M E M O R A N D U M

Date:		December 29, 2006

To:		Lester Wyborny, U.S. Environmental Protection Agency

From:		Jeff Kolb, Abt Associates Inc.

Subject:	Procedures for Calibrating the Benzene Spreadsheet Model

			[WA 0-01, EPA Contract EP-C-06-094]

			

This memorandum describes some of the procedures used to calibrate the
benzene spreadsheet model to the 2004 and 2012 baselines.  

The baseline for 2004 is defined by process capacity reported by
refineries in the OGJ survey for that year (supplemented in some
instances by data from DOE) and by the gasoline volumes and properties
reported by refineries to EPA, averaged across the year.  

The baseline for 2012 is defined by (1): refinery-level process capacity
for 2004, adjusted to account for PADD-level changes in process capacity
utilization, as yielded by refinery modeling; and (2) gasoline volumes
and properties projected by adjusting the 2004 reported volumes and
properties by projected PADD-level increases in total gasoline
out-turns,

Procedures for Calibrating Gasoline Out-turns

The first step in calibrating the benzene spreadsheet model was to
calibrate gasoline volumes for 2004, i.e., to match closely (in most
instances) modeled gasoline out-turns with reported gasoline volumes for
2004.  A number of procedures were used to accomplish this.

Refinery utilization of all process capacity was adjusted, generally
downward, but sometimes slightly upward.

Purchases of natural gasoline were set on a refinery-by-refinery basis
depending on modeled gasoline volume and isomerization capacity
utilization (taking account of aggreagate PADD-level reported
purchases).

PADD-level sales of naphtha to the petrochemical industry (assumed to be
light naphtha) were allocated to each refinery based on its share of
crude distillation capacity, except for refineries that otherwise would
be “short” of target gasoline volume.

Ethanol use in 2004 was determined by gasoline property data reported to
EPA.  

Straight run naphtha (C5 to 325°) volumes were increased or decreased
to take account of modeled versus reported gasoline out-turns, modeled
reformer capacity utilization, and modeled versus reported aromatics
content of gasoline.

Reformer heavy naphtha feed (infrequently) was reduced or supplemented
to account for reformer capacity utilization, modeled gasoline
out-turns, and modeled gasoline aromatics levels.

In a few instances, reported gasoline volumes were out-of-line with
gasoline volumes that could be produced with reported refining process
capacity (either substantially higher or lower than a refinery’s
gasoline production capability).  In such cases, we allowed the model to
“over- or under- produce” gasoline, subject to PADD-level aggregate
gasoline production matching reported PADD-level gasoline out-turns.  We
presume that this reflects the sale of gasoline blendstocks among the
refineries.

Calibrating the model for 2012 required that we first: (1) estimate
target gasoline volumes for each refinery; and (2) develop estimates of
PADD-level process capacity utilization.

We estimated refinery-level gasoline out-turns for 2012 by increasing
2004 reported gasoline volumes by the projected percentage growth in
PADD-level gasoline out-turns.  We developed estimates of PADD-level
process capacity utilization through refinery modeling. We then
calculated the ratios of PADD-level process capacity use in 2012 to that
in 2004 and proportionately adjusted process capacities for individual
refineries.

The calibration assumptions made for 2004 generally were applied
directly to 2012.  However, in some instances, we modified the general
capacity utilization factor to match modeled with projected gasoline
volumes. We also reduced the percentage of heavy naphtha feed available
for reforming for PADD 4 based on refinery modeling results.  We
allocated ethanol to individual refineries such that all RFG was blended
with ethanol at 10% and CG was blended with ethanol according to the
PADD average for CG. 

Procedures for Calibrating Baseline Benzene Controls

We used different calibration procedures for establishing baseline
benzene control for 2004 depending on whether an individual refinery:
(1) produced aromatics for the petrochemical industry; (2) did not have
Pen/Hex isomerization capacity; or (3) had Pen/Hex isomerization
capacity.

Aromatics refineries.  We set the percentage of benzene extracted from
reformate so as to match modeled and reported benzene levels in
gasoline.  However, if this percentage exceeded 90%, we assumed that all
benzene was extracted from reformate.  In this latter case (or when
benzene extraction at 100% did not bring modeled benzene levels down to
reported levels), we raised the benzene levels assumed for other
blendstocks (naphthas, hydrocrackate, and FCC naphtha) such that modeled
and reported benzene levels were the same.

Refineries without Pen/Hex units.  We set the percentage of C6-rich
light naphtha blended to gasoline (rather than being reformed) to match
modeled with reported benzene levels in gasoline.  However, if this
percentage exceeded 85%, we set it at 100% and made compensating
increases in the assumed benzene levels of other gasoline blendstocks. 
If modeled gasoline benzene levels, with 100% of C6-rich light naphtha
blended to gasoline, were higher than reported levels, we reduced the
assumed benzene levels of other gasoline blendstocks.  Finally, if
modeled gasoline benzene levels, with all C6-rich light naphtha being
reformed, were lower than reported levels, we increased the assumed
benzene content of other gasoline blendstocks.

Refineries with Pen/Hex units.  For the most part, we assumed that any
C6-rich light naphtha cut diverted from reforming would be pooled with
other naphthas that were candidates for isomerization, i.e., the
percentage of diverted C6-rich naphtha isomerized was the same as the
percentage of all naphthas isomerized.  The fraction of C6-rich light
naphtha diverted from reforming was then set so that modeled and
reported gasoline benzene levels matched.

We generally used the same set of benzene controls for the 2012 baseline
calibration that we developed for the 2004 calibration.

Calibration Procedures for Benzene Spreadsheet Model

 PAGE   

 PAGE   2 

December 29, 2006		Abt Associates Inc.

 																			         				     Abt Associates Inc.

