3. Hardship and Small Refiner Provisions
a. Hardship
The hardship provisions and requirements are extensively discussed
in section VII.E.2, and thus are only briefly addressed here. We are
proposing that a refiner for any of its refineries could seek temporary
relief from meeting the proposed benzene standard due to unusual
circumstances, including those situations, such as a natural disaster,
which would clearly be outside the control of the refiner. A refiner
would have to apply to EPA for this temporary relief, and EPA could
deny the application or approve it for an appropriate period of time.
However, given the existence of a flexible ABT program, EPA expects
that, prior to requesting hardship relief, the refiner would have made
best efforts to obtain credits in order to comply with the proposed
benzene standard. In past rulemakings, for example the gasoline sulfur
rule, the hurdle for receiving a hardship was very high, with very few
granted. While we are proposing these provisions again here, the
expectation is that the hurdle would be even higher. Given the
existence and flexibility afforded by the ABT program and the more
limited cost of the benzene standard, it is our expectation that as
long as a viable credit market existed, it would be difficult to
justify granting a hardship. Furthermore, the form of any relief we are
proposing is in the form of additional time to demonstrate compliance
via credits as opposed to any waiver of the standards.
b. Small Refiners
As discussed earlier, we are proposing to allow small refiners to
meet the proposed benzene standard beginning with the 2015 averaging
period, which is four years later than non-small refiners and
importers. Small refiners could also generate both early and standard
credits if they can meet the requirements of those programs. A refiner
would have to apply to EPA by December 31, 2007 in order to be
considered a small refiner under this proposed rule even if the entity
was or had been considered a small refiner under other EPA rules. The
requirements for small refiners under this rule are detailed in section
VII.E.
4. Administrative and Enforcement Related Provisions
a. Sampling/Testing
As under the Tier 2 program where a sulfur concentration must be
determined for every batch of gasoline, we are proposing that a benzene
concentration value also be determined for every batch of gasoline
produced or imported. Thus, as gasoline samples are taken for sulfur
measurement, they would also be taken for benzene measurement. The RFG
program, which has both a toxics emissions requirement and a per-gallon
benzene cap, already requires a benzene value to be determined for
every batch of gasoline. The Anti-dumping program, which has only a
toxics emissions requirement, allows benzene values to be determined
from composite samples. See 40 CFR 80.101(i). Thus, the proposed
sampling requirement would be a change from the current sampling
methodology allowed under the Anti-dumping provisions but makes it
consistent with the ongoing Tier 2 sulfur program. However, unlike the
gasoline sulfur requirements, this every batch testing requirement for
conventional gasoline benzene would not have to occur prior to the
batch leaving the refinery. Additionally, the batch numbering system
would be the same as that used for conventional gasoline sulfur.
We are not proposing any changes to the benzene test methodology.
See 40 CFR 80.46(e). We are proposing sample retention requirements
similar to those in the gasoline sulfur provisions. See 40 CFR 80.335.
b. Recordkeeping/Reporting
We are proposing to require that records be kept for each averaging
period in order to accommodate the proposed benzene standard and the
accompanying credit trading program. These records would include: the
benzene baseline calculation, if applicable; the number of early
credits generated, if applicable; the actual average benzene
concentration of gasoline produced or imported; the compliance benzene
value; any deficit; the number of credits generated; and records of any
credit transfers to or from the refinery or importer, including price
of the credits and dates of transactions. All of this information, and
any other information that EPA may require, such as information similar
to that proposed below for inclusion in the pre-compliance reports,
would be submitted in a refiner's or importer's annual report to the
Agency. Since we are proposing that the regulatory provisions for the
benzene control program would become the single regulatory mechanism
covering RFG and Anti-dumping annual average toxics requirements once
the benzene standard is in effect, and would replace the MSAT1
requirements, we expect to be able to streamline several of the current
reporting forms once the proposed program is fully implemented in 2015.
As mentioned, we are also proposing to require that refiners and
importers submit pre-compliance reports in order to provide information
as to the likely number of benzene credits needed and

[[Page 15894]]

available, and how the refiner or importer plans to achieve compliance
with the proposed benzene requirements. These reports would be required
annually each June 1 from 2001 through 2011 (or through 2015 for small
refiners). In addition to information regarding gasoline production and
the number of credits expected to be used or produced, the pre-
compliance reports would include information regarding the benzene
reduction technology expected to be used, any capital commitments, and
information on the progress of the installation of the technology. We
are also proposing that these reports include price and quantity
information for any credits bought or sold. The reports would include
updates from the previous year's estimates, and comparison of previous
year actual production to the projected values.
c. Attest Engagements, Violations, Penalties
We are proposing to require attest engagements for generation of
both early and other credits, credit use, and compliance with the
proposed program, using the usual procedures for attest engagements.
The violation and penalty provisions applicable to this proposed
benzene program would be very similar to the provisions currently in
effect in other gasoline programs. We request comment on the need for
additional attest engagement, violation or penalty provisions specific
to the proposed benzene program.
5. How Would Compliance With the Provisions of the Proposed Benzene
Program Affect Compliance With Other Gasoline Toxics Programs?
As discussed above, we expect that virtually all refineries will
reduce benzene from their current levels, and no refineries will
increase it. This impact on benzene levels, combined with the pre-
existing gasoline controls in sulfur, RVP, and VOC performance, means
that compliance with the benzene content provisions is also expected to
lead to compliance with the annual average requirements on benzene and
toxics performance for reformulated gasoline and the annual average
Anti-dumping toxics performance for conventional gasoline. EPA is
therefore proposing that upon full implementation in 2011 the
regulatory provisions for the benzene control program would become the
single regulatory mechanism used to implement these RFG and Anti-
dumping annual average toxics requirements, replacing the current RFG
and Anti-dumping annual average toxics standards as unnecessary. The
proposed benzene control program would also replace the MSAT1
requirements. However, we propose the RFG per gallon benzene cap of 1.3
vol% remain in effect; we are requesting comment on the need to retain
this requirement for RFG. Note that compliance with the proposed
benzene standard would ensure compliance with the aforementioned RFG,
Anti-dumping and MSAT1 requirements beginning with the 2011 averaging
period, or the 2015 averaging period for small refiners. Thus, during
the early credit generation period, 2007 through 2010, all entities
would still be required to comply with their applicable RFG, Anti-
dumping and MSAT1 requirements. In addition, from 2011 through 2014,
small refiners would have to continue to meet their applicable RFG,
Anti-dumping and MSAT1 requirements. As discussed earlier in section
VII.E.2, we are also requesting comment on the option of allowing some
refineries to meet the proposed benzene standard early, thus replacing
the current RFG and Anti-dumping annual average toxics provisions and
replacing MSAT1 requirements for these refineries.

VIII. Gas Cans

Gas cans are consumer products people use to refuel a wide variety
of gasoline-powered equipment. Their most frequent use is for refueling
lawn and garden equipment such as lawn mowers, trimmers, and chainsaws.
They are also routinely used for recreational equipment such as all-
terrain vehicles and snowmobiles, and for passenger vehicles which have
run out of gas. The gas cans are red, per ASTM specifications, and
about 95 percent of them are made of plastic (high density polyethelene
(HDPE)). There are approximately 20 million gas cans sold annually and
about 80 million cans are in use nationwide. The average lifetime of a
gas can is about 5 years.
California has established an emissions control program for gas
cans which began in 2001. Since then, some other states have adopted
the California requirements. Last year, California adopted a revised
program which is very similar to the one we are proposing in this
rulemaking. Manufacturers are required to meet the new requirements in
California by July 1, 2007 at the latest. State programs are discussed
further in section VIII.A.3., below.

A. Why Are We Proposing an Emissions Control Program for Gas Cans?

1. VOC Emissions
We are proposing standards to control VOCs as an ozone precursor
and also to minimize exposure to VOC-based toxics such as benzene and
toluene. Gasoline is highly volatile and evaporates easily from
containers that are not sealed or closed properly. Although an
individual gas can is a relatively modest emission source, the
cumulative VOC emissions from gas cans are quite significant. We
estimate that containers currently emit about 315,000 tons of VOC
annually nationwide, which is equal to about 5 percent of the
nationwide mobile source inventory (see section V.A.). Left
uncontrolled, a gas can's evaporative emissions are up to 60 times the
VOC of a new Tier 2 vehicle evaporative control system. Gas can
emissions are primarily of three types: evaporative emissions from
unsealed or open containers; permeation emissions from gasoline passing
through the walls of the plastic containers; and evaporative emissions
from gasoline spillage during use.
As discussed in section IV. above, ozone continues to be a
significant air quality concern, and gas cans are currently an
uncontrolled source of VOC emissions in many areas of the country.
Section 183(e) of the Clean Air Act directs EPA to study, list, and
regulate consumer and commercial products that are significant sources
of VOC emissions. In 1995, after conducting a study and submitting a
Report to Congress on VOC emissions from consumer and commercial
products, EPA published an initial list of product categories to be
regulated under section 183(e). Based on criteria that we established
pursuant to section 183(e)(2)(B), we listed for regulation those
consumer and commercial products that we considered at the time to be
significant contributors to the ozone nonattainment problem, but we did
not include gas can emissions.\275\ After analyzing the emissions
inventory impacts of gas cans, EPA plans to publish a Federal Register
notice that would add portable gasoline containers to the list of
consumer products to be regulated and explain the rationale for this
action in detail. EPA will afford interested persons the opportunity to
comment on the data underlying the listing before taking final action
on today's proposal. In today's notice, EPA is proposing that the
standards for

[[Page 15895]]

portable gasoline containers represent ``best available controls'' as
required by section 183(e)(3)(A). Determination of the ``best available
controls'' requires EPA to determine the degree of reduction achievable
through use of the most effective control measures (which includes
chemical reformulation, and other measures) after considering
technological and economic feasibility, as well as health, energy, and
environmental impacts.\276\
---------------------------------------------------------------------------

\275\ 60 FR 15264 ``Consumer and Commercial Products: Schedule
for Regulation,'' March 23, 1995.
\276\ See section 183(e)(1); see also section 183(e)(4)
providing broad authority to include ``systems of regulation'' in
controlling VOC emissions from consumer products.
---------------------------------------------------------------------------

2. Technological Opportunities to Reduce Emissions From Gas Cans
Gas can manufacturers have already developed and applied emissions
controls in response to California requirements. Traditional gas cans
typically have a spout for pouring fuel and a vent at the rear of the
can to allow air to flow into the cans when in use. About 70 percent of
emissions from gas cans are due to evaporative losses from caps being
left off one or both of these openings. The primary way to reduce these
emissions is to design cans that are not easily left open. To
accomplish this, gas can manufacturers have developed spouts that
incorporate a spring mechanism to close cans automatically when not in
use. Many spout designs are opened by consumers pushing the spout
against the equipment fuel tank. Some designs incorporate a button or
trigger mechanism that the consumer pushes to start fuel flow and then
releases when done refueling. Also, some cans are made without rear
vents, incorporating venting into the spouts and thus eliminating one
potential emission point. The consumer still must remove the spout to
refill the cans but would replace the spout once the can is full in
order to prevent spillage during transport.
The auto-closing spouts reduce spillage by giving consumers greater
control over the fuel flow. The spouts allow consumers to place the can
in position before activating or opening the cans. Once the receiving
fuel tank is full, consumers can easily release the mechanism to stop
the fuel flow. This reduces spillage during the positioning and removal
of the can and reduces overall spillage by about half. Consumers
generally appreciate the greater control over the refueling event.
Blow-molding is used to manufacture gas cans. Typically, blow-
molding is performed by creating a hollow tube, known as a parison, by
pushing high-density polyethylene (HDPE) through an extruder with a
screw. The parison is then pinched in a mold and inflated with an inert
gas. The HDPE plastics used for gas cans allow gasoline molecules to
permeate (i.e., pass through) the walls of the container. This
contributes to overall emission losses from the containers. There are
several effective permeation barriers that can be incorporated into the
can walls. Gas can manufacturers have used several of these methods to
meet California program requirements. The technologies were initially
developed to meet automotive evaporative emissions standards and are
now also being used for other types of fuel tanks. The barriers are
either incorporated as part of the manufacturing process of the can
(either as a layer or by mixing the barrier materials with the
plastics) or are applied to the cans after they are manufactured. These
barriers typically achieve reductions of 85 percent or better compared
to untreated cans.
Some gas can manufacturers have produced non-permeable plastic gas
cans by blow molding a layer of ethylene vinyl alcohol (EVOH) or nylon
between two layers of polyethylene. This process is called coextrusion
and requires at least five layers: The barrier layer, adhesive layers
on either side of the barrier layer, and HDPE as the outside layers
which make up most of the thickness of the gas can walls. However, this
blow-molding process requires two additional extruder screws, which
significantly increases its cost.
An alternative to coextrusion is to blend a low-permeability resin
with the HDPE and extrude it with a single screw to create barrier
platelets. The trade name typically used for this permeation control
strategy is Selar. The low-permeability resin, typically EVOH or nylon,
creates non-continuous platelets in the HDPE gas can which reduce
permeation by creating long, tortuous pathways that the hydrocarbon
molecules must navigate to pass through the gas can walls. Although the
barrier is not continuous, this strategy can still achieve greater than
a 90-percent reduction in permeation of gasoline. EVOH has much higher
permeation resistance to alcohol than nylon; therefore, it would be the
preferred material to use for meeting our proposed standard (described
at Section B., below), which is based on testing with a 10-percent
ethanol fuel.
Another type of low permeation technology for HDPE gas cans is
treating the surfaces of plastic gas cans with a barrier layer. Two
ways of achieving this are known as fluorination and sulfonation. The
fluorination process causes a chemical reaction where exposed hydrogen
atoms are replaced by larger fluorine atoms, creating a barrier on the
surface of the gas can. In this process, a batch of gas cans is
generally processed post production by stacking them in a steel
container. The container is then voided of air and flooded with
fluorine gas. By pulling a vacuum in the container, the fluorine gas is
forced into every crevice in the gas can. As a result of this process,
both the inside and outside surfaces of the gas can would be treated.
As an alternative, gas cans can be fluorinated on the manufacturing
line by exposing the inside surface of the gas can to fluorine during
the blow molding process. However, this method may not prove as
effective as off-line fluorination, which treats the inside and outside
surfaces.
Sulfonation is another surface treatment technology. In this
process, sulfur trioxide reacts with the exposed polyethylene to form
sulfonic acid groups on the surface. Current practices for sulfonation
are to place a gas can on a small assembly line and expose the inner
surfaces to sulfur trioxide, then rinse with a neutralizing agent.
However, sulfonation can also be performed using a batch method. Either
of these processes can be used to reduce gasoline permeation by more
than 95 percent.
3. State Experiences Regulating Gas Cans
California established an emissions control program for gas cans
that began in 2001.\277\ Twelve other states and the District of
Columbia have adopted the California program in recent years. These
states include Delaware, Maine, Maryland, Pennsylvania, New York,
Connecticut, Massachusetts, New Jersey, Rhode Island, Vermont,
Virginia, Washington, DC, and Texas.
---------------------------------------------------------------------------

\277\ Portable Fuel Container Spillage Control Regulations,
Final Statement of Reasons, State of California Environmental
Protection Agency Air Resources Board, June 2000.
---------------------------------------------------------------------------

Last year, California adopted a revised program that is very
similar to the one we are proposing in this rulemaking.\278\
California's new program goes into effect on July 1, 2007. California
addressed several deficiencies they observed in their first program by
adding new enhanced diurnal standards, new testing requirements, and
new certification requirements, and by removing automatic shut-off
requirements that lead to designs that do not work well in the field.

[[Page 15896]]

California's original program contained several design specifications
which limited manufacturer flexibility and resulted, in many cases, in
products that were difficult for consumers to use. California has
removed most of these design specifications from their revised program.
---------------------------------------------------------------------------

\278\ Public Hearing to Consider Amendments to the Regulations
for Portable Fuel Containers, Final Statement of Reasons, California
Air Resources Board, October 2005.
---------------------------------------------------------------------------

California's original program included an automatic shut-off
requirement intended to reduce spillage caused by overfilling the
receiving fuel tank. The spouts were required to be designed to stop
fuel flow when the fuel reached the tip of the spout, similar to how
gas pumps shut off when refueling a vehicle. California specified a
test fixture, the height of the fuel in the receiving tank at which
point the fuel flow must stop, and the minimum fuel flow rate. The gas
cans were designed by manufacturers to work well with the test fixture,
but the automatic shut-off failed in use a significant amount of the
time. California found that the design of the equipment fuel tank had a
big impact on the performance of the automatic shut-off. Due to the
wide variety of fuel tank designs, the automatic shut-off worked on a
relatively small percentage of equipment. In addition, many of the
spout designs were not compatible with passenger vehicles. This is
especially critical because the cans are customarily used by consumers
when their vehicles run out of gas.
These problems led to many consumer complaints to both the
manufacturers and to the California Air Resources Board. It also led to
increased spillage in many cases. It was also found that many consumers
did not understand how the spouts were supposed to operate. Even in
cases where the spouts would have stopped the flow of fuel in time,
consumers did not use the cans properly. Consumers are used to actively
controlling the flow of fuel. For these reasons, California removed the
automatic shut-off requirements from their program for all cans.

B. What Emissions Standard Is EPA Proposing, and Why?

1. Description of Emissions Standard
We are proposing a performance-based standard of 0.3 grams per
gallon per day (g/gal/day) of HC to control evaporative and permeation
losses. The standard would be measured based on the emissions from the
can over a diurnal test cycle. The cans would be tested as a system
with their spouts attached. Manufacturers would test the cans by
placing them in an environmental chamber which simulates summertime
ambient temperature conditions and cycling the cans through the 24-hour
temperature profile (72-96[deg]
F), as discussed below. The test
procedures, which are described in more detail below, would ensure that
gas cans meet the emission standard over a range of in-use conditions
such as different temperatures, different fuels, and taking into
consideration factors affecting durability.
2. Determination of Best Available Control
The 0.3 g/gal/day emissions standard and associated test procedures
reflect the performance of the best available control technologies
discussed above, including durable permeation barriers, auto-closing
spouts, and a can that is well-sealed to reduce evaporative losses. The
standard is both economically and technologically feasible. As
discussed above, to comply with California's program, gas can
manufacturers have developed gas cans with low VOC emissions at a
reasonable cost (see section IX. for costs). Testing of cans designed
to meet CARB standards has shown the proposed standards to be
technologically feasible. When tested over cycles very similar to those
we are proposing, emissions from these cans have been in the range of
0.2-0.3 g/gal/day.\279\ These cans have been produced with permeation
barriers representing a high level of control (over 90 percent
reductions) and with auto-closing spouts, which are technologies that
represent best available controls for gas cans. Establishing the
standard at 0.3 g/gal/day would require the use of best available
technologies. We are proposing a level at the upper end of the tested
performance range to account for product performance variability. In
addition, we believe that any of the current best designs can achieve
these levels, so we do not believe that the proposed standard
forecloses use of any of the existing performing product designs. Our
detailed feasibility analysis is provided in the Regulatory Impact
Analysis. We request comment on the level of the standard and on its
feasibility. We request that commenters provide detail and data where
possible.
---------------------------------------------------------------------------

\279\ ``Quantification of Permeation and Evaporative Emissions
From Portable Fuel Container'', California Air Resources Board, June 2004.
---------------------------------------------------------------------------

In addition to considering technological and economic feasibility,
section 183(e)(1)(A) requires us to consider ``health, environmental,
and energy impacts'' in assessing best available controls.
Environmental and health impacts are discussed in section IV. Moreover,
control of spillage from gas cans may reduce fire hazards as well
because cans would stay tightly closed if tipped over. We expect the
energy impacts of gas can control to be positive, because the standards
will reduce evaporative fuel losses.
3. Emissions Performance vs. Design Standard
We are proposing an emissions performance standard rather than
mandating that gas cans be of any specified design. Rather than
proposing to require that gas cans only have one opening, or other
design-based requirements, we believe that it is sufficient to require
gas cans to meet an emissions performance standard. A performance
standard allows flexibility in can design while ensuring the overall
emissions performance of the cans. We are reluctant to specify design
standards for consumer products in order not to limit manufacturer (and
ultimately consumer) choice. The market will encourage manufacturers to
offer products that work well for consumers, and design-based
requirements could unnecessarily limit manufacturer design flexibility.
4. Automatic Shut-Off
We are not requiring automatic shut-off as a design-based standard,
or considering it to be a ``best available control.'' As described in
section VIII.A.3. above, the automatic shut-off has been shown to be
problematic for consumers for several reasons, and we believe that
including requirements for automatic shut-off would be
counterproductive. Automatic shut-off is supposed to stop the flow of
fuel when the fuel reaches the top of the receiving tank in order to
prevent over-filling. However, due to a wide variety of receiving fuel
tank designs, the auto shut off spouts do not work well with a variety
of equipment types. In California, this problem led to spillage and
consumer dissatisfaction. We want to avoid cases where spills occur
even when consumers are using the products properly due to a mismatch
between the spout design and the design of the receiving fuel tank
being filled. Excessive consumer difficulties in using new cans would
likely lead to some consumers defeating the low emissions features of
the cans by removing the spouts and using other means such as funnels
to refuel equipment. Any additional emissions reductions provided by
automatic shut-off in cases where it worked properly would likely be
largely or completely offset by increased spillage due to cases where

[[Page 15897]]

consumers defeated the designs or the designs failed to work properly.
We believe that the automatic closing cans, even without automatic
shut-off requirements, will lead to reduced spillage. As discussed
above, automatic closure keeps the cans closed when they are not in use
and provides more control to the consumer during use.
Some additional reduction in spillage is likely possible in some
cases with automatic shut-off, but may not be feasible across the wide
array of gas can usage. It is possible to design a spout that works
well on some equipment but not for all equipment. It might also be
possible to cover more uses by having multiple spouts, but we believe
that having multiple spouts would lead to confusion and would also
require consumers to have multiple cans depending on the types of
equipment that they refuel. We request comment on automatic shut-off
requirements and on ways to establish an automatic shut-off requirement
that would reduce spillage, be feasible for manufacturers, and be
practical for consumers.
5. Consideration of Retrofits of Existing Gas Cans
Clean Air Act section 183(e) provides authority to consider
retrofitting gasoline containers as an approach for controlling
emissions. We do not believe, however, that requiring the retrofit of
existing gas cans would be a feasible approach for controlling gas can
emissions, either technically or economically. This would likely entail
manufacturers first developing retrofit systems (including spouts for
various previous gas can designs), testing them for emissions
performance, and certifying them with EPA. Manufacturers would need
time to develop and certify systems and also to develop an
implementation strategy, considering that there are millions of cans in
use. Manufacturers would then likely need to collect gas cans from
consumers, recondition the cans, permanently close vents, incorporate
permeation barriers, and incorporate new spouts. We believe that this
process would lead to costs that far exceed the cost of newly
manufactured gas cans. In addition, emissions reductions would depend
on consumer participation, which would be highly uncertain given that
gas cans are relatively low-cost consumer products. In fact, we believe
that consumers who are concerned about emissions would be more likely
to discard old gas cans and purchase new cans meeting emissions
standards. For all these reasons, we do not believe that a retrofitting
approach makes sense for gas cans.
6. Consideration of Diesel, Kerosene and Utility Containers
We are requesting comment on but not proposing applying emissions
control requirements to diesel, kerosene, and utility containers. Due
to the low volatility of diesel and kerosene, the evaporative losses
from diesel and kerosene cans would be minimal when used with the
designated fuels. California has included diesel and kerosene cans in
their regulations largely due to the concern that they would be
purchased as substitutes for gasoline containers. California also
included utility containers in their portable fuel container program
due to concerns that these containers would be used for gasoline. We
believe that manufacturers can minimize this incentive by designing
gasoline cans and spouts that are easy to use and beneficial to the
consumer. However, storing gasoline in diesel, kerosene, and utility
containers would result in a loss of emissions reductions and therefore
we are requesting comment on including them in the program. The costs
for these containers would be similar to the costs estimated for
gasoline containers. We request comment on the potential for diesel,
kerosene, and utility containers to be used as a substitute for
regulated gas cans, and the cost and other implications of including
them in the program.

C. Timing of Standard

As an aspect of considering the proposed standard's technological
feasibility, we are proposing to require manufacturers to meet the
standard beginning January 1, 2009. Manufacturers have developed the
primary technologies to reduce emissions from gas cans but will need a
few years of lead time to certify products and ramp up production to a
national scale. The certification process would take at least six
months due to the required durability demonstrations described below,
and manufacturers would need time to procure and install the tooling
needed to produce gas cans with permeation barriers for nationwide sales.
The standards would apply to gas cans manufactured on or after the
start date of the program and would not affect cans produced before the
start date. We propose that as of July 1, 2009, manufacturers and
importers must not enter into U.S. commerce any products not meeting
the emissions standards. This provides manufacturers with a 6-month
period to clear any stocks of gas cans manufactured prior to the
January 1, 2009 start of the program, allowing the normal sell through
of these cans to the retail level. Retailers would be able to sell
their stocks of gas cans through the course of normal business without
restriction. Gas cans are currently stamped with their production date,
which would allow EPA to determine which cans are required to meet the
new standards.
We believe that the 2009 time frame is feasible, but recognize that
it could be a challenge for manufacturers with high volume sales to
ramp up production. We request comment on the economic feasibility of
the proposed timing and also on whether or not a phase-in of the
standards would ease the transition to a national program. We encourage
commenters to provide detailed rationale and data where possible to
support their comments.

D. What Test Procedures Would Be Used?

As part of the proposed system of regulations for gas cans, we are
proposing test conditions designed to assure that the intended emission
reductions occur over a range of in-use conditions such as operating at
different temperatures, with different fuels, and considering factors
affecting durability. These proposed test procedures implement section
183(e)(4), which authorizes EPA to develop appropriate standards
relating to product use. Emission testing on all gas cans that
manufacturers produce is not feasible due to the high volumes of gas
cans produced every year and the cost and time involved with emissions
testing. Instead, we are proposing that before the gas cans are
introduced into commerce, EPA would need to certify gas cans to the
emissions standards based on manufacturers' applications for
certification. Manufacturers would submit test data on a sample of gas
cans that are prototypes of the products manufacturers intend to
produce. Manufacturers would also need to certify that their production
cans would not deviate in materials or design from the prototype gas
cans that are tested. Manufacturers would need to obtain approval of
their certification from EPA prior to introducing their products into
commerce. The proposed test procedures and certification requirements
are described in detail below.
We are proposing that manufacturers would test cans in their most
likely storage configuration. The key to reducing evaporative losses
from gas cans is to ensure that there are no openings on the cans that
could be left open by the consumer. Traditional cans

[[Page 15898]]

have vent caps and spout caps that are easily lost or left off cans,
which leads to very high evaporative emissions. We expect manufacturers
to meet the evaporative standards by using automatic closing spouts and
by removing other openings that consumers could leave open. However, if
manufacturers choose to design cans with an opening that does not close
automatically, we are proposing to require that containers be tested in
their open condition. If the gas cans have any openings that consumers
could leave open (for example, vents with caps), these openings thus
would need to be left open during testing. This would apply to any
opening other than where the spout attaches to the can. We believe it
is important to take this approach because these openings could be a
significant source of in-use emissions and there is a realistic
possibility that these openings would be inadvertently left open in use.
We propose that spouts would be in place during testing because
this would be the most likely storage configuration for the emissions
compliant cans. Spouts would still be removable so that consumers would
be able to refill the cans, but we would expect the containers to be
resealed by consumers after being refilled in order to prevent spillage
during transport. We do not believe that consumers would routinely
leave spouts off cans because spouts are integral to the cans' use and
it is obvious that they need to be sealed.
1. Diurnal Test
We are proposing a test procedure for diurnal emissions testing
where manufacturers (or others conducting the testing) place gas cans
in an environmental chamber or a Sealed Housing for Evaporative
Determination (SHED), vary the temperature over a prescribed
temperature and time profile, and measure the hydrocarbons escaping
from the gas can. We are proposing that gas cans would be tested over
the same 72-96 [deg]F (22.2-35.6 [deg]C) temperature profile used for
automotive applications. This temperature profile represents a hot
summer day when ground level ozone emissions (formed from hydrocarbons
and oxides of nitrogen) would be highest. We propose that three
containers would be tested, each over a three-day test. We are
proposing that three cans would be tested for certification in order to
address variability in products or test measurements. All three cans
would have to individually meet the proposed standard. As noted above,
gas cans would be tested in their most likely storage configuration.
The final result would be reported in grams per gallon, where the
grams are the mass of hydrocarbons escaping from the gas can over 24
hours and the gallons are the nominal gas can capacity. The daily
emissions would then be averaged for each can to demonstrate compliance
with the standard. This test would capture hydrocarbons lost through
permeation and any other evaporative losses from the gas can as a
whole. We are proposing that the grams of hydrocarbons lost would be
determined by either weighing the gas can before and after the diurnal
test cycle or measuring emissions directly using the SHED instrumentation.
Consistent with the automotive test procedures, we are proposing
that the testing take place using 9 pounds per square inch (psi) Reid
Vapor Pressure (RVP) certification gasoline, which is the same fuel
required by EPA to be used in its other evaporative test programs. We
are proposing for this testing to use E10 fuel (10% ethanol blended
with the gasoline described above) in this testing to help ensure in-
use emission reductions on ethanol-gasoline blends, which tend to have
increased evaporative emissions with certain permeation barrier
materials. We believe including ethanol in the test fuel will lead to
the selection of materials by manufacturers that are consistent with
``best available control'' requirements for all likely contained
gasolines, and is clearly appropriate given the expected increase over
time of the use of ethanol blends of gasoline under the renewable fuel
provisions of the Energy Policy Act of 2005. Diurnal emissions are not
only a function of temperature and fuel volatility, but of the size of
the vapor space in the container as well. We are proposing that the
fill level at the start of the test be 50% of the nominal capacity of
the gas can. This would likely be the average fuel level of the gas can
in-use. Nominal capacity of the gas cans would be defined as the volume
of fuel, specified by the manufacturer, to which the gas can could be
filled when sitting on level ground. The vapor space that normally
occurs in a gas can, even when ``full,'' would not be considered in the
nominal capacity of the gas can. All of these test requirements are
proposed to represent typical in-use storage conditions for gas cans,
on which EPA can base its emissions standards. These provisions are
proposed as a way to implement the standards effectively, which will
lead to the use of best available technology at a reasonable cost.
Before testing for certification, the gas cans would be run through
the durability tests described below. Within 8 hours of the end of the
soak period contained in the durability cycle, the gas cans would be
drained and refilled to 50 percent nominal capacity with fresh fuel,
and then the spouts re-attached. When the gas can is drained, it would
have to be immediately refilled to prevent it from drying out. The
timing of these steps is needed to ensure that the stabilized
permeation emissions levels are retained. The can will then be weighed
and placed in the environmental chamber for the diurnal test. After
each diurnal, the can would be re-weighed. In lieu of weighing the gas
cans, we propose that manufacturers could opt to measure emissions from
the SHED directly. For any in-use testing of gas cans, the durability
procedures would not be run prior to testing.
California's test procedures are very similar to those described
above. However, the California procedure contains a more severe
temperature profile of 65-105 [deg]F. We propose to allow manufacturers
to use this temperature profile to test gas cans as long as other parts
of the EPA test procedures are followed, including the durability
provisions below. We request comment on these test procedures,
including ways the procedures may be further streamlined without impacting
the overall emissions measurements and performance of the gas cans.
2. Preconditioning To Ensure Durable In-Use Control
a. Durability Cycles
To determine permeation emission deterioration rates, we are
specifying three durability aging cycles: Slosh, pressure-vacuum
cycling, and ultraviolet exposure. They represent conditions that are
likely to occur in-use for gas cans, especially for those cans used for
commercial purposes and carried on truck beds or trailers. The purpose
of these deterioration cycles is to help ensure that the technology
chosen by manufacturers is durable in-use, representing best available
control, and the measured emissions are representative of in-use
permeation rates. Fuel slosh, pressure cycling, and ultraviolet (UV)
exposure each impact the durability of certain permeation barriers, and
we believe these cycles are needed to ensure long-term emissions
control. Without these durability cycles, manufacturers could choose to
use materials that meet the certification standard but have degraded
performance in-use, leading to higher emissions. We do not expect these
procedures to adversely impact the feasibility of the standards, because

[[Page 15899]]

there are permeation barriers available at a reasonable cost that do
not deteriorate significantly under these conditions (which permeation
barriers are examples of best available controls). As described above,
we believe including these cycles as part of the certification test is
preferable to a design-based requirement.
For slosh and pressure cycling, we are proposing to use durability
tests that are based on draft recommended SAE practice for evaluating
permeation barriers.\280\ For slosh testing, the gas can would be
filled to 40 percent capacity with E10 fuel and rocked for 1 million
cycles. The pressure-vacuum testing contains 10,000 cycles from -0.5 to
2.0 psi. The third durability test is intended to assess potential
impacts of ultraviolet (UV) sunlight (0.2 [mu]m-0.4 [mu]m) on the
durability of a surface treatment. In this test, the gas cans must be
exposed to a UV light of at least 0.40 Watt-hour/meter\2\ /minute on
the gas can surface for 15 hours per day for 30 days. Alternatively,
gas cans could be exposed to direct natural sunlight for an equivalent
period of time. We have also established these same durability
requirements as part of our program to control permeation emissions
from recreational vehicle fuel tanks.\281\ While there are obvious
differences in the use of gas cans compared to the use of recreational
vehicle fuel tanks, we believe the test procedures offer assurance that
permeation controls used by manufacturers will be robust and will
continue to perform as intended when in use. We request comments on the
use of these procedures for gas cans to help ensure permeation control
in-use.
---------------------------------------------------------------------------

\280\ Draft SAE Information Report J1769, ``Test Protocol for
Evaluation of Long Term Permeation Barrier Durability on Non-
Metallic Fuel Tanks,'' (Docket A-2000-01, document IV-A-24).
\281\ Final Rule, ``Control of Emissions from Nonroad Large
Spark-ignition engines, and Recreational Engines (Marine and Land-
based)'', 67 FR 68287, November 8, 2002.
---------------------------------------------------------------------------

We also propose to allow manufacturers to do an engineering
evaluation, based on data from testing on their permeation barrier, to
demonstrate that one or more of these factors (slosh, UV exposure, and
pressure cycle) do not impact the permeation rates of their gas cans
and therefore that the durability cycles are not needed. Manufacturers
would use data collected previously on gas cans or other similar
containers made with the same materials and processes to demonstrate
that the emissions performance of the materials does not degrade when
exposed to slosh, UV, and/or pressure cycling. The test data would have
to be collected under equivalent or more severe conditions as those
noted above.
b. Preconditioning Fuel Soak
It takes time for fuel to permeate through the walls of containers.
Permeation emissions will increase over time as fuel slowly permeates
through the container wall, until the permeation finally stabilizes
when the saturation point is reached. We want to evaluate emissions
performance once permeation emissions have stabilized, to ensure that
the emissions standard is met in-use. Therefore, we are proposing that
prior to testing the gas cans, the cans would need to be preconditioned
by allowing the cans to sit with fuel in them until the hydrocarbon
permeation rate has stabilized. Under this step, the gas can would be
filled with a 10-percent ethanol blend in gasoline (E10), sealed, and
soaked for 20 weeks at a temperature of 28 ± 5[deg] C. As an
alternative, we are proposing that the fuel soak could be performed for
10 weeks at 43 ± 5[deg]C to shorten the test time. During
this fuel soak, the gas cans would be sealed with the spout attached.
This is representative of how the gas cans would be stored in-use. We
have established these soak temperatures and durations based on
protocols EPA has established to measure permeation from fuel tanks
made of HDPE.\282\ These soak times should be sufficient to achieve
stabilized permeation emission rates. However, if a longer time period
is necessary to achieve a stabilized rate for a given gas can, we would
expect the manufacturer to use a longer soak period (and/or higher
temperature) consistent with good engineering judgment.
---------------------------------------------------------------------------

\282\ Final Rule, ``Control of Emissions from Nonroad Large
Spark-ignition engines, and Recreational Engines (Marine and Land-
based)'', 67 FR 68287, November 8, 2002.
---------------------------------------------------------------------------

Durability testing that is performed with fuel in the gas can may
be considered part of the fuel soak provided that the gas can
continuously has fuel in it. This approach would shorten the total test
time. For example, the length of the UV and slosh tests could be
considered as part of the fuel soak provided that the gas can is not
drained between these tests and the beginning of the fuel soak.
c. Spout Actuation
In its recently revised program for gas cans, California included a
durability demonstration for spouts. We are proposing a durability
demonstration consistent with California's procedures. Automatically
closing spouts are a key part of the emissions controls expected to be
used to meet the proposed standards. If these spouts stick or
deteriorate, in-use emissions could remain very high (essentially
uncontrolled). We are interested in ways to ensure during the
certification procedures that the spouts also remain effective in use.
California requires manufacturers to actuate the spouts 200 times prior
to the soak period and 200 times near the conclusion of the soak period
to simulate spout use. The spouts' internal components would be
required to be exposed to fuel by tipping the can between each cycle.
Spouts that stick open or leak during these cycles would be considered
failed. The total of 400 spout actuations represents about 1.5
actuations per week on average over the average container life of 5
years. In the absence of data, we believe this number of actuations
appears to reasonably replicate the number that can occur in-use for
high end usage and will help ensure quality spout designs that do not
fail in-use. We also believe that proposing requirements consistent
with California will help manufacturers to avoid duplicate testing. We
request comment on the above approach for demonstrating spout durability.

E. What Certification and In-Use Compliance Provisions Is EPA Proposing?

1. Certification
Section 183(e)(4) authorizes EPA to adopt appropriate systems of
regulations to implement the program, including requirements ranging
from registration and self-monitoring of products, to prohibitions,
limitations, economic incentives and restrictions on product use. We
are proposing a certification mechanism pursuant to these authorities.
Manufacturers would be required to go through the certification process
specified in the proposed regulations before entering their containers
into commerce. To certify products, manufacturers would first define
their emission families. This is generally based on selecting groups of
products that have similar emissions. For example, co-extruded gas cans
of various geometries could be grouped together. The manufacturer would
select a worst-case configuration for testing, such as the thinnest-
walled gas can. These determinations may be made using good engineering
judgment and would be subject to EPA review. Testing with those
products, as specified above, would need to show compliance with
emission standards. The manufacturers would then send us an application
for certification. We propose to define the

[[Page 15900]]

manufacturer as the entity that is in day-to-day control of the
manufacturing process (either directly or through contracts with
component suppliers) and responsible for ensuring that components meet
emissions-related specifications. Importers would not be considered a
manufacturer and thus would not be certifying entities; the
manufacturers of the cans they import would have to certify the cans.
Importers would only be able to import gas cans that are certified.
After reviewing the information in the application, we would issue
a certificate of conformity allowing manufacturers to introduce into
commerce the gas cans from the certified emission family. EPA review
would typically take about 90 days or less, but could be longer if we
have questions regarding the application. The certificate of conformity
would be for a production period of up to five years. Manufacturers
could carry over certification test data if no changes are made to
their products that would affect emissions performance. Changes to the
certified products that would affect emissions would require
reapplication for certification. Manufacturers wanting to make changes
without doing testing would be required to present an engineering
evaluation demonstrating that emissions are not affected by the change.
The certifying manufacturer accepts the responsibility for meeting
applicable emission standards. While we are proposing no requirement
for manufacturers to conduct production-line testing, we may pursue EPA
in-use testing of certified products to evaluate compliance with
emission standards. If we find that gas cans do not meet emissions
standards in use, we would consider the new information during future
product certification. Also, we may require certification prior to the
end of the five-year production period otherwise allowed between
certifications. The details of the proposed certification process are
provided in the proposed regulatory text. We request comments on the
certification process we are proposing.
2. Emissions Warranty and In-Use Compliance
We are proposing a warranty period of one year to be provided by
the manufacturer of the gas can to the consumer. The warranty would
cover emissions-related materials defects and breakage under normal
use. For example, the warranty would cover failures related to the
proper operation of the auto-closing spout or defects with the
permeation barriers. We are also proposing to require that
manufacturers submit a warranty and defect report documenting
successful warranty claims and the reason for the claim to EPA annually
so that EPA may monitor the program. Unsuccessful claims would not need
to be submitted. We believe that this warranty will encourage designs
that work well for consumer and are durable. Although it does not fully
cover the average life of the product, it is not typical for very long
warranties to be offered with products and therefore we believe a one
year warranty is reasonable. Also, the warranty period is more similar
to the expected life of gas cans when used in commercial operations,
which would need to be considered by the manufacturers in their
designs. We request comment on the warranty period.
EPA views this aspect of the proposal as another part of the
``system of regulation'' it is proposing to control VOC emissions from
gas cans, which system may include ``requirements for registration and
labeling * * * use, or consumption * * * of the product'' pursuant to
section 183(e)(4) the Act. A warranty will promote the objective of the
proposed rule by assuring that manufacturers will ``stand behind''
their product, thus improving product design and performance.
Similarly, the proposed defect reporting requirement will promote
product integrity by allowing EPA to readily monitor in-use performance
by tracking successful warranty claims.
Gas cans have a typical life of about five years on average before
they are scrapped. We are proposing durability provisions as part of
certification testing to help ensure containers perform well in use (a
system of regulation for ``use'' of the product, pursuant to section
183(e)(4)). Under the proposal, we could test gas cans within their
five-year useful life period to monitor in-use performance and take
steps to correct in-use failures, including denying certification, for
container designs that are consistently failing to meet emissions
standards. (This proposed provision thus would work in tandem with the
warranty claim reporting provision proposed in the preceding paragraph.)
We are not proposing any recall provisions for gas cans.
Manufacturers do not have registration programs for gas cans and
implementing such a program for a low-cost consumer product may be
overly burdensome, and have a very low participation rate. Also, we
would not expect a high participation rate from consumers in a recall,
in any event, due to the nature of gas cans as a consumer product. We
believe, however, that by having the authority to test products in use,
along with the possible repercussions of in-use noncompliance, will
encourage manufacturers to develop robust designs.
3. Labeling
Since the requirements will be effective based on the date of
manufacture of the gas can, we propose that the date of manufacture
must be indelibly marked on the can. This is consistent with current
industry practices. This is needed so that we and others can recognize
whether a unit is regulated or not. In addition, we propose to require
a label providing the manufacturer name and contact information, a
statement that the can is EPA certified, citation of EPA regulations,
and a statement that it is warranted for one year from the date of
purchase. The manufacturer name and contact information is necessary to
verify certification. Indicating that a 1 year warranty applies will
ensure that consumers have knowledge of the warranty and a way to
contact the manufacturer. Enforcement of the warranty is critical to
the defect reporting system. In proposing this labeling requirement, we
further believe, pursuant to section 183(e)(8), that these labeling
requirements would be useful in meeting the NAAQS for ozone. They
provide necessary means of implementing the various measures described
above which help ensure that VOC emission reductions from the proposed
standard will in fact occur in use.

F. How Would State Programs Be Affected by EPA Standards?

As described in section VIII.A.3. above, several states have
adopted emissions control programs for gas cans. California implemented
an emissions control program for gas cans in 2001. Thirteen other
states, mostly in the northeast, have adopted the California program in
recent years.\283\ Last year, California adopted a revised program,
which will go into effect on July 1, 2007. The revised California
program is very similar to the program we are proposing. We believe
that although a few aspects of the program we are proposing are
different, manufacturers will be able to meet both EPA and CARB
requirements with the same gas can designs and therefore sell a single
product in all 50

[[Page 15901]]

states. In most cases, we believe manufacturers will take this
approach. By closely aligning with California where possible, we will
allow manufacturers to minimize research and development (R&D) and
emissions testing, while potentially achieving better economies of
scale. It may also reduce administrative burdens and market logistics
from having to track the sale of multiple can designs. We consider
these to be important factor under CAA section 183(e) which requires us
to consider economic feasibility of controls.
---------------------------------------------------------------------------

\283\ Delaware, Maine, Maryland, Pennsylvania, New York,
Connecticut, Massachusetts, New Jersey, Rhode Island, Vermont,
Virginia, Washington DC, and Texas.
---------------------------------------------------------------------------

States that have adopted the original California program will
likely choose to either adopt the new California program or eliminate
their state program in favor of the federal program. Because the
programs are similar, we expect that most states will eventually choose
the EPA program rather than continue their own program. We expect very
little difference in the emissions reductions provided by the EPA and
California programs in the long term. In addition, if EPA's program
starts in 2009, as discussed above, this would be the same timing
states would likely target in their program revisions.

G. Provisions for Small Gas Can Manufacturers

As discussed in previous sections, prior to issuing a proposal for
this proposed rulemaking, we analyzed the potential impacts of these
regulations on small entities. As a part of this analysis, we convened
a Small Business Advocacy Review Panel (SBAR Panel, or ``the Panel'').
During the Panel process, we gathered information and recommendations
from Small Entity Representatives (SERs) on how to reduce the impact of
the rule on small entities, and those comments are detailed in the
Final Panel Report which is located in the public record for this
rulemaking (Docket EPA-HQ-OAR-2005-0036). Based upon these comments, we
propose to include flexibility and hardship provisions for gas can
manufacturers. Since nearly all gas can manufacturers (3 of 5
manufacturers as defined by SBA) are small entities and they account
for about 60 percent of sales, the Panel recommended to extend the
flexibility options and hardship provisions to all gas can
manufacturers. (Our proposal today is consistent with that
recommendation.) Moreover, implementation of the program would be much
simpler by doing so. The flexibility provisions are incorporated into
the program requirements described earlier in sections VIII.C through
VIII.E. The hardship provisions are described below. For further
discussion of the Panel process, see section XII.C of this proposed
rule and/or the Final Panel Report.
The Panel recommended that two types of hardship provisions be
extended to gas can manufacturers. These entities could, on a case-by-
case basis, face hardship, and we are proposing these provisions to
provide what could prove to be needed safety valves for these entities.
Thus, the propose hardship provisions are as follows:
1. First Type of Hardship Provision
Gas can manufacturers would be able to petition EPA for limited
additional lead-time to comply with the standards. A manufacturer would
have to demonstrate that it has taken all possible business, technical,
and economic steps to comply but the burden of compliance costs or
would have a significant adverse effect on the company's solvency.
Hardship relief could include requirements for interim emission reductions.
2. Second Type of Hardship Provision
Gas can manufacturers would be permitted to apply for hardship
relief if circumstances outside their control cause the failure to
comply (i.e. supply contract broken by parts supplier), and if failure
to sell the subject containers would have a major impact on the
company's solvency. The terms and timeframe of the relief would depend
on the specific circumstances of the company and the situation involved.
For both types of hardship provisions, the length of the hardship
relief would be established during the initial review for not more than
one year and would be reviewed annually thereafter as needed. As part
of its application, a company would be required to provide a compliance
plan detailing when and how it would achieve compliance with the standards.

IX. What Are the Estimated Impacts of the Proposal?

A. Refinery Costs of Gasoline Benzene Reduction

The proposed 0.62 volume percent benzene standard would generally
result in many refiners investing in benzene control hardware and
changing the operations in their refineries to reduce their gasoline
benzene levels. The proposed ABT program would allow refiners to
optimize their investments, which we believe would maximize the benzene
reductions at the lowest possible cost. We have estimated that the
capital and operating costs that we believe would result from the
proposed program would average 0.13 cents per gallon of gasoline.
In this section we summarize the methodology used to estimate the
costs of benzene control, the scenarios we evaluated, and our estimated
costs for the program. We also summarize the results of our analyses of
other potential MSAT control programs. A detailed discussion of all of
these analyses is found in Chapter 9 of the RIA.
1. Tools and Methodology
a. Linear Programming Cost Model
We considered performing our cost assessments for this proposed
program using a linear programming (LP) cost model. LP cost models are
based on a set of complex mathematical representations of refineries
which, for national analyses, are usually conducted on a regional
basis. This type of refining cost model has been used by the government
and the refining industry for many years for estimating the cost and
other implications of changes to fuel quality.
The design of LP models lends itself to modeling situations where
every refinery in a region is expected to use the same control strategy
and/or has the same process capabilities. As we began to develop a
gasoline benzene control program with an ABT program, it became clear
that LP modeling was not well suited for evaluating such a program.
Because refiners would be choosing a variety of technologies for
controlling benzene, and because the program would be national and
would include an ABT program, we initiated development of a more
appropriate cost model, as described below. However, the LP model
remained important for providing many of the inputs into the new model,
and for performing analyses of other potential programs.
b. Refiner-by-Refinery Cost Model
In contrast to LP models, refinery-by-refinery cost models are
useful when individual refineries would respond to program requirements
in different ways and/or have significantly different process
capabilities. Thus, in the case of today's proposed gasoline benzene
control program, we needed a model that would accurately simulate the
variety of decisions refiners would make at different refineries,
especially in the context of a nationwide ABT program. For this and
other related reasons, we developed a refinery-by-refinery cost model
specifically to evaluate the proposed benzene control program.
Our benzene cost model incorporates the capacities of all the major
units in

[[Page 15902]]

each refinery in the country, as reported by the Energy Information
Administration and in the Oil and Gas Journal. Regarding operational
information, we know less about how the various units are used to
produce gasoline and such factors as octane and hydrogen costs for
individual refineries. We used the LP model to estimate these factors
on a regional basis, and we applied the average regional result to each
refinery in that region (PADD). We calibrated the model for each
individual refinery based on 2003 gasoline volumes and benzene levels,
which was the most recent year for which data was available, and found
that the model simulated the actual situation well. We also compared
cost estimates of similar benzene control cases from both the refinery-
by-refinery model and the LP model, and the results were in close agreement.
Refinery-by-refinery cost models have been used in the past by both
EPA and the oil industry for such programs as the highway and nonroad
diesel fuel sulfur standards, and they are a proven means for
estimating the cost of compliance for fuel control programs. For the
specific benzene cost model, we have initiated a peer review process,
and have received some comments on the design of our model. Although we
did not receive these comments in time to respond to them in this
proposal, we plan to address all peer review comments in the
development of the final rule. (Based on our initial assessment of
these comments, we do not believe that the changes suggested would
significantly affect the projected costs of the program. See Chapter 9
of the RIA for our initial responses to these peer-review comments.)
Based on our understanding of the primary benzene control
technologies (see section VII.F above), the cost model assumes that
four technologies would be used, as appropriate, for reducing benzene
levels. All of these technologies focus on addressing benzene in the
reformate stream. They are (1) routing the benzene precursors around
the reformer; (2) routing benzene precursors to an existing
isomerization unit, if available; (3) benzene extraction (extractive
distillation); and (4) benzene saturation. There are several
restrictions on the use of these various technologies (such as the
assumption that benzene extraction would only be expanded in areas with
strong benzene chemical markets) and these are incorporated into the model.
For the proposed benzene control program, the associated nationwide
ABT program is intended to optimize benzene reduction by allowing each
refinery to individually choose the most cost-effective means of
complying with the program. To model this phenomenon, we first
establish an estimated cost for the set of technologies required for
each refinery to meet the standard. We then rank the refineries in
order from lowest to highest control cost per gallon of gasoline. The
model then follows this ranking, starting with the lowest-cost
refineries, and adds refineries and their associated control
technologies one by one until the projected national average benzene
level reaches 0.62 volume percent. This establishes which refineries we
expect to apply control technologies to comply, as well as those that
would generate credits and those that would use credits in lieu of
investing in control. The sum of the costs of the refineries expected
to invest in control provides the projected overall cost of the program.
c. Price of Chemical Grade Benzene
The price of chemical grade benzene is critical to the proposed
program because it defines the opportunity cost for benzene removed
using benzene extraction and sold into the chemicals market. According
to 2004 World Benzene Analysis produced by Chemical Market Associates
Incorporated (CMAI), during the consecutive five year period ending
with 2004, the price of benzene averaged 24 dollars per barrel higher
than regular grade gasoline. During the three consecutive year period
ending with 2004, the price of benzene averaged 28 dollars per barrel
higher than regular grade gasoline. However, during the first part of
2004, the price of benzene relative to gasoline rose steeply, primarily
because of high energy prices adding to the cost of extracting benzene.
The projected benzene price for 2004 indicated that the benzene price
averaged 38 dollars per barrel higher than regular grade gasoline.
For the future, CMAI projects that the price of benzene relative to
gasoline will return to more historic levels or lower, in the range of
$20 per barrel higher than regular grade gasoline. We have based our
modeling on this value. However, we have also examined the sensitivity
of the projected overall program costs for a case where the cost of
benzene control remains at $38 higher than gasoline into the future.
d. Applying the Cost Model to Special Cases
For the comparative cases we modeled that involve a maximum-average
(max-avg) standard in addition to an average benzene standard, modeling
the costs requires a different modeling methodology. Refineries that
the model estimates would have benzene levels above the max-avg
standard are assumed to apply the most cost-effective benzene reduction
technologies that the model shows would reduce benzene levels to below
the max-avg standard. The benzene reductions associated with meeting
the max-avg standard may or may not be sufficient for also meeting the
average standard, depending on how stringent the max-avg standard is
relative to the average standard. If the model indicates that
additional benzene reduction would be necessary, these additional
benzene reductions are modeled in the same way as the case of an
average standard only, as described above.
We also evaluated a limited number of cases that did not include an
ABT program. In such cases, the model assumes that all the refineries
with benzene levels below the standard would maintain the same benzene
level, while each refinery with benzene levels above the standard would
take all the necessary steps to reduce their benzene levels down to the
standard. If the model shows that capital investments are needed to
achieve the necessary benzene reduction, we assume that the refiner
installs a full sized unit to treat the entire stream and then operates
the unit only to the extent necessary to meet the standard.
2. Summary of Costs
a. Nationwide Costs of the Proposed Program
We have used the refinery-by-refinery cost model to estimate the
costs of the proposed program, with an average gasoline benzene content
standard of 0.62 volume percent and the proposed ABT program. In
general, the cost model indicates that among the four primary
reformate-based technologies, benzene extraction would be the most cost
effective. The next most cost effective technologies are benzene
precursor rerouting, and rerouting coupled with isomerization. The
model indicates that benzene saturation would be the least cost-
effective, but only marginally so in the larger refineries.
Our refinery-by-refinery model estimates that 92 refineries of the
total 115 gasoline-producing refineries in the U.S. would have to put
in new capital equipment or change their refining operations to reduce
the benzene levels in their gasoline. Of these refineries 25 would use
benzene precursor removal, 32 refineries would use benzene precursor
removal coupled with isomerization, 24 would use extraction,

[[Continued on page 15903]]

Notices For
2009
2008
2007
2006
2005
2004
2003
2002
2001
2000
1999
1998
1997
1996
1995
1994

Control of Hazardous Air Pollutants From Mobile Sources

Local Navigation