TO:	EPA Docket No. EPA-HQ-OAR-2003-0146

		

FROM:	Bob Lucas, EPA/SPPD 

DATE:	August 20, 2007

SUBJECT:	Collection of Detailed Benzene Emissions Data from 22 Petroleum
Refineries

I.	Purpose

This memorandum documents the detailed benzene emissions data collected
to support the residual risk assessment for petroleum refineries. 

II.	Background

Section 112(f) of the Clean Air Act Amendments (CAA) directs the U.S.
Environmental Protection Agency (EPA) to assess source categories
regulated under Section 112(d) of the CAA and determine whether any
human health or environmental risks remain from the continued emissions
of hazardous air pollutants (HAPs) following implementation of maximum
achievable control technology (MACT) standards.  The CAA further states
that if the MACT standards do not reduce lifetime excess cancer risk to
the most exposed individual to less than one in one million, EPA must
set additional standards to protect human health and the environment, in
accordance with the interpretation set forth in the Benzene NESHAP. 
Additionally, the EPA is required to review these technology-based
standards and to revise them “as necessary (taking into account
developments in practices, processes and control technologies)” no
less frequently than every 8 years, under CAA section 112(d)(6).  The
Refinery MACT 1 (40 CFR part 63, subpart CC) was promulgated over 8
years ago and is now being reviewed.   

In 2002, EPA conducted a preliminary, screening risk assessment using a
refinery emission model (REM) and model plant configurations (RTI,
2002).  During this risk assessment, benzene dominated the inhalation
cancer risk from Refinery MACT 1 emission sources.  Based on this
preliminary risk assessment, a targeted data collection effort was
initiated to obtain detailed emissions and source characterization data
for Refinery MACT 1 sources of benzene.  Through the efforts of American
Petroleum Institute (API) and the National Petrochemicals and Refiners
Association (NPRA), twenty-two petroleum refineries voluntarily
submitted detailed emissions data for the 2004 calendar year.  This data
collection effort, along with the subsequent data analysis, is referred
to as “the 22 Refinery Study.”  One of the refineries in this study
(at the time Motiva – Norco refinery; now Shell – Norco) has been
divided by into two plants:  one containing refinery operations and one
containing chemical production operations.  The refinery and chemical
plant are integrated and there is no physical boundary between many of
the process units.  While the overall refinery and petrochemical complex
was treated as a single entity, these facilities are permitted
separately.  Consequently, this study has also been referred to as
“the 23 Refinery Study.”

III.	Selection of the 22 Refineries

The results of the screening risk assessment were used to identify the
facilities targeted for additional data gathering.  First, the 10
highest risk facilities from the screening survey were selected for the
study.  The remaining 12 facilities were selected to provide a
representative sampling of refineries.  The refineries were generally
classified as small, medium, and large, and the initial risk results
were classified as low, midrange, and high.  As the 10 highest risk
facilities were already selected, the remaining 12 were predominately
selected from the midrange and low risk groups.  Additionally, the
refineries were selected so that any one corporate entity did not
complete more than two inventories.

IV.	Summary of the Types of Data Collected

The 22 facility study specifically focused on improving the emission
estimates for benzene; no other HAP emissions data were collected. 
Spreadsheet templates were developed to ensure the appropriate
information and data were collected.  Separate spreadsheets for each of
the primary Refinery MACT 1 emission sources were developed:  equipment
leaks, wastewater, storage tanks, miscellaneous process vents, gasoline
loading racks, marine vessel loading, and cooling towers.  An additional
spreadsheet, labeled “Other,” was used to collect data from other
miscellaneous emission sources, such as flares, the FCCU vent, etc. 
Detailed data were requested regarding the emission release
characteristics (emission point-specific latitudes and longitudes,
diameter and velocity of stack emissions or area of fugitive releases,
and release height).

A minimal amount of process information was requested to help
characterize the process, such as wastewater flow rate and cooling water
recirculation rate.  General information was also requested regarding
the emission estimation methodology used in developing the reported
benzene emissions.  For example, a common response for the tank emission
estimates is that the refinery used TANKS software; some facilities
stated that they used the speciation profile TANKS default while other
used site-specific composition data.  While this information is not
sufficient to reproduce the calculations from the data provided, these
data often provided insight into the refinery’s management practices
and emission estimation methodologies.  For example, the respondents
typically reported the leak definitions that were used in their
equipment leak detection and repair (LDAR) program.  Also, some
refineries used component counts and default component leak rates, which
would only provide a crude estimate of their equipment leak emissions. 
Other refineries used component-specific leak readings and the screening
value correlation for that component’s reading; these values were then
summed to estimate equipment leak emissions for the plant.  This
detailed approach is expected to provide a more accurate emission
estimate than the default factors.  The spreadsheets provided by the
respondents are available in EPA Docket No. EPA-HQ-OAR-2003-0146.

V.	Summary Results of Data Review

The benzene emission subtotals of each of the primary emission sources
for each of the 22 refineries were compiled from the questionnaire
responses (these data are provided in Table A-1 of Attachment A).  The
same emissions information, to the extent available, was also compiled
for these refineries from the 2002 National Emissions Inventory (NEI),
the 2004 Toxics Release Inventory (TRI), and the 2004 Texas Commission
on Environmental Quality (TCEQ) emissions inventory.  A summary of this
comparison is presented in Figure 1.  The annual benzene emissions were
normalized by the annual crude throughput of the refineries so that the
data could be more easily compared.  Comparing the 22 Refinery study
residual risk survey (RR survey) results with the 2002 NEI reveals that:
 the survey emissions are higher than the NEI for eight of the 22
refineries; the estimates effectively match for nine of the 22
refineries; and the survey emissions are less than the NEI for five of
the 22 refineries.  Comparing the 22 Refinery study residual risk survey
(RR survey) results with the 2004 TRI reveals that:  the survey
emissions are higher than the TRI for four of the 22 refineries; the
estimates effectively match for 13 of the 22 refineries; and the survey
emissions are less than the TRI for five of the 22 refineries.

Figure 1.  Comparison of refinery-wide benzene emission estimates.

 

Figure 1 also presents the Canadian average refinery emission factor and
revised REM emission estimates.  The only revisions made to the original
REM emission estimation methodology were to revise the equipment leak
emission estimates to account for refineries with lower leak definitions
and to use a “controlled” cooling tower emission factor as the
default.  The remainder of this section provides overall observations
regarding specific emission sources, presents additional detail
regarding the revisions made to the REM, and discusses other potential
improvements to the REM.  Once the REM is tuned and validated, the model
can be used to assess the variability and uncertainty in the emission
estimates.

Equipment Leaks

One general concern regarding the equipment leak emission estimates is
the proper application of Method 21 in doing the component leak
screening.  Moving the probe too quickly around the component or holding
the probe too far from the potential leak areas (e.g., valve shaft seal)
can significantly reduce the number of leaking components detected. 
Method 21 monitoring data collected by the EPA’s National Enforcement
Investigation Center (NEIC) showed the number of leaking components was
generally underestimated by the refinery’s monitoring efforts (EPA,
1999a), often times by a factor of two or more.  

In preparing impacts for the review of the New Source Performance
Standards (NSPS) for equipment leaks (40 CFR part 60, subparts VV and
GGG), the monitoring data compiled by Taback (1997) were used with the
screening value correlations (EPA, 1995b) to develop more accurate
emission factors for refineries with effective LDAR programs at leak
definitions of 500/1,000 parts per million by volume (ppmv) (for valves,
pumps, and flanges).  In the NSPS analysis, a small correction factor
was used to scale the emissions to account for this discrepancy; the
correction factor used was the 10 percentile ratio of the NEIC measured
percent leakers to the refinery measured percent leakers (that is, 90%
of the refineries tested had by NEIC had a higher ratio than the one
used).  However, for the purposes of the residual risk analysis, a more
central tendency or upper-end value for this correction should be used. 
Therefore, new equipment leak emission factors were generated for the
REM using the Taback data, the median NEIC correction factor, and the
screening value correlations (EPA, 1995b).

Revised equipment leak benzene emission factors were developed for three
categories of LDAR programs:  (1) high leak definitions (10,000 ppmv);
(2) low leak definitions (500 or 1,000 ppmv) but no direct monitoring of
connectors; and (3) low leak definition that includes monitoring
connectors.  The revised emission factors assume that there are no leaks
at the start of the monitoring period (all leaking components are
repaired), and estimates the average leak rate between monitoring
periods using one-half the number components monitored above the repair
threshold at the end of the monitoring period.  The revised emission
factor approach will underestimate emissions from valves on the delay of
repair list (as these valves are leaking during the entire period
between monitoring events rather than half the time), but provides a
better average emission rate over the year than using the
end-of-monitoring period results directly.  Each refinery was assigned
to one of the three LDAR program categories based on their LDAR program
requirements.  

Table 1 shows the comparison of benzene equipment leak emission
estimates as reported by the survey respondents, as projected using the
original REM model, and as estimated using the revised equipment leak
emission factors.  The revised emission factor approach compares well
with (and generally slightly exceeds) the emissions reported by the
refineries except for the refineries in Kansas, New Mexico, and two of
the three refineries in both Louisiana and Texas.  The high emissions
reported for the two Texas refineries appear to be caused by the use of
the Texas Commission on Environmental Quality (TCEQ) emission factors,
which are based on average component leak rates and assumed LDAR control
efficiency rather than refinery-specific screening values.  For the
Kansas refineries, the revised emission factors underestimate the
reported emissions by approximately a factor of 2.  This might be caused
because the refineries used the monitored screening values directly for
estimating the equipment leak emissions (rather than “averaging” the
emissions between the monitoring events). 

Table 1.  Comparison of Equipment Leak Emission Estimates

RTI ID	Facility Name	Benzene Emissions (tpy)



Survey Response	Original REM	Revised Emission Factors

15	Chevron USA Inc. -- El Segundo, CA	0.112	6.9	0.23

20	ExxonMobil Corp. -- Torrance, CA	0.133	6.0	0.19

30.1	ConocoPhillips -- Rodeo, CA	0.173	4.7	0.24

31	Tesoro -- Martinez, CA	0.038	7.7	0.26

35	Suncor Energy -- Commerce City, CO	0.007	4.3	0.26

51	National Cooperative Refinery Association -- McPherson, KS	2.472	6.1
0.94

52	Frontier Oil Corp. -- El Dorado, KS	3.188	9.2	1.2

64	ExxonMobil Corp. -- Chalmette, LA	7.031	10.3	1.4

66	Motiva Enterprises -- Convent, LA	0.145	7.0	0.24

67	Motiva Enterprises -- Norco, LA	10.202	8.6	0.60

88	Sunoco, Inc. -- Westville, NJ	0.099	7.4	0.49

92	Giant Refining Co. -- Gallup, NM	4.93	4.4	0.60

93	Navajo Refining Co. -- Artesia, NM	2.643	5.6	0.38

96	Marathon Petroleum Company LLC -- Canton, OH	0.093	5.8	0.39

100	Gary-Williams Energy Corp. -- Wynnewood, OK	1.15	6.2	0.72

105	Sunoco, Inc. -- Marcus Hook, PA	0.731	6.0	0.91

113	BP -- Texas City, TX	34.156	11.0	1.5

116	Valero Energy Corp. -- Corpus Christi, TX (East)	10.123	11.0	0.74

133	Valero Energy Corp. -- Houston, TX	0.02812	4.9	0.35

145	ConocoPhillips -- Ferndale, WA	0.079	5.8	0.75

146	US Oil & Refining Co. -- Tacoma, WA	0.62	5.3	0.74

151	Sinclair Oil Corp. -- Sinclair, WY	0.164	4.3	0.47



While some of the high reported equipment leak emissions may be based on
the methodologies used, others are more difficult to reconcile.  The
Motiva Norco refinery (#67) includes emissions from the Shell chemical
plant processes, which explains some of the high emission estimate. 
However, some of this refinery’s “high” leak emissions result from
an intentional effort to account for emissions from unmonitored
components; these emissions dominated the refinery’s equipment leak
emission estimate.  The question is:  Are the emissions from unmonitored
components overestimated by this refinery or underestimated by other
refineries?  Furthermore, Giant Refining (#92) provided one of the more
detailed explanations of the methodology used to calculate the equipment
leak emissions (using direct measurement of benzene concentrations in
the various process streams, providing direct table references rather
than just referring to the Protocol document, and providing overall
percent leaking components).  Even though this refinery uses a leak
definition of 10,000 ppmv, it is difficult to understand why this
emission estimate is so much greater than the revised emission factors
estimate.  As the revised emission factors were based on monitoring data
from California refineries with low leak definitions, it could be that
the methodology used to extrapolate the leak profiles for refineries
with a leak definition of 10,000 ppmv underestimates the emissions for
refineries using the 10,000 ppmv leak definition.  If this is true, then
the emission reductions achieved when implementing a lower leak
definition may be larger than projected in the NSPS equipment leak
analysis.

Storage Tanks

The REM model storage tank emissions were developed from a limited
number of Louisiana refineries.  The model appears to do a reasonable
job of estimating storage tank emissions over a wide size-range of
refineries.  Using direct throughput correlation, the REM model may be
expected to slightly underestimate storage tank emissions for small
throughput refineries and overestimate emissions from larger refineries.
 However, without tank-specific input data to directly use TANKS (EPA,
1999b), the REM modeling approach appears to provide reasonable emission
estimates from storage tanks.

Wastewater Collection and Treatment

The REM model used wastewater generation factors and benzene
concentrations reported by EPA (1998a) to estimate the total wastewater
benzene load for each refinery.  These factors were also used to
estimate the refineries’ total annual benzene (TAB).  A simple
correlation was developed to estimate the “control efficiency” of
the Benzene Waste Operations NESHAP (BWON) based on initial TAB reports
when the BWON was first promulgated.  Based on the emission estimates
provided by the residual risk survey respondents, it appears that this
correlation consistently overestimates the benzene emissions from
wastewater collection and treatment systems.  First, the TAB estimates
developed using the EPA wastewater generation factors and benzene
concentrations were compared with the TAB reported by the refineries. 
Table 2 presents the comparison of the reported and predicted TAB for
the 22 refineries.  There are significant differences (both high and
low) between the reported and predicted TAB.  Based on the data
presented in Table 2, it appears that there is significant variability
among refineries regarding their wastewater benzene loads.  This may be
caused by differences in the concentration of benzene in the crude oil
or by differences in the process operations.  However, it is unlikely
that this accounts for the factor of 10 differences observed for some
refineries.  Data from the individual TAB reports submitted by the
refineries indicate that waste high in oil content is responsible for
the high TAB values.  Remediation waste must also be included in the
refinery’s TAB, but these are not (and cannot be) included in the
predicted TAB.  It is possible that wastes such as these are
contributing to some of the TAB values. 

Table 2.  Comparison of Predicted and Measured Total Annual Benzene
(TAB)

RTI ID	Facility Name	Benzene Quantities (Mg/yr)



Reported 2004 TAB	Predicted TAB	Predicted Total WW loada

15	Chevron USA Inc. -- El Segundo, CA	755	82.6	91.1

20	ExxonMobil Corp. -- Torrance, CA	284	54.4	58.7

30.1	ConocoPhillips -- Rodeo, CA	6.3	31.5	34.0

31	Tesoro -- Martinez, CA	64.7	61.2	66.0

35	Suncor Energy -- Commerce City, CO	5.8	18.9	19.8

51	National Cooperative Refinery Association -- McPherson, KS	6.6	26.6
29.9

52	Frontier Oil Corp. -- El Dorado, KS	73.4	32.6	36.8

64	ExxonMobil Corp. -- Chalmette, LA	605b	62.4	68.3

66	Motiva Enterprises -- Convent, LA	275	56.0	62.7

67	Motiva Enterprises -- Norco, LA	600	65.1	69.5

88	Sunoco, Inc. -- Westville, NJ	468	35.2	38.8

92	Giant Refining Co. -- Gallup, NM	3.1	5.1	5.6

93	Navajo Refining Co. -- Artesia, NM	4.9	19.9	22.0

96	Marathon Petroleum Company LLC -- Canton, OH	33c	23.8	25.5

100	Gary-Williams Energy Corp. -- Wynnewood, OK	<10d	18.1	18.9

105	Sunoco, Inc. -- Marcus Hook, PA	131	38.1	41.9

113	BP -- Texas City, TX	>10e	159	174

116	Valero Energy Corp. -- Corpus Christi, TX (East)	14.6	39.5f	43.6f

133	Valero Energy Corp. -- Houston, TX	8.6	29.5	30.9

145	ConocoPhillips -- Ferndale, WA	54.3	21.8	24.3

146	US Oil & Refining Co. -- Tacoma, WA	5.4	11.2	11.8

151	Sinclair Oil Corp. -- Sinclair, WY	24.4	19.2	20.4

aThe wastewater (WW) load includes streams less than 10 ppmw benzene
that are not included in the TAB.

bTAB not provided with initial response; reported TAB is for 2006.

cTAB total not provided; based on 6 month TAB totals of 16.5 Mg/yr,
assumed 2004 TAB to be 33 Mg/yr.

dTAB not provided; refinery only indicated that their TAB was less than
10 Mg/yr.

eTAB not provided; refinery only indicated that their TAB was greater
than 10 Mg/yr.

fValero purchased the Coastal – Corpus Christi refinery and now
operates 2 refineries in Corpus Christi; EIA (2006) combines the
capacity for these two (“East” and “West”) refineries.  Assumed
the predicted TAB for the east refinery was one-half the TAB for the
combined refinery.

Most of the emission estimates provided by the respondents used AP-42
emission factors (EPA, 1995a) or Water9 modeling (EPA, 1994; EPA, 2001).
 The reported emissions data were always less than the simple
correlation originally used to simulate BWON control efficiency.  While
many refinery wastewater emission estimates that appear to neglect
potential emission sources (e.g., assuming the activated sludge tank has
zero benzene emissions), the low emissions for some refineries appears
remarkable.  For example, Chevron – El Segundo refinery has a TAB of
755 Mg/yr (830 tpy), but reported wastewater emissions of only 0.06 tpy
for an apparent control efficiency of greater the 99.99 percent.  This
level of control efficiency seems unlikely.  Nonetheless, considering
all of the reported data, an improved correlation or methodology appears
to be needed to more accurately assess the emissions from a refinery
subject to the control requirements of the BWON rule.  

Cooling Towers

Cooling tower emissions were assumed to be “uncontrolled” (i.e.,
unmonitored) in the original REM.  Based on the responses received,
nearly all refineries perform at least periodic monitoring of chemical
use rates, oxidation/reduction potential, or water sampling and analysis
which provide the operators an indication or direct evidence of a
hydrocarbon leak into the cooling tower water as part of their standard
operating procedures.  Therefore, nearly every refinery used of the
“controlled” emission factors from AP-42 (EPA, 1995a) to estimate
their cooling tower emissions; a few refineries based their cooling
tower emission estimate on direct cooling water sampling and analysis
results.  There has been some controversy over the use of the controlled
emission factors when using surrogate monitoring techniques (such as
monitoring chlorine use rates) because of concerns regarding the ability
of these techniques to detect small leaks (TCEQ, 2006; Corsi, 2004). 
Nonetheless, based on the control cost analysis conducted for cooling
towers (Lucas, 2007), it appears that refineries have an economic
incentive to find and repair leaks (provided it does not interfere with
process operations).  Therefore, the REM cooling tower emission
estimates have been revised to use the controlled AP-42 emission factor
for cooling towers.    

Combustion Sources

In the original REM model, combustion sources (flares, heaters, boilers,
catalyst regenerators) provided only a minor contribution to the
refinery’s benzene emissions inventory.  However, with significant
reductions in the emissions from other sources, combustion sources may
play a larger role in the overall emission inventory.  Furthermore, many
refineries, especially smaller refineries, do not include combustion
sources in their inventory.  For the refineries that do include
combustion sources, the emission factors used by the refineries are the
same as those used in the REM.  Given the general agreement with the
reported and modeled emission estimates, no changes are recommended for
the REM regarding combustion sources.

VI.	Uncertainty and Biases in the Emissions Data

The original REM model consistently provided higher emission estimates
than reported by the facilities in either the residual risk survey or
the TRI.  As such, this conservative model provides an excellent
screening tool.  The revised model provides a better correlation with
the refinery reported emissions while maintaining a level of
conservatism.  If additional changes in the model are made for
wastewater or other sources, it will be impossible to categorically
state that refineries modeled as having risk below a given level are
indeed below that risk threshold.  Additionally, miscellaneous sources,
such as spills and vacuum trucks, are not accounted for in the modeled
sources.  As the emission estimates for the directly modeled sources
become less conservative, the importance of adding emission estimates
for these sources increases.  Additionally, emissions caused by
start-up, shutdown, or malfunction are not included in the model
estimate, but may significantly contribute to human exposure.  None of
the refineries provided emission estimates that included start-up,
shutdown, or malfunction emissions.  These emissions are highly variable
and difficult to monitor.  When using highly conservative emission
models, these miscellaneous sources or short-term emission events can be
considered to be included in the model estimates.  However, if the model
estimates are based on less conservative assumptions, the model output
cannot be considered to include these releases and may result in
underestimating actual exposure in subsequent risk analyses.

In addition to these concerns, some of the reported emissions may
include hidden assumptions that may or may not be valid.  For example,
emissions estimated by TANKS (EPA, 1999b) may be accurate assuming all
primary and secondary seals, if present, are in good condition.  On
March 11, 2003, the South Coast Air Quality Management District (SC
AQMD) filed suit against BP West Coast Products, LLC (Whetzel, 2003). 
Most of the allegations accuse the company of failing to properly
inspect and maintain 26 storage tanks equipped with floating roofs, as
required under SC AQMD Rule 463.  SC AQMD inspections revealed that more
than 80 percent of the tanks had numerous leaks, gaps, torn seals, and
other defects that caused excess emissions.  

Additionally, optical fugitive emission measurements performed at a
140,000 bbl/day capacity refinery in Alberta indicated that storage
tanks, coker unit emissions, and cooling towers were the largest sources
of VOC emissions (Breen, 2004).  The cooling tower emissions were not
speciated for benzene; nonetheless, the benzene emissions that were
measured from the plant were projected to be 44 tpy, almost a factor of
20 higher than the emission rate reported in Canada’s National
Pollutant Release Inventory (NPRI).  Sixty-four percent of the measured
benzene emissions came from storage tanks; this is 28 tpy of benzene
emissions from storage tanks alone at a mid-sized refinery.  These are
higher benzene emissions than were reported for any refinery except for
BP – Texas City (#113) with a capacity of 446,500 bbl/day.  The
Alberta report did not indicate the condition of the floating roof cover
or how the inventory estimates were derived, but it is evident that the
emissions measured during this study were significantly higher than
estimated from storage tanks at this refinery (or other U.S. refineries
of similar size that responded to the residual risk survey).

Similar concerns may be warranted regarding the wastewater treatment
emission estimates.  A recent Bay Area (BA) AQMD study evaluated
collection system emissions for five Bay Area refineries (Breen, 2004). 
Utilizing extensive sampling, flow measurements, and detailed TOXCHEM+
modeling, the study showed that four of the five refineries
underestimated the VOC emissions from their wastewater collection
system.  Two refinery estimates were within a factor of 2 of the BA AQMD
estimate (one higher and one lower), but one refinery had underestimated
its emissions by a factor of 40 and another refinery underestimated its
emissions by a factor of 1,400.  In reviewing the emission estimates
reported by the residual risk survey respondents for wastewater
collection and treatment systems, it appears that the reported
wastewater treatment emissions may also be understated.  Therefore,
developing a refined wastewater treatment emission estimate from these
data that is both reasonably accurate and conservative may be difficult.

VII.	References

Breen, D.  2004.  Proposed Revision of Regulation 8, Rule 8:  Wastewater
Collection Systems.  Staff Report.  Prepared for Bay Area Air Quality
Management District, San Francisco, CA.  March 17, 2004.

Corsi, R, et.al., 2004.  Assessment of Selected Leak Detection,
Sampling, Testing, Measurement and Monitoring Methods for Estimating
Emissions of Highly Reactive Volatile Organic Compounds from Industrial
Cooling Waters.  University of Texas at Austin for TCEQ.  September
2004.

Chambers, A., and M. Strosher.  2006.  Refinery Demonstration of Optical
Technologies for Measurement of Fugitive Emissions and for Leak
Detection.  Final Report.  Prepared for:  Roy McArthur, Environment
Canada; Ontario Ministry of Environment; and Alberta Environment.  March
31, 2006.

Lucas, B.  2007.  Memorandum from B. Lucas, EPA/SPPD, to Project Docket
File (EPA Docket No. EPA-HQ-OAR-2003-0146).  Cooling Towers:  Control
Options and Impact Estimates.  August 17, 2007.

RTI.  2002.  Petroleum Refinery Source Characterization and Emission
Model for Residual Risk Assessment.  Prepared for Robert Lucas, U.S.
Environmental Protection Agency, Office of Air Quality Planning and
Standards, Research Triangle Park, NC.  July 1, 2002.

Taback, H.  1997.  Analysis of Refinery Screening Data.  Prepared for
the American Petroleum Institute, Health and Environmental Affairs
Department.  API Publication No. 310.  November.

TCEQ (Texas Commission on Environmental Quality).  2000.  Air Permit
Technical Guidance for Chemical Sources:  Equipment Leak Fugitives. 
October.

TCEQ (Texas Commission on Environmental Quality).  2006.  2005 Emission
Inventory Guidelines  - Technical Supplement 2:  Cooling Towers.  TCEQ
Publication RG-360/05, Revised.  January.

U.S. EPA (Environmental Protection Agency).  1994.  Air Emissions Models
for Waste and Wastewater.  EPA-453/R-94-080A.  Office of Air Quality
Planning and Standards, Research Triangle Park, NC.

U.S. EPA (Environmental Protection Agency).  1995a.  Compilation of Air
Pollutant Emission Factors.  AP-42.  Office of Air Quality Planning and
Standards, Research Triangle Park, NC. 

U.S. EPA (Environmental Protection Agency).  1995b.  Protocol for
Equipment Leak Emission Estimates.  EPA-453/R-95-017.  Office of Air
Quality Planning and Standards, Research Triangle Park, NC.  November.

U.S. EPA (Environmental Protection Agency).  1998.  Locating and
Estimating Air Emissions from Sources of Benzene.  EPA-454/R-98-011. 
Office of Air Quality Planning and Standards, Research Triangle Park,
NC.  

U.S. EPA (Environmental Protection Agency).  1999a.  “Proper
Monitoring Essential to Reducing “Fugitive Emissions” under Leak
Detection and Repair Programs.”  Enforcement Alert.  2(9):1-4. 
October.

U.S. EPA (Environmental Protection Agency).  1999b.  User’s Guide to
TANKS - Storage Tank Emissions Calculation Software, Version 4.0. 
Office of Air Quality Planning and Standards, Research Triangle Park,
NC.

U.S. EPA (Environmental Protection Agency).  2001.  User’s Guide for
WATER9 Software, Version 1.0.0.  Office of Air Quality Planning and
Standards, Research Triangle Park, NC.

Whetzel, C., 2003.  “South Coast Air District Seeks $319 For
Violations at Los Angeles Area Refinery.”  The Bureau of National
Affairs, Inc., Washington D.C   SEQ CHAPTER \h \r 1 



Table A-1.  Reported Benzene Emissions by Source Sorted by Facility ID
Number

RTI ID	Facility Name	Crude Capacity

(bbl/cd)	Cooling Towers	Equipment Leaks	Tanks	Waste-water	Misc. Vents
Marine Vessel Loading	Gasoline Rack	Other	totals	EmF

15	Chevron USA Inc. -- El Segundo, CA	 260,000 	0	0.112	0.242	0.056	0	0
0	0.061	0.47	 9.92 

20	ExxonMobil Corp. (formerly Mobil) -- Torrance, CA	 149,500 	0.001
0.133	0.954	0.008	0.013	0	0.014	1.824	2.95	108.00 

30.1	ConocoPhillips (formerly Tosco, formerly Unocal Corp.) -- Rodeo, CA
 68,000 	0	0.173	0.437	0.836	0	0	0	0.145	1.59	128.36 

31	Tesoro (formerly Ultramar Diamond Shamrock, formerly Tosco Corp. -
Avon) -- Martinez, CA	 161,000 	0.017	0.038	1.051	0.241	0	0.011	0.058
1.715	3.13	106.60 

35	Suncor Energy (formerly Conoco Inc.) -- Commerce City, CO	 60,000 
0.004	0.007	0.391	0.092	0	0	0.045	0.328	0.87	 79.05 

51	National Cooperative Refinery Association -- McPherson, KS	 79,000 
1.376	2.472	0.656	1.820	0	0	0.850	0.923	8.10	561.62 

52	Frontier Oil Corp. (formerly El Dorado Refining, formerly Texaco
Refining & Marketing Inc.) -- El Dorado, KS	 110,000 	0.224	3.188	7.021
0.927	0	0	0.456	0.532	12.35	615.12 

64	ExxonMobil Corp. (formerly Exxon) -- Chalmette, LA	 187,000 	1.759
7.031	5.534	2.186	0	0	0.010	4.234	20.93	613.23 

66	Motiva Enterprises (formerly Star) -- Convent, LA	235,000 	0.090
0.145	0.435	0.013	0	0.044	0	0.069	0.80	 18.54 

67	Motiva Enterprises (formerly Shell Oil Co.) -- Norco, LA	220,000 
0.405	10.202	7.636	0.480	0	0.067	0	3.310	22.10	550.44 

88	Sunoco, Inc. (formerly Coastal Eagle Point Oil Co.) -- Westville, NJ
150,000 	0.015	0.099	4.178	0.987	0	0.061	0.218	0	5.56	203.02 

92	Giant Refining Co. -- Gallup, NM	26,000 	0.016	4.93	2.573	0.885	0	0
0.022	0	8.43	1,775.79

93	Navajo Refining Co. -- Artesia, NM	60,000 	0.303	2.643	0.299	0.387	0
0	0.018	0	3.65	333.41 

96	Marathon Petroleum Company LLC -- Canton, OH	73,000 	0.065	0.093
1.489	1.944	0	0	0.122	0	3.71	    278.48 

100	Gary-Williams Energy Corp. -- Wynnewood, OK	52,500 	1.410	1.15	1.47
2.648	0	0	0.03	0	6.71	700.14 

105	Sunoco, Inc. -- Marcus Hook, PA	175,000 	0.001	0.731	6.314	"no data"
0	3.047	0.096	0	10.19	319.03 

113	BP (formerly Amoco Oil Co.) -- Texas City, TX	446,500 	1.604	34.156
24.44	1.817	0	1.263	0	0	63.28	776.57 

116	Valero Energy Corp. (formerly Coastal Refining & Marketing Inc.) --
Corpus Christi, TX (East)	100,000 	0	10.123	5.001	4.285	9.368	1.244
0.002	0	30.02	1,645.11 

133	Valero Energy Corp. (formerly Basis Petroleum, Inc.) -- Houston, TX
90,000 	0.0311	0.0281	1.119	1.226	0.891	0.2495	0	0	3.54	215.76 

145	ConocoPhillips (formerly Tosco Refining Co.) -- Ferndale, WA	93,000 
0.11	0.079	0.779	0.279	0	0.775	0.003	0	2.02	119.27 

146	US Oil & Refining Co. -- Tacoma, WA	35,800 	0	0.62	0.08	0.04	0	0.1
0.05	0	0.89	136.22 

151	Sinclair Oil Corp. -- Sinclair, WY	72,000 	0.689	0.164	1.327	0.558
0.316	0	0.001	0.404	3.46	263.31 



Technical Memorandum – 22 Facility Study

August 20, 2007

Page   PAGE  3 

 PAGE   

Technical Memorandum

A- PAGE   2 

Appendix A.  Reported Benzene Emissions for 22 Refinery Study

A-  PAGE  1 

