Fuel Program Costs 

Methodology

This chapter provides a summary of the methodology used and the results
obtained from our cost analyses of the proposed gasoline sulfur control.
 We start by summarizing the refinery models used for our analysis.  We
then describe our detailed methodology for estimating the sulfur control
costs for our proposed sulfur program followed by the results.  We
present the results from our energy and supply analyses for our proposed
gasoline sulfur program.  Finally, we discuss and compare the results of
several cost analyses for various sulfur programs.  

Overview

When we began our planning for estimating the cost of additional
reductions in gasoline sulfur, we considered two different options.  One
option for estimating the costs would be to utilize a linear programming
(LP) model, while the second option would be to develop a
refinery-by-refinery cost model.  While the LP refinery models are
necessary and appropriate for many analyses, they also have several
important limitations of relevance here.  When used to model the cost of
nationwide fuel control programs on the entire refining industry, LP
models are usually used to model groups of refineries in geographic
regions called Petroleum Administration for Defense Districts (PADDs). 
The LP refinery model averages the costs over the refineries represented
in the PADDs; however, the technology chosen by the refinery model would
normally be the lowest cost technology found by the refinery model. 
This may represent an unreasonable choice of technologies for individual
refineries because of how refineries are configured and based on the
sulfur control technologies installed for compliance with the Tier 2
gasoline sulfur program.  While the choice of technologies can be
limited based on an approximate analysis of what mix of technologies
would best suit the group of refineries modeled in each PADD, this would
only provide an approximate estimate of the cost incurred.  Based on the
quality of input data to these LP models and the assumptions made for
complying with a regulatory requirement, LP refinery models may
overestimate or underestimate the program costs.  For example, Also an
the LP refinery model would not be a sensible tool for estimating the
credit averaging and trading between refineries.  This could be
partially overcome by iterating between PADD refinery model runs, thus
estimating the number of credits traded between PADDs and estimating the
level of sulfur control in each PADD.  However, the need to make
multiple runs per PADD for each case, coupled with the need to run
multiple control cases for different sulfur standards, would be very
time consuming, costly and still would only result in approximate
estimates of the sulfur levels achieved and the cost incurred.

For this reason, EPA developed a refinery-by-refinery cost model which
models the capability for each refinery to revamp existing or install
new sulfur control technologies available to them to reduce their
gasoline sulfur levels.  Rather than start from scratch, we started from
a refinery-by-refinery cost model developed by APT (Mathpro) for EPA to
estimate the cost of benzene control under MSAT2.  However, instead of
using the representations of benzene control technology contained in the
model, we obtained information about gasoline desulfurization and
represented the cost of this desulfurization in the refinery-by-refinery
cost model.  

We believe the refinery-by-refinery cost model best estimates the cost
of individual refineries, especially when considering an averaging,
banking and trading (ABT) program and therefore is the best analysis
tool for estimating nationwide costs.  However, the refinery-by-refinery
cost model cannot estimate certain inputs necessary for estimating
costs.  Because the refinery-specific information is not publicly
available, it was necessary to find another way to estimate this
information.  The inputs and outputs from LP refinery cost modeling
provide this needed information and it was utilized in the
refinery-by-refinery cost model.  The information from LP refinery
modeling used in the refinery-by-refinery cost model is described in
Section   REF _Ref308076513 \r \h  5.1.3 .

 Since the refinery-by-refinery cost model contains confidential
business information for each refinery, we could not publish the model
or present some of the details of the model here.  Therefore, to ensure
its viability the refinery-by-refinery cost model was subjected to peer
review by two refinery industry consultants.  Our review of most of the
suggested changes recommended by the peer reviewers suggested that there
would be little to no change in our desulfurization cost estimate (some
of the changes would increase the estimated costs, while others would
reduce the estimated costs).  Also, we anticipate making other
improvements to the cost analysis conducted for the final rule, which
would necessitate a second round of peer review.  Therefore, the peer
review comments will be addressed prior to undertaking the cost analysis
for the final rulemaking  along with the other changes that we will be
making to our cost analysis.  The peer review comments are contained in
two reports submitted to the docket. 

  The refinery-by-refinery cost model focuses on reducing sulfur from
the FCC naphtha because of its high sulfur content.  To comply with the
30-ppm Tier 2 sulfur control program, most refiners installed FCC
naphtha hydrotreaters (referred to as FCC postreaters) or FCC feed
hydrotreaters (referred to as FCC pretreaters) to reduce that unit’s
sulfur contribution to their gasoline pool.  If refiners installed an
FCC postreater under Tier 2, we modeled refiners revamping those units. 
However, if refiners relied on FCC pretreaters to comply with Tier 2, we
assumed that grassroots FCC postreaters would have to be installed at
those refineries to reduce its gasoline pool down to 10 ppm.  However,
since adding grassroots FCC postreaters is relatively expensive for the
amount of sulfur reduction obtained, the ABT analysis we conducted
avoided many of these types of investments.  Refineries with both pre
and postreaters today could achieve further gasoline sulfur reductions
less than 10 ppm at a relatively low incremental cost and sell the
credits to those refiners who are operating refineries which would
otherwise be faced with grassroots postreater investments.  In addition
to addressing the sulfur in the FCC naphtha, we believe that some
refineries may need to reduce the sulfur in light straight run (LSR)
naphtha.  Some refineries might also need to reduce sulfur in butane,
although we don’t expect refiners to need to address butane sulfur
unless they are pursuing a very stringent gasoline sulfur standard,
e.g., 5 ppm.

To better understand the desulfurization costs, we evaluated several
different scenarios or cases.  For a 10-ppm average sulfur standard, we
assessed the costs based on each refinery achieving the 10-ppm standard
with no averaging among refineries, an averaging program which assumed
intra-company transfers of sulfur credits, and an averaging program
which assumed nationwide transfers of sulfur credits.  To provide
credits for averaging and trading under the 10-ppm average standard, we
also evaluated refiners reducing their gasoline sulfur down to 5 ppm. 
Since we had estimated costs for each refinery to get to 5 ppm sulfur,
we also report out the cost for a 5 ppm average gasoline sulfur standard
assuming no averaging between refineries.  The costs for the proposed
sulfur program are based on a 10-ppm sulfur standard with intra-company
credit transfers.  These different scenarios are summarized in   REF
_Ref309223592  Table 5-1 .

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  1  Sulfur Control
Cases Evaluated for the Proposal

10-ppm Standard	5-ppm Standard

No ABT Program	No ABT Program

ABT Program with Intra-Company Credit Transfers (Proposed Rule Costs)
N/A

ABT Program with Nationwide Credit Transfers	N/A

LP Refinery Modeling Methodology and Results

Although we used the refinery-by-refinery cost model to estimate
gasoline desulfurization costs, certain input information was needed to
estimate the costs with refinery-by-refinery cost model, and without
access to detailed refinery-specific information, we relied on outputs
from our LP refinery modeling.  Perhaps the most important input is the
cost for making up the octane loss that occurs with desulfurization. 
Certain refinery operations information from the LP refinery model was
used for estimating the volume of gasoline produced in the
refinery-by-refinery model, including the utilization factors of
individual refinery units, and the percentage that straight run naphtha,
FCC naphtha and hydrocrackate comprises of the feed volume of their
respective units.  

  LP refinery models are detailed mathematical representations of
refineries.  They are used by individual refining companies to project
how best to operate their refineries.  They are also used by government
agencies, such as EPA and DOE, as well as by refining industry
associations and individual companies, to estimate the cost and supply
impacts of fuel quality changes.  LP refinery models have been used for
these purposes for decades and a certain protocol has been established
to conduct these studies.  

Two different sets of refinery modeling runs from two different LP
refinery models were used as inputs into the refinery-by-refinery cost
model.  The refinery-by-refinery cost model already contained the
utilization factors and gasoline production volumes for individual
refinery units from the analysis conducted by Mathpro for the MSAT2, and
we continued to use that information for this cost analysis.  The
gasoline demand is expected to be fairly flat in the future, so using
the previous refinery modeling work for these inputs will likely have
little impact on the cost estimate.  We plan on updating these inputs
for the final rule to reflect more recent refinery modeling work.  

Additional refinery modeling was conducted using the Haverly GRTMPS
refinery model.  The primary reason for conducting new LP refinery
modeling analysis was to estimate the cost of making up the octane loss
associated with desulfurization as well as estimate how gasoline
qualities would be affected by the octane recovery to feed into the
emissions inventory impact analysis discussed in Chapter 7.  While the
gasoline demand and production volumes are not expected to change in the
future, the cost of octane is expected to decrease dramatically due to
expected much larger use of ethanol under the RFS2 rulemaking.  

The first step in conducting an LP refinery modeling analysis was the
development of a base case.  The base case is a refinery modeling case
that calibrates the refinery model based on actual refinery unit
capacity and input and output data.  The base year for this study was
the year 2000 for the Mathpro model and the year 2004 for the Haverly
model.  Because much of the information available for establishing the
base case is only available for PADDs of refineries, the LP refinery
modeling was conducted on a PADD-wide basis.  Refinery capacity
information from the Oil and Gas Journal was aggregated by PADD and
entered into the LP refinery model.  The feedstock volumes, including
crude oil and gasoline blendstocks, were obtained from the Energy
Information Administration (EIA) and entered into each PADD’s model. 
Similarly, product volumes such as gasoline, jet fuel, and diesel fuel
were obtained from EIA and entered into the cost model.  The
environmental and ASTM fuel quality constraints in effect in the base
year were imposed on the products.  This includes the Reformulated
Gasoline program and the 500-ppm highway diesel fuel sulfur standard,
and for the Haverly LP refinery modeling, the first year of the Tier 2
gasoline sulfur standard.  This information was input into the LP
refinery cost model for each PADD and each PADD model was run to model
the U.S. refinery industry for the base year.  The gasoline quality for
each PADD refinery model was then compared to the actual gasoline
quality for conventional and reformulated gasoline which is available
from the RFG database.  Each model was calibrated to closely approximate
the gasoline quality of each PADD.  

The second step in modeling is the development of a reference case.  The
purpose of the reference case is to model the refining industry
operations and cost in a future year, which is the year that the control
program is modeled to be in effect (serving as a point of reference to
the control cases for estimating costs and other impacts).  The
reference year for the Mathpro LP refinery modeling was 2012 while the
reference year for the Haverly refinery modeling was 2017 and 2030.  We
developed two reference cases with the Haverly model to model different
control case scenarios.  The reference case was created by starting with
the base cases for each PADD and adjusting each base case to model the
future year, accounting for the changes between the two years.  

Two different types of adjustments were made to the base case refinery
models to enable modeling the refining industry for the reference case. 
First, the change in certain inputs such as product volumes and energy
prices need to be accounted for U.S. refinery gasoline, diesel fuel and
jet fuel demands are projected year-by-year by EIA in its Annual Energy
Outlook (AEO); the projections from the AEO for the reference case are
used in the refinery modeling analysis.  The Mathpro LP refinery
modeling relied on AEO 2006 while the Haverly LP refinery modeling
relied on AEO 2011.  This growth in demand is used to project refinery
production for each PADD to meet that increased demand.  This projected
growth in U.S. refinery production was entered into the reference case
version of the LP refinery model.  The utility and crude oil and other
feedstock prices which are projected by EIA for the future year being
modeled were also entered into the refinery model as well as the
estimated product prices.  

The second adjustment made to model the reference cases was the
application of fuel quality changes.  Environmental programs which have
been implemented or which will largely be implemented by the time that
the prospective fuels control program would take effect were modeled in
the reference case.  These fuel quality changes include limits such as
the 30-ppm average gasoline sulfur standard, 15-ppm caps on highway and
nonroad diesel fuel and the MSAT2 benzene control program, in addition
to the environmental programs which were already being modeled in the
base case.  For the Mathpro refinery modeling, which was conducted
before the nonroad diesel fuel program and MSAT2 benzene control
programs were finalized, those fuels control programs were not modeled
in the reference case.  Also, the implementation of EPAct required a
large increase in the amount of ethanol to be blended into gasoline to
comply with the renewable fuels standard (RFS), but not RFS2.  In its
AEO 2006, EIA projected that the volume of ethanol blended into gasoline
exceeded the RFS required amounts, resulting in 9.6 billion gallons of
ethanol blended into gasoline by 2012.  Other provisions of EPAct that
were modeled with both the Mathpro and Haverly models included a de
facto ban on MTBE and rescinding the RFG oxygenate requirement.  The
reference case unit throughputs and gasoline blendstock volumes were
used in the refinery-by-refinery cost model.  For the Haverly refinery
modeling work, in addition to the EPAct provisions, the RFS2 renewable
fuels volumes were modeled for 2017.  For the 2017 reference case, 17.8
billon gallons of ethanol were assumed to be blended into gasoline, and
3.9 billion gallons of renewable and cellulosic diesel fuel and
biodiesel were assumed to be blended into diesel fuel for the control
case.  For gasoline, the ethanol volume beyond the E10 blendwall was
assumed to be blended as E15.  For the 2030 reference case, we modeled
22.2 billion gallons of corn and cellulosic ethanol, and 8.3 billion
gallons of renewable diesel and biodiesel. 

The third step in conducting the LP refinery modeling was to run the
various control cases.  The control cases are created by applying a
specific fuel control standard to each PADD reference case.  To single
out a specific cost or other impact, the sole difference between the
control case and the reference case is the parameter change being
studied.  

For the Haverly modeling, a control case was run to model the octane
loss associated with desulfurization using 2017 as the year of analysis.
 Since we solely wanted to identify the cost of recovering lost octane
for the refinery-by-refinery modeling, this case was run by reducing the
octane value of the FCC naphtha by one octane number, and this was the
sole change relative to the reference case.  The control case was run
with capital costs evaluated at a 15 percent rate of return on
investment (ROI) after taxes.  The octane cost estimated by the LP cost
model is 0.76 cents per octane number per gallon of FCC naphtha. 
Because the octane loss associated with a specific technology may be
lower or higher than 1 octane number, we scaled the octane cost based on
the relative estimated octane loss on the FCC naphtha (i.e., a ½ octane
loss of the FCC naphtha was estimated to cost 0.38 cents per gallon of
FCC naphtha.    REF _Ref308540950 \h  Table 5-35  at the end of this
chapter summarizes the data output from the refinery modeling from which
we calculated the octane cost for using in the refinery-by refinery cost
model.  

It was necessary to estimate the gasoline qualities for estimating the
emissions impact of the proposed Tier 3 program.  This was conducted in
two separate steps.  First it was necessary to estimate the gasoline
qualities of the 2017 and 2030 reference cases relative to the gasoline
qualities of a revised base case.  The sole differences that we modeled
between the 2005 revised basecase and the 2017 and 2030 reference cases
was the phase out of MTBE and the addition of ethanol.  For the 2005
revised basecase we modeled 1.7 billion gallons of MTBE and 4.1 billion
gallons of ethanol.  For the 2017 and 2030 reference and control cases,
we modeled 17.8 and 22.2 billion gallons of ethanol, respectively.  In
2017, we estimated that approximately half the gasoline would be 10
percent ethanol and the about the other half would be 15 percent.  To
model the emissions impact of the different ethanol blends, we modeled
two reference cases, one with 100 percent E10 and the other with 100
percent E15.  These two ethanol cases were modeled in 2030 and we used
the results for 2017 as well.  The gasoline qualities for the reference
and two ethanol cases are summarized in   REF _Ref309051307 \h  Table
5-57  to   REF _Ref309051334 \h  Table 5-61  in the appendix at the end
of this chapter.  The changes in gasoline quality are summarized in  
REF _Ref309223669  Table 5-2 .  Because of the tendency for the LP
refinery model to shift gasoline blendstocks around resulting in odd
gasoline quality changes in individual PADDs, we solely used the
national average change in gasoline qualities and applied those changes
for all E10 or E15 gasoline for the emissions analysis.

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  2  Difference in
Gasoline Qualities between E10 and E15 Control Cases with the Reference
Case

The second step for estimating gasoline qualities was to model the
impact of desulfurization on gasoline qualities.  The total impact of
desulfurization on gasoline qualities is comprised of the reduction in
gasoline sulfur, the associated reduction in olefins and the impacts of
recovering the lost octane.  The sulfur reduction is fixed by the
standard and the olefins reduction is a function of the selectivity of
the desulfurization technologies.  We reviewed the information that we
had obtained for the gasoline desulfurization technologies and estimated
that desulfurizing gasoline from 30 ppm to 10 ppm would result in a 1
percent reduction in olefin level.  Since we estimated the cost of
making up lost octane using the LP refinery model, we used that case for
estimating the impact of octane recovery on gasoline qualities.  The
gasoline qualities for the reference case and the control case which
reflects a 1 octane number loss in the FCC naphtha pool are summarized
in   REF _Ref309051385 \h  Table 5-47  to   REF _Ref309051409 \h  Table
5-51  at the end of this chapter.  The difference in gasoline qualities
between the reference and control cases is summarized in   REF
_Ref309223718  Table 5-3 .  Because of the tendency for the LP refinery
model to shift gasoline blendstocks around resulting in odd gasoline
quality changes in individual PADDs, we solely used the national average
change in gasoline qualities and applied those changes for all gasoline
for the emissions analysis.  After we integrated the gasoline
desulfurization information into the refinery-by-refinery cost model, we
estimated that desulfurizing gasoline from 30 ppm down to 10 ppm would
result in about a one-half reduction in FCC naphtha octane ((R+M)/2)
number.  To estimate the changes in gasoline quality from a one-half
octane number loss in FCC naphtha that we estimated, we divided the
gasoline quality changes for one octane number in the FCC naphtha by a
factor of two resulting in half the changes in gasoline quality that we
estimated for a one octane number change in FCC naphtha.  The second set
of columns in   REF _Ref311636537 \h  Table 5-3  summarizes the gasoline
quality changes that we estimated for reducing the gasoline sulfur
levels from 30 to 10 ppm.  

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  3  Differences in
Gasoline Qualities Between the Control and Reference Cases

Summary of Refinery-by-Refinery Model Methodology

The purpose of the refinery-by-refinery cost model is to project how
each refinery would reduce the sulfur in its gasoline pool to 10 ppm or
lower and to estimate the cost for doing so.  To do this we created a
U.S. refining industry refinery-by-refinery spreadsheet cost model using
inputs from an LP refinery model case to allow us to better understand
the gasoline sulfur control costs to individual refineries.  This
spreadsheet cost model also allowed us to model how costs would be
affected by an ABT program.  

The building of the refinery-by-refinery model consisted of two major
steps.  The first step was to estimate baseline operating conditions for
each refinery.  This involves estimating the volumes and sulfur levels
of the gasoline blendstocks that comprise each refinery’s gasoline. 
We chose to use information from 2009 for modeling the baseline
operating conditions for the refineries as it’s the latest year we had
data for refiner operations and yields.  Additionally, EIA projections
indicated that gasoline demand is expected to be essentially flat
between 2009 and 2017, alleviating the need to adjust refinery operating
throughputs and yields for future changes in gasoline demand.  Because
of these factors, the 2009 gasoline production volumes and refinery
operating conditions can reasonably be projected to be at the same level
in 2017 (the first year of implementation of the Tier 3 program) in
estimating costs and refinery impacts.  As a final adjustment to our
estimated gasoline volumes and sulfur levels, we calibrated the model to
actual refinery gasoline volume and sulfur levels to ensure our
model’s accuracy.

To estimate the cost for each refiner to lower its gasoline pool down to
10 ppm, we used our refinery-by-refinery model to estimate the FCC
naphtha volume, the sulfur level of the FCC naphtha, and the amount of
sulfur reduction needed in FCC naphtha to meet a 10-ppm sulfur standard
at each refinery.  We also incorporated in our refinery-by-refinery
model the impacts that FCC pretreaters have on FCC naphtha yields and
sulfur levels, as well as the impact of refinery-specific crude oil
sulfur levels on FCC naphtha yields.  Similarly, we also used the
refinery-by-refinery cost model to estimate the volume levels of light
straight run naphtha (LSR) and natural gas liquids (NGL) that require
additional hydrotreating, as well as butane volumes that are directly
blended into the gasoline pool.

The second step involves applying the various sulfur control
technologies to each refinery as necessary to meet the 10-ppm sulfur
standard.  We expect that the majority of the sulfur reductions
necessary to comply with a 10-ppm gasoline sulfur standard will come
from reducing the sulfur level in their FCC naphtha.  Using our
refinery-by-refinery model we also estimate that a few refineries will
add additional LSR/NGL hydrotreating capacity.  We also evaluate each
refiner’s cost to install new butane Merox extraction equipment to
lower the sulfur level of butane that is directly blended to the
gasoline pool. Reducing the sulfur content of butane was assumed
necessary to meet a 5-ppm sulfur standard for our ABT cases.  This
assumption is conservative as many refiners may already have this
equipment, or may purchase low-sulfur butanes that have already been
treated by their supplier.

This allows us to generate a cost estimate for the sulfur control
technology in each refinery.  The capital costs for installing the
sulfur control technologies in each refinery were evaluated based on a 7
percent return on investment (ROI) before taxes.  In the following
sections, we present the various steps that were used in this
refinery-by-refinery modeling analysis.

Estimating Individual Refinery Gasoline Blendstock Volumes

In order to develop a baseline for our refinery-by-refinery analysis, it
was necessary to understand the sulfur levels and volumes of the various
blendstocks which make up each refinery’s gasoline.  Each refinery
blends up its gasoline pool from the various gasoline blendstocks that
are produced from the refinery units installed at each refinery. 
However, information on the volumes and sulfur levels of each gasoline
blendstock produced by each refinery is not publicly available, so it
was necessary to estimate them.  Estimating each refinery’s gasoline
blendstock volumes was accomplished using actual 2009 refinery specific
throughput rates that we obtained from EIA for crude, FCC, cokers and
hydrocracking units, and  published refinery unit capacity information
for the other refinery units.  We used this information to estimate the
extent that each refinery process unit is utilized, followed by a
unit-specific analysis for estimating how each refinery unit produces
material for blending into gasoline.  After the unit-by-unit estimates
are completed, we performed an overall check by comparing our estimated
gasoline volumes with reported gasoline volumes for each refinery, using
EPA’s RFG database

The model requires the total gasoline volume and each gasoline
blendstock volume for each refinery as an input.  Although the model
does estimate this volume of gasoline produced by each refinery based on
the estimated volumes of each gasoline blendstock, we chose to use
actual 2009 gasoline production data reported by refiners as for the
total gasoline volume for each refinery in our cost calculations. To
comply with the RFG program, refiners report gasoline production volumes
and sulfur levels for reformulated and conventional gasoline to EPA.  We
used this data and imputed each refiner’s 2009 total gasoline
production and corresponding sulfur levels into our model.

 In the end, our completed refinery-by-refinery modeling estimates of
gasoline produced on a national basis, correlated very well with the
actual refinery production volume in 2009, with our estimated volumes
having an overall error of approximately 0.5 percent relative to the
reported refinery production volumes.  In order to minimize the impact
of this error, we forced the estimated total refinery gate gasoline
volume to match actual reported 2009 gasoline production volume across
all the refineries.  The unadjusted refinery-by-refinery estimates of
FCC and LSR gasoline volumes, however, are used for estimating necessary
equipment modifications and costs for sulfur removal.  This is due to
the fact that the refinery by refinery models estimates for each
refiner’s FCC gasoline are likely to be very close to actual FCC
production, as we use actual refinery specific FCC charge throughput
rates and account for the effects of FCC feed pretreating on FCC
gasoline yields.  Additionally, the cost for treating FCC gasoline in
our Tier 3 programs, comprise over 85 percent of the total costs, while
LSR comprises the bulk of the remaining costs.   

Principal Refinery Unit Volumes

To estimate the production volumes for each of the refinery’s gasoline
blendstocks, the refinery-by-refinery model needs process capacity
information.  The Oil and Gas Journal (OGJ) publishes and the EIA
reports unit capacities for the principal refinery units for each
refinery in the U.S.,  We updated our database from these two sources to
reflect capacity that was in place in 2009, the base year for the model.
 Where differences between the two databases existed, we used the
information that was judged best overall from the two sources and
entered it into the refinery-by-refinery cost model.  These unit
capacities indicate the maximum throughput rate for each individual
unit, not the actual unit throughput rates for each facility, as this is
proprietary business information and not publicly available.  In order
to enhance our model, we obtained from EIA the actual 2009 annual unit
throughput rates for each refiner’s crude and major refinery units
(FCC, cokers and hydrocracker units).  With this information, the
refinery-by-refinery model was fine-tuned to reflect each refineries
gasoline blendstocks yields.  Our use of this information significantly
improved our model’s ability to estimate FCC naphtha, as well as other
gasoline blendstocks that each refinery makes.  The FCC, coker and
hydrocracker unit’s throughputs versus actual capacity that we
obtained from EIA for each domestic refinery on a PADD average basis are
listed in   REF _Ref307316492  Table 5-4 .  This information is
presented on a PADD average to protect CBI.

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  4  Process
Capacity Utilizationa

	Crude Throughput	FCC Throughput	Coker Throughput	Hydrocracker
Throughput

Total U.S.	0.843	0.840	0.761	0.768

PADD 1b	0.779	0.754	0.643	0.707

PADD 2	0.859	0.814	0.810	0.774

PADD 3	0.858	0.880	0.782	0.637

PADDs 4/5 excluding California	0.817	0.794	0.824	0.903

aActual unit throughput rates as a fraction of maximum unit capacity on
a PADD basis

bPADD 1 data includes Hovensa, VI

In the model, we also adjusted the refinery capacity information to
account for refinery expansions or refinery shutdowns that we were aware
of and are scheduled to occur over the next several years.  Refinery
expansions include those announced for WRB Refinery in Wood River
Illinois, the Valero Refinery in Norco Louisiana, and the Marathon
Refinery in Garyville, Louisiana.  For these expansions, there is
limited public data on which of the specific process unit capacities
would be increased, though each expansion project has information on the
crude unit capacity increase.  Since the data was limited, we increased
all of the existing individual process unit capacities by the fractional
increase in crude oil unit capacity at each of the expanding refineries.
 Refiners that we believe are permanently shutdown in PADD 1 were
removed from our analysis but, consistent with recent import/export
trends, we allowed PADD 3 to supply any lost capacity to PADD 1 as a
result of this lost production.  PADD 1 refiners that were presumed to
be permanently shutdown are; Giant refinery located in Yorktown,
Virginia,  Sunoco refinery in Westville, New Jersey, and Shell Oil
refinery in Bakersfield, California.

Other Refinery Unit Volumes

The next step was to calculate actual unit throughput rates for the
other refinery processes that produce gasoline blendstocks. These units
include alkylation, dimerization, polymerization, isomerization, naphtha
reforming.  All of these processes feedstocks are primarily supplied by
the crude and FCC Units.  Since this data is similarly not publicly
available we tuned these units to the EIA throughputs rates for crude
and FCC units at each facility, with alkylation units running at the
same throughput rates as the FCC and the remaining units running at the
crude oil throughput ratesThe results of the capacity utilizations of
these downstream units are summarized in   REF _Ref307316582  Table 5-5 
below.

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  5  Other Unit
Process Capacity Utilizationa

	Reformer Throughput	Alkylation Throughput	Isomerization Throughput
Poly/Dimersol Throughput

PADD 1	0.774	0.886	0.931	1.000

PADD 2	0.859	0.878	0.859	0.859

PADD 3	0.858	0.880	0.858	0.345

PADDs 4/5 excluding California	0.817	0.794	0.714	0.100

a Actual unit throughput rates as a fraction of maximum unit capacity on
a PADD basis

With these inputs the refinery-by-refinery model now contained estimates
of the feedstock charge rates for all of the gasoline blendstock
producing units, though estimating refinery unit capacity and capacity
utilization may or may not translate directly into the gasoline
blendstock volume produced by a specific refinery unit.  This is because
some refinery units may also produce products other than gasoline
blendstock.  Additionally, some processes have volume loss of feedstock
due to process reactions and conversions that take place that increase
or decrease the density and therefore the volume of products.  To take
this into account, a gasoline fraction yield factor has to be applied to
each process to convert the process charge rate into the yield of
gasoline blendstocks.  The process fractional yields that were used in
our refinery by refinery model were taken from our MSAT2 final rule LP
refinery modeling work, which represented the U.S. refining industry on
a PADD basis.  The FCC unit process yields of naphtha blendstock are
different for units with an FCC feed pretreater, versus those without
feed pretreating.  In our modeling we accounted for this by adjusting
yields and sulfur levels of FCC units with pretreaters and those without
a pretreater.  The fractional yields of gasoline blendstock for the
major process units and the 2009 throughputs for each of these units
used in our model are summarized below in   REF _Ref307316673  Table 5-6
 and   REF _Ref312058732 \h  Table 5-7 .

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  6  Gasoline
Blendstock Fraction Yields Per Process Unit Charge

	Crude	FCC Units Average	Coker	Hydrocracker

PADD 1	0.190	0.560	0.234	0.369

PADD 2	0.211	0.570	0.234	0.311

PADD 3	0.188	0.554	0.239	0.212

PADDs 4/5 excluding California	0.183	0.565	0.234	0.276



Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  7  2009 Refinery
Unit Throughputs (1,000 BPSD)

	Crude	FCC Units	Coker	Hydrocracker

PADD 1	1,624	652	<3 a	<3 a

PADD 2	3,193	1,017	322	223

PADD 3	7,262	2,604	1,043	500

PADDs 4/5 excluding California	1,363	263	130	94

a  Since there are less than three refiners in this PADD with these
units, the data was not reported to protect CBI information.

The FCC unit produces significant volumes of naphtha, a gasoline
blendstock.  The conversion percentage to naphtha is affected by the
severity of the operation of the FCC unit.  As shown in   REF
_Ref307316673  Table 5-6  above, the portion of FCC feedstock converted
to naphtha ranged from 55 to 57 percent across the various PADDs.  The
range among individual refineries can be quite large, but we didn’t
have access to refinery specific data for this.  However, as a group
there is expected to be differences between refineries with and without
FCC pretreaters.  Therefore, rather than simply use the PADD average
conversion of FCC feedstock to naphtha for all refineries in a given
PADD, the refinery-by-refinery model differentiates between refineries
that have an FCC feedstock pretreater and those that do not.  We have
also quantified the gasoline blendstock fraction yield for FCC units
that have both feed pretreater and postreater units.

 Historically, refiners have installed FCC feed pretreaters for economic
reasons, as pretreaters increase FCC unit conversion to high value
gasoline blendstock  while decreasing the production of low value light
cycle oils and residual material from FCC units.  FCC feed pretreaters
also have the benefit of reducing sulfur from the FCC feedstocks,
resulting in the production of lower sulfur FCC naphtha and ultimately
lower sulfur gasoline.  In developing our refinery-by-refinery model, we
quantified the impact FCC feed pretreating and postreating has on FCC
naphtha yields and sulfur levels based on our evaluation of information
we received from technology vendors.  The results of this analysis are
shown in   REF _Ref307316805  Table 5-8  below.

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  8  FCC Unit
Gasoline Blendstock Fraction Yields

	Average of All FCC Units	FCC Units with No Pretreater	FCC Units with a
Pretreater Only	FCC Units with a Pretreater and Postreater

PADD 1	0.560	0.558	0.638	0.607

PADD 2	0.570	0.533	0.648	0.617

PADD 3	0.554	0.520	0.630	0.600

PADD 4/5 excluding California	0.565	0.548	0.642	0.612

Poly Gas and Alkylate

For the polymerization and alkylation units the capacity of the unit
coupled with its estimated utilization rates listed in   REF
_Ref307316582  Table 5-5  is sufficient to establish the volume of
gasoline blendstock produced by these units.  For example, a particular
refinery unit in PADD 1 might have a 10,000 barrel per day alkylation
unit.  If the alkylation units in PADD 1 are estimated to be operating
at 56 percent of its listed capacity in 2017, the alkylate production is
projected to be 5,600 barrels per day at that refinery.  Each of the
refineries within a given PADD was assumed to have the same utilization
rate for any alkylation units.

Light Straight Run Naphtha

The remaining gasoline blendstocks, including light straight run naphtha
(LSR), coker naphtha and hydrocrackate cannot be estimated simply using
the unit capacity and unit utilization rate.  In order to determine the
volume of gasoline blendstock produced by each of these units,
additional steps are required.  LSR naphtha is principally comprised of
five- and six-carbon hydrocarbons which come directly from crude oil. 
Thus the volume of LSR for each refinery was based on the volume of
crude oil processed by each refinery as determined in Section   REF
_Ref309737486 \r  5.1.3.1.1 , as well as the percentage of that crude
oil that is LSR.  The fraction of LSR in each refinery’s crude oil was
estimated on a PADD average basis using the LP refinery model since it
is not available on a refinery-by-refinery basis.  This percentage is
based on the types and quality of crude oil processed by all the
refineries in each PADD from our LP model.  LSR as a percentage of crude
oil is estimated to vary from 4 to 5 percent across the PADDs.  These
PADD level results are shown in   REF _Ref307316865  Table 5-9  below.

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  9  LSR as a
Percentage of Crude Oil by PADD

	PADD 1	PADD 2	PADD 3	PADDs 4/5a

LSR as a Percentage of Crude Oil	4.5%	5.0%	4.4%	4.4%

	a Excluding California

 After we calculated how much LSR is produced at each refinery we
determined how much of the LSR is used as a gasoline blendstock.  LSR
has several possible destinations that vary from refinery to refinery. 
For each refinery, with the exception of those located in PADD 2, a
portion of the LSR is designated to be sold into the petrochemicals
market where it is processed into other hydrocarbon compounds.  EIA
publishes the volume of naphtha which is sold into the petrochemicals
market in each PADD.  This information is summarized in   REF
_Ref307326393  Table 5-10  below.  Since this information is not
publicly available on a refinery-by-refinery basis, we assumed that the
volume of LSR naphtha sold into the petrochemicals market by each
refinery is proportional to the refinery’s percentage of the total
volume of crude oil processed in the PADD in which the refinery is
located.  After accounting for the volume of LSR naphtha sold to the
petrochemicals market, the balance of LSR naphtha is used as a feedstock
for the isomerization unit if the refinery has one.  If a refinery does
not have an isomerization unit, all of the LSR not sold to the
petrochemical market is assumed to be used as a gasoline blendstock. 
Any volume of LSR at a given refinery that exceeds the capacity of the
isomerization unit at the facility is also assumed to be used as a
gasoline blendstock.   However, if a refiner does not have enough naptha
hydrotreating capacity to process all of the refiner’s LSR volume, we
assumed that the refiner would use excess capacity in their FCC
postreater, to reduce the sulfur content of the LSR blendstock. 

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  10  Refiner Sales
of Naphtha in 2009 (1,000 BPSD)

	PADD 1	PADD 2	PADD 3	PADD 4/5a

Naphtha Sold to the Petrochemical Industry	12.2	22.4	161.2	0

Sales of Special Naphtha	0.8	0	31.4	0

a Excluding California data

For further clarity on gasoline blendstock yields from the model, the
gasoline blendstock volumes of LSR and naphtha from the naphtha splitter
overhead tower are adjusted to subtract sales of these blendstocks that
are sent to the petrochemicals market.   The values listed in   REF
_Ref310596107  Table 5-11  for LSR and naphtha splitter overhead are the
volumes sent to gasoline, as a fraction of crude throughput.  

 Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  11  PADD Average
Gasoline Blendstock Yields per Fraction of Crude Input

	PADD 1	PADD 2	PADD 3	PADD’s 4/5a

LSR to Gasoline	0.0309	0.0286	0.0187	0.0381

Naphtha from Light Naphtha Splitter Overhead	0.0021	0.0081	0.0072	0.0093

a Excluding California

In refineries with an isomerization unit, much of the LSR is processed
into isomerate, the product produced by the isomerization unit.  The
volume of isomerate produced is dependent on the volume of feedstock
processed by the isomerization unit up to its capacity.  As described
above, all of the LSR that is not assumed to be sold into the
petrochemical markets is assumed to be sent to the isomerization unit,
up to the maximum capacity of the isomerization unit.  The isomerization
unit produces a blendstock with a slightly higher energy density and
smaller volume compared to the feedstock volume.  To account for this
effect, the volume of isomerate produced is estimated to be 1.6 percent
less than the volume of LSR feedstock to the isomerization unit.
Hydrocrackate and Coker Naphtha

The hydrocracker and coker units also produce some light naphtha
material which is blended into gasoline.  Heavy naphtha is also produced
in these units, which is feed to the reformer, as discussed in the next
section.  The light naphtha material produced by the hydrocracker and
coker are termed light hydrocrackate and light coker naphtha,
respectively.  Based on LP refinery modeling work done for the MSAT2
rule we estimated that the portion of the feedstock processed by each of
these units converted to light coker naphtha and light hydrocrackate was
5 percent for coker units across all the PADDs, and ranges from 23 to 32
percent for hydrocracker units depending on the PADD. The light coker
naphtha is poor in quality and require hydrotreating to removes sulfur,
olefins and other impurities, before sending them to an isomerization
unit, if a refiner has one.    REF _Ref307316905  Table 5-12  below
summarizes the percentage of the feedstock to these units that is
converted to light naphtha and blended into gasoline.

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  12  Isomerization
Unit Feed Rates by PADD

	PADD 1	PADD 2	PADD3	PADD 4/5a

Light Coker Naphtha

(% of Coker Feed)	5.0%	5.0%	5.0%	5.0%

Light Hydrocrackate 

(% of Hydrocracker Feed)	28.7%	32.0%	23.3%	27.2%

a Excluding California

Reformate

The volume of reformate produced by the reformer was estimated based on
the volume of feed to the reformer as limited by each unit’s capacity.
 The feed to the reformer comes from various sources depending on the
refinery configuration.  For virtually all refineries, the heavy part of
the straight run naphtha from the atmospheric crude tower is sent to the
reformer, while the light naphtha is generally processed in the
isomerization unit or blended directly into gasoline as discussed above.
 Those refineries with a hydrocracker or a coker will send the heavy
naphtha from these units to the reformer as well.  This reformate feed
naphtha contains the six, seven, eight and usually the nine carbon
compounds from these various sources.  In some cases, the six carbon
compounds are separated from the rest of the reformate feedstock to
reduce the benzene in the final reformate.  The volume of the feed to
the reformer is estimated based on a fractions of the material processed
in the atmospheric crude tower, hydrocracker and coker on a PADD by PADD
basis using information from the LP refinery model.  

The fraction of crude oil that is fed to the reformer from the
atmospheric crude tower ranges from about 13 to 16 percent of the crude
oil input depending on the PADD.  About 18 percent of the material
processed in the coker unit is estimated to end up as feedstock to the
reformer.  The percentage of the feedstock processed in the hydrocracker
that is fed to the reformer ranges from 30 to 50 percent depending on
the PADD in which the refinery is located.  The variance in the fraction
of hydrocracker material sent to the reformer is due to the significant
flexibility that the hydrocracker has for producing either gasoline or
diesel fuel.  In certain PADDs, such as PADD 4 and 5, there is a higher
relative demand for diesel fuel compared to gasoline so there is a lower
conversion to naphtha than in other PADDs.  The product from the
reformer experiences a volume decrease of about 18 percent relative to
the volume of feed, due to the conversion of straight chain and cyclical
hydrocarbons to energy dense aromatics and other light products.  This
volume reduction and conversion to lighter products increases with the
severity and thus the conversion of the reformer unit.  All the
refineries in each PADD are assumed to be operating their reformers at
the same severity as estimated by the LP refinery model.  Each of the
values discussed in this paragraph are shown on a PADD by PADD basis in 
 REF _Ref307316967  Table 5-13  below.

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  13  Reformer Feed
Rates and Volume Loss

	PADD 1	PADD 2	PADD3	PADD 4/5a

Medium/Heavy Straight Run Naphtha (% of Crude Input)	13.8%	16.2%	14.0%
13.6%

Medium/Heavy Coker Naphtha 

(% of Coker Feed)	18.4%	18.4%	18.4%	18.4%

Medium/Heavy Hydrocrackate 

(% of Hydrocracker Feed)	35.4%	43.4%	50.2%	33.3%

Volume Loss in Reformer	18%	17%	18%	19%

a Excluding California

Purchased Blendstocks

Some gasoline blendstocks are purchased and blended into gasoline.  The
gasoline blendstocks typically purchased include natural gasoline,
alkylate, isooctene and ethanol.  We did not have information on the
volume of these gasoline blendstocks purchased and blended into gasoline
by each refinery, so we again relied on the information from EIA, which
reports the consumption of these blendstocks on a PADD basis.  The EIA
information on the amount of pentane plus, naphtha’s and NGLs
purchased in each PADD are listed in   REF _Ref312134122 \h  Table 5-14 
below.  Our RFG database has each refiners amount of ethanol blended
into RFG, but does not contain the amount of ethanol that is splash
blended into CG at terminals.  We accounted for ethanol blended into CG,
as well as the purchase of other gasoline blendstocks, by assuming that
each refinery purchased a volume of any given gasoline blendstock
purchased within their respective PADD proportional to that refinery’s
crude oil consumption within the PADD.  In the 2009 RFG database, the
ethanol volumes only averaged 2.7 percent of refiner’s gasoline
production, which results in an over estimation of our refinery and
program costs in this NPRM. In our NPRM analysis, we did not include any
desulfurization costs for Pentane plus and naphtha and lighter
blendstocks, since we did not know the extent that they were being
treated today.  However, the addition of these blendstocks, results in
very negligible increases in demand for additional naphtha
hydrotreating.  We will evaluate whether we need to include any costs
for treating these streams in the final rule analysis.  If there are
costs, because the streams are so small, the costs would be negligible. 


Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  14  Refiner
Purchases in 2009 (1,000 BPSD)c

	PADD 1	PADD 2	PADD 3	PADD 4/5a

Natural Gas Liquids 	17.5	115.9	272.4	15.5

Naphtha’s and lighter	24.6	39.3	53.7	3.8

Pentanes Plus 	0	41.8	94.9	25.7

Ethanol b	57.3	66.9	77.3	0.40

Notes:

a Excluding California

b Ethanol from EPA RFG database, excluding volumes that are splash
blended into conventional gasoline

c Natural Gas Liquids and Pentanes Plus are different names for the same
hydrocarbon stream and we inadvertently found two different volumes for
the same hydrocarbon stream and added them both as inputs into our
refinery cost model.  We will correct this in the final rulemaking
analysis.Butane Volumes 

To estimate the butane volumes in our refinery-by-refinery model we used
an RVP balance equation.  This equation states that the product of the
overall RVP and volume of the gasoline pool is equal to the sum of the
product of the RVP and volumes of the non-butane components plus the
product of the RVP and the volume of the butane blendstocks.  This
equation can be rearranged to solve for the volume of butane blendstocks
as shown in   REF _Ref349803662  \* MERGEFORMAT  Equation 5-1  below.  

Equation   STYLEREF 1 \s  5 -  SEQ Equation \* ARABIC \s 1  1  RVP
Butane Balance Equation

Butane = (A*D-B*D)/(C-A)

Where:

Butane = Volume of Butane added in each refinery in BPSD

A = Blended gasoline RVP average

B = Non-butane blendstock RVP average

C = Butane RVP

D = Volume of gasoline produced

The gasoline production volumes and RVP of the blended gasoline are
reported to EPA by refiners for each refinery and were used for the A
and D terms in   REF _Ref349803662  Equation 5-1 .  To calculate the RVP
of the butane used as gasoline blendstock we first had to consider the
relative proportion of isobutane versus n-butane being used as a
gasoline blendstock as their RVP values differ.  This ratio was
estimated on a PADD by PADD basis from the LP modeling work.  We then
used a volume weighted average to calculate the RVP of the mixed butane
stream blended into gasoline in each PADD.  The information for these
calculations is shown in   REF _Ref349803738  Table 5-15  below.  The
non-butane blendstock RVP was estimated by multiplying each individual
gasoline blendstock RVP times the gasoline blendstocks volume fraction
of each refineries gasoline pool (CG and RFG) using 2009 ethanol volumes
and taking the sum of all of these values.  The RVP value for each of
these streams is shown in   REF _Ref349803760  Table 5-16  below.  With
this information we were then able to estimate the volume of butane
added to the gasoline blendstock at each refinery.  The annual volumes
of butane added by refineries on a PADD level are listed in   REF
_Ref349803767  Table 5-17 .  The volume of butane blended into gasoline
at each individual refinery varies based on the annual average gasoline
RVP that the refinery produces ( the RVP of CG and RFG gasoline are
volume weighted together), as well the variance in gasoline blendstock
streams that a particular refinery uses to produce CG and RFG gasoline.

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  15  PADD Average
Composition of Mixed Butanes Added to Gasoline

	PADD 1	PADD 2	PADD 3	PADD’s 4/5a

Isobutane %	96%	32%	53%	66%

N-butane %	4%	68%	47%	34%

Mixed Butane RVP (C), psi	71.376	58.192	62.518	65.196

a Excluding California

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  16  PADD Average
RVP’s of Gasoline Blendstocks

	PADD 1	PADD 2	PADD 3	PADD’s 4/5a

LSR	12.0	12.0	12.0	12.0

Naphtha from Light Naphtha Splitter Overhead	3.0	3.0	3.0	3.0

Reformate	4.5	6.6	5.0	6.2

FCC Naphtha	4.6	4.6	4.6	4.6

Coker Naphtha	13.0	13.0	13.0	13.0

Isomerate C5	13.0	13.0	13.0	13.0

Isomerate C6	7.2	7.2	7.2	7.2

Natural Gasoline (NGL)	12.6	12.6	12.6	12.6

Polymerization Gasoline	2.8	2.8	2.8	2.8

Light Hydrocrackate	9.2	9.2	9.2	9.2

Alkylate, C3	3.6	3.6	3.6	3.6

Alkylate, C4	3.2	3.2	3.2	3.2

Dimersol	5.8	5.8	5.8	5.8

Ethanol	10.7	10.7	10.7	10.7

a Excluding California data

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  17  PADD Average
Gasoline Data

	PADD 1	PADD 2	PADD 3	PADD’s 4/5a

Non-butane blendstock RVP (B), psi	6.4	6.3	5.7	6.3

Gasoline Pool Volume (D), BPSD	738.2	1744	3487.8	500.3

Volume Butane Added, BPSD	21.4	95.1	143.1	25.7

Blended Gasoline RVP average (A), psi	8.5	9	8	9.1

a Excluding California data

Calibrating the Blendstock Volumes in the Refinery-By-Refinery Model

After calculating gasoline volume estimates for each refinery in the
refinery-by-refinery cost model, we calibrated these values against
their reported gasoline blendstock volumes.  Refiners report their
production volumes for both conventional and reformulated gasoline to
EPA to comply with the gasoline reporting requirements.  We used these
reported volumes from 2009, along with LP modeling results from our
MSAT2 Rule, to calibrate the refinery-by-refinery model.  Before making
any adjustments, the refinery-by-refinery modeling estimates for
gasoline produced on a national basis correlated very well with the
reported refinery production volume in 2009, with volumes differing by
less than 0.5 percent from actual production.  In order to eliminate
this discrepancy we modified each refiner’s yields in our
refinery-by-refinery analysis based on the 2009 data as reported by the
refineries.  The volume of each of the gasoline blendstocks, excluding
the light straight run (LSR) and FCC gasoline streams, were increased or
decreased proportionally in order to align the aggregated national
finished gasoline production volumes in our refinery-by-refinery model
with the aggregated national finished gasoline production volume
reported by US refineries in 2009.

In making adjustments to the refinery-by-refinery analysis to better
align its volumes with the reported gasoline volumes and sulfur levels,
we did not make any changes to the production volumes of LSR or FCC
gasoline.  The LSR and FCC gasoline volumes are left unchanged as these
volumes are based on actual refinery-specific FCC charge throughput
rates.  This accounts for the effects of FCC feed pretreating on FCC
gasoline yields and is therefore likely to accurately reflect the
production volumes for each individual refinery.  Additionally, these
volumes are of central importance to our analysis as they are used for
estimating the equipment modifications necessary for complying with new
Tier 3 sulfur standards and the costs associated with additional sulfur
removal.  

With these calibrated volumes for each of the gasoline blendstocks, the
refinery-by-refinery model can now be used to estimate the sulfur level
that refiners must achieve in the FCC naphtha to meet the current sulfur
limit under the Tier 2 standards.  These volumes also allow us to model
what modifications to existing equipment and refinery operations will be
required to comply with the new Tier 3 sulfur standards

Refinery Blendstock Sulfur Levels

After determining the volume of each gasoline blendstock stream, we next
estimated the sulfur level of each of the gasoline blendstocks for our
modeling analysis using information we collected from literature reviews
and discussions with refinery consultants and technology providers.  We
also considered the blendstock sulfur levels estimated for the MSAT2
rule and the estimates derived from our refinery-by-refinery model to
estimate the sulfur levels of the blendstock streams.  Establishing
these sulfur levels is important as this sets a baseline for the
refinery-by-refinery model that represents our estimate for the current
operations of each refinery.  This allows us to project what changes
refiners would have to make in their refineries to comply with the Tier
3 standards, and project the new investments and operating costs
associated with these changes.  The following section contains further
details on how the sulfur content of each of the blendstocks was
estimated.  The results of this analysis can be found in   REF
_Ref307490687  Table 5-18  at the end of this section.

 The first stream we considered was the butanes that are used as a
gasoline blendstock.  The butanes used as gasoline blendstock within a
refinery come from a variety of sources.  Much of the butane used as a
gasoline blendstock is distilled from the crude oil or other blendstock
streams within the refinery.  Refiners remove the butanes from crude oil
and sometimes gasoline blendstocks which contain some butane (i.e., FCC
naphtha, hydrocrackate) and then blend them back into the gasoline up to
the RVP or vapor/liquid limit applicable to the gasoline market that the
gasoline is being sold into.  During the summer months refiners usually
have excess butane which cannot be blended into the gasoline pool
because of the tighter RVP standards.  Many refiners store the excess
butanes and then blend them back into gasoline in the winter months when
the volatility limits for gasoline are less stringent.  Other sources of
butanes used as gasoline blendstocks are natural gas processers and
crude oil drilling operations.  The butanes from these sources are
produced in downstream units which separate the various hydrocarbon
components.  Most of these downstream units “sweeten” the butanes
using a Merox unit prior to shipping them in pipelines or selling them
directly to refiners.  The sweetening process reacts the hydrocarbon
mercaptan compounds to disulfide compounds reducing their odor and
corrosivity.  The sweetening process, however, does not lower the sulfur
level.  If the source natural gas well is very high in sulfur, the
operator may need to use an extractive Merox treatment technology which
actually removes the sulfur from the butane stream.  This treatment
generally lowers the sulfur level of the butanes to under  5ppm. 
Butanes that are blended into gasoline have a sulfur limit of 30 ppm and
those that are shipped through pipelines, regardless of their end use,
have a limit of 140 ppm.  Furthermore, many refiners have Merox units on
site that are capable of removing sulfur from butanes that are either
purchased or generated internally from refinery units.  We were,
however, unable to evaluate existing butane Merox treating capacity at
NGL processers, crude drilling operations, or in refineries as there was
no information available in the OGJ, from EIA, or other publically
available sources.  Because we do not know the prevalence of these
units, we conservatively assumed in our baseline case that refiners are
adding treated butanes with a sulfur content of 10 ppm to their gasoline
pool.  

For hydrocrackate, dimersol, and poly gas blendstock streams, we used
the same sulfur levels that we estimatd for our MSAT2 rulemaking. The
sulfur levels for these streams are inherently low due to the dynamics
of process reactions in the hydrocracker, dimersol and polymerization
units.  Furthermore, it is unlikely that refiners have altered these
processes in their refineries since our analysis for the MSAT2 rule was
completed.

Alkylate blendstocks usually have a small amount of sulfur, usually less
than 15 ppm.  The primary source of sulfur in alkylate is the sulfuric
acid that is used as a catalyst in the alkylation process.  Finished
product coalescers and knockout drums are used by refiners to remove
impurities, including sulfuric compounds, from the alkylate product as
it leaves the alkylation unit.  This separation is imperfect, and a
small quantity of the sulfuric compounds which remain in the alkylate
account for the majority of its sulfur content.  Prior to the enactment
of the Tier 2 standards, the alkylate produced by most refineries
contained 10 to 15 ppm sulfur which assumes that there was some
carryover of sulfuric compounds into the alkylate.  Based on our
discussions with gasoline desulfurization technology vendors, however,
refiners have installed new acid coalescers and knock out drums in
recent years.  These new units improved the removal of residual sulfuric
compounds and can produce an alkylate blendstock with a 5-ppm sulfur
level.  This adjustment by refiners seems to be a low cost method for
reducing the sulfur content of alkylate.  For our refinery-by-refinery
baseline analysis, we assumed that refiners have already installed
improved acid knockout drums and are currently producing a 5-ppm
alkylate.  We also assumed that Hydrofluoric Acid (HF) alkylation
processes had the same alkylate yield per feedstock throughput as a
sulfuric acid alkylation unit in our refinery by refinery model.  We
assumed that the sulfur level of alkylate from an HF units also averages
5 ppm sulfur, even though HF processing units use hydrofluoric acid as
the processing catalyst, instead of using sulfuric acid.  

The coker unit produces a gasoline blendstock with a significant amount
of sulfur.  These units convert the heavy portion of crude oil, called
residuals, into gasoline and diesel blendstocks through the use of heat
and pressure.  The gasoline blendstock produced by the coker can contain
more than 3,000 ppm sulfur.  This stream is normally split into two
different streams.  The stream which contains the six to nine carbon
hydrocarbons is processed in the naphtha hydrotreater, which reduces the
sulfur level of this blendstock to below 1 ppm.  This stream is then
routed to the reformer for octane improvement.  The five and six carbon
hydrocarbon portion of coker naphtha is called light coker naphtha and
usually contains on the order of several hundred ppm sulfur.  Because of
the instability of this stream due to its high olefin content, it is
generally processed by the naphtha hydrotreater and sent to the
isomerization unit if the refinery has one.  After being processed in
the hydrotreater, the sulfur content of this stream is reduced to
approximately 1 ppm.  These treating pathways were assumed for each
refinery in the refinery-by-refinery baseline analysis.

Straight run naphtha is a gasoline blendstock which contains a moderate
amount of sulfur.  Straight run naphtha is the product stream from the
atmospheric crude oil tower with a boiling point that falls within the
boiling range of gasoline.  The heaviest portion of straight run naphtha
is higher in sulfur relative to the lighter portion of the straight run
naphtha.  The heavy portion of straight run naphtha is normally
processed by the reformer in order to improve its octane before being
blended into gasoline.  After this processing, the reformate has a
sulfur level of less than 1 ppm.  The light straight run naphtha (LSR)
contains the five and part of the six carbon hydrocarbons and has on the
order of 100 ppm sulfur before any hydrotreating.  LSR that is routed as
feedstock to isomerization units has its sulfur lowered to 1 ppm by
processing in the naphtha hydrotreater.  This hydrotreating is necessary
to allow this material to be processed in the isomerization unit, as the
catalysts in these units require low sulfur feedstocks to function
properly.  Some refiners, however, do not have isomerization units or
they produce LSR volumes that are greater than the capacity of their
isomerization units.  Even cases where there is insufficient capacity in
the isomerization units it is still desirable for refiners to hydrotreat
as much of the LSR as possible since it is more cost-effective to reduce
the sulfur content of the LSR than the FCC naphtha.  Refiners can either
hydrotreat this volume of LSR in the naphtha hydrotreaters or in FCC
naphtha postreaters.

Natural Gas Liquids (NGL) have a composition that is similar to LSR, as
it is comprised primarily of pentanes and hexanes.  NGLs are produced
from natural gas processers and crude oil drilling operations and the
sulfur content of the NGLs can vary depending on its source, although we
estimate that this stream averages about 100.  While some of the NGLs
are treated to remove sulfur by the NGL producers before being purchased
by the refineries we did not have sufficient information to be able to
determine the extent to which NGLs are treated before arriving at the
refinery.  For the baseline case in our refinery-by-refinery model we
assumed that NGL liquids are purchased with a sulfur content of 100 ppm
and hydrotreated based on capacity availability at refineries in a
similar manner as LSR.  Based on the gasoline blendstock volumes and
hydrotreating capacities as discussed in the previous sections, we
estimated in the baseline case for our refinery-by-refinery analysis
that refiners are hydrotreating 66 percent of the volumes of LSR and
NGLs produced and purchased for gasoline blendstock usage on a national
average basis.  Our hydrotreating capacity evaluation for each refinery
is discussed in more detail in Section   REF _Ref307490724 \r  5.1.3.4 .
 As a result of the proposed sulfur standards under the Tier 3 program,
we anticipate that refiners will revamp existing hydrotreaters and add
new hydrotreating capacity to allow them to hydrotreat all of their LSR
and NGL material.   

We also assumed that all ethanol blended into gasoline has a sulfur
content of 5 ppm.  Ethanol produced at ethanol plants naturally has a
negligible amount of sulfur.  Before being sold, however, a denaturant
is added to the ethanol.  This denaturant most commonly used is natural
gasoline, a C5 to C7 naphtha produced during natural gas processing. 
Natural gasoline has a sulfur content that ranges anywhere from a few
parts per million to a couple hundred parts per million sulfur.  We
assumed that the natural gasoline used as an ethanol denaturant is not
hydrotreated and has an average sulfur level of 250 ppm.  Ethanol
contains 2 percent denaturant, which results in denatured ethanol having
a sulfur level of 5 ppm.

After determining the sulfur level for each of the gasoline blendstock
streams as discussed above we can use this information, along with the
gasoline production volumes and sulfur levels for the United States in
2009, to determine the sulfur level of the FCC naphtha stream on a
national average basis.  To do this we used the following equation,
referred to as   REF _Ref308599170  Equation 5-2  hereafter:

FCC Naphtha Sulfur ppm = [(A*B) –
(C*D+E*F+G*H+I*J+K*L+M*N+O*P+Q*R+S*T)] / Z

Where:

A = Refinery Total Gasoline Yield, BPSD

B = Refinery Total Gasoline Sulfur level, ppm

C = Butane to Gasoline, BPSD

D = Butane Sulfur, ppm

E = Alkylate BPSD

F = Alkylate Sulfur, ppm

G= Reformate BPSD

H= Reformate Sulfur, ppm

I = Coker Naphtha, BPSD

J = Coker Naphtha Sulfur, ppm

K= Hydro-crackate BPSD

L= Hydro-crackate Sulfur, ppm

M= Light Straight Run (LSR) and Natural Gas Liquids (NGL), BPSD

N =LSR and NGL Sulfur, ppm

O= Dimersol, BPSD

P= DimersolSulfur, ppm

Q= Polymerization BPSD

R= Polymerization Sulfur, ppm

S=  Ethanol, BPSD

T = Ethanol Sulfur, ppm

Z= FCC Gasoline Yield, BPSD

Equation   STYLEREF 1 \s  5 -  SEQ Equation \* ARABIC \s 1  2 
Calculating FCC Naphtha Sulfur Content for Refinery-By-Refinery Model

We used this equation to assess two cases; a baseline case where the
30-ppm Tier 2 sulfur standards were fully implemented and a control case
that reflects the proposed 10-ppm Tier 3 sulfur standards.  The only
terms in   REF _Ref308599170  Equation 5-2  that change between the two
cases are the national average sulfur level and the sulfur levels of the
LSR, NGL, and FCC naphtha streams.  The national average sulfur levels
for the two cases were set at the sulfur limits under the Tier 2 and
Tier 3 programs -- 30 ppm and 10 ppm, respectively.  For the baseline
case we assumed that the sulfur level of the NGL and LSR streams was 34
ppm.  This reflects our assessment of how these streams are currently
being handled as discussed earlier in this section.  We estimate that 
66 percent of the volume of NGL and LSR are hydrotreated before being
blended into gasoline and have a very low sulfur content of
approximately 1 ppm.  The remaining 34 percent are untreated and have a
sulfur content of approximately 100 ppm.  For the Tier 3 control case we
assumed that all of the NGLs and LSR were hydrotreated and therefore had
an average sulfur content of 1 ppm.  This information allowed us to
solve   REF _Ref308599170  Equation 5-2  for the FCC naphtha content. 
The resulting FCC naphtha sulfur numbers, along with our estimation of
the gasoline blendstock sulfur levels and percent of total gasoline
volume made up by each blendstock are shown in   REF _Ref307490687 
Table 5-18  below.

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  18  Sulfur Levels
for Gasoline Blendstocks in the Refinery-By-Refinery Model

Gasoline Blendstocks	Baseline Tier 2 Case	Proposed Tier 3 Case Year 2017
Proposed Tier 3 Case Year 2030

	Percent of Total Volume	Sulfur Levels 30 ppm	Percent of Total Volume
Sulfur Levels 10 ppm	Percent of Total Volume	Sulfur Levels 10 ppm

FCC Naphtha	   37.2	80a	36.0	21a	35.0	21a

Reformate	22.5	0.5	21.8	0.5	21.2	0.5

Alkylate	12.7	5	12.5	5	12.1	5

Isomerate	3.2	0.5	3.1	0.5	3.1	0.5

Butane	4.0	10	4.0	10	3.8	10

Light Straight Run Naphtha (LSR) and Natural gas Liquids (NGL)	5.2	34
4.9	1	4.8	1

Hydrocrackate	3.0	8	2.9	8	2.8	8

Ethanol	9.9	5	12.5	5	15	5

Coker Naphtha	2.2	0.5	2.1	0.5	2.0	0.5

Other Gasoline Blendstocks	0.2

	10	0.2

	10	0.2	10

Total/Sulfur Average	100	30	100	10	100	10

a These values are calculated using   REF _Ref308599170  Equation 5-2 ;
all other sulfur levels are assumed

The numbers in the table above represent national averages.  While this
is useful information, it is insufficient for us to be able to model the
implications of the proposed Tier 3 standards for an individual
refinery.  Each refinery has a unique combination of processing units
that will determine the cost and operational changes necessary for that
refiner to comply with our proposed sulfur limit.  While each of these
processing units may impact the cost for refiners to lower the sulfur
content of the gasoline they produce we believe these costs will be
dominated by the units responsible for the desulfurization of the FCC
naphtha, and to a lesser extent the NGLs and LSR.  This is because these
are the only streams we anticipate would see significant sulfur
reduction under the proposed Tier 3 sulfur standards.  The units that
are used to desulfurize these streams include the FCC unit pre- and
postreaters and the naphtha hydrotreaters.  It is important, therefore,
to have a good understanding of which of these units are in place in
each refinery, as well as the type and capacity of these units, in order
to allow us to most accurately estimate the cost of the Tier 3 sulfur
standards to the refining industry.  We used the above FCC naphtha
sulfur balance information as the basis of our vendor request for
refiner modifications to FCC postreaters under Tier 3.  However, for the
vendor requests, we used a preliminary model, where the FCC naphtha
levels under Tier 2 averaged 75 ppm, while FCC naphtha levels under Tier
3 averaged 25 ppm for 10-ppm sulfur gasoline, representing a 50 ppm
sulfur reduction, close to the same delta presented in the table above. 
The following section discusses our assessment of the desulfurization
equipment currently being used in refineries.

Assessment of Refineries’ Existing Desulfurization Equipment

Since the desulfurization cost of the Tier 3 program is largely impacted
by the cost of lowering sulfur in FCC gasoline, it is important to
understand what refiners are already doing to lower the sulfur content
of the FCC gasoline blendstock to meet the Tier 2 sulfur standards. 
This was important to our analysis of the cost for each individual
refiner to reduce the sulfur content of their gasoline to meet the
proposed Tier 3 sulfur standard.  Refiners that already have an FCC
pretreater or postreater can revamp these units for a lower cost than
installing grass roots units.  It was also important to determine which
refineries have an FCC feed pretreater, since these units increase the
refineries FCC conversion and production of FCC naphtha and also lower
the sulfur level of the FCC naphtha.  To compile this information we
analyzed capacity information for FCC naphtha pretreaters and
postreaters for each refinery listed in the OGJ and the EIA database. 
If one of the databases showed that a refinery had FCC pretreating
and/or post-treating capacity, while the other did not, we assumed that
the refinery did have the units listed with a capacity as reported.

For refineries that have FCC naphtha postreaters we next determined
which vendor’s FCC naphtha desulfurization technology is installed in
each refinery.  To do this we conducted a public database search using
OGJ, company web postings and, other refinery publications.  To
supplement this data we also had extensive discussions with many
refiners to obtain confidential data from many of them on type and
capacity of the desulfurization technology currently installed in their
refineries as well as how their operations might be adjusted to meet the
new Tier 3 sulfur standards.  The various FCC naphtha desulfurization
technologies that we identified as currently being used by refiners are
CD Tech’s Cd Hydro and CDHDS, Axens Prime G and Prime G+, UOP’s ISAL
and Selectfining, Exxon’s Scanfining I or II and Sinopec’s S-Zorb. 
For refiners that we could not find or obtain information on the type of
desulfurization they were using, Axens was chosen as the default as they
have the largest market share of desulfurization units in the U.S.  To
confirm the accuracy of our work we reviewed our assessments with one of
the main technology vendors.  Our desulfurization technology selection
assumptions were adjusted based on feedback from the vendor.  The
aggregated results of this assessment are summarized in   REF
_Ref307497080  Table 5-19  below.

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  19  Postreater
Technologies Used By Refineries

	CD Tech	IFP/Axens	Scanfining	UOP ISAL	S-Zorb

Refiners with Existing Postreater	15	40	16	1	4

The next step of our analysis was to determine which refineries use FCC
feed hydrotreating technology (pretreaters) in addition to post-treating
units.  FCC feed hydroteating was primarily installed at refineries not
as a sulfur control technology, but because of the economic benefits
that can be obtained from hydro-treating FCC feed.  Hydrotreating the
FCC feed increases the crackability of this stream by saturating the
components with hydrogen resulting in a higher paraffin content in the
feed stream.  Hydrotreating also removes FCC feed impurities such as
nitrogen, metals, con-carbon and sulfur, which improve FCC unit catalyst
effects.  An additional benefit of FCC feed pretreating is that it
reduces the sulfur content of the FCC feedstock by 70 to 90 percent,
resulting in the production of FCC naphtha with lower sulfur levels than
what would be produced using FCC feed that is not hydrotreated.  

Our analysis indicates that approximately 53 refiners are currently
using FCC feed pretreaters.  Of the 53 refineries with pretreaters, 35
of also have FCC postreaters installed to comply with the Tier 2
gasoline sulfur standard.  The technologies used by these 35 refineries
are shown in   REF _Ref309115176  Table 5-20  below.  FCC naphtha
produced using only an FCC pretreater operating at standard severity
generally produces a gasoline with a sulfur content that exceeds the
Tier 2 standards.  According to information from vendors, the average
FCC naphtha sulfur level of refineries with an FCC feed pretreater
operating at standard conditions without a postreater ranges from 200 to
500 ppm.  Further reductions in the sulfur level of the FCC naphtha are
possible using only an FCC pretreater by operating the pretreater at a
higher severity or higher pressure (if the unit is designed to do so). 
These high pressure FCC pretreating units were designed to be able to
run at a high severity to further increase the crackability of the FCC
feed and therefore increase the conversion rate of the FCC unit.  These
more severe conditions also further reduce the sulfur level of the FCC
naphtha.  The naphtha produced from these units operating with high
severity or high pressure has an average sulfur content ranging from 75
to 100 ppm, allowing these refineries to produce gasoline that meets the
Tier 2 sulfur standards.  Operating FCC pretreaters at the high
severities necessary to meet the Tier 2 standards, however, also results
in increased operating cost, as the pretreater requires more frequent
catalyst changeouts, consumes more hydrogen, and operates higher
temperatures than pretreaters operating under standard conditions.

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  20  Technologies
Used By Refiners with FCC Pre and Postreaters

	CD Tech	IFP/Axens	Scanfining	UOP ISAL	S-Zorb

Refiners with FCC Pretreater and Naphtha Postreater	9	20	5	0	1

Our analysis also showed that there are several refineries that have an
FCC unit but have installed neither an FCC naphtha postreater nor an FCC
feed pretreater.  These are small refineries, or refineries that produce
a refinery gate gasoline with a sulfur level below the Tier 2 cap of 80
ppm sulfur, but above the 30-ppm average.  These refiners are relying on
buying or sharing sulfur credits from other refineries that are
over-complying with Tier 2 and make gasoline with a sulfur level less
than 30 ppm.  

Finally, some refineries do not have an FCC unit and therefore have not
installed FCC postreaters to comply with the Tier 2 sulfur standards. 
These refiners primarily use reformate, alkylate, LSR, butanes, and
pentanes to make gasoline.  Since these blendstocks all have low sulfur
content this allows refiners to produce gasoline with a low enough
sulfur content to meet the Tier 2 sulfur standards.  

A summary of the number refineries which fall into differing categories
of how they are complying with Tier 2 is shown in   REF _Ref307499166 
Table 5-21  below.

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  21  Refinery FCC
Naphtha Desulfurization Unit Characterization

FCC Treatment Units Installed	Number of Refineries

No FCC Unit	14

FCC Unit, No Pretreater or Postreater	7

FCC Unit With Postreater Only	38

FCC Unit With Pretreater Only	17

FCC Unit With Pretreater and Postreater	35



Crude Oil and FCC Feed Sulfur Levels

After we had determined the desulfurization technology in place at each
refinery, we sought to calculate the sulfur content of the FCC
feedstock.  Knowing this is important as it allows us to determine how
far the sulfur level of the FCC naphtha, and ultimately the gasoline,
produced at any given refinery can be reduced using the units currently
in place at each refinery.  It also helps us understand the extent to
which the existing hydrotreaters are being taxed to comply with the Tier
2 gasoline sulfur standard.  Some refineries may have excess capacity in
their FCC naphtha pretreater or postreaters that would allow them to
produce gasoline that would meet the proposed Tier 3 standards without
having to revamp existing units or add grass roots units.  These
refineries will have much lower cost impacts than refineries that have
to make more significant capital investments.

The sulfur level of the FCC feedstock is primarily dependent on the
sulfur level of the crude oil being processed by the refinery and
whether or not the refinery has an FCC feed pretreater.  The first step,
therefore, in determining the sulfur level of the FCC feedstock was to
input the crude sulfur level for each refinery into our
refinery-by-refinery model.  For this, we obtained confidential business
information (CBI) from EIA on the annual average crude sulfur levels
that each refinery processed in 2009. This data, which is reported to
EIA for each refinery, was used as the baseline crude sulfur level in
our refinery-by-refinery analysis. Using this data, we then determined
what each refiners FCC feed charge rate sulfur level would be, using a
regression co-relation we built from data on crude sulfur levels and FCC
feedstock material, as discussed below.  We assumed that refineries
today are primarily processing heavy gas oils (HGO) and vacuum gas oils
(VGO) produced from each refinery’s crude and coker units.  To
determine the volume of feedstock processed by the FCC units we assumed
that after distillation HGO makes up 20.5 percent (by volume) of the
processed crude and VGO makes up an additional 15.6 percent of the
crude.  Together, these two streams comprise the FCC feed.  

The boiling point range that we assumed for VGO also contained some
residual material, representing FCC feed with residual content.  This
was done to reflect the imperfect distillation cuts in crude towers and
that some refiners use small amounts of residual material as FCC
feedstock.  The balance of the residual material, however, was excluded
from the feed to FCCs since this material makes a poor feedstock due to
its high aromatics, metals and concarbon content.  Each of these
materials negatively affects the FCC gasoline conversion yields.  Most
refiners today do not directly use residuals as feedstock to their FCC
units, but instead send them to be processed in coker units or use the
residual material for fuel oil and asphalt production.  The boiling
point ranges that we used for HGO and VGO are listed   REF _Ref307837456
 Table 5-22  below.

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  22  Boiling Ranges
of FCC Feedstocks

	TBP Initial	TBP Final

Heavy Gas Oil (HGO)	600°F	800°F

Vacuum Gas Oil (VGO)a	800°F	1,000°F

a Contains some residual material

For our FCC feed sulfur regression, we used data obtained various crude
oil assays that we obtained from Jacobs Engineering for work that Jacobs
conducted for us.    We used data from five specific crude types,
including West Texas intermediate ,Bonny LightSaudi Heavy, Alaskan North
Slope, and Mayan, and three blended crude assays.  The equation for this
regression, along with the estimated FCC feed sulfur contents for
various crude oils are shown in   REF _Ref312134039 \h  Equation 5-3 
and   REF _Ref307837875  Table 5-23  below.

Equation   STYLEREF 1 \s  5 -  SEQ Equation \* ARABIC \s 1  3  FCC Feed
Sulfur Content Based on Crude Sulfur Content

FCC Feed Sulfur Weight Percent = (Crude Sulfur Weight Percent)0.8 *
1.1858 + 0.0409

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  23  FCC Feed
Sulfur For Various Crude Sulfur Levels

Crude Sulfur Level (Weight %)	FCC Feed Sulfur Level (Weight %)

0.11%	0.24%

0.28%	0.47%

0.85%	1.08%

1.3%	1.50%

1.4%	1.59%

1.6%	1.77%

2.8%	2.74%

3.04%	2.93%

Impact of FCC Feed Pretreaters on FCC Feed Sulfur Levels

With FCC feed sulfur estimated, the next step in our analysis was to
consider the impact of pretreaters on the FCC feed sulfur levels for
refineries that have these units.  There are several factors that must
be considered to determine the impact of the pretreaters on the FCC
sulfur level, including the pressure at which the unit operates, the
severity at which it is run, and whether or not the FCC naphtha will be
postreated.  

To inform our understanding of how FCC pretreaters operate, we obtained
guidance from technology and catalyst providers.  From these discussions
we learned that the capability for FCC pretreaters to remove sulfur from
the gas oil feed varies significantly depending on the pressure at which
the unit operates.  FCC pretreaters can generally be subdivided into
high pressure units (1400 psi and above), medium pressure units (900 to
1,400 psi), and low pressure units (below 900 psi).  High pressure FCC
pretreaters are capable of removing about 90 percent of the sulfur
contained in the gas oil feedstock to the FCC unit, while low and medium
pressure units are capable of removing 65 to 80 percent of the feed
sulfur.  Information we received from the vendors also indicated that
refiners with both a pretreater and a postreater are producing FCC
naphtha that ranges from 200 to 450 ppm before being processed by the
postreater.  Having a postreater allows these refineries to not have to
operate their pretreaters at a high severity as the sulfur will further
be reduced to levels necessary to meet the applicable standards in the
post-treating units.  Refiners with only a pretreater are making lower
sulfur FCC naphtha in the 75 to 100 ppm range, according to vendor
estimates.

With this information we used our refinery-by-refinery model to estimate
the pretreater desulfurization rates required to get FCC naphtha sulfur
levels within the ranges specified.  We estimated that FCC units with a
pretreater and a naphtha postreater are operating their pretreaters at a
severity which results in a 76 percent desulfurization of the FCC feed
stream.  This number represents the national average.  While the actual
severity at which the pretreating units are run varies on a
refinery-by-refinery basis this average was used in our modeling for all
refineries with both pretreating and postreating units due to a lack of
refinery-specific information.  For FCC units with a feed pretreater but
no postreater we calculated the FCC naphtha sulfur level required by
refiner to make a refinery gate gasoline that meets the Tier 2 standard.
 To do this calculation we used the gasoline yields from our
refinery-by-refinery model along with the gasoline blendstock sulfur
levels discussed above and shown in   REF _Ref307490687  Table 5-18 . 
These calculations showed that refiners with FCC feed pretreating units,
but no postreaters, need to produce FCC naphtha that averages about 85
ppm on a national level.  This sulfur level corresponded to these
refiners operating their pretreaters at a severity that results in a
reduction of sulfur in the FCC feed stream of approximately 91-92
percent.  This number is close to the estimate we received from the
vendors for this category of refineries and therefore was used in our
refinery-by-refinery model to determine the FCC feed sulfur level for
refiners with pretreaters.

After we have calculated the sulfur level of the FCC feed we must then
take into consideration the impact the FCC unit itself has on the sulfur
level of the FCC naphtha.  We reviewed several literature sources, and
found that the FCC naphtha sulfur level can be accurately determined by
dividing the FCC feed sulfur level by 20 for refineries with an FCC feed
pretreating unit.  For refineries without an FCC feed pretreater, the
FCC naphtha sulfur levels can be calculated by dividing the desulfurized
FCC feed sulfur level by 10.  In these cases the effect of the FCC unit
itself on the sulfur level of the FCC naphtha is lower, as the FCC feed
has already been through a desulfurization process.  These factors, when
combined with the sulfur levels of the FCC streams as discussed above,
allow us to calculate the sulfur level of the FCC naphtha before any
post-treating operations on a refinery-by-refinery basis.  The results
of this analysis are summarized in   REF _Ref307922771  Table 5-24 
below.

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  24  FCC Naphtha
Sulfur Levels for Various Refinery Configurations

	PADD 1	PADD 2	PADD 3	PADD 4/5

No Pretreater or Postreater	N.A.a	-<3 b	606	784

Pretreater Only	N.A.a	73	74	82

Postreater Only	1348	1157	705	1036

Pretreater and Postreater	49	<3 b	278	230

a N.A. – not applicable, no units of this type in the PADD

b Since there are less than three refiners in this PADD with the
described configuration, the data was removed to protect potential CBI
concerns.

This information, along with the information described in previous
sections (e.g., gasoline blendstock volumes and sulfur levels,
desulfurization equipment currently in place at refineries, and crude
oil sulfur levels) allows us to conduct the best analysis for the
baseline case for our refinery-by-refinery model.  This baseline case
reflects what we believe the current operating conditions are at each
refinery, including any modifications they have made to meet the Tier 2
sulfur standards.  The next step of our analysis was to project what
further changes, either to operations, adding new equipment or revamping
existing units, each refiner would have to make to meet the proposed
Tier 3 standards.  After these changes are estimated, we can then
estimate the cost associated with each of these changes, and ultimately
the cost of the program.

Cost Inputs for the Sulfur Control Technologies

After we determined the sulfur levels of the gasoline blendstocks for
each refinery and the sulfur levels that these blendstocks would have to
achieve to meet the proposed Tier 3 sulfur standards, the next step in
our refinery-by-refinery analysis was to project the changes to refinery
units and operations each refinery would have to make in order to comply
with the proposed Tier 3 sulfur standard.  The first step refiners would
take for further gasoline sulfur control would be to desulfurize the
light straight run, natural gas liquids, and butane blendstocks.  The
costs to reduce the sulfur content of these streams is relatively low
and would therefore be the most cost effective way to further reduce the
sulfur content of the finished gasoline. We also projected in our
analysis for RFS2 that 50 percent of all gasoline produced by refiners
in 2017 would contain 15 percent ethanol.  In 2030 we projected in our
analysis for RFS 2 that almost all gasoline contains 15 percent ethanol.
 Because ethanol tends to be a relatively low sulfur blendstock (assumed
to be 5 ppm for our refinery-by-refinery model) increasing the amount of
ethanol in the gasoline pool lowers the overall sulfur content of the
gasoline.  Reducing the sulfur content of the LSR, NGL, and butane
streams and increasing the ethanol content, however, would be
insufficient to allow refiners to comply with the proposed Tier 3
standards.   Refineries with an FCC unit would still have to reduce the
sulfur content of their FCC naphtha blendstock in order to meet the Tier
3 standards.

For each refinery we considered two cases.  In the first case each
refinery had to meet the proposed Tier 3 gasoline sulfur standard of 10
ppm.  In order to meet this standard, as discussed in   REF
_Ref309028998 \r  0 , we determined that they would have to reduce the
sulfur level of their FCC naphtha stream to 25 ppm.  We also considered
a case where each refinery would reduce the sulfur level of their
gasoline to 5 ppm.  We assumed that to meet a 5-ppm sulfur standard,
refiners would desulfurize the butane blendstock to 10ppm in addition to
further lowering sulfur in the FCC naphtha to 10 ppm.  This information
was used to help us determine which refineries might reduce the sulfur
level of their gasoline below our proposed 10-ppm standard to earn
credits for our ABT scenarios discussed in Sections   REF _Ref308083693
\r  5.2.1.2  and   REF _Ref308083698 \r  5.2.1.3 .

Our refinery-by-refinery model assumed that reducing the sulfur content
of the FCC naphtha to 25 ppm and 10 ppm for the two cases discussed
above would require that each refinery that produces FCC naphtha have an
FCC naphtha postreater.  In our final rule we plan to investigate the
ability of refineries with high pressure pretreaters to meet the
required standards without the addition of postreaters.  For companies
that already have an FCC naphtha postreater we assumed that all that
would be necessary to meet the proposed sulfur standards was to revamp
their existing FCC postreating units.  We received cost information from
several vendors for revamping FCC postreating units and assumed a revamp
cost for each refinery in line with the cost projections quoted by the
vendor of the technology already in place in their refinery.  We assumed
that refineries with FCC units that currently do not have an FCC
postreater would have no choice other than to add a new grass roots FCC
postreating unit.  We ultimately only received cost information from one
vendor for the cost of adding a new grass roots FCC postreating unit
that fit the sulfur reduction requirements of the proposed Tier 3
program.  We therefore assumed the cost for each refinery that would
need to add a new FCC post-treating unit would be in line with this
estimate.  More details on the costs used in our refinery-by-refinery
model for the desulfurization of LSR, NGL, and butanes as well as the
new FCC post-treating units and revamps can be found in the following
sections.

FCC Naphtha Desulfurization Costs

To estimate the cost for revamping existing FCC postreating units or for
adding new postreating capacity, we contacted several technology vendors
for cost estimates and reviewed literature, including cost information
provided for the Tier 2 rulemaking.  Because no two refineries are
exactly the same, the cost for new FCC postreater units or revamps to
existing units will vary significantly from refinery to refinery.  Some
of the factors that have the most significant impact on the cost of FCC
postreaters are the technology that the refiner used to comply with Tier
2, the volume of FCC naphtha, the sulfur content of the FCC unit feed
and the level of desulfurization in the existing postreater, and the
location of the refinery.  Based on feedback from vendors we considered
three categories of FCC postreaters based on whether the FCC naphtha
(the feed for the existing Tier 2 postreater) contained low (0 – 400
ppm) medium (400 – 1,200 ppm) or high (>1,200 ppm) levels of sulfur.  
Specific cost factors applicable to estimating unit revamps or grass
units were also taken into account as described below.  The following
sections discuss in greater detail how the cost estimates we received
from the vendors were used in our refinery-by-refinery analysis.

Cost to Revamp Existing FCC Naphtha Postreaters

We obtained information from several technology providers for the revamp
costs of existing FCC postreaters.  One of the estimates submitted was
deemed not credible as inadequate information was provided and the
capital and operating costs were extraordinarily high relative to the
rest of the cost information we received.  We believe it is likely that
this cost estimate represented not only a grass roots FCC postreater,
but also significant refinery investment in other refinery processes
such as FCC feed pretreating, coker unit expansion etc.  As these costs
did not seem to be a reasonable representation of the revamp costs for
FCC post treating, this estimate was not considered in our FCC
postreater cost analysis.  Some of the submitted information only had
cost information for our medium sulfur (400 – 1,200 ppm) FCC feed
case.  Another technology provider did not provide cost estimates for
producing an FCC naphtha with a sulfur level of 25 ppm, corresponding to
a finished gasoline with 10 ppm sulfur, therefore, we needed to
interpolate their cost information.  Because the cost information
provided by the technology providers was labeled CBI, this cost
information cannot be listed individually, however we aggregated  the
cost information we received for FCC postreater revamps to meet 10 ppm
and 5 ppm sulfur levels in gasoline.  The aggregated information is
summarized in   REF _Ref311717664 \h  Table 5-26  and   REF
_Ref311717674 \h  Table 5-27 .

One of the vendors we contacted for a cost estimate for FCC naphtha
desulfurization technology provided information for several potential
FCC postreater revamp cases.  The first case was a no capital costs case
where refiners made no equipment modifications, but rather solely made
operational changes using their existing equipment installed for Tier 2.
 The second case we requested was one where refiners would incur only
minor capital costs and was intended to be used for analyzing program
options with moderate octane costs.  The third case we requested was one
where refiners were willing to incur greater capital costs in order to
minimize operating costs and octane loss.  The majority of the vendors
only supplied cost estimates for the third case, which included adding
an additional catalyst reactor bed to the existing FCC postreater unit. 
This ensures that refiners will be able to run their existing FCC
postreater at 4 to 5 year catalyst cycle lengths, which is a critical
feature for FCC unit operations.

The costs for the FCC postreater revamps submitted by one of the
vendors, however showed that for low (0 – 400 ppm) and medium (400 –
1,200 ppm) sulfur FCC naphtha sulfur levels, second case, with low
capital costs, resulted in the lowest cents per gallon costs for meeting
the proposed 10-ppm Tier 3 standards.  According to this vendor these
cases also had a 4 to 5 year catalyst cycle length, equivalent to the
higher capital cost cases even though a second stage reactor was not
required.  We therefore assumed that refineries using this vendor’s
technology would choose the minor capital cost pathway for meeting the
10-ppm Tier 3 standard when they had low or medium sulfur levels in
their FCC feed.  The high capital cost cases for producing gasoline to
meet a 5 ppm sulfur standard from low and medium sulfur FCC feeds were
found to have the lowest cost on a cents per gallon basis and were
therefore selected by our model for these cases.

One vendor only submitted information for postreater revamp cost
estimates for FCC naphtha in the 400 – 1,200 ppm sulfur category that
produced a 5-ppm sulfur gasoline.  In our refinery-by-refinery model,
however, we had multiple refineries with FCC feed sulfur levels in the 0
– 400 and >1,200 ppm categories that use this vendor’s postreating
technology.  In order to apply this vendor’s cost estimate to cases of
low (0 – 400 ppm) and high (>1,200 ppm) sulfur feed categories we
adjusted this vendors 400 – 1,200 ppm postreater revamp cost based on
the  cost differentials between the three FCC naphtha sulfur levels in
the other vendors’ revamp estimates.  We similarly derived a
postreater revamp cost estimate to produce a 10 ppm gasoline for this
vendor using cost differentials between the 10 and 5 ppm cases from
other vendors.  For refineries currently employing technology by other
vendors for which we had no specific cost information, we used an
average of all of the vendors’ estimates to represent FCC postreater
revamp costs for refiners using this particular technology in our
refinery-by-refiner model.

After we had determined cost estimates for the FCC postreater revamps
based on information from the vendors the next step was to scale these
costs based on the size of the FCC postreating unit present in each
refinery.  The vendor estimates submitted for revamp costs were based on
various FCC postreater design volumes ranging from 10,000 BPSD to 45,000
BPSD depending on the base unit size used by the vendor. To determine
how to apply these vendor costs to each refinery, we first calculated
each refinery’s maximum FCC naphtha production.  The maximum
production was derived by assuming each refiner runs its FCC unit at its
maximum nameplate throughput capacity (barrels per stream day) with the
FCC naphtha yield rates discussed in Section   REF _Ref309033191 \r 
5.1.3.1.2 .  We then increased the size of the FCC postreater by 7.5
percent above each refinery’s maximum FCC gasoline production rate as
an over design factor to account for excess capacity that refiners
generally design into their unit for processing additional flows of FCC
naphtha (i.e, rerunning high sulfur FCC naphtha batches either from that
refinery or from neighboring refineries).  After sizing the FCC
postreater that would be required for each refinery we then scaled the
costs given by the vendors using the six-tenths rule as shown below. 
This is a “rule of thumb” cost estimating tool commonly used for
cost estimating by the refining and petrochemical industries for
estimating the cost of a process unit based on a similar unit of
differing size.  

Cost to Revamp an FCC Unit= A * (B/C)0.6

Where:

		A = Cost Estimate Received from the Vendor

		B = Size of the FCC Unit in the Refinery

C = Size of the FCC Unit in the Vendor’s Estimate

Equation   STYLEREF 1 \s  5 -  SEQ Equation \* ARABIC \s 1  4 
Six-Tenths Rule for Estimating Capital Cost

We also adjusted the costs submitted by the technology providers based
on the location of each refinery.  We assumed that each vendor’s
estimate was based on revamping an FCC postreater in PADD 3 (Gulf
Coast), which is the lowest cost region for installing new capital in
refineries.  The cost for refineries that are not located in PADD 3 were
adjusted upwards based on a ratio of the cost of refinery capital
projects in the PADD in which they are located relative to PADD 3.  An
additional factor was applied to account for the “offsite” costs
that are incurred when installing new capital in refineries.  When
vendors provide a cost estimate for their technology, this estimated
cost is called the inside battery limits (ISBL) cost and it is solely
for the immediate unit of interest.  However, refiners may need to
install peripheral capital to support the new unit, such as electrical
switchgear, a control room, storage for feed, intermediate or unit
products, and longer than anticipated pipeline runs – these costs are
usually considered Outside battery limit (OSBL) costs, or offsite costs.
 In some cases, OSBL costs may include hydrogen and sulfur plant costs,
although, for our analysis, we separately estimated the cost for
providing additional hydrogen and for processing the removed sulfur and
included this cost in our cost analysis.  We estimate that the
grassroots offsiste cost factors for very modest levels of
desulfurization inherent in this analysis are in the 1.2 to 1.3 range,
and those for these very simple revamps are in the 1.1 to 1.15 range
(the ISBL costs are multiplied by these factors to derive a total cost).
 These cost factors, as well as the utility prices that we used in our
refinery-by-refinery cost model, are shown in   REF _Ref312134282 \h 
Table 5-25  below:

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  25  Cost Factors
for Various PADDs

	PADD 1	PADD 2	PADD 3	PADD 4	PADD 5a

Capital Cost Factor	1.5	1.3	1.0	1.4	1.2

Natural Gas ($/MMBTU)	7.88	7.86	5.64	7.06	7.44

Electricity (¢/kW-hr)	7.98	5.47	5.78	4.69	8.36

Steam ($/1,000 lb)	12.9	12.6	9.20	11.6	11.9

Offsite Capital Cost Factor – New Units	1.26	1.26	1.20	1.30	1.30

Offsite Capital Cost Factor – Unit Revamps	1.13	1.13	1.10	1.15	1.15

a Excluding California

The volume-weighted cost estimates for revamping FCC postreaters across
the entire refining industry as calculated by our refinery-by-refinery
model are shown in   REF _Ref311717664 \h  Table 5-26  and   REF
_Ref311717674 \h  Table 5-27  below.  These costs are aggregated cost
estimates for the FCC revamp costs used in our refinery-by-refinery
model.  In our model, we paired vendor cost data with refineries that
are already using that particular vendor’s technology for their FCC
postreating units.  We further tailored the information provided by the
vendors to match the specific refinery configuration to the extent
possible.  The information that we received from the vendors and the
individual refinery capital costs, however, cannot be shown due to CBI
concerns.

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  26  Revamp Cost
for a 30,000 BPSD FCC Postreater for 10-ppm Standard

FCC Feed Sulfur Levela	0 – 400 ppm	400 – 1,200 ppm	>1,200 ppm	Volume
Weighted Average b

Capital Cost ($/B ISBL)	235	382	505	265

Hydrogen (scf/bbl)	36.8	25.9	25.9	31.3

Fuel Gas (kBTU/bbl)	15.5	15.4	8.51	14.4

Electricity (kWh/bbl)	0.14	0.27	0.52	0.223

Octane Loss (R+M)/2	0.56	0.46	0.42	0.492

Olefin Decrease (vol%)	2.67	2.14	1.97	2.39

Catalyst Cost ($/B)	0.01	0.01	0.01	0.010

Steam (lb/bbl)	0.61	30.88	0.61	13.44

a  $/B = dollars per barrel, scf/bbl = standard cubic feet per barrel;
kBTU/bbl = thousand BTU per barrel; kWh/bbl = kilowatt-hours per barrel;
(R+M)/2 = (research octane + motor octane)/2; vol% = volume percent; $/B
= dollars per barrel; lb/bbl = pounds of steam per barrel of feed.

b  Of the refineries that are expected to revamp their FCC naphtha
hydrotreater for the no ABT case, 36 have FCC naphtha sulfur levels in
the 0 – 400 ppm range, 20 have FCC naphtha sulfur levels in the 400
– 1200 ppm range and 13 have FCC naphtha sulfur levels greater than
1200 ppm.    

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  27  Revamp Cost
for a 30,000 BPSD FCC Postreater for 5 ppm Standarda

FCC Feed Sulfur Level	0 – 400 ppm	400 – 1,200 ppm	>1,200 ppm	Volume
Weighted Average

Capital Cost ($/B ISBL)	400	578	600	444

Hydrogen (scf/bbl)	36.2	52.6	56.3	38

Fuel Gas (KBTU/bbl)	22.7	14.6	13.2	17.2

Electricity (kWh/bbl)	0.30	0.52	0.52	0.42?

Octane Loss (R+M)/2	0.60	0.77	0.90	0.62

Olefin Decrease	2.81	3.38	3.96	2.37

Catalyst Cost ($/B)	0.00	0.01	0.00	0.004

Steam (lb/bbl)	23.61	30.88	0.61	26.9

a Assumes every refinery is complying with a 5 ppm gasoline sulfur
standard

We also found that there were 11 refineries that had an existing FCC
postreaters that were not sized large enough to process their maximum
FCC naphtha production volume.  For these refineries we assigned
additional capital costs to debottleneck the existing first stage
reactor in order to increase the postreater capacity so that it could
accommodate maximum FCC naphtha production.  For each refinery with an
existing unit that could not process more than 70 percent of our
estimate of a refiner’s maximum FCC naphtha production we added
capital costs to revamp and expand the first stage to increase its
capacity to allow the postreater to process 100 percent of its maximum
FCC naphtha rate.  For the capital costs for this debottlenecking we
used 35 percent of the cost of a new grass roots unit (discussed below)
for the volume of the expansion.  We once again used the six-tenths rule
to adjust the capital cost for the volume expansion needed versus the
cost for the 30,000 BPSD grass roots treater used for technology vendor
estimates.

Cost for Grassroots FCC Postreaters

While all refineries that already have FCC postreaters should be able to
meet the proposed Tier 3 standards by revamping their existing
postreaters, refineries that do not currently have an FCC postreater
would have to add a grass roots FCC postreater.  To determine the cost
of building grass roots FCC postreating units at a refinery we similarly
requested cost estimates from vendors.  Only one of the vendors that
supplied FCC postreating equipment submitted information on the cost of
a grass roots FCC postreating unit for desulfurizing FCC naphtha with a
feed sulfur content of 100 ppm.  Based on the calculation methodology
shown in   REF _Ref308599170  Equation 5-2 , we estimate that refineries
that require a grass roots postreater will already have an FCC feed
sulfur level that averages between 85 and 100 ppm as these refineries
already have FCC feed pretreaters.  The other grassroots vendor estimate
we received, as well as those we received for Tier 2, represented a
grass roots postreater with an FCC feed sulfur content of about 800 ppm.
 These estimates were deemed to be not representative of the costs to
refineries that would be installing grass roots postreating units as the
capital, hydrogen, and other operating costs would be much higher for an
FCC feed sulfur of 800 ppm vs. 100 ppm.  We did not consider this other
vendor’s cost estimate for a grass roots postreater and therefore
relies on a single vendor’s cost estimate for grass roots FCC
postreating units for the Tier 3 program.  In our FRM analysis we
attempt to incorporate additional information from other vendors and
literature sources.  This one vendor cost estimate seemed reasonable
relative to the other cost data that we have for higher levels of
desulfurization.

The vendor estimate submitted for a grass roots postreater was based on
a postreater with a capacity of 30,000 BPSD capable of producing an FCC
naphtha with a sulfur level of 10 ppm, corresponding to a gasoline
sulfur level of 5 ppm.  To scale the cost submitted by the technology
vendor to be applicable to a specific refinery, we used a similar
methodology to that which was used for postreater revamps.  We first
determined the appropriate size for each unit based on each refiners
maximum FCC naphtha production rate, adding 7.5 percent above each
refinery’s maximum FCC gasoline production rate as an over design
factor.  We then used the six-tenths rule (  REF _Ref312134400 \h 
Equation 5-4 ) to scale the cost reported by the vendor up or down as
appropriate based on the relative volume of the grass roots unit
required by the refinery and the size on which the vendor’s cost
estimate was based.  We once again assumed that the capital cost from
the technology vendor was representative of a refinery in PADD 3 and
adjusted the cost based on the cost of refinery capital projects in the
PADD in which they are located relative to PADD 3.  Finally, we used a
new unit offsite adjustment factor as listed in   REF _Ref312134282 \h 
Table 5-25  to determine the final cost of a grass roots FCC postreater
for each refinery.  The costs to produce an FCC naphtha with a sulfur
level of 25 ppm (corresponding to a 10-ppm gasoline) were estimated
based on the grass roots postreater unit that makes FCC naphtha for the
5-ppm standard.  These costs are summarized in   REF _Ref312134876 \h 
Table 5-28  and   REF _Ref312134882 \h  Table 5-29  below.

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  28  Cost for a
30,000 BPSD Grass Roots FCC Postreater for 10-ppm Standard

FCC Feed Sulfur Level	100 ppm

Capital Cost ($/B ISBL)	1500

Hydrogen (scf/bbl)	62.1

Fuel Gas (KBTU/bbl)	4.95

Electricity (kWh/bbl)	1.06

Octane Loss (R+M)/2	0.55

Olefin Decrease	2.65

Catalyst Cost ($/B)	0.04

Steam (lb/bbl)	20.0



Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  29  Cost for a
30,000 BPSD Grass Roots FCC Postreater for 5-ppm Standard

FCC Feed Sulfur Level	100 ppm

Capital Cost ($/B ISBL)	1500

Hydrogen (scf/bbl)	133

Fuel Gas (kBTU/bbl)	9

Electricity (kWh/bbl)	1.06

Octane Loss (R+M)/2	1.00

Olefin Decrease	5.15

Catalyst Cost ($/B)	0.04

Steam (lb/bbl)	20.0

Light Straight Run and Natural Gas Liquids Desulfurization Costs

Another action refiners may need to take to reduce the sulfur content of
their gasoline is to desulfurize their light straight run naphtha (LSR)
and natural gas liquids (NGL) blendstocks.  While these blendstocks have
lower sulfur contents than the FCC naphtha, they may be cheaper to
desulfurize for refineries that are not already treating these streams. 
Many refineries can desulfurize some or all of these blendstocks using
existing excess capacity in their naphtha hydrotreaters or FCC naphtha
postreaters.  Further, as opposed to hydrotreating FCC naphtha which
contains olefins, the LSR and NGL blendstocks contain no olefins and
therefore, hydrotreating them does not result in octane loss and has a
lower hydrogen consumption.  The combination of the potential for using
excess capacity in existing units and low operating costs result in the
relatively low desulfurization costs for the LSR and NGL blendstocks. 
From our discussions with refiners, several refineries indicated that
they would install new standalone hydrotreaters for processing LSR and
NGL blendstocks, though it is unclear which other refineries will have
to add equipment to desulfurize LSR and NGL.  For our Tier 3 cost
estimates, however, we have conservatively used our refinery-by-refinery
model to estimate the costs for other refiners to hydrotreat all refiner
volumes of LSR and NGL by installing additional hydrotreating equipment.
 To determine the cost to desulfurize the LSR and NGL blendstocks we
first had to determine the volume of blendstock that requires
desulfurization.  Our determination of the quantity of LSR and NGL used
as gasoline blendstock at each refinery is discussed in Section   REF
_Ref309033191 \r  5.1.3.1.2 .  From this total we then subtracted the
volume of LSR processed in the isomerization unit.  Because the
isomerization units require a low sulfur feedstock we assume all the
feed to this unit is treated by the naphtha hydrotreater.

The next step in our assessment of the desulfurization costs of the LSR
and NGL blendstocks was to estimate the extent to which these
blendstocks are already being treated at refineries to meet the existing
Tier 2 sulfur standards.  Based on our discussion with refining
consultants, vendors and refiners, it appears that in response to the
Tier 2 standards refiners have altered their operations to use excess
capacity in their FCC naphtha postreaters and naphtha hydrotreaters to
reduce the sulfur content of LSR and NGL blendstocks.  Since information
on the extent to which these streams are currently being hydrotreated is
not publicly available we estimated these volumes using the capacities
of the FCC postreaters and reformer feed hydrotreaters under normal
refiner crude throughputs and yields from the refinery-by-refinery
model.

We evaluated each refinery’s capacity to hydrotreat LSR and NGL using
existing equipment by first determining the volume that can be processed
in their naphtha hydrotreaters.  We assumed that every refinery’s
production volume of straight run and coker naphthas, as determined in
Section   REF _Ref306713020 \r  5.1.3.1 , are processed in the
refinery’s reformer feed hydrotreater and that any capacity up to 85
percent of the unit’s maximum capacity can be used to treat LSR or
NGL.  If a refinery had insufficient excess capacity in their naphtha
hydrotreater to treat all of the LSR and NGL volumes we next determined
if there was excess capacity in that refinery’s FCC postreater.  We
allowed LSR and NGL to be processed using excess FCC postreating
capacity in refineries where the capacity of the FCC postreater exceeds
120 percent of that required to process a refinery’s maximum FCC
naphtha yield as determined by the refinery-by-refinery model.  Only
several refiners had excess FCC postreating capacity available for the
treating of LSR or NGL feedstocks, as the capacity of most FCC
postreaters was less than 120 percent of the maximum FCC naphtha
production rate.  We assumed that refiners are currently using any
excess hydrotreating capacity in their naphtha hydrotreating and FCC
postreating units to desulfurize LSR and NGL in response to the Tier 2
sulfur standards.  To meet the proposed Tier 3 sulfur standard 10 ppm
and the 5 ppm level we considered for the ABT cases, we assumed that
refiners would desulfurize all of the LSR and NGL blendstocks.  We
assumed that LSR and NGL purchased by refiners would be hydrotreated in
addition to that processed by the isomerization unit and that blended
directly into the gasoline.  To determine which of the refineries in our
refinery-by-refinery model would need to add hydrotreating capacity, we
evaluated each refinery based on their maximum crude processing rate and
yields.  We used our model to increase each refinery’s yield of coker
naphtha, reformer naphtha, isomerate, LSR and NGL to correspond to a
crude utilization rate of 100 percent.  We then used these maximum
production volumes to evaluate whether or not each refinery could
process all of the LSR and NGL using excess naphtha hydrotreater and FCC
postreating capacity.  If a refinery did not have sufficient excess
hydrotreating capacity for all of the LSR and NGL in these units we
assumed the refinery would have to either revamp their existing
equipment or add new hydrotreating capacity.  If the additional capacity
needed at any given refinery exceeded the existing naphtha hydrotreater
capacity by less than 30 percent we assumed the necessary capacity could
be added by revamping the existing unit. (For our FRM, we will also
evaluate if refiners could also use the caustic Merox extraction process
to lower the sulfur level in LSR and NGL blendstocks, and if so, are
they already doing that today).  If, however, the additional capacity
required exceeded the existing reformer feed hydrotreater capacity by
more than 30 percent we assumed the refinery would install a new
stand-alone hydrotreater to desulfurize the excess LSR and NGL. Based on
available capacity in our refinery-by-refinery model and 2009 crude
throughput data, we estimated that refiners are already hydrotreating 66
percent of the LSR and NGL that are directly blended into gasoline
(excluding LSR processed in the isomerization units).   The results of
this assessment are shown in   REF _Ref309141836  Table 5-30  below.

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  30  Refineries
Adding Hydrotreating Capacity for LSR and NGL

	New Hydrotreater (No FCC Unit)	New Hydrotreater (with FCC Unit)
Revamped Hydrotreater (No FCC Unit)	Revamped Hydrotreater (with FCC
Unit)

Number of Refineries	4	11	1	7

We conservatively evaluated the capital cost required for hydrotreater
revamps and new units by assuming that refiners will size their
hydrotreater equipment needs to treat all production volumes of LSR and
NGL based on each refinery’s maximum crude run rate.  The operating
costs used in our refinery-by-refinery model, however, are based on LSR
and NGL blendstock rates from the models yields at the 2009 operational
crude throughputs as discussed in Section   REF _Ref306713020 \r 
5.1.3.1 .  Sizing the equipment this way allows each refiner to have
excess hydrotreater capacity utilization, which is beneficial in the
event of process unit shutdowns and to reprocess blendstocks from
abnormal operations.

Our estimate for the cost of adding a new hydrotreater at a refinery was
obtained from Gary and Handework’s Petroleum Refining Technology and
Economics, page 182-183, Curve C, Table 9.1, 30,000 BSD unit.  The
capital cost for a grass roots hydrotreater listed by this source was
for a hydrotreater with a capacity of 30,000 BPSD and was based on 1999
dollars.  We multiplied this cost by 1.534 to determine the equivalent
cost in 2010 dollars based on the relative increase in the Nelson
Refining Construction index from 1999 to 2010 (listed as 1497 and 2296
respectively).  We used the six-tenths rule to scale the capital cost
listed in Petroleum Refining Technology and Economics to those of
differing capacities based on relative size of the desired unit.  We
assumed a hydrogen consumption of 40 SCF/Bbl for the processing of LSR
and NGL blendstocks which we obtained from the Jacobs Refining LP
modeling database for naphtha hydrotreating as this information was not
presented in the literature source.  For refineries that only required a
revamp of existing units we assumed a capital cost equivalent to 40
percent of the cost of a new hydrotreating unit of equal size.  We
assumed equivalent operating costs for new hydrotreating units and
revamped units.  The capital and operating costs for these hydrotreating
units that were used in our model is shown in   REF _Ref309142410  Table
5-31  below.

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  31  Capital and
Operating Costs for LSR and NGL Hydrotreaters

	New Hydrotreating Units	Revamped Hydrotreating Units

Capital ($/BBL ISBL)	1380	550

Hydrogen (SCF/BBL)	40	40

Fuel Gas (kBTU/BBL)	100	100

Electricity (kWh/BBL)	2.0	2.0

Octane Loss (R+M)/2	0.0	0.0

Change in Olefins (vol%)	0.0	0.0

Steam (lb/BBL)	6.0	6.0

Catalyst Cost ($/BBL)	0.03	0.03

Butane Desulfurization Costs

After considering the desulfurization of LSR and NGL we then estimated
the cost to each refinery for desulfurizing the butane that is blended
into gasoline.  For the baseline case without ABT where each refinery
had to meet the 10-ppm sulfur standard we did not assume any cost
associated with butane desulfurization, as we assumed that any refiners
likely to reduce the sulfur content of the butane stream already had the
necessary equipment.  To estimate the costs for each refinery to produce
a 5-ppm gasoline for our ABT cases, however, we assumed refineries would
reduce the sulfur content of any butane blended into gasoline using
extractive desulfurization units (such as Merox or Merichem) which
extract mercaptan sulfur (the bulk of sulfur species in butanes) from
butane volumes that are added to each refiner’s gasoline pool.  We
accounted for butane treating cost in our refinery-by-refinery model by
adding capital to install Merox treating units to each refinery.  This
is likely a conservative estimate based on our lack of information on
which refineries may already have these units.  We set up the model to
size the Merox treater based on each refinery’s maximum addition of
butane volume to the gasoline pool, using the amount that is blended in
the winter, as winter grade gasoline has the highest seasonal gasoline
RVP limits and therefore allows for the highest levels of butane
blending.  As with some of the LSR and NGL desulfurization units, it is
unclear if additional butane desulfurization units will be added to meet
the Tier 3 standards.  As a conservative estimate, though, we have
included the cost of adding these units to our program costs. 

To calculate the volume of butanes added to winter grade gasoline we
used the refinery-by-refinery model estimate of the full year addition
of butane to each refiners CG and RFG gasoline grades which was based on
the EPA annual gasoline database.  The winter addition of butane volumes
was assumed to be 66 percent of the total annual volume of butanes added
to a refiner’s gasoline pool on a yearly basis.  The capital cost we
used was for a 10,000 BPSD Merox unit  and was listed in 1995 dollars. 
We multiplied this cost by 1.64 to determine the equivalent cost in 2010
dollars based on the relative increase in the Nelson Refining
Construction index from 1995 to 2010.  We then multiplied the maximum
winter daily butane volume by 1.08 to account for an over design factor
to size the equipment.  We next used the six-tenths rule to scale the
capital cost listed in Petroleum Refining Technology and Economics to
those of differing capacities based on relative size of the desired unit
in BPSD divided by 10,000 (the size of the unit the cost is based on). 
Finally, we applied offsite and PADD location factor adjustments as
listed in   REF _Ref312134282 \h  Table 5-25 .  For the operating cost
we assumed that each unit had an annual fixed operating cost of 6.7
percent of the total installed capital cost.  To this we added an
operating cost of 0.03¢ per gallon based on an operating cost figure
from literature listed in 1995 dollars and scaled to 2010 dollars using
appropriate factors.  Operating costs were presumed to represent all
process energy and utility costs, as well as costs associated with
purchasing caustic. 

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  32   Merox
Treating cost for Sulfur Removal from Butane

Size, BPSD	Capital costs, $ MM	Operating Costs, ¢/gal

10,000	3.28	0.03

Overall Cost Methodology

Sulfur Costs without ABT Program

While we are proposing an ABT program as described in Section V.E.1 of
the preamble, we also evaluated a scenario with no averaging, banking,
or trading between refineries.  Under this scenario, every individual
refinery would comply with the 10-ppm annual average sulfur standard,
and none would lower their annual average sulfur below 10 ppm.  This
evaluation provided a reference point for determining the impacts of an
ABT program on compliance costs.  

In complying with the 10-ppm sulfur standard, our refinery-by-refinery
analysis concluded that some refineries would do so with revamps to
existing equipment while others would need to install new grassroots
equipment.  Those refineries whose average sulfur levels are already at
or below the 10-ppm standard would not need to do anything.  Because the
most recently available data was for 2009 when some phase-ins under Tier
2 were still effective, we made adjustments to ensure that
refinery-specific sulfur levels represented levels after all Tier 2
phase-ins had expired.  For instance, if any refinery exhibited an
annual average sulfur level above the 80-ppm cap based on 2009 data, we
assumed that their current annual average sulfur level was 30 ppm. 
Further discussion of our refinery-by-refinery analysis can be found in
Section   REF _Ref308076513 \r  5.1.3 .

To determine the impacts on cost of the 10-ppm sulfur standard without
ABT, we volume-weighted the refinery-specific ¢/gal costs calculated
according to the refinery-by-refinery modeling using 2009 gasoline
production data for each refinery.  Likewise, we determined nationwide
capital costs by summing the individual revamp and/or grassroots
equipment costs across all refineries.  In these calculations, we
assumed that all early credit generation and phase-ins had expired,
consistent with our approach to the ABT analysis.  The results of our
analysis of costs under a scenario without ABT are provided in Section  
REF _Ref308076594 \r  5.2.1.1  below.

Sulfur Costs with ABT Program

As described in Section V.E.1 of the preamble, we are proposing an ABT
program that is designed to ease the overall burden on the industry
while still achieving the 10-ppm annual average sulfur standard for the
nation as a whole.  Under the proposed ABT program, refineries that can
reduce sulfur below 10 ppm at a relatively low cost can generate credits
which can then be sold to refineries for whom the cost of attaining the
10 ppm sulfur standard would be relatively high.  The net effect of this
credit trading would be expected to reduce the overall cost of the
program.

To estimate the impact that the ABT program could have on nationwide
average fuel costs, we began with the refinery-by-refinery costs
described in Section 5.2.1.1 for sulfur reductions down to either 10 ppm
or 5 ppm.  We then determined the lowest cost option among three
alternatives for each refinery:

1.  	The refinery reduces its sulfur to 10 ppm.

2.  	The refinery reduces its sulfur to 5 ppm and generates credits (in
ppm-gal) for the increment between 10 ppm and 5 ppm.

3.  	The refinery does not lower sulfur, but instead obtains credits to
comply with the 10-ppm standard.

A fourth category applied to refineries whose average gasoline sulfur
levels are already below 10 ppm.  All such refineries were assumed to
generate credits for the increment between 10 ppm and their current
sulfur level.

To optimize the nationwide average costs under an ABT program, we
determined the credit price at which the total number of credits
generated was equal to the total number of credits consumed.  To
simplify the modeling of how an ABT program might operate, we focused on
the circumstances that refineries would face in the longer term,
specifically after 2020.  This approach meant that the ABT program
modeling did not consider the impact on gasoline sulfur levels of
delayed compliance for small refiners and small volume refineries, nor
did it consider the generation and use of any early sulfur credits. 
Moreover, our ABT modeling considered only gasoline sold for use outside
of California, and only gasoline produced by domestic refineries (not
imports).  

Since the cost information available for this NPRM was limited to the
costs of reducing sulfur to either 10 ppm or 5 ppm (under which FCC
gasoline sulfur levels would be reduced to either 25 or 10 ppm,
respectively), we were not able to estimate refinery-specific costs of
reducing sulfur to other levels.  As a result, our ABT modeling could
not account for scenarios in which a refinery makes some capital
investments to lower sulfur to some interim level, such as 20 ppm, and
then obtains credits in order to demonstrate compliance with the 10-ppm
standard.  Our ABT analysis also could not account for credit generation
at sulfur levels other than 5 ppm.  Our ABT analysis, then, most likely
underestimates the cost savings that could occur due to ABT since the
greatest efficiencies are achieved when every refinery has the option of
using any combination of capital investments and credits generation or
use.  For the final rule, we may investigate methods for expanding our
ABT analysis to examine these types of scenarios.

We evaluated two ABT scenarios designed to bound the likely outcomes. 
In the first scenario, we assumed perfect nationwide credit trading in
which all credit generators make their credits available to any refinery
that needs them, regardless of their ownership.  As described in Section
V.E.1 of the preamble, the proposed ABT program is designed such that it
could operate in this way.  However, such perfect nationwide trading may
not be realistic.  Under Tier 2 today, there is still a considerable
amount of inter-company trading occurring, but a significant fraction of
Tier 2 sulfur credits are bought and sold within companies.  Under Tier
3, it will be more difficult to generate credits, and also more
difficult to make up for deficits.  Consequently, we also investigated a
second ABT scenario in which trading between companies does not occur
and averaging would only occur within companies that own more than one
refinery.  Under this second scenario, individual companies might decide
to bank credits for their own use, declining to make credits that they
generate at one of their facilities available to other companies.  They
might do this, for instance, to address unplanned equipment downtime or
other circumstances that could make future compliance more difficult. 
While we do anticipate some trading to occur between companies, we
believe that this second, more limited scenario more closely represents
how our proposed ABT program might actually operate.  As a result, the
within-company credit trading scenario is the primary scenario that we
used to estimate overall costs for our proposed program.  It represents
a somewhat conservative scenario with respect to estimation of costs. 
The results of both scenarios are described in Sections   REF
_Ref308083693 \w  5.2.1.2  and   REF _Ref308083698 \w  5.2.1.3  below.

Since an ABT program would allow some refineries to continue producing
gasoline with an average sulfur level above 10 ppm, we also investigated
whether any areas of the country might experience inordinately high
sulfur levels as a result, and if so whether those higher sulfur levels
might be problematic in terms of either vehicle performance or local air
quality.  While we were not able to model the distribution of all
gasoline from the point of production to the point of consumption, we
did compare average sulfur levels at each refinery's location to the
location of nearby ozone nonattainment areas (based on the designation
status with respect to the 1997 ozone NAAQS).  A discussion of our
analysis of sulfur levels by refinery and location can be found in
Section 5.2.1 below.

Estimated Tier 3 Sulfur Control Costs

Sulfur Cost Results

Cost of the Sulfur Program without ABT

Without ABT, all refineries would be required to meet the 10-ppm annual
average sulfur standard.  We estimate that 13 refineries are already at
or below 10 ppm, and thus their costs of compliance would be zero.  The
remaining refineries would incur ¢/gal costs at various levels
depending on their particular configurations and the steps they would
need to take.  While we estimate that the nationwide average cost of
compliance would be 0.97 ¢/gal under this scenario, two refineries
would incur costs above 6 ¢/gal.  The distribution of costs is shown in
  REF _Ref308083854  Figure 5-1  as a function of the number of
refineries.

Figure   STYLEREF 1 \s  5 -  SEQ Figure \* ARABIC \s 1  1  Distribution
of Compliance Costs by Refinery for 10 ppm without ABT Program

In order to enable evaluation of the costs of the program with ABT, we
also needed to estimate the costs for each refiner to reach 5 ppm.  The
distribution of costs for this scenario is shown below.

Figure   STYLEREF 1 \s  5 -  SEQ Figure \* ARABIC \s 1  2  Distribution
of Compliance Costs by Refinery for 5 ppm without ABT Program

Cost of Sulfur Control Program with ABT

Under a scenario in which credit trading only occurs between refineries
within the same company, we determined that about 23 percent of
refineries would be involved in either generating or consuming credits. 
  REF _Ref308084037  Table 5-33  summarizes our estimate of the amount
of credit trading that would occur under this scenario.

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  33  Impacts of ABT
Program with Intra-Company Credit Transfers

Number of refineries whose sulfur is already below 10 ppm, and which
generate creditsa	3

Number of refineries which lower their sulfur to 5 ppm and generate
credits	15

Number of refineries that do not lower their sulfur but instead consume
credits	8

Number of refineries that lower sulfur to 10 ppm and neither generate
nor consume credits	85

a Additional refineries also have sulfur levels below 10 ppm, but do not
generate credits under this scenario.

The nationwide average cost of compliance with the 10-ppm sulfur
standard would be reduced from 0.97 ¢/gal without ABT to 0.89 ¢/gal
under company-only trading.  However, there would continue to be a
significant variation in costs by refinery as illustrated in   REF
_Ref308084154  Figure 5-3 .

Figure   STYLEREF 1 \s  5 -  SEQ Figure \* ARABIC \s 1  3  Distribution
of Compliance Costs by Refinery for 10-ppm Average Standard with ABT
Program and Intra-Company Credit Transfers

 In this figure, refineries with negative costs are those whose sulfur
levels were already below 10 ppm and who would generate and sell
credits.  While there are several refineries with costs above 3 ¢/gal
in this scenario, in practice we would expect such refineries to incur
lower costs by pursuing some combination of lowering their gasoline
sulfur to some level above 10 ppm and also purchasing credits to cover
the remaining sulfur reductions needed to reach 10 ppm.  The analysis we
conducted for this draft RIA was unable to capture such nuances, but we
will be investigating them for the final rule.

  REF _Ref308084237  Figure 5-4  illustrates how the cumulative
distribution of costs by volume would change under this scenario.  In
general, the distribution would shift downward slightly in comparison to
a scenario in which there was a 10 ppm average standard but no ABT
program.

Figure   STYLEREF 1 \s  5 -  SEQ Figure \* ARABIC \s 1  4  Distribution
of Costs by Volume for 10-ppm Average Standard with ABT Program and
Intra-Company Credit Transfers

Since we estimate that 85 out of 111 refineries would meet the 10-ppm
sulfur standard without the use of credits under the intra-company
trading scenario, only about 5 percent of gasoline would continue to
have an annual average sulfur level above 10 ppm.  These refineries tend
to be smaller than average.  The distribution of sulfur levels under
this scenario is shown in   REF _Ref308084298  Figure 5-5 .

Figure   STYLEREF 1 \s  5 -  SEQ Figure \* ARABIC \s 1  5  Distribution
of Sulfur Levels by Refinery with ABT Program and Intra-Company Credit
Transfers

Of the 5 percent of gasoline volume that would have average sulfur
levels above 10 ppm under this scenario, nearly half would be below 20
ppm, and a majority would be below 30 ppm.  Moreover, as noted above, we
would expect refineries with the highest sulfur levels to pursue some
combination of lowering their gasoline sulfur to some level above 10 ppm
and also purchasing credits to cover the remaining sulfur reductions
needed to attain the 10-ppm standard.

We also investigated whether the existence of refineries producing
gasoline with average sulfur levels higher than 10 ppm after 2020 might
contribute to local areas where the average sulfur level is higher than
10 ppm, and whether this might be problematic in terms of either vehicle
performance/emissions or local air quality.  To do this, we compared the
location of refineries with sulfur levels higher than 10 ppm to the
location of other refineries serving similar areas.  We also compared
the location of refineries to areas that have historically been in
nonattainment for ozone.

As shown above in   REF _Ref308084037  Table 5-33 , we estimated that
eight refineries would have annual average sulfur levels above 10 ppm
under company-only credit trading.  Three of these refineries serve
areas that are, at a minimum, several hundred miles away from historical
ozone nonattainment areas.  The remaining five refineries with average
sulfur levels above 10 ppm supply areas that are also supplied by other
refineries whose average sulfur levels would be 10 ppm or lower.  As a
result, gasoline with higher sulfur levels would likely be diluted by
gasoline with lower sulfur levels, and we would not expect any ozone
non-attainment area to receive gasoline with an average sulfur level
higher than 20 ppm.  

Cost of Sulfur Control Program with ABT and Nationwide Credit Transfers

Although we believe that credit trading will occur primarily between
refineries within the same company, we also investigated a scenario in
which credit transfers would occur between all refineries regardless of
location or company.  For this scenario, we determined that more than 60
percent of refineries would be involved in either generating or
consuming credits.    REF _Ref308084524  Table 5-34  summarizes our
estimate of the amount of credit trading that would occur under this
scenario.

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  34  Impacts of ABT
Program with Nationwide Credit Transfers

Number of refineries whose sulfur is already below 10 ppm, and which
generate credits	12

Number of refineries which lower their sulfur to 5 ppm and generate
credits	34

Number of refineries that do not lower their sulfur but instead consume
credits	25

Number of refineries that lower sulfur to 10 ppm and neither generate
nor consume credits	40

The nationwide average cost of compliance with the 10-ppm sulfur
standard would be reduced from 0.97 ¢/gal without any ABT program to
0.79 ¢/gal under nationwide trading.  There would continue to be some
variation in costs by refinery as illustrated in   REF _Ref308084621 
Figure 5-6 , though the distribution would not be as wide as under the
intra-company credit transfer scenario.

Figure   STYLEREF 1 \s  5 -  SEQ Figure \* ARABIC \s 1  6  Distribution
of Refinery Compliance Costs with ABT Program and Nationwide Credit
Transfers

  REF _Ref308084678  Figure 5-7  illustrates how the cumulative
distribution of costs by volume would change under this scenario.  In
comparison to a scenario wherein credit trading occurs only between
refineries within the same company, the downward shift in the
distribution would be more pronounced.

Figure   STYLEREF 1 \s  5 -  SEQ Figure \* ARABIC \s 1  7  Distribution
of Compliance Costs by Volume with ABT Program and Nationwide Credit
Transfers

Under a scenario in which credit transfers occur nationwide, we estimate
that 25 out of 111 refineries, or about 23 percent, would not lower
their sulfur levels and would instead obtain credits.  These refineries
tend to be smaller than average, and represent about 17 percent of
gasoline volume.  The distribution of sulfur levels under this scenario
is shown in   REF _Ref308084748  Figure 5-8 .

Figure   STYLEREF 1 \s  5 -  SEQ Figure \* ARABIC \s 1  8  Distribution
of Sulfur Levels by Refinery with ABT Program and Nationwide Credit
Transfers

Of the 17 percent of gasoline volume that would have average sulfur
levels above 10 ppm under this scenario, the vast majority would be
below 30 ppm.  

As for the previous scenario under which credit transfers would occur
only within companies, we investigated the localized impact of higher
sulfur levels under a scenario in which credit transfers occur
nationwide and thus there are more refineries with average sulfur levels
above 10 ppm.  Under this scenario, most refineries with average sulfur
levels higher than 10 ppm supply areas that are also supplied by
refineries with sulfur levels at 10 ppm or below.  We expect the
dilution of higher sulfur with lower sulfur in such areas to result in
annual average sulfur levels no higher than about 20 ppm.  Of those
refineries projected to have average sulfur levels higher than 10 ppm
under this scenario and which are the primary or only suppliers of
gasoline to a particular area, none of the affected areas have
historically been in nonattainment for ozone.  

Other Cost Studies

Other other cost studies were recently conducted to estimate the cost of
additional reduction in gasoline sulfur.  We evaluated each of these
studies and compare them to our own cost analysis.  

The International Council for Clean Transportation (ICCT) retained
Mathpro in October 2011 to study the cost of a 10 ppm average gasoline
sulfur standard as well as a 1 psi reduction in RVP.  Since the lower
RVP standard was modeled as a separate step from the low sulfur
standard, we were able to isolate the gasoline sulfur reduction costs
from the low RVP costs.

ICCT’s estimated cost for a 10 ppm average gasoline sulfur standard is
0.8 cents per gallon which reflects the capital costs amortized assuming
a before-tax 7 percent rate of return on investment.  This cost reflects
an assumption that the capital cost for revamps of FCC postreaters is 30
percent of the capital costs for a grassroots FCC postreater .  Mathpro
also analyzed costs assuming that the capital cost for revamps of FCC
postreaters are 50 percent of a grassroots FCC postreaters, which is 1.1
cents per gallon.  ICCT’s cost estimate for complying with a 10 ppm
average gasoline sulfur standard is very close to ours.  

In 2008 The Alliance retained Mathpro to use its LP refinery cost model
to estimate the costs of what they termed National Clean Gasoline (NCG)
in PADDs 1, 2 and 3 (generally speaking, this is the part of the U.S.
east of the Rocky Mountains).  Achieving NCG would entail reducing
gasoline sulfur to 5 ppm under a 10-ppm cap standard and the reduction
of gasoline RVP to 7 psi.  For the low-RVP standard, a 1-psi waiver was
allowed for conventional gasoline, but not for current low-RVP areas. 
The study also evaluated two sensitivity cases which increases the
stringency of the distillation index (DI) from 1250 to 1200.  The
Alliance study also evaluated crude oil price as a second sensitivity
case, evaluating crude oil prices at $51 /bbl and $125/bbl.  

The Alliance studied three different cases.  The first case applied the
10-ppm sulfur cap to RFG.  The second case applied the 10-ppm sulfur cap
and the 7.0-psi low-RVP standard to RFG as well as 7.0- and 7.8-psi
low-RVP gasoline.  The third case applies the 10-ppm sulfur cap and
7.0-psi RVP standard to all RFG and CG.  Of these three cases, the first
case is most relevant because applying the fuels changes to RFG solely
applies the 10-ppm sulfur cap to RFG and does not involve any changes in
RVP.  However, the 10-ppm sulfur cap standard studied by the Alliance is
still 5 ppm more stringent than the 10-ppm average standard that we are
proposing. 

The Alliance cost estimate for Case 1 is 1.6 cents per gallon for RFG in
PADDs 1, 2 and 3.  This cost estimate is based on amortizing the capital
costs on a 10 percent after-tax return on investment (ROI).  We adjusted
the cost estimate to amortize the capital costs based on a before tax 7
percent ROI and adjusted the costs to 2010 dollars which increases the
costs to 1.75 cents per gallon.  The 1.75 ¢/gal cost estimate is based
on a crude oil price of $51/bbl.  The Alliance estimated the cost of a
10-ppm sulfur cap standard on RFG assuming that crude oil is priced at
$125/bbl.  At the $125/bbl crude oil price, the Alliance study estimates
that it costs 2.50 ¢/gal to require that RFG comply with a 10-ppm
sulfur cap standard.  Adjusting the Alliance costs to reflect a 7
percent before tax ROI and 2010 dollars increases the Alliance costs
based on a $125/bbl crude oil price to 2.69 ¢/gal.  

For our cost analysis we analyzed the cost of sulfur control assuming
that crude oil is priced at $91.8/bbl.  We can interpolate between the
Alliance costs based on $51 and $125 per barrel crude oil prices, which
results in a single cost which is 2.3 cents per gallon.  We also
estimated a cost for refiners lowering their gasoline sulfur to 5 ppm
using the refinery-by-refinery cost model and our cost is 1.38 ¢/gal.  

In response to the Alliance study, API retained Baker and O’Brien
(BOB) in 2010 to study the cost of additional gasoline sulfur control
and an increase in RVP using a refinery-by-refinery cost approach with
BOB’s Prism model.  The Prism model is largely a spreadsheet cost
model with blending optimization.  The primary case analyzed by the API
study is the cost of reducing gasoline sulfur to an average of 10 ppm
and reducing gasoline RVP to 7.0 psi without a 1-psi waiver for blending
10 percent ethanol.  The study also analyzed three other sensitivity
cases:  1) a 5-ppm average gasoline sulfur standard with 7 psi RVP limit
on conventional gasoline without a 1-psi waiver; 2) a 10-ppm average
gasoline sulfur standard and a 7.8-psi RVP limit on conventional
gasoline without a 1-psi waiver; and 3) a 10-ppm average gasoline sulfur
standard with a 7.8-psi RVP limit on all conventional gasoline with a
1-psi waiver.  

In an addendum to its fuels study report released in 2011, API
contracted with Baker and O’Brien to study a sensitivity case 4, which
is a sulfur only case, using its PRISM refinery model.  From our
understanding of the study, the study parameters seemed to be about the
same as the original study, except that API solely studied a 10 ppm
average gasoline sulfur standard (not including any RVP control), the
same sulfur standard that we are proposing.  However, API also assumed
that a 20 ppm cap standard would also be in place which would not allow
the application of an averaging, banking and trading (ABT) program to
optimize refinery investments and minimize overall costs.  

API made a series of conclusions based on the study.  Perhaps the most
important difference with the original study is that API concluded that
not a single refinery would shut down as a result of the proposed 10 ppm
gasoline sulfur control standard, even though API did not study the
flexibilities of an ABT program and used excessively high capital costs
for a grassroots FCC postreater (see below).  Like the original study,
API did not report average costs, but reported the marginal costs for
the cost study.  Marginal costs reflect the cost of the program to the
refinery or refineries which would incur the highest costs, assuming
that the highest cost refineries would set the price (or in this case,
the price increase) of gasoline.  The report concluded that marginal
costs after the imposition of a 10 ppm gasoline sulfur program would
increase the price of gasoline by 6 to 9 cents per gallon in most
markets.  API did not define how its statement “in most markets”
would apply to the US gasoline supply.  API also did not provide any
justification why it assumed that the refineries that would experience
the highest desulfurization cost under Tier 3 would also be the same
refineries which sets the gasoline price in the gasoline market today. 

Although API did not provide an average gasoline desulfurization cost in
its report, we could calculate an average cost based on the gasoline
volume and total annual costs provided.  The total cost reported in the
report for the 10 ppm average gasoline sulfur standard is $2390MM/yr and
the non-California gasoline volume is 7343 thousand barrels per day. 
This results in an average per-gallon desulfurization cost of $0.89/bbl
or 2.12 c/gal.  The difference between the average cost and marginal
cost (price increase) that API is projecting is profit.  API is
projecting that the oil industry would profit from 10 ppm low sulfur
standard by the roughly 4 to 7 cents per gallon difference between the
average cost and the two marginal price values.  That per-gallon profit
translates into $4 to $8 billion dollars per year in profit.  

 The average cost of the 10 ppm average gasoline sulfur standard was
calculated using API’s methodology for amortizing capital investments.
 To facilitate a fairer comparison between the API cost study and our
cost study, we adjusted the API costs to be on a similar basis as our
costs.  We adjusted the API costs to reflect a before-tax 7 percent
return on investment (ROI) for capital invested for the hydrotreaters
and hydrogen plants instead of the after-tax 15 percent ROI used by API.
 This lowered the API estimated costs from 2.12 c/gal to 1.58 c/gal. 
API’s 1.58 cents per gallon cost is still higher than our 0.89 c/gal
cost with an ABT program that assumes intercompany trading of credits,
and higher than our 0.97 c/gal for the case which assumes no ABT
program.  Thus comparing “apples-to-apples” to the to the extent
possible, API’s 1.58 c/gal estimated cost for complying with a 10 ppm
average gasoline sulfur standard compares very favorably with our own
cost estimates, and not at all near the 25 c/gal value that was
sometimes quoted from the first API study.  The remaining cost
difference between our estimated costs and those by API are the capital
cost assumptions that API used, as discussed below.  While little detail
is is provided by API about what hardware comprises their
desulfurization units, the inside battery limits (ISBL) and total
capital costs for the FCC postreaters and FCC pretreaters are provided
in API’s report.  API’s FCC pretreaters capital costs are consistent
with the capital costs that we have used for this unit.  However, the
FCC postreater costs used by API are much higher than what we used and
have been used in the past by others.  API’s capital cost for a
grassroots FCC postreater is $228 million for a 35,000 bbl/day unit, or
$6540 per/bbl per day.  API’s capital cost includes the outside
battery limit (OSBL) costs.  In contrast, the ISBL capital cost that we
used for a grassroots FCC postreater is $1500/bbl-day for a 30,000
bbl/day grassroots unit, which increases to $1875/bbl/day when the
offsite costs are added on.  Thus, the API capital costs are about 3 ½
times higher than the capital costs that we are using for a grassroots
FCC postreater.  To check our capital costs, we found other capital cost
estimates to which we could compare our costs, including the capital
costs used by the National Petroleum Council when it studied the cost of
gasoline desulfurization prior to Tier 2.    REF _Ref308540950  Table
5-35  contains a cost comparison of ISBL, and ISBL and OSBL FCC
postreater capital costs.

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  35  Capital Cost
Comparison

Technology	EPA (Tier 3)	API	Mathpro (ICCT)	Jacobs	Exxon Scanfining (NPC)
IFP Prime G (NPC)	CDTech (NPC)	Phillips S-Zorb (NPC)

ISBL Capital Cost ($/bbl/day)	1500	-	-	2440	1045	1410	960	860

ISBL and OSBL Capital Cost ($/bbl/day)	1875	6540	1800	3538	1360	1833
1248	1118

  REF _Ref308540950  Table 5-35  shows that, compared to the average of
the rest of the capital cost estimates, the API capital cost for FCC
postreater is about four times higher.  Compared to the next highest
cost estimate, which is the FCC postreater capital cost from the Jacobs
data base in the Haverly refinery cost model that we use, the API
capital costs are almost two times higher.  

An important distinction must be made with respect to the severity of
desulfurization for the capital cost comparison made in   REF
_Ref311717674 \h  Table 5-27 .  For complying with the Tier 2 gasoline
sulfur standard (Jacobs and NPC costs), a typical refinery would have
installed an FCC postreater to desulfurize the FCC naphtha from about
800 ppm down to about 75 ppm, a 725 ppm, or a 91 percent sulfur
reduction.  In the case of a grassroots postreater that would be
installed for Tier 3, the postreater would treat FCC naphtha already low
in sulfur due to the pretreater installed before the FCC unit (these
refineries are currently complying with Tier 2 using an FCC pretreater).
Thus, the new grassroots FCC postreater would only have to reduce the
FCC naphtha from 100 ppm to 25 ppm, a much smaller 75 ppm or 75 percent
sulfur reduction.  A grassroots FCC postreater installed for Tier 2
would typically remove 10 times more sulfur than one installed for Tier
3.  This is important because a significant portion of the FCC
postreater capital cost is devoted to avoiding the recombination
reactions which occur when hydrogen sulfide concentrations are high and
react with the olefins contained in the FCC naphtha.  Thus, a grassroots
FCC postreater installed for Tier 3 would be expected to be
significantly lower in capital cost compared to a Tier 2 FCC postreater.
 When API presented the costs to us, they stated that their grassroots
capital costs were based on an actual installation for the Tier 2
program.  This could be one reason why the capital costs used by API for
its cost study of the Tier 3 program are so high.  Another way to assess
the API capital cost for the FCC postreaters is to compare it to the FCC
pretreater cost that API is using.  FCC pretreaters are much higher
pressure units and use more expensive metallurgy than FCC postreaters
and, for these two reasons, are much more expensive than FCC postreaters
on a per-barrel basis.  However, API’s FCC postreater capital costs
are about 50 percent more expensive than its FCC pretreater capital
costs, which is inconsistent with the design requirements of the units. 


API’s estimated range of capital cost for revamping an FCC postreater
is also higher than our range of capital cost for revamping an FCC
postreater, when assessing the revamped costs as a percentage of the
capital cost for a grassroots unit.  API estimates that revamping an FCC
postreater would cost 30 to 70 percent of the capital cost for a
grassroots FCC unit.  Our capital cost estimate for revamping an FCC
naphtha postreaters from 17 to 50 percent of the capital cost for a
grassroots FCC postreater, however, most of the revamps are estimated to
cost at the lower end of that range.  

The Emissions Control Technology Association (ECTA) retained personnel
within Navigant Economics to study the costs of a 10 ppm average
gasoline sulfur standard and assess the ICCT and API cost studies.  The
authors made a number of conclusions.  After reviewing both the ICCT and
API studies, the authors found that a primary difference in estimated
costs between the two studies was the capital costs.  The authors
contacted vendor companies that license FCC postreater technologies and
surveyed the companies to find out what the capital costs are for a FCC
postreater.  As a result of the survey, the report authors concluded
that API’s capital costs were too high, and those used in the ICCT
study were about right.  The authors found that Baker and O’Brien has
a history of exaggerating the economic impacts of EPA rules, citing the
costs and other impacts of its analysis of the 2007 on-highway
heavy-duty proposed rulemaking.  The authors concluded that the impact
of a 10 ppm gasoline sulfur standard on the average refining cost would
likely be closer to the 1 cent per gallon estimate by the ICCT study.
Furthermore, the report’s authors also pointed out that the marginal
cost analysis conducted by API did not consider the proposed averaging
banking and trading (ABT) program that we were expected to propose,
which would reduce the marginal costs of the Tier 3 proposed rule.
Because API’s addendums to its original report came out many months
after its original report, we originally assessed the most similar case
to our proposal from API’s original study.  This was sensitivity case
3, which studied a 10-ppm average gasoline sulfur standard with a
7.8-psi RVP limit on all conventional gasoline with a 1-psi waiver.  One
of the most important conclusions by API with respect to sensitivity
case 3 was the projected 623 thousand barrels per day (about 6.5 percent
of total gasoline demand) reduction in U.S. gasoline production due to
reduced blending of light hydrocarbons in response to RVP control and
the closure of 4 refineries caused by the investment hurdle of complying
with both the sulfur and RVP standards evaluated.  Because of API’s
significant projected impacts, which differed from those made by the
Alliance and by ECTA in their studies, we looked closely at this
projected impact.  We found that this projection for a large shortfall
in supply is similar to previous projections made by API for earlier
rulemakings.  Baker and O’brien made a very dire projection for diesel
fuel supply when analyzing the final highway and proposed nonroad diesel
fuel sulfur regulations for the proposed nonroad diesel rulemaking back
in 2003.  API’s projection was that 12 refineries would be shutdown
and that U.S. refiners would exit the diesel market resulting in a
shortage of diesel fuel supply of 639 thousand barrels per day and to
make up the shortfall, the distillate market would need to be supplied
by imports.  Since that projection was made back in 2003 and the highway
diesel fuel program is fully implemented and the nonroad diesel fuel
program is mostly implemented, we can look at the impacts to see how
well API’s projection played out.  

First, with respect to imports, the distillate import/export market was
fairly stable at about 200 thousand barrels per day of net imports
during the period 2003 to 2005, before the highway and nonroad programs
started.  If we add the expected increase in imports estimated by API
caused by the ULSD highway rule, API projected the imports to increase
to 549 thousand barrels per day in 2006.  When the temporary compliance
option of the ULSD highway rule ended in 2010 and nonroad diesel
(excluding locomotive and marine diesel because those requirements
don’t start until 2012) must comply with the ULSD nonroad diesel rule
in 2010, API estimated that imports will increase to 829 thousand
barrels per day including the baseline imports.  However, what actually
occurred is that imports not only did not increase, but they actually
decreased to the point that the U.S. became net exporters of 428
thousand barrels per day of distillate.   Thus, API was off by over 1200
thousand barrels per day in its estimate of distillate production in
2010 by U.S. refiners.    REF _Ref309215910 \h  Figure 5-9  summarizes
the imports and exports of diesel fuel and gasoline during the period
when the clean fuels regulations were being phased in.

Figure   STYLEREF 1 \s  5 -  SEQ Figure \* ARABIC \s 1  9  Petroleum
Product Imports and Exports During the Implementation of Clean Fuels
Regulations

One reason why API was so far off in its projection that U.S. distillate
supply would decrease significantly is that it had projected that 12
refineries would shut down due to the highway and nonroad rulemakings,
but, in fact, there were very few refinery shutdowns.  Between 2003 and
2011 when the highway and nonroad diesel fuel ULSD programs were phasing
in (as well as Tier 2, MSAT2, RFS1 and RFS2), there was a total of 5
refinery shutdowns of refineries which produce transportation fuels. 
However, also during this time period there were reactivations of three
previously shutdown transportation fuel producing refineries. 
Considering both the shutdown refineries and reactivated refineries,
there was a net shutdown of two transportation fuel producing refinery
closures.  It is unclear why those 5 transportation fuel producing
refineries shutdown, but there is no evidence to suggest they were
caused by the diesel rules.  There are many factors which cause refinery
shutdowns.  

To understand these factors it is necessary to revisit the history of
the U.S. refining industry.  During the second crude oil embargo which
occurred during the Carter Administration, subsidies were established
that incentivized the construction of many small refineries.  After
crude oil prices dropped, those subsidies ended and many of those
refineries were not economically competitive with larger refineries. 
Over time the number of refineries decreased and the remaining
refineries were expanded to supply the U.S. market.  This is part of a
rationalization that most any industry experiences as the industry
matures.    REF _Ref309303921 \h  Figure 5-10  and   REF _Ref312057882
\h  Table 5-36  summarize the changes in the number of refineries, the
average size of refineries, and the total production capacity of the
U.S. refining industry based on data from the Energy Information
Administration (EIA).  

Figure   STYLEREF 1 \s  5 -  SEQ Figure \* ARABIC \s 1  10   Refining
Industry Statistics

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  36  U.S. Refinery
Industry Statistics

Year	1982 - 1992	1993 – 2002	2003 - 2011

Net Change in the Number of Operable Refineries	-102	-50	-2

Change in Average Refinery Size (kbbl/day)	+18,400	+31,800	+11,000

Change in Total U.S. Operating Refinery Capacity (MMbbl/day)	-1.2	+1.6
+0.4

  REF _Ref309303921 \h  Figure 5-10  and   REF _Ref312057882 \h  Table
5-36  show that there were many refinery closures during the 1980s and
1990s, although over time the number of refinery closures was
diminishing.  Therefore, the refinery closures which occurred during the
2003 to 2011 time frame could very well be due to this rationalization
process reflecting a maturing industry.  In fact, examining the trend,
the number of net refinery shutdowns is virtually flat during the 2003
to 2011 time frame relative to the previous periods.  This is clearly a
rationalization of the refining industry because as refineries were
closing, the average refinery size increased over each of the periods. 
Finally, the total U.S. refinery capacity decreased and increased back
to its 1982 levels during the 1980s and early 90s (decrease was likely
due to the corporate average fuel economy standards), but showed a
modest increase from the late 1990’s through 2011. 

Projected Energy Impacts and Impacts on Permitting

Our refinery-by-refinery model was also used to determine the impact the
Tier 3 standards would have on the energy related CO2 emissions and
permitting of existing refineries.  While the Tier 3 proposal will
reduce emissions from vehicles, the addition of grass roots units and
revamping of existing units which we project will happen as a result of
the Tier 3 sulfur standards are likely to result in some increased
emissions of regulated air pollutants at refineries.  Refinery projects
designed to meet the new fuel standards could trigger preconstruction
air permitting requirements under the Clean Air Act and EPA’s New
Source Review (NSR) regulations.  To address this concern, we used our
refinery-by-refinery model to estimate the likely process and equipment
changes that may be required to meet the Tier 3 gasoline standards. 
This information was submitted to EPA’s Office of Air Quality Planning
and Standards (OAQPS) to provide the inputs that are necessary for the
modeling and analysis of the refinery emissions and permitting impacts
of the Tier 3 fuel standards.

Using our refinery-by-refinery model we generated refinery-specific
estimates of the increased energy, hydrogen, and gasoline octane demands
that we estimate will result from the proposed Tier 3 standards.  We
also estimated the increase in sulfur plant recovery unit (SRU)
loading/operations for the 111 U.S. refineries that we modeled in our
analysis.  Energy demand includes fuel that is needed to generate
refinery process heat, steam and electricity.  Hydrogen demand is
associated with increased hydrotreating of Fluid Catalytic Cracking
(FCC) naphtha and light straight run (LSR) streams.  Increased gasoline
octane demand results from refineries having to replace octane that is
lost due to increased FCC naphtha hydrotreating.  Increased SRU loading
results from the increased fuel desulfurization and associated H2S
generation.  All of these incremental demands will be referred to as
“demands” in the following sections.  We used our
refinery-by-refinery model to calculate the increase in these various
demands for several scenarios where sulfur averaging, banking, and
trading (ABT) was not allowed and each refinery had to meet the 10ppm
standard, as well as scenarios that allowed ABT between refineries owned
by the same parent company to minimize the cost of compliance with the
Tier 3 standards.

Emissions Impacts of Different Production Volumes

In addition to considering scenarios with and without ABT we also
considered the impacts on emissions and permitting of different gasoline
production volumes for each refinery.  In the first case, called the
normal case, we considered the incremental demands for each refinery
assuming no change in gasoline production volume.  We also considered a
case, called the maximum demand case, where each refinery maximized
gasoline production based on currently existing refinery capacity and
equipment.

Normal Case

The normal case was estimated using each refinery’s predicted yields
of FCC naphtha and LSR from our refinery-by-refinery model, along with
each refinery’s total gasoline production volume from EPA’s RFG
database.  For each refinery the refinery-by-refinery model generated
specific Tier 3 demands for hydrogen, steam, fuel gas, electricity and
gasoline octane based on the desulfurization technology used by each
refinery for any FCC postreating and LSR hydrotreating.  To determine
the FCC postreating demands the model considers each refinery’s volume
of FCC naphtha under normal operations, the FCC naphtha sulfur level at
the refinery prior to postreating, and the process requirements of the
FCC postreater technology used by that refinery.  The demands are
calculated by multiplying the FCC naphtha volume by the demands from the
use of the associated FCC postreating technology.    REF _Ref311717664 
Table 5-26  through   REF _Ref312134882  Table 5-29  show the FCC
postreater technology demand averages as applied to refineries on a
national basis for the 5 and 10 ppm gasoline sulfur standards.  Note
that the demands vary significantly with the FCC naphtha sulfur level
prior to postreating.

Similarly, the normal case demands for any LSR blendstocks that require
additional hydrotreating as a result of the Tier 3 standards were
determined based on each refinery’s yield of LSR blendstock under
normal operations and the demands for the additional LSR hydrotreating. 
These demands, on a national average basis, are listed in   REF
_Ref309142410  Table 5-31 .  The normal case demands for FCC postreating
and LSR hydrotreating were then summed to determine the increase in
energy, hydrogen, and octane demand.  To determine the additional sulfur
removed from gasoline we first calculated the difference between the
current gasoline sulfur level of the gasoline produced at each refinery
according to their compliance reports to EPA and the proposed Tier 3
standard.  This difference was multiplied by the refinery’s gasoline
production volume and divided by the number of days of operation to
calculate the additional sulfur removal level at each refinery.  This
sulfur removal information was then used to determine the increase in
SRU loading on a fractional basis by dividing the additional sulfur
removal as a result of the Tier 3 standards (in tons of sulfur per day)
by the refineries SRU process capacity.

Maximum Case

We also considered a second demand case, called the maximum case, in
which we calculated the demands that result from the Tier 3 standards if
each refinery maximizes gasoline production based on currently existing
refinery capacity.  For this case we first determined each refiners FCC
unit process capacity utilization rate in the normal case.  The annual
FCC unit feedstock charge rate for each refinery as reported in the 2009
EIA data was divided by the FCC unit design capacity as reported in the
Oil and Gas Journal (OGJ) to calculate the capacity utilization rate for
the normal case.  These normal capacity utilization rates were then
scaled up to reflect maximum capacity utilization rates, and further
adjusted using an overdesign factor.

For refineries projected to meet the proposed Tier 3 standards by
revamping existing FCC postreating units we assumed that their maximum
gasoline production rate was equal to the rate produced running the FCC
unit at 92% of the refinery’s maximum FCC design capacity.  There were
several refineries that are currently operating their FCC unit greater
than a 92% capacity utilization rate.  We assumed that these refineries
were already operating at their maximum annual capacity utilization
rate.  For refineries projected to install a new FCC postreater we
similarly assumed that the new unit would be scaled to process the
output of the FCC unit operating with a 92% utilization rate.  For new
FCC postreaters, however, we increased the results by 15% as an
overdesign factor and adjusted the results accordingly.  A similar
sizing approach was taken for refineries we projected would revamp or
add new LSR hydrotreating capacity to comply with the proposed Tier 3
standards.

The results represent a “maximum” annual gasoline production case
for each refinery under the Tier 3 standards based on each refiners FCC
unit design capacity.  These cases represent a scenario where each
refinery projects emissions based on the maximum achievable annual
production rate for their existing processing units.  These cases
reflect each refiner’s potential emissions impacts as a result of the
proposed Tier 3 standards, rather than the existing Tier 2 standards,
when operating at maximum FCC rates as opposed to normal operation more
indicative of national gasoline demand.

Refinery Demand Sourcing

After determining the increased demands for each refinery as a result of
the proposed Tier 3 regulations we next developed cases for each
refinery demand scenario that represented different options for sourcing
these demands.  Some refiners may choose to produce all of the required
hydrogen and electricity, as well as make up for all of the lost octane
at their refinery.  Others may choose to purchase some or all of the
hydrogen, electricity, and high octane blendstocks that they would need
to comply with the Tier 3 standards from external suppliers.  These
decisions have a significant impact on the emissions and permitting
impacts of the proposed Tier 3 regulations.  In order to bound all
possible scenarios we considered both high and low impact cases for each
refinery demand scenario.  In the high impact scenarios we assume that
each refinery produces all of the required hydrogen, and electricity
needs while making up for any gasoline octane loss at their refinery. 
In the low impact scenarios we assume that all the necessary hydrogen,
electricity, and high octane blendstocks are purchased from an external
supplier.

In both cases we assumed that fuel gas demands would increase to meet
the increased thermal demands at the refinery.  In the high impact
scenarios the refinery’s fuel gas needs would be further increased to
produce the needed hydrogen and electricity while replacing the lost
octane in their gasoline.  We consulted literature sources to determine
the conversion factors from MBTU fuel gas to 1,000 standard cubic feet
(scf) of hydrogen and 1,000 pounds of steam that are typical for
refineries.  We also assumed a standard conversion efficiency from fuel
gas to electricity for our modeling.  These conversion factors are shown
in   REF _Ref315775997  Table 5-37  below.

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  37  Fuel Gas
Required to Produce Hydrotreater Utilities

Utility	Fuel Gas Required (M BTU)

Hydrogen (1,000 SCF)	239

Steam (1,000 lbs)	1530

Electricity (1 kWh)	5.1

We also had to determine the fuel gas demands that would be required to
make up for lost octane in our high impacts scenario.  For this analysis
we assumed that all lost octane in the gasoline that results from
increased FCC postreating to meet the proposed Tier 3 standards would be
recovered by running the reformers at the refineries at a higher
severity as opposed to sourcing it from ethanol or other means.  This
would increase the octane of the reformate and offset the octane losses
from the FCC naphtha, resulting in no change in the octane of the
overall gasoline pool.  Running the reformer at a higher severity
requires a higher reactor processing temperature and an increased
volumetric rate for the pumps and compressors.  These changes further
increase the fuel gas and electricity demands of the refinery.  We
estimated these demands using information we obtained from our Jacob’s
refinery LP modeling analysis.  The estimates we used for the fuel gas
and electricity demands for increasing gasoline octane are shown in  
REF _Ref315774510  Table 5-38  below.  These demand increases were only
used for the high impact scenarios.  For the low impact scenarios we
assumed that lost octane was recovered by purchasing high octane
blendstocks resulting in no fuel gas or electricity demand increases at
the refinery.

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  38  Energy
Required to Increase Gasoline Octane

Utility	Amount Required to Increase Octanea 1 Point (per barrel
gasoline)

Fuel Gas (M BTU)	1.87

Electricity (kWh)	0.057

Note:

a (R + M)/2 method

Refinery Demand Impacts

Using our refinery-by-refinery model, along with the technology vendor
data for new and revamped FCC postreating and LSR hydrotreating data
(shown in   REF _Ref311717664  Table 5-26  through   REF _Ref312134882 
Table 5-29  and   REF _Ref309142410  Table 5-31 ) we were able to
calculate the increases in refinery demands as a result of our proposed
Tier 3 regulations for each of the various scenarios outlined in the
previous sections.  This information is summarized in   REF
_Ref315774433  Table 5-39  and   REF _Ref315774442  Table 5-40  below. 
The refineries have been identified by randomly assigned numbers to
protect confidential business information (CBI).  This information was
submitted to OAQPS to serve as the basis for their emissions and
permitting analysis of the Tier 3 regulations.  Based on the data
provided by EPA, OAQPS determined that the increased demands that result
from the Tier 3 regulations would exceed the permitting thresholds under
the New Source Review (NSR) regulations, including the Prevention of
Significant Deterioration (PSD) and/or the Nonattainment New Source
Review (NA NSR) for 16 refineries in the low impact scenarios and 22
refineries in the high impact scenarios.  An additional 7 refineries in
the low impacts scenarios, and 13 refineries in the high impact
scenarios, would exceed the threshold to trigger reviews for greenhouse
gas emissions only.  A technical memorandum describing the OAQPS
analysis and results is in the public docket for this proposal.

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  39  Tier 3
Refinery Energy, Hydrogen and Sulfur Plant Demand Increases (10 ppm, No
ABT Cases)

Refinery Number	Demand Estimates

	Fuel Gas Demands	Sulfur Plant Production	Hydrogen

	Low Impact Cases (Million BTU/Yr)	High Impact Cases (Million BTU/Yr)
Sulfur Production Increase (Tons Sulfur/Day)	Sulfur Plant Capacity
Increase (Percent of Existing Facility)	Hydrogen Demand Increase
(million scf/year)

	Normal Volumes	Maximum Volumes	Normal Volumes	Maximum Volumes



	1	117,423	117,423	243,241	243,241	0.28	0.02	426

2	0	0	0	0	0.00	0.00	0

3	2,239,110	2,739,637	2,566,562	3,140,286	0.00	0.00	863

4	173,672	173,672	569,009	569,009	0.43	0.03	1,628

5	46,380	58,279	261,528	328,624	0.13	0.04	655

6	4,433	7,291	14,523	23,888	0.04	1.96	42

7	60,923	69,776	199,605	228,609	0.10	0.03	571

8	23,941	29,792	134,995	167,988	0.17	0.07	338

9	107,543	135,777	222,775	281,262	0.20	0.18	390

10	25,924	29,812	146,179	168,106	0.10	0.08	366

11	13,984	16,250	78,855	91,630	0.09	0.05	198

12	76,663	92,786	251,175	303,997	0.43	0.07	719

13	51,432	51,432	168,509	168,509	0.47	1.18	482

14	0	0	6,444	6,444	0.00	0.00	27

15	16,876	16,876	89,463	89,463	0.07	0.13	219

16	831,027	1,244,900	899,341	1,347,235	0.26	0.22	218

17	31,964	31,964	82,442	82,442	2.19	0.96	206

18	150,568	150,568	311,900	311,900	0.22	0.03	547

19	27,087	44,061	152,736	248,452	0.26	a	383

20	43,909	46,106	143,861	151,058	0.37	0.08	412

21	165,539	212,719	343,091	440,875	0.17	0.04	538

22	14,886	16,690	38,395	43,048	0.01	0.02	96

23	83,222	83,222	272,663	272,663	0.41	0.12	780

24	190,870	210,535	206,560	227,841	0.33	0.29	50

25	702,808	702,808	798,012	798,012	0.10	0.09	256

26	20,503	22,001	67,175	72,082	0.90	0.60	192

27	0	0	0	0	0.00	0.00	0

28	3,145,454	3,528,532	3,223,899	3,616,532	0.31	0.07	201

29	0	0	0	0	0.00	0.00	0

30	64,080	66,959	165,279	172,703	0.15	0.01	414

31	0	0	0	0	0.00	0.00	0

32	62,002	62,002	159,918	159,918	0.35	3.46	400

33	1,693,620	1,948,219	1,832,841	2,108,370	0.27	0.03	444

34	16,574	24,879	93,456	140,290	0.15	0.21	234

35	10,469	10,469	12,340	12,340	0.04	0.96	6

36	337,961	411,915	365,743	445,776	0.15	a	89

37	5,175	5,175	29,180	29,180	0.13	3.30	73

38	0	0	2,907	2,907	0.00	0.00	12

39	21,773	21,773	71,336	71,336	0.58	0.93	204

40	19,171	23,883	91,207	113,621	0.04	0.01	214

41	177,093	203,657	636,542	732,024	0.30	0.03	1,288

42	3,288	4,683	10,773	15,343	0.03	0.97	31

43	41,705	43,146	107,567	111,284	0.52	0.26	269

44	3,609,133	4,265,613	3,905,818	4,616,262	0.66	0.22	946

45	27,946	41,431	105,829	156,897	0.06	a	221

46	6,846	8,054	38,603	45,416	0.05	0.55	97

47	1,416,139	1,416,139	1,532,550	1,532,550	0.25	0.03	371

48	21,918	41,206	45,403	85,358	0.02	0.01	80

49	7,024	7,191	18,116	18,547	0.19	2.16	45

50	2,170,692	2,170,692	2,349,131	2,349,131	0.61	0.23	569

51	0	0	0	0	0.00	0.00	0

52	3,945,526	3,984,109	4,269,863	4,311,617	0.61	0.03	1,034

53	102,473	124,044	264,305	319,941	0.46	a	662

54	40,155	51,299	108,185	138,209	0.11	0.09	236

55	21,405	26,758	120,695	150,884	0.16	0.13	302

56	26,807	28,515	70,005	74,463	0.04	0.13	144

57	91,964	107,033	518,563	603,534	0.00   	0.03	1,299

58	0	0	0	0	0.00	0.00	0

59	52,160	62,683	108,050	129,848	0.10	0.03	189

60	109,050	109,050	357,285	357,285	0.62	0.11	1,022

61	23,773	23,773	36,151	36,151	0.04	0.00	52

62	9,233	9,655	30,251	31,632	0.04	1.96	87

63	136,167	175,790	282,068	364,148	0.08	0.01	494

64	54,151	60,794	165,033	185,278	0.24	0.12	221

65	360,695	402,762	390,346	435,870	0.05	0.19	95

66	8,113	8,589	20,926	22,152	0.04	0.39	52

67	26,289	34,157	67,806	88,100	0.19	0.17	170

68	89,955	103,448	507,237	583,323	0.20	0.02	1,271

69	30,259	33,479	99,139	109,690	0.14	0.14	284

70	9,369	20,299	61,521	133,298	0.20	0.03	103

71	37,188	37,188	121,841	121,841	1.31	0.44	349

72	1,079,231	1,299,822	1,167,948	1,406,672	0.09	0.14	283

73	15,845	17,618	51,915	57,724	0.03	0.00	149

74	38,525	40,291	101,764	106,429	0.09	0.04	215

75	0	0	0	0	0.00	0.00	0

76	14,588	14,588	47,795	47,795	0.03	0.00	137

77	16,875	20,135	43,524	51,932	0.13	0.70	109

78	139,558	139,558	289,093	289,093	0.16	0.02	507

79	0	0	0	0	0.00	0.00	0

80	64,517	76,071	65,729	77,500	0.07	0.10	234

81	2,009,982	2,009,982	2,280,122	2,280,122	0.24	0.05	729

82	58,475	60,441	191,583	198,026	0.26	a	548

83	2,008,357	2,021,700	2,173,451	2,187,891	0.48	0.09	526

84	0	0	0	0	0.00	0.00	0

85	23,472	27,526	132,354	155,214	0.27	0.14	332

86	20,570	24,731	115,988	139,454	0.00	0.00	291

87	97,698	112,353	363,505	418,031	0.35	0.10	752

88	4,171	5,743	27,391	37,710	0.08	0.12	46

89	147,823	147,823	306,331	306,331	0.13	0.01	495

90	786,104	984,013	850,725	1,064,903	0.30	0.18	206

91	786,104	984,013	850,725	1,064,903	0.41	1.47	206

92	25,608	25,608	53,152	53,152	0.03	0.08	56

93	35,993	47,565	36,669	48,459	0.05	a	131

94	4,151,496	4,231,505	4,606,346	4,695,121	0.65	0.08	1,307

95	1,232	2,692	8,090	17,675	0.01	0.03	14

96	2,457	3,308	13,417	18,060	0.02	a	33

97	28,695	28,695	74,012	74,012	0.52	a	185

98	0	0	0	0	0.00	0.00	0

99	46,468	50,408	152,246	165,155	0.26	0.66	436

100	2,591,010	2,708,025	2,804,001	2,930,635	0.83	0.13	679

101	0	0	0	0	0.00	0.00	0

102	32,339	57,437	111,894	198,734	0.12	0.06	221

103	0	0	0	0	0.00	0.00	0

104	74,538	74,538	154,710	154,710	0.00	0.00	162

105	5,400	5,668	35,458	37,218	0.00	0.00	59

106	139,029	139,029	455,507	455,507	0.25	0.02	1,303

107	0	0	11,540	11,540	0.05	a	48

108	8,314	8,879	17,223	18,393	0.00	0.00	30

109	1,485,414	1,704,414	1,607,521	1,844,524	0.32	0.05	389

110	122,762	122,762	279,765	279,765	0.22	0.06	515

111	27,548	33,475	155,338	188,757	0.10	0.07	389

a The refinery did not have published information on the capacity of the
existing sulfur plant.

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  40  Tier 3
Refinery Energy, Hydrogen and Sulfur Plant Demand Increase

(10 ppm, ABT Cases)

Refinery Number	Demand Estimates

	Fuel Gas Demands	Sulfur Plant Production	Hydrogen

	Low Impact Cases (Million BTU/Yr)	High Impact Cases (Million BTU/Yr)
Sulfur Production Increase (Tons Sulfur/Day)	Sulfur Plant Capacity
Increase (Percent of Existing Facility)	Hydrogen Demand Increase
(million scf/day)

	Normal Volumes	Maximum Volumes	Normal Volumes	Maximum Volumes



	1	20,503	22,001	67,175	72,082	0.90	0.60	192

2	64,080	66,959	229,725	240,045	0.22	0.02	467

3	323,504	388,955	476,338	572,710	0.00	0.00	528

4	89,955	103,448	507,237	583,323	0.20	0.02	1,271

5	0	0	0	0	0.00	0.00	0

6	426,001	692,966	627,257	1,020,346	0.29	a	696

7	0	0	0	0	0.00	0.00	0

8	26,807	28,515	70,005	74,463	0.04	0.13	144

9	117,423	117,423	243,241	243,241	0.28	0.02	426

10	5,400	5,668	35,458	37,218	0.00	0.00	59

11	1,525,387	1,525,387	1,558,831	1,558,831	0.51	1.28	64

12	8,113	8,589	20,926	22,152	0.04	0.39	52

13	7,024	7,191	18,116	18,547	0.19	2.16	45

14	74,538	74,538	154,710	154,710	0.00	0.00	162

15	0	0	0	0	0.00	0.00	0

16	2,591,010	2,708,025	2,804,001	2,930,635	0.83	0.13	679

17	177,093	203,657	636,542	732,024	0.30	0.03	1,288

18	0	0	0	0	0.00	0.00	0

19	19,171	23,883	91,207	113,621	0.04	0.01	214

20	26,289	34,157	67,806	88,100	0.19	0.17	170

21	2,239,110	2,739,637	2,566,562	3,140,286	0.00	0.00	863

22	83,222	83,222	272,663	272,663	0.41	0.12	780

23	147,823	147,823	306,331	306,331	0.13	0.01	495

24	0	0	2,907	2,907	0.00	0.00	12

25	2,273,706	2,751,864	2,323,556	2,812,199	0.50	0.08	96

26	0	0	0	0	0.00	0.00	0

27	107,543	135,777	222,775	281,262	0.20	0.18	390

28	16,875	20,135	43,524	51,932	0.13	0.70	109

29	150,568	150,568	311,900	311,900	0.22	0.03	547

30	0	0	0	0	0.00	0.00	0

31	30,259	33,479	99,139	109,690	0.14	0.14	284

32	0	0	0	0	0.00	0.00	0

33	139,558	139,558	289,093	289,093	0.16	0.02	507

34	9,369	20,299	61,521	133,298	0.20	0.03	103

35	0	0	6,444	6,444	0.00	0.00	27

36	0	0	0	0	0.00	0.00	0

37	109,050	109,050	357,285	357,285	0.62	0.11	1,022

38	0	0	0	0	0.00	0.00	0

39	10,469	10,469	12,340	12,340	0.04	0.96	6

40	210,772	269,511	319,729	408,832	0.07	0.00	379

41	173,672	173,672	569,009	569,009	0.43	0.03	1,628

42	21,773	21,773	71,336	71,336	0.58	0.93	204

43	3,145,454	3,528,532	3,223,899	3,616,532	0.31	0.07	201

44	337,961	411,915	365,743	445,776	0.15	a	89

45	3,945,526	3,984,109	4,269,863	4,311,617	0.61	0.03	1,034

46	97,698	112,353	363,505	418,031	0.35	0.10	752

47	1,416,139	1,416,139	1,532,550	1,532,550	0.25	0.03	371

48	16,876	16,876	89,463	89,463	0.07	0.13	219

49	433,258	526,468	637,943	775,189	0.14	0.09	707

50	4,433	7,291	14,523	23,888	0.04	1.96	42

51	58,475	60,441	191,583	198,026	0.26	a	548

52	2,170,692	2,170,692	2,349,131	2,349,131	0.61	0.23	569

53	8,497	8,497	12,247	12,247	0.00	0.00	16

54	31,964	31,964	82,442	82,442	2.19	0.96	206

55	336,635	420,834	495,673	619,649	0.19	0.15	550

56	32,339	57,437	111,894	198,734	0.12	0.06	221

57	38,525	40,291	101,764	106,429	0.09	0.04	215

58	54,151	60,794	165,033	185,278	0.24	0.12	221

59	0	0	0	0	0.00	0.00	0

60	23,773	23,773	36,151	36,151	0.04	0.00	52

61	190,870	210,535	206,560	227,841	0.33	0.29	50

62	3,288	4,683	10,773	15,343	0.03	0.97	31

63	0	0	0	0	0.00	0.00	0

64	2,008,357	2,021,700	2,173,451	2,187,891	0.48	0.09	526

65	2,457	3,308	13,417	18,060	0.02	a	33

66	41,705	43,146	149,509	154,676	0.63	0.31	304

67	46,468	50,408	152,246	165,155	0.26	0.66	436

68	1,079,231	1,299,822	1,167,948	1,406,672	0.09	0.14	283

69	3,071	3,325	7,294	7,896	0.00	0.00	17

70	4,123,377	4,123,377	4,213,781	4,213,781	0.38	0.03	174

71	786,104	984,013	850,725	1,064,903	0.30	0.18	206

72	37,188	37,188	121,841	121,841	1.31	0.44	349

73	0	0	0	0	0.00	0.00	0

74	40,155	51,299	108,185	138,209	0.11	0.09	236

75	1,485,414	1,704,414	1,607,521	1,844,524	0.32	0.05	389

76	122,762	122,762	279,765	279,765	0.22	0.06	515

77	102,473	124,044	264,305	319,941	0.46	a	662

78	0	0	0	0	0.00	0.00	0

79	3,609,133	4,265,613	3,905,818	4,616,262	0.66	0.22	946

80	43,909	46,106	143,861	151,058	0.37	0.08	412

81	1,693,620	1,948,219	1,832,841	2,108,370	0.27	0.03	444

82	0	0	0	0	0.00	0.00	0

83	60,923	69,776	199,605	228,609	0.10	0.03	571

84	4,171	5,743	27,391	37,710	0.08	0.12	46

85	4,151,496	4,231,505	4,606,346	4,695,121	0.65	0.08	1,307

86	8,314	8,879	17,223	18,393	0.00	0.00	30

87	5,175	5,175	29,180	29,180	0.13	3.30	73

88	96,113	113,326	98,454	116,085	0.13	0.17	543

89	988,379	1,480,617	1,109,185	1,661,587	0.31	0.26	412

90	0	0	0	0	0.00	0.00	0

91	9,233	9,655	30,251	31,632	0.04	1.96	87

92	21,918	41,206	45,403	85,358	0.02	0.01	80

93	14,886	16,690	38,395	43,048	0.01	0.02	96

94	35,993	47,565	36,669	48,459	0.05	a	131

95	202,851	261,880	506,929	654,442	0.15	0.01	1,146

96	786,104	984,013	850,725	1,064,903	0.41	1.47	206

97	25,924	29,812	146,179	168,106	0.10	0.08	366

98	1,232	2,692	8,090	17,675	0.01	0.03	14

99	369,152	432,912	543,551	637,434	0.30	0.15	603

100	0	0	0	0	0.00	0.00	0

101	28,695	28,695	74,012	74,012	0.52	a	185

102	828,747	828,747	965,965	965,965	0.13	0.11	412

103	0	0	11,540	11,540	0.06	a	48

104	13,984	16,250	78,855	91,630	0.09	0.05	198

105	62,002	62,002	159,918	159,918	0.35	3.46	400

106	260,661	391,286	383,805	576,142	0.17	0.25	426

107	194,048	194,048	294,359	294,359	0.07	0.00	349

108	0	0	0	0	0.00	0.00	0

109	428,992	479,023	481,425	537,572	0.08	0.27	179

110	52,160	62,683	108,050	129,848	0.10	0.03	189

111	0	0	0	0	0.00	0.00	0

a The refinery did not have published information on the capacity of the
existing sulfur plant.



Chapter 5 Appendix

LP Refinery Modeling Output Tables

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  41  Volume and
Cost Information Used for Estimating the Cost of Octane

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  42  PADD 1 Unit
Capacity and Throughput Volumes from LP Refinery Modeling (Thousand
bbl/day)

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  43  PADD 2 Unit
Capacity and Throughput Volumes from LP Refinery Modeling (Thousand
bbl/day)

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  44  PADD 3 Unit
Capacity and Throughput Volumes from LP Refinery Modeling (Thousand
bbl/day)

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  45  PADD 4 and
PADD 5OC Unit Capacity and Throughput Volumes from LP Refinery Modeling
(Thousand bbl/day)

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  46  U.S. (except
CA) Unit Capacity and Throughput Volumes from LP Refinery Modeling
(Thousand bbl/day)

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  47  PADD 1
Gasoline Qualities Estimated by LP Refinery Modeling

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  48  PADD 2
Gasoline Qualities Estimated by LP Refinery Modeling

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  49  PADD 3
Gasoline Qualities Estimated by LP Refinery Modeling

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  50  PADD 4 and 5OC
Gasoline Qualities Estimated by LP Refinery Modeling

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  51  U.S. (except
CA) Gasoline Qualities Estimated by Refinery Modeling

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  52  PADD 1 Unit
Capacity and Throughput Volumes from LP Refinery Modeling (Thousand
bbl/day)

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  53  PADD 2 Unit
Capacity and Throughput Volumes from LP Refinery Modeling (Thousand
bbl/day)

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  54  PADD 3 Unit
Capacity and Throughput Volumes from LP Refinery Modeling (Thousand
bbl/day)

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  55  PADD 4 and 5OC
Unit Capacity and Throughput Volumes from LP Refinery Modeling (Thousand
bbl/day)

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  56  U.S. (except
CA) Unit Capacity and Throughput Volumes from LP Refinery Modeling
(Thousand bbl/day)

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  57  PADD 1
Gasoline Qualities for the E10 and E15 Cases relative to the Reference
Case

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  58  PADD 2
Gasoline Qualities for the E10 and E15 Cases relative to the Reference
Case

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  59  PADD 3
Gasoline Qualities for the E10 and E15 Cases relative to the Reference
Case

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  60  PADD 4 and 5OC
Gasoline Qualities for the E10 and E15 Cases relative to the Reference
Case

Table   STYLEREF 1 \s  5 -  SEQ Table \* ARABIC \s 1  61  U.S. (except
CA) Gasoline Qualities for the E10 and E15 Cases relative to the
Reference Case

References

  Normally we conduct the refinery modeling assuming an after-tax 15%
ROI and adjust the costs to reflect a before-tax 7% ROI to report the
costs.  However, in this case because the new capital investments were
so minimal, we omitted the capital cost amortization adjustment because
its effect on costs was judged to be negligible.  

 Since we completed the LP refinery modeling to estimate the cost for
recovering the lost octane and the associated changes in gasoline
quality, we found that other Tier 3 refinery modeling studies did not
show the same increase in aromatics and decrease in E300 (see also
7.1.3.2).  We then discovered that the LP refinery model that we have
licensed to use required some improvements in how the refinery model was
characterizing both the light-cut and the heavy-cut naphtha from the
reformer streams to more accurately estimate the E300 and aromatics
content of these streams.  We have subsequently worked with a contractor
to make these improvements to the LP refinery model and will reassess
the changes in gasoline quality for the final rule analysis.  Thus,
while our modeling results shown in Table 5.3 show a meaningful impact
on aromatics and E300, we believe there will in fact be little or no
change.  Note that these improvements are not expected to have any
impact on the cost estimates made by the refinery model.   

 Since we conducted the cost analysis for the proposed rulemaking, we
put in place an additional round of greenhouse emission reductions (2017
– 2025) for light duty cars and trucks that will reduce future
gasoline demand.  When we model the costs for the final rulemaking, we
will incorporate this reduction in gasoline demand in our costs
analyses.  

 We did not account for any undercutting of the heavy FCC naphtha into
jet and diesel fuel, nor did we account for the removal of any pentanes
that might be occurring in refineries to comply with stringent
summertime RVP standards, therefore our analysis is likely somewhat
conservative and overestimates the costs. 

 Naphtha Splitter towers separate the naphtha feed stream into a light
and heavy streams, whereby the heavy stream is typically  reformer
feedstock, while the light stream is blend stock lighter than reformer
feed. 

 After we completed our cost analysis, we met with UOP staff, including
those who market their Merox technology for removing mercaptans from
gasoline streams.  The UOP staff said that pretty much all butanes are
already being treated by Merox (or similar) extraction units.  Thus,
there would be no additional cost for treating butanes for complying
with Tier 3.  We will update our cost analysis to reflect this for the
final rule cost analysis. 

 In our Tier 3 cost analysis, we inadvertently only accounted for the
ethanol volumes that are contained in the 2009 RFG database, not those
mandated by RFS 2.  For our final rulemaking cost analyiss, we will
adjust our model to account for higher ethanol use in the future, which
will slightly reduce our Tier 3 cost estimates, due to the dilution
effects of ethanol.  

 No refinery compliance margins were included in this analysis.

 Based on Tier 2 sulfur compliance data, of the 26 companies that
purchased sulfur credits in 2010, eight bought credits only from their
own company, and another five bought credits both from their own company
and from other companies. The remainder bought credits only from other
companies.

  Approximately 55% of this desululfurization cost is comprised of the
variable cost, and the majority of that is due to the cost of recovering
the octane lost when the hydrotreater unit saturates the octane-rich
olefins contained in the FCC naphtha.  Another 30% of the cost is due to
the capital cost amortized over the volume of gasoline.  Finally, about
15% of the cost is due to the fixed operating cost which includes the
maintenance of the new equipment and taxes.  

  The installed capital cost for an FCC postreater from the Jacobs data
base was adjusted to current year dollars.  This estimated installed
capital cost is several years old and may not represent Jacobs current
cost estimate for a FCC postreater. 

Chapter 1 

*** E.O. 12866 Review – Revised Version – Do Not Cite, Quote, or
Release During Review ***

Page   PAGE  2 

  PAGE   \* MERGEFORMAT  5-1 

 Annual Energy Outlook 2006, Energy Information Administration,
Department of Energy

 Worldwide Report, Oil and Gas Journal.    HYPERLINK
"http://www.ogjonline.com"  www.ogjonline.com , December 22, 2005.

 Petroleum Supply Annual 2004, Volume 1, Table 38.  Energy Information
Administration, Department of Energy.

 Petroleum Supply Annual 2004, Volume 1, Tables 22 – 25.  Energy
Information Administration, Department of Energy.

 Petroleum Supply Annual 2005, Volume 1, Table 17.  Energy Information
Administration, Department of Energy.

 Petroleum Refining Technology and Economics 4th Edition.  James Gary
and Glenn Handerwork.  Page 65, Table 4.7

 The Challenges & Opportunities of 10 ppm Sulfur Gasoline.  Jay Ross,
Delphine Largeteau,  Marc Laborde, Larry Wisdom.  NPRA Annual Meeting,
AM-11-57, March, 2011, page 6

 The Benefits of Cat Feed Hydrotreating and the Impact of Feed Nitrogen
on Catalyst Stability.  Brian Moyse at Haldor Topso.  NPRA Annual
Meeting, AM-10-167, March 2010, Page 2

 Handbook of Petroleum Refining Processes, Third Edition.  Robert
Meyers, page 11.39

 Refining Economics of a National Low Sulfur, Low RVP Gasoline Standard,
Performed for The International Council for Clean Transportation by
Mathpro, October 25, 2011.

 Refining Economics of a National Clean Gasoline Standard for PADDs 1-3;
for The Alliance of Automobile Manufacturers by Mathpro, June 27, 2008. 


 Potential Supply and Cost Impacts of Lower Sulfur, Lower RVP Gasoline;
prepared for The American Petroleum Institute by Baker and O’Brien;
July 2011.

 Schink, George R., Singer, Hal J., Economic Analysis of the
Implications of Implementing EPA’s Tier 3 Rules, prepared for the
Emissions Control Technology Association, June 14, 2012.

 An Assessment of the Impact of Nonroad Diesel Fuel Sulfur Regulation on
Distillate Fuel Production and Availability in the U.S., prepared for
the American Petroleum Institute by Baker and O’Brien, July 2003.

 Petroleum Supply Annual 2003 – 2010; Table 24,  Imports of Crude Oil
and Petroleum Products, Energy Information Administration.  

 Petroleum Supply Annual 2003 – 2010; Table 31, Exports of Crude Oil
and Petroleum Products by PAD Destination.

 Petroleum Supply Annual 2011, Refinery Capacity Report, Table 13
Refineries Permanently Shutdown by PAD District Between January 1, 1990
and January 1, 2010; Energy Information Administration.

 Petroleum Supply Annual 2003 – 2011, Refinery Capacity Report, Table
11 New, Shutdown and Reactivated Refineries During 2010; Energy
Information Administration.

 Petroleum and Other Liquids; Number and Capacity of Petroleum
Refineries, Data, Total Number of Operable Refineries and Total Number
of Operating Refineries, 1990 – 2010; Energy Information
Administration.

 Gary, J., & Handewerk, G. (2001). Petroleum Refining: Technology and
Economics. (4th ed.). CRC Press. p 256.

 Gary, J., & Handewerk, G. (2001). Petroleum Refining: Technology and
Economics. (4th ed.). CRC Press. p 337.

 Keller, P. (2012, January). New Source Review Permitting Impact
Analysis for Proposed Tier 3 Gasoline Program. Memorandum to the docket.

