United States Office of Air Quality EPA-450/3-84-003
Environmental Protection Planning and Standards March 1984
Agency Research Triangle Park NC 27711
Air ——
4>EPA Benzene Emissions
From Ethylbenzene/
Styrene Plants —
Background
Information for
Proposal to Withdraw
Proposed Standards
-------
EPA-450/3-84-003
Benzene Emissions from
Ethylbenzene/Styrene Plants —
Background Information for
Proposal to Withdraw Proposed Standards
Emission Standards and Engineering Division
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Air and Radiation
Office of Air Quality Planning Standards
Research Triangle Park, North Carolina 27711
March 1984
-------
This report has been reviewed by the Emission Standards and Engineering Division of the Office of Air
Quality Planning and Standards, EPA, and approved for publication. Mention of trade names or commercial
products is not intended to constitute endorsement or recommendation for use. Copies of this report are
available through the Library Services Office (MD-35), U.S. Environmental Protection Agency, Research
Triangle Park, N.C. 27711, or, for a fee, from the National Technical Information Services, 5285 Port Royal
Road, Springfield, Virginia 22161.
-------
ENVIRONMENTAL PROTECTION AGENCY
Background Information
for Ethyl benzene/Styrene Plants
Prepared by:
Jack R. Farmer - (Date)
Director, Emission Standards and Engineering Division
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
1. The Federal Register notice proposes withdrawal of proposed national
emission standards (45 FR 83448; December 18, 1980) for benzene emissions
from existing and new ethyl benzene/styrene plants.
2. Copies of this document have been sent to the following Federal Depart-
ments: Labor, Health and Human Services, Defense, Transportation,
Agriculture, Commerce, Interior, and Energy; the National Science
Foundation; the Council on Environmental Quality; State and Territorial
Air Pollution Program Administrators; EPA Regional Administrators;
Local Air Pollution Control Officials; Office of Management and Budget;
and other interested parties.
3. The comment period for review of this document is 30 days from date of
proposal in the Federal Register. Mr. Gilbert H. Wood may be contacted
at (919) 541-5578 regarding the date of the comment period.
4. For additional information contact:
Gilbert H. Wood
Standards Development Branch (MD-13)
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Telephone: (919) 541-5578
5. Copies of this document may be obtained from:
U.S. Environmental Protection Agency Library (MD-35)
Research Triangle Park, NC 27711
Telephone: (919) 541-2777
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
-------
TABLE OF CONTENTS
Section
Page
List of Tables vi
1 Summary 1-1
1.1 Summary of Industry Changes Since Proposal 1-1
1.2 Summary of Proposal to Withdraw the Proposed
Standards 1-2
2 Summary of Public Comments 2-1
2.1 Selection of Ethylbenzene/Styrene Plants for
Regulation 2-1
2.1.1 Appropriateness of Regulating EB/S Plants
Under Section 112 2-1
2.1.2 Criteria for Unreasonableness of Residual
Risks 2-9
2.2 Health and Environmental Impacts 2-9
2.2.1 Estimated Current Emissions and Risks 2-9
2.2.2 Flare Control Efficiency 2-12
Appendix A. Emission Data and Environmental Impacts A-l
Appendix B. Methodology for Estimating Leukemia Incidence
and Maximum Lifetime Risk from Exposure
to Benzene Emissions from Ethylbenzene/Styrene
Process Vents B-l
-------
LIST OF TABLES
Number
Page
1-1 Changes in Industry Impacts 1-2
2-1 List of Commenters on the Proposed National Emission
Standard for Benzene Emissions from EB/S Plants 2-2
2-2 Current Estimated Impacts 2-8
2-3 Plant-by-Plant Baseline Emission Rates for Benzene
i n 1978 and 1981 2-13
2-4 Flare Emission Studies Complete as of October 1982 2-15
VI
-------
1. SUMMARY
The U.S. Environmental Protection Agency (EPA), on December 18,
1980, proposed a national emission standard for hazardous air pollutants
(NESHAP) (45 FR 83448) that would regulate benzene emissions from
ethylbenzene/styrene (EB/S) plants under the authority of Section 112
of the Clean Air Act as amended. Public comments were requested on
the proposal in the Federal Register publication. Twelve comment
letters were received, and six interested parties testified at the
public hearing. These comments were made by EB/S manufacturers, the
Chemical Manufacturers Association (CMA), and State and Federal
Government offices. Comments submitted relevant to the withdrawal
decision, along with the responses to these comments, are summarized
in this document. The summary of comments and responses serves as the
basis for the proposal to withdraw the proposed standards.
1.1 SUMMARY OF INDUSTRY CHANGES SINCE PROPOSAL
Since the standards for benzene emissions from EB/S plants were
proposed (December 18, 1980; 45 FR 83448), benzene emissions from this
source category have declined considerably. This reduction is due to
revised emission estimates, a revised estimate of flare efficiency
(from 60 to 98 percent), installation of controls, industry capacity
reductions, and better maintenance of existing product recovery units.
These changes are described in more detail in Section 2.2.1 of this
document. Table 1-1 compares the nationwide baseline benzene emission
and health impacts due to EB/S process vents at proposal with current
estimated impacts.
1-1
-------
TABLE 1-1. CHANGES IN INDUSTRY IMPACTS
Impact
Benzene emissions (Mg/yr)
Leukemia incidence
At proposal
2,400
0.027 to 0.20
Current
210
0.0057
(cases/yr)
Maximum lifetime risk
6.2 x io"J to
4.4 x 10 *
1.4 x 10
-4
1.2 SUMMARY OF PROPOSAL TO WITHDRAW THE PROPOSED STANDARDS
The Administrator is proposing to withdraw proposal of the benzene
standards for EB/S plants. This decision is based on several factors,
including the broad amount of control currently within the source
category, the relatively small amount of emissions, the small estimated
leukemia incidence and maximum lifetime risk at current control levels,
and the small reduction in these health risks achievable with add-on
controls. This decision is discussed in greater detail in Section 2.1.1.
1-2
-------
2. SUMMARY OF PUBLIC COMMENTS
The commenters and docket numbers are listed in Table 2-1.
Twelve letters contained comments and six people testified at the
public hearing on the proposed standard and its background information
document (BID). Because the proposed standards are being proposed for
withdrawal, only comments relevant to that decision are addressed in
this document. These comments have been combined into the following
two categories:
1. Selection of Ethyl benzene/Styrene (EB/S) Plants for Regulation
2. Health and Environmental Impacts
2.1 SELECTION OF ETHYLBENZENE/STYRENE PLANTS FOR REGULATION
2.1.1 Appropriateness of Regulating EB/S Plants Under Section 112
Comment: Several commenters stated that EB/S plants do not
present the magnitude of risk to health for which a national emission
standard for hazardous air pollutants (NESHAP) is appropriate. They
felt that the risk from benzene emissions from EB/S plants is far
below the threshold level for induction of benzene-related leukemia,
and therefore, emissions from EB/S plants do not present the kind of
health risk that Congress foresaw when it enacted Section 112 to deal
with the limited number of very hazardous pollutants (IV-F-1; IV-F-2;
IV-F-5; IV-D-10).
Response: At the time the standard was proposed, total EB/S
industry-wide process benzene emissions were estimated to be 2,100 Mg/yr
based on 100 percent production capacity and actual plant-by-plant
control levels. (After publication of the proposal preamble (45 FR 83448),
this number was revised to 2,400 Mg/yr because of miscalculations in
the industry-wide total.) It was estimated that 2.5 million people
live within 20 kilometers of EB/S plants. As a result of exposure to
2-1
-------
TABLE 2-1. LIST OF COMMENTERS ON THE PROPOSED NATIONAL
EMISSION STANDARD FOR BENZENE EMISSIONS
FROM EB/S PLANTS
Commented Affiliation
IV-F-1 Public Hearing Transcript
IV-F-2 Chemical Manufacturers Association (CMA)
IV-F-3 Oxirane Chemical/ARCO Chemical
IV-F-4b Dow Chemical
IV-F-5 Applied Meteorology, Inc.
IV-F-6b Cos-Mar
IV-D-1 Wilmer, Cutler, and Pickering (CMA)
IV-D-2 Dow Chemical, U.S.A.
IV-D-3 Federal Energy Regulatory Commission (FERC)
IV-D-4 Department of Environmental Regulation, State of
Florida
IV-D-5 Public Health Service, Department of Health and
Human Services
IV-D-6 Wilmer, Cutler, and Pickering (CMA)
IV-D-7 Air Products and Chemicals, Inc.
IV-D-8 Wilmer, Cutler, and Pickering (CMA)
IV-D-9 American Petroleum Institute
IV-D-10 Wilmer, Cutler, and Pickering (CMA)
IV-D-11 Dow Chemical, U.S.A.
IV-D-12 Gulf Oil Chemicals
These designaters represent docket entry numbers for Docket Number
A-79-49. These docket entries are available for public inspection
at:
Central Docket Section
West Tower Lobby, Gallery 1
Waterside Mall
401 M Street, S.W.
Washington, DC 20460
These references are transcripts submitted by commenters at the
public hearing and are essentially identical to their oral testimony.
2-2
-------
these benzene concentrations, maximum lifetime risk was estimated to
-4 -3
be within a range of 6.2 x 10 to 4.4 x 10 . In addition, a range
of leukemia incidence of 0.027 to 0.20 cases per year within this
population due to benzene exposure from EB/S plants was estimated.
Maximum lifetime risk is the probability of someone within the assumed
exposed population contracting leukemia if he or she were exposed to
the highest maximum annual average benzene concentration predicted
from dispersion modeling during an entire lifetime (70 years).
These ranges represent 95 percent confidence limits on two sources
of uncertainty in the benzene risk estimates (II-A-8). One source
derives from the variations in dose/response among the three occupational
studies upon which the benzene unit risk factor is based. A second
source involves the uncertainties in the estimates of ambient exposure.
In the former case, the confidence limits are based on the assumption
that the slopes of the dose/response relationships are unbiased estimates
of the true slope and that the estimates are log normally distributed.
In the latter case, the limits are based on the assumption that actual
exposure levels may vary by a factor of two from the estimates obtained
by dispersion modeling (assuming that source-specific input data are
accurate).
Several other uncertainties are associated with the estimated
health numbers not quantified in the proposal ranges. EPA has extra-
polated the leukemia risks identified for occupationally exposed
populations (generally healthy, white males) to the general population
for whom susceptibility to a carcinogenic insult could differ. The
presence of more or less susceptible subgroups within the general
population would result in an occupationally-derived risk factor that
may underestimate or overestimate actual risks. To the extent that
there are more susceptible subgroups within the general population,
the maximum individual lifetime risks may be underestimated.
On the other hand, general population exposures to benzene are
much lower than those experienced by the exposed workers in the occupa-
tional studies, often by several orders of magnitude. In relating the
occupational experience to the general population, EPA has applied a
linear, nonthreshold model that assumes that the leukemia response is
2-3
-------
linearly related to benzene dose, even at very low levels of exposure.
There are biological data supporting this approach, particularly for
carcinogens. However, there are also data which suggest that, for
some toxic chemicals, dose/response curves are not linear, with response
decreasing faster than dose at low levels of exposure. At such levels,
the nonlinear models tend to produce smaller risk factors than the
linear model. The data for benzene do not conclusively support either
hypothesis. EPA has elected to use the linear model for benzene
because this model is generally considered to be conservative compared
to the nonlinear alternatives. This choice may result in an overesti-
mate of the actual leukemia risks.
EPA estimates ambient benzene concentrations in the vicinity of
emitting sources through the use of atmospheric dispersion models.
EPA believes that its ambient dispersion modeling provides a reasonable
estimate of the maximum ambient levels of benzene to which the public
could be exposed. The models accept emissions estimates, plant param-
eters, and meteorology as inputs and predict ambient concentrations at
specified locations, conditional upon certain assumptions. For example,
emissions and plant parameters often must be estimated rather than
measured, particularly in determining the magnitude of fugitive emissions
and where there are large numbers of sources. This can lead to overesti-
mates or underestimates of exposure. Similarly, meteorological data
often are not available at the plant site but only from distant weather
stations that may not be representative of the meteorology of the
plant vicinity.
EPA's dispersion models normally assume that the terrain in the
vicinity of the sources is flat. For sources located in complex
terrain, this assumption would tend to underestimate the maximum
annual concentration although estimates of aggregate population exposure
would be less affected. On the other hand, EPA's benzene exposure
models assume that the exposed population is immobile and outdoors at
their residence, continuously exposed for a lifetime to the predicted
concentrations. To the extent that benzene levels indoors are lower
and that people do not reside in the same area for a lifetime, these
assumptions will tend to overpredict exposure.
2-4
-------
Upon reconsideration, EPA has concluded that the presentation of
the risk estimates as ranges does not offer significant advantages over
the presentation as the associated point estimates of the risk.
Further, the proposal ranges for benzene make risk comparisons among
source categories more difficult and tend to create a false impression
that the bounds of the risks are known with certainty. For these
reasons, the benzene risks in this rulemaking are presented as point
estimates of the leukemia risk. EPA believes that these risk numbers
represent plausible, if conservative, estimates of the magnitude of
the actual cancer risk posed by benzene emitted from the source cate-
gories evaluated. For comparison, the proposal ranges may be converted
into rough point estimates by multiplying the lower end of the range
by a factor of 2.6.
When possible, given sufficient health data and appropriate
assumptions regarding hazardous air pollutants, estimates are made of
cancer incidence and risk in the general population due to ambient
exposure around the sources of emissions. Quantitative risk estimates
at ambient concentrations involve an analysis of the effects of the
substance in high-dose epidemiological or animal studies and extrapo-
lation of these high-dose results to relevant human exposure routes at
low doses. The mathematical models used for such extrapolations are
based on observed dose-response relationships for carcinogens and
assumptions about such relationships as the dose approaches very low
levels or zero.
The risks to public health from a carcinogen's emissions may be
estimated when the dose-response relationship obtained from this
carcinogenicity strength calculation is combined with an analysis of
the extent of population exposure to the substance through the ambient
air. Exposure in this context is a function of both a substance's
concentration and the length of time the concentration is encountered.
Exposure analyses are based on air quality models, available estimates
of emissions from EB/S plants, and approximations of population distri-
butions near their sources.
The air quality models used estimated exposures of up to 20 kilo-
meters, and population and growth statistics were examined. Along
2-5
-------
with the existing carcinogenic strength determinations, the information
collected was used to estimate the degree of risk to individuals and
the range of increased cancer incidence expected from ambient air
exposures associated with EB/S plants at various possible emission
levels.
The assumptions necessary to estimate benzene health risks and
the underlying uncertainties have led some commenters on EPA's proposed
rules to suggest that the risk estimates are inappropriate for use in
regulatory decisionmaking. Although EPA acknowledges the potential
for error in such estimates, the Agency has concluded that both the
unit risk factor for benzene and the evaluation of public exposure
represent plausible, if conservative, estimates of actual conditions.
Combining these quantities to produce estimates of the leukemia risks
to exposed populations implies that the risk estimates obtained are
also conservative in nature; that is, the actual leukemia risks from
benzene exposure are not likely to be higher than those estimated. In
this context, EPA believes that such estimates of the health hazard
can and should play an important role in the regulation of hazardous
pollutants.
Based on the magnitude of benzene exposures from this source
category, the resulting estimated maximum individual risks and leukemia
incidence in the exposed population, and consideration of the uncer-
tainties of quantitative risk assessment, the Administrator determined
at proposal that benzene emissions from EB/S plants pose a significant
cancer risk.
After noting the control status of changes in the EB/S industry
since proposal (see Response 2.2.1 of this document), EPA reassessed
whether EB/S process vents continue to pose a significant risk of
leukemia and whether a benzene standard is warranted under Section 112.
EB/S process vents are currently estimated to emit about 210 megagrams
of benzene annually from 13 plants. This amount is less than 0.4
percent of total benzene emissions from stationary sources. Estimated
lifetime risk due to these emissions is about 1.4 x 10 for the most
exposed individuals, and over the total exposed population (within
2-6
-------
20 km of each plant) about 0.0057 leukemia cases per year is estimated
to occur. The current (or baseline) impacts are presented in Table 2-2.
For comparison, at proposal the industry was estimated to emit
2,400 Mg/yr, resulting in a range of leukemia cases per year of about
-4
0.027 to 0.20 and a range of maximum lifetime risk of 6.2 x 10 to
-3
4.4 x 10 . Thus, since proposal, estimated benzene emissions, esti-
mated annual leukemia incidence, and the estimated maximum lifetime
risk have all declined by over 90 percent.
Combustion control techniques applicable to EB/S process vents
were described at proposal and include boilers, incinerators, and
flares which can reduce emissions by atf least 98 percent. Although at
proposal flares were assumed to achieve only 60 percent reduction,
this estimate has been revised upward to 98 percent due to the results
from recent flare studies (see Section 2.2.2) (IV-A-4; IV-A-5; IV-D-17;
IV-J-2; IV-J-3; IV-J-7). These studies indicate that a flare meeting
certain operating conditions can achieve emission reductions in this
range. Applying these combustion control techniques could reduce
nationwide benzene emission from these sources over baseline by about
70 percent.
The current estimated leukemia incidence and maximum lifetime
risk represent small risks to public health. By both expressions of
health risk, the extent of the hazard posed by this source category is
more than an order of magnitude smaller than for benzene source cate-
gories for which standards are being developed. Using combustion
control techniques, leukemia incidence could be reduced to roughly
0.001 cases per year and maximum lifetime risk to roughly 9.2 x 10 .
Although a large percentage reduction could be achieved in the health
risks (about 80 to 90 percent), the absolute amount is also small.
Therefore, in light of the extent of control now exhibited by the
industry, the small amount of benzene emissions from these sources and
the small portion (less than 0.4 percent) of the total benzene emissions
from stationary sources that these sources represent, the small leukemia
incidence and maximum lifetime risk estimated at current levels, and
the small incremental reductions in these health risks achievable with
2-7
-------
TABLE 2-2. CURRENT ESTIMATED IMPACTS
Plant
American Hoechst, LA
American Hoechst, TX
Amoco, TX
ARCO, PA
Cos Mar, LA
Dow, MI
Dow, TX (A)
Dow, TX (B)
El Paso Products, TX
Gulf Oil, LA
Monsanto, TX
Oxirane, TX
Koch, TX
U.S. Steel, TX
Total
Benzene
emissions3
(Mg/yr)
0.72
6.25
15.0
39.4
17.5
57.3
1.34
36.1
1.85
9.90
19.8
1.20
0.58
0.44
208
Leukemia
incidence
(cases/yr
x 10"4)
0.093
0.62
1.4
15.0
0.44
24.0
0.19
8.3
0.32
0.24
6.3
0.36
0.030
0.19
57.0
Maximum
lifetime risk
(x 10"5)
0.11
0.12
6.6
14.0
2.7
140.0
1.5
7.7
0.54
0.25
9.5
0.21
0.035
0.0057
aD . . cr.. , .
lated from CMA data package (IV-D-13)
(IV-D-34, IV-D-35, IV-D-36).
estimates after proposal. Calcu-
and subsequent industry submittals
2-8
-------
combustion control technologies, the Administrator has concluded that
benzene emissions from EB/S process vents do not warrant Federal
regulatory action under Section 112.
2.1.2 Criteria for Unreasonableness of Residual Risks
Comment: EB/S plants, relative to other stationary sources of
benzene emissions, present an insignificant risk to the population.
The risk EPA found would be "not unreasonable" after implementation of
a national emission standard for benzene emissions from the maleic
anhydride industry is almost eight times greater than the current risk
from EB/S process vents. Also, leukemia risk from benzene emissions
from stationary sources is insignificant because 80 percent of benzene
emissions are derived from mobile sources not subject to Section 112
regulation. Therefore, control of EB/S plants represents an inconsis-
tency in EPA's application of risk assessments to the determination of
acceptable and unacceptable risks in current proposals for the control
of benzene (IV-F-1; IV-F-2; IV-F-3; IV-F-6; IV-D-7; IV-D-10; IV-D-12).
Response: As described in the previous response, the Administrator
has reevaluated the health risk posed by this source category and
concluded that Federal regulatory action is no longer warranted.
As pointed out by the commenter, mobile sources cannot be regulated
under Section 112, and, consequently, no determination has been made
concerning the unreasonableness of risk due to exposure to benzene
from those sources. However, even if mobile sources could be regulated
under Section 112, catalytic converters already being installed seem
to be an effective control technology.
2.2 HEALTH AND ENVIRONMENTAL IMPACTS
2.2.1 Estimated Current Emissions and Risks
Comment: According to several commenters, CMA used (1) plant-
specific emissions data, (2) local meteorological data, (3) refined
population data, (4) a ring-sector approach, which accounts for direc-
tional variability in population densities, and (5) the Lamm risk factor
to estimate the current risk from benzene emissions from EB/S process
vents to be "the same as the 0.00034 risk EPA calculated would exist if
the proposed standard had been adopted." Also, EPA has overstated
current emissions by a factor of three after correcting for capacity
2-9
-------
and existing controls. Therefore, EPA has overestimated both individual
and industry-wide emissions; population extent and distribution; and
as a consequence, risk from EB/S benzene emissions (IV-F-1; IV-F-2;
IV-F-5; IV-F-6; IV-D-10; IV-D-12).
Response: At proposal, environmental impacts were calculated
based on data industry supplied from 1972 to 1978 in response to EPA
"114" letters (information requests provided for by Section 114 of the
Clean Air Act). The responses to the "114" letters provided detailed
information from the entire industry. The data submitted were compiled
to develop a "typical" EB/S plant that was used to determine impacts.
While this model plant did not conform to any particular plant in the
industry, it was considered representative for purposes of character-
izing the baseline. EPA estimated total nationwide benzene emissions
from all EB/S process vents to be 2,100 Mg/yr based on 1978 control
levels and actual capacity utilization of nameplate production rates.
After publication in the proposal preamble (45 FR 83448), this number
was revised to 2,400 Mg/yr to correct miscalculations in the industry-
wide total (see Table 2-2).
The industrial source complex (ISC) dispersion model was used to
estimate annual average benzene concentrations for distances and
directions out to 20 kilometers from the model plant. Houston meteoro-
logical data, which best represented weather conditions for most EB/S
plants, were used in the dispersion model. Each actual plant was
"modeled" by scaling the model plant's emissions to the actual plant's
emissions based on known or assumed production capacity and control
level. The population residing within 20 kilometers of each EB/S
plant was determined from the 1970 Bureau of the Census Master Enumera-
tion District (MED) List. Each plant site was located by latitude and
longitude on a grid system having grids of approximately 10 square
kilometers. The population was determined from the MED List for each
grid block within 20 kilometers of each plant site. Each plant's
populations were then coupled with the scaled concentrations derived
from the dispersion modeling at the respective distance intervals and
the risk factor developed by CAG for the beyond best available technol-
ogy (BAT) analysis and to describe the standard's health impact.
These numbers did not play a role in selection of BAT.
2-10
-------
After proposal, CMA supplied detailed data regarding current
controls and emissions (IV-D-13; IV-D-34; IV-D-35; IV-D-36). EPA
subsequently reanalyzed and revised the baseline environmental impacts
on a plant-specific basis using the CMA package and more current
information obtained from the industry (IV-D-34; IV-D-35; IV-D-36).
These submittals give vent-by-vent emissions based on measurements
or site-specific engineering calculations. In several cases, actual
control efficiencies, vent stream measurements, and reduced plant
capacities or capacity utilization result in several-fold lower total
emissions than reported at proposal. Monsanto has installed a control
system that routes nearly all continuous emissions to a boiler (IV-F-2;
IV-D-13). Production capacity has been reduced at the American Hoechst,
Louisiana, plant, and repairs and new recovery equipment installed
since proposal have further reduced baseline emissions at this plant
(IV-F-2; IV-D-37). ARCO, Pennsylvania, has significantly reduced
production capacity. Most vents at the Oxirane, Texas, plant are
controlled before flaring by product recovery devices, giving a greater
level of control than assumed by EPA at proposal (IV-F-2; IV-D-13).
Likewise, site-specific information on capacities and product recovery
at the Dow, Texas, plant have resulted in lower emission estimates
that at proposal.
In addition, changes have occurred in the number of operating
plants. The ARCO, Texas, ethylbenzene unit has been converted to
other uses (IV-F-2), and the Koch (formerly Sun Oil), Texas, styrene
unit has been shut down (IV-C-28). Two new plants have come on line,
American Hoechst, Texas, and Cos-Mar, Louisiana (a second Cos-Mar
plant) (IV-F-2). At the American Hoechst, Texas, plant all process
emissions are routed to a boiler, resulting in low benzene emissions
(IV-F-2).
Finally, the revised flare efficiency (98 percent compared with
60 percent used at proposal; see Section 2.2.2) results in significantly
lower emission estimates in an industry that relies heavily on flaring
as a control technique. Plants with significantly reduced, continuous
emission estimates because of the revised flare efficiency are:
Oxirane, Texas; Gulf, Louisiana; Cos-Mar, Louisiana; El Paso, Texas;
2-11
-------
and Dow, Texas. Based on the more recent information and revised
analysis, current nationwide benzene emissions from EB/S process vents
are estimated to be about 208 Mg/yr (see Table 2-3 and Appendix A).
The ISC dispersion model was used again to compute annual average
concentrations at various distances and directions from the plant.
However, each plant was modeled individually according to local meteoro-
logical data. Geographical coordinates supplied by the plant were
then used to match concentrations with corresponding populations on
both a distance and directional basis from the plant. Additionally,
the CAG unit risk factor was revised based on comments. A more detailed
discussion of this methodology can be found in Appendix B of this
document, which describes the methodology used to calculate the health
impacts. The revised leukemia incidence is estimated to be about 0.0057
-4
cases per year and the maximum lifetime risk about 1.4 x 10 .
2.2.2 Flare Control Efficiency
Comment: Three commenters maintained that in the proposed standard,
EPA avoids specifying the type of control equipment to be used to
limit continuous process emissions. However, the proposed standard
was based on the control level attainable when emissions are combusted
in a boiler or process heater, and if a plant chooses to control
emissions by a method other than combustion in a boiler or process
heater, it is required to monitor emissions continuously using a gas
chromatograph and flame ionization detector. Because proven technology
does not exist for measuring emissions continuously from a flare, the
net effect of these requirements is to preclude the use of flares to
control continuous process emissions, either as the sole control
device or in conjunction with absorbers or condensers. Hydroscience
(now IT Enviroscience) (II-B-23) estimated flare efficiency to range
from 75 to 98 percent. Small flares (2 to 10 inches in diameter) may
have over 99 percent efficiency under optimum conditions. Siege! in
Germany tested 1,300 samples to achieve 99 percent efficiency 100 per-
cent of the time. The Battelle study for EPA demonstrated 95 percent
2-12
-------
TABLE 2-3. PLANT-BY-PLANT BASELINE EMISSION RATES
FOR BENZENE IN 1978 and 1981
(Mg/yr)
Plant/location
American Hoechst
Baton Rouge, LA
Bayport, TX
Amoco
Texas City, TX
ARCO
Port Arthur, TX
Kobuta, PA
Cos-Mar, Inc.
Carville, LA
Dow Chemical, USA
Midland, MI
Freeport, TX
El Paso Products
Odessa, TX
Gulf Oil
Oonaldsville, LA
Monsanto Company
Texas City, TX
Oxirane
Channelview, TX
Koch (formerly Sun Oil)
Corpus Christi , TX
U.S. Steel
Houston, TX
Subtotal
Plus H2 separation
Plus excess emissions
TOTAL
1978
Total
process
emissions
320
0
26
300
70
420
70
180
16
180
320
290
15
1
2,208
110
113
2,431
1981
Continuous
process ,
emissions
0.47
5.89
14.4
0
37.0
17.2
57.1
35.8
1.34
9.85
4.35
1.20
0.54
0.42
186
-
-
TOTAL
Excess
emissions
0.25
0.36
0.59
0
2.45
0.27
0.20
1.64
0.51
0.05
15.5
0
0.04
0.02
22
-
-
208
Assumes 100 percent capacity utilization and actual plant-by-plant
control levels. Derived from proposal BID (EPA-450/3-79-035a).
Includes H2 separation unit operations.
package (IV-D-13).
Derived from CMA data
2-13
-------
efficiency. Therefore, EPA's decision to preclude the use of flares
is arbitrary, capricious, and cannot be sustained (IV-F-1; IV-F-3;
IV-D-10.)
Response: At proposal, no conclusive data were available on
flare destruction efficiency. Calculations and limited test data
showed benzene destruction efficiencies ranging from 60 to 99 percent.
These studies included engineering evaluations by Hydroscience, Inc.,
(II-B-23), calculations based on residence time, limited and unverified
test data from a flare manufacturer on a flare burning methane (II-I-29),
and EPA in-house estimates. For the purposes of determining impacts,
a 60-percent flare efficiency was chosen to enable assessment of
"worst-case" emissions from flaring based on limited data and vent
stream conditions within the EB/S industry.
Since proposal, more information on flare design and destruction
efficiency has become available. In response, EPA has examined recent
flare studies to develop a more accurate destruction efficiency estimate
for calculating emission impacts from EB/S smokeless flares. The
number of flare studies is limited and no data specifically address
conditions present within EB/S vent streams. Table 2-4 summarizes
four flare studies that provided information on flare gas composition,
flow rate, and destruction efficiency. Each study can be found in
complete form in the docket (IV-D-17; IV-J-2; IV-J-3; IV-J-7; IV-A-4;
IV-A-5). The following discussion reviews the experimental flare
systems and operating conditions used in these studies.
Palmer (1972) experimented with a 13-millimeter inside diameter
flare head, the tip of which was located 1.2 meter from the ground.
Ethylene was flared at 15 to 76 m/s at the exit, with a heat release
rate of 0.4 to 2.2 GJ/hr (0.4 x 106 to 2.1 x 106 Btu/hr). Helium was
added to the ethylene as a tracer at 1 to 3 volume percent, and the
effect of steam injection was investigated in some experiments. Four
sets of operating conditions were investigated; destruction efficiency
was measured as greater than 99.9 percent for three sets and 97.8 percent
for the fourth. The author questioned the validity of the 97.8-percent
result due to possible sampling and analytical errors. He recommended
further sampling and analytical techniques development before conducting
further flare evaluations.
2-14
-------
TABLE 2-4. FLARE EMISSIONS STUDIES COMPLETE AS OF OCTOBER 1982
rv>
i
Investigator
Palmer (1972)
Lee and Whipple
Sponsor
E. I. du Pont
Union Carbide
Flare tip design
13 mm diameter
Discrete holes in
Flared gas
Ethyl ene
Propane
Throughput
(GJ/hr)
0.4 - 2.2
0.3
Flare
efficiency
(%)
>97.8
96 - 100
(1981)
Siegel (1980)
McDaniel et al.
(1982)
Ph.D. dissertation
University of Karlsruhe
Howes et al. (1981) EPA
CMA-EPA
51-mm diameter cap
Commercial design
(700 mm diameter steam)
Commercial design
(150 mm diameter air
assist)
Commercial design H.P.
(3 tips @ 100 mm
diameter)
Commercial design
(460 mm diameter
air assist)
Commercial design
(200 mm diameter
steam assist)
* 50% H2
plus light
hydrocarbons
Propane
52 - 188
46
Natural gas 30 (per tip)
Propylene
Propylene
0.003 - 61
0.01 - 60
97 - >99
92.6 - 100
>99
61.9 - 100
68.9 - 100
Sources: Docket items IV-A-4, IV-A-5.
-------
Siege! (1980) made the first comprehensive study of a commercial
flare system. He studied burning of refinery gas on a commercial
flare head manufactured by Flaregas Company. The flare gases used
consisted primarily of hydrogen (45.4 to 69.3 percent by volume) and
light paraffins (methane to butane). Traces of H2S were also present
in some runs. The flare was operated from 130 to 2,900 kg/hr of fuel,
and the maximum heat release rate was approximately 248 GJ/hr (235 x
106 Btu/hr). Combustion efficiency was greater than 99 percent for
1,294 points and greater than 99 percent for all points except one
that had a 97-percent efficiency. The author attributed the 97-percent
result to excessive steam addition.
Lee and Whipple (1981) studied a bench-scale propane flare. The
flare head was 51 millimeters in diameter with one 21-millimeter
center hole surrounded by two rings of sixteen 3.2-millimeter holes
and two rings of sixteen 4.8-millimeter holes. This configuration had
an open area of 57.1 percent. The velocity through the head was
approximately 0.9 m/s and the heating rate was 0.3 GJ/hr (0.3 x
106 Btu/hr). Effects of steam and crosswind were not investigated in
this study. Destruction efficiencies were greater than 99.9 percent
for three of four tests. A 97.8-percent result was obtained in the
only test where the probe was located off the center line of the
flame. The author did not believe this probe location provided a
valid gas sample for analysis.
Howes et al. (1981) studied two commerical flare heads at John
Zink's flare test facility. The primary purpose of this test (which
was sponsored by EPA) was to develop a flare testing procedure. The
commercial flare heads were an LH air assisted head and an LRGO (Linear
Relief Gas Oxidizer) head manufactured by John Zink Company. The LH
flare burned 1,000 kg/hr of commercial propane. The exit gas velocity
based on the pipe diameter was 8 m/sec and the firing rate was 46 GJ/hr
(44 x 106 Btu/hr). The LRGO flare consisted of three burner heads
located 0.9 m apart. The three burners combined fired 1,900 kg/hr of
natural gas. This corresponds to a firing rate of 88 GJ/hr (83.7 x
106 Btu/hr). Steam was not used for either flare, but the LH flare
head was assisted by a forced draft fan in some trials. In 4 of 5
2-16
-------
tests, combustion efficiency was determined to be greater than
99 percent when sampling height was sufficient to ensure the combustion
process was complete. One test resulted in combustion efficiency as
low as 92.6 percent when the flare was operated under smoking conditions.
An excellent, detailed review of all four studies was done by
Payne et al. in January 1982, and a summary of the studies is given in
Table 2-4. A fifth study, by McDaniel et al. 1982, determined the
influence on flare performance of mixing, flare gas energy content and
gas flow velocity. Steam-assisted and air-assisted flares were tested
at the John Zink facility using the procedures developed by Howes.
The test was sponsored by CMA with the cooperation and support of EPA.
All tests were with an 80 percent propylene, 20 percent propane mixture
diluted as required with nitrogen to give different flare gas energy
contents. This was the first work which determined flare efficiencies
at a variety of "nonideal" conditions where lower efficiencies had
been predicted. All previous tests were of flares which burned gases
which were very easily combustible and did not tend to soot. This was
also the first test which used the sampling and chemical analysis
methods developed for EPA by Howes.
The steam assisted flare was tested with exit flow velocities up
to 19 m/sec, with gas heat contents from 11.0 to 81.3 MJ/stdm3 (294 to
2,183 Btu/standard cubic foot [scf]) and with steam to gas (weight)
ratios varying from 0 (no steam) to 6.86. Flares without assist were
tested down to 7.15 MJ/stdm3 (192 Btu/scf). All of these tests,
except for those with very high steam to gas ratios, showed combustion
efficiencies of over 98 percent. Flares with high steam to gas ratios
(about 10 times more steam than that required for smokeless operation)
had lower efficiencies (69 to 82 percent) when combusting 81.3 MJ/stdm3
(2,183 Btu/scf) of gas.
The air assisted flare was tested with flow velocities up to
66.4 m/s and with gas heat contents from 3.1 to 81.3 MJ/stdm3 (83 to
2,183 Btu/scf). Tests at 10.5 MJ/stdm3 (282 Btu/scf) and above gave
over 98 percent efficiency. Tests at 6.26 MJ/stdm3 (168 Btu/scf) gave
55 percent efficiency.
2-17
-------
After consideration of the results of these five tests, EPA has
concluded that 98 percent combustion efficiency can be achieved by
steam-assisted, smokeless flares with exit flow velocities less than
18 m/s and flared gas heat contents of at least 11.2 MJ/stdm3 (300 Btu/
scf) and by flares operated without assist with exit flow velocities
less than 18 m/sec and flared gas heat contents of at least 7.45 MJ/stdm3
(200 Btu/scf). Ninety-eight percent combustion efficiency can also be
achieved by air-assisted, smokeless flares with flared-gas heat contents
over 11.2 MJ/stdm3 (300 Btu/scf) and with exit flow velocities below a
calculated value dependent upon the flared gas heat content. These
are the only conditions for which EPA has data supporting 98 percent
emission reduction by flares. Flares are not normally operated at the
very high steam to gas ratios that resulted in low efficiency in some
tests because steam is expensive and operators make every effort to
keep steam consumption low. Flares with high steam rates are also
noisy and may be a neighborhood nuisance.
Consequently, since the 98-percent value is considered to be a
reasonable estimate of flare efficiency for EB/S plants based on the
limited data available in the literature to date, EPA assigned a
98-percent destruction efficiency to smokeless flares within the EB/S
industry in order to calculate impacts. This efficiency estimate does
not suggest that all flares used in the EB/S industry are only 98 per-
cent efficient but represents a reasonable level of flare efficiencies
at EB/S plants in order to estimate emissions.
2-18
-------
APPENDIX A
EMISSION DATA AND ENVIRONMENTAL IMPACTS
-------
APPENDIX A
EMISSION DATA AND ENVIRONMENTAL IMPACTS
A.I INTRODUCTION
The purpose of this appendix is to show how the baseline benzene
emissions were developed. An explanation of assumptions used in
emission estimations accompanies the discussion. Subsection A.2
discusses the assumptions used in determining baseline benzene
emissions. This subsection also summarizes the benzene emission data
used to determine the environmental impacts of these emissions.
A.2 ORIGINAL AND REVISED BASELINE BENZENE EMISSIONS
A.2.1 U.S. Environmental Protection Agency (EPA) Baseline Benzene
Emission Estimates at Proposal
The emission rates presented at proposal were derived from an
integrated and totally uncontrolled model EB/S plant. Emission rates
were first estimated assuming a constant nameplate production capacity,
100 percent capacity utilization, and no control for four process
emission sources: (1) alkylation reactor vents; (2) atmospheric/
pressure column vents; (3) vacuum column vents; and (4) H2 separation
vents. Variations in emissions potential among plants were averaged,
assuming a constant capacity and no control, to yield a "typical"
emission rate for each emission source in a plant. Emission rates
for the four sources described above were scaled to actual plant
control levels, actual capacity use, and 8,000 operating hr/yr to
derive plant-by-plant emission estimates. In plants where only
ethylbenzene or styrene was produced, the model plant's corresponding
process vents were used to estimate the contribution of that source.
The plant-by-plant emission controls for continuous process
vents used at proposal are listed in Table A-l. At proposal, these
A-l
-------
TABLE A-l. EMISSIONS CONTROLS PRESENT ON EB/S PLANTS AT PROPOSAL, LISTED BY VENT TYPE
i
ro
Emission controls on various vent types
Plant/location
American Hoechst, LA "A"
American Hoechst, TX "B"
Amoco, TX
ARCO, PA
ARCO, TX
Cos-Mar, LA
Dow Chemical, TX "A"
Dow Chemical, TX "B"
Dow Chemical, MI
El Paso Products, TX
Gulf, LA
Monsanto, TX
Oxirane, TX
Sun Oil, TX
U.S. Steel, TX
Alkylation reactor
area vents
LD FLR
RC/RS to SO FLR
RC to BLR via Cmp
-
No C
LD FLR
No C
BLR
-
BLR
LD FLR
BLR
SD FLR
BLR
-
Atmospheric/pressure
column vents
No C
RC/RS to SD FLR
BLR
-
No C
LD FLR
SD FLR
BLR
-
BLR
LD FLR
RC to BLR via Cmp
SD FLR
BLR
BLR via NGE
Benzene/toluene
column vents
No C
SD FLR via NGE
RC to BLR via Cmp
No C
-
RC
No C
RS to SD FLR
No C
SD FLR
RC
RC to BLR via Cmp
-
No C
BLR via NGE
Other vacuum
column vents
No C
SD FLR via NGE
No C
No C
-
No C
No C
RS to SD FLR
No C
SD FLR
RC
RC to BLR via Cmp
-
No C
BLR via NGE
Information from EB/S Docket Item No. II-B-18.
Control device key: FLR -
BLR -
RC -
RS -
Cmp •
Flare (60% efficiency)
Boiler (99% efficiency)
Refrigerated condenser (85% efficiency)
Refrigerated scrubber (85% efficiency)
Compressor
NGE - Natural gas ejector
SD - Small diameter
LD - Large diameter
No C - No control
- Denotes no vents of this type in
that particular plant.
-------
data (II-B-18) were derived from EPA 114 letters, plant visits, and
telephone calls. The estimation of pi ant-by-plant emissions for each
vent involved multiplication of plant nameplate capacity by the sum of
the appropriate emission factors times the specific control achieved.
The estimation of emissions from all plants for a given vent involved
multiplication of the emission factor by the sum of each plant's
nameplate capacity times the fraction of control for the vent at that
plant. The estimation of emissions from the hydrogen separation vent
involved multiplication of the entire industry nameplate capacity by
the emission factor times the fraction of control. This was applic-
able because 99 percent control was assumed for the entire industry.
The data used to estimate emissions are listed in Table A-2.
A.2.1.1 Continuous Emission Estimation--Plant-By-Plant Emissions.
The following example shows how emissions from the American Hoechst "B,"
Bayport, Texas, facility were estimated. PI ant-by-plant emissions
were estimated through the following calculations.
The nameplate capacity was multiplied by the sum of the products
of the emission factor times (Infraction of control]) for each vent
type to determine a plant's emissions. For the American Hoechst "B"
plant, the calculation would be as follows:
(0.67) x (450 x IQ3 Mg/yr) x [0.3 Mg/103 Mg x (1-0.6) x
(1-0.85) + 1.2 Mg/103 Mg x (1-0.6) x (1-0.85) + 0.3 Mg/103 Mg x
(1-0.6) + 0.05 Mg/103 Mg x (1-0.6)] = 70 Mg/yr ,
where
(0.67) x (450 x 103) = utilized fraction of nameplate production
capacity
(0.3/103) x (1-0.6) x (1-0.85) = uncontrolled emission factor
from alkylation reactor area
vents
(1.2/103 ) x (1-0.6) x (1-0.85) = uncontrolled emission factor
from atmospheric/pressure
column vents
(0.3/103) x (1-0.6) = uncontrolled emission factor from benzene/
toluene column vents
(0.05/103) x (1-0.6) = uncontrolled emission factor from other
vacuum column vents.
A-3
-------
TABLE A-2.
NAMEPLATE CAPACITIES, EMISSIONS FACTORS, AND PERCENT CONTROL USED AS THE BASIS
FOR PLANT-BY-PLANT BENZENE EMISSIONS ESTIMATED AT PROPOSAL3
Percent control by vent type (emissions factor )
Alkylation reactor
Capacity area vents
Plant/location (103 Mg/yr) (0.3)
American Hoechst, LA "A" (0.
American Hoechst, TX "B" (0.
Amoco, TX
ARCO, PA
ARCO, TX (1.
Cos-Mar, LA
Dow Chemical , TX "A" (0.
Dow Chemical, TX "B" (0.
Dow Chemical , MI
El Paso Products, TX
Gulf, LA
Monsanto, TX
Oxirane, TX
Sun Oil , TX
U.S Steel, TX
Total emissions per vent
type (Mg/yr)
Plus H2 separation
Plus excess emissions
Total emissions
33X450) 0.6
67)(450) (0.85X0.6)
375 0.99
200
15X227) 0
600 0.6
2) (700) 0
8) (700) 0.99
200
100 0.99
275 0.6
600 0
550 0.6
40 0.99
54
375
Atmospheric/pressure Benzene/ toluene
column vents column vents
(1.2) (0.3)
0 0
(0.85X0.6) 0.6
0.99 0.99
0.
0
0.6 0.85
0.6 0
0.99 (0.8S)(0.6)
0
0.99 0.6
0.6 0.85
0.85 0.85
0.6
0.99 0
0.99
1,380 425
Other vacuum
column vents
(0.05)
0
0.6
0
0
-
0
0
(0.85X0.6)
0
0.6
0.85
0.85
-
0
0.99
100
Total emissions
(Mg/yr)a
250
70
26
70
300
420
160
20
70
16
180
320
290
15
<1
2,208
110
113
2,451
Information from EB/S Docket Item No. II-B-18.
Emissions factors are in Mg/10;! Mg.
Capacity presented in utilization fraction times nameplate capacity, where nameplate capacity alone is given, utilization fraction equals 1
Assumes 8,000 hr/yr of operation.
eAssumes uncontrolled emissions factoi of 2 1 Mg/10'1 Mg, nameplate capacity and 99% control for H2 separation continuous vents at all plants
Excess emissions from the proposal background information document (BID) (EPA-450/3-37-03Sa; p AA3) based on nameplate production capacity
existing controls.
(II-B-18)
and
-------
A.2.1.2 Continuous Emission Estimation—Industry-Wide Emissions
For a Given Vent. The following example shows how industry-wide
emissions from the atmospheric/pressure column vents were estimated.
Each vent's industry-wide emissions were estimated through the following
calculations.
To determine industry-wide emissions for a given vent, multiply
the emission factor by the summed products of the plant capacity
times (l-[fraction of control]) for each vent of that type at each
plant. For the atmospheric/pressure column vents, the calculation
would be as follows:
1.2 x [0.33 x (450) + 0.67 x (450) x (1-0.6) x (1-0.85) +
(375) x (1-0.99) + (1.15) x (227) + 600 x (1-0.6) + 0.2 x
(700) x (1-0.6) + 0.8 x (700) x (1-0.99) + 100 x (1-0.99) +
275 x (1-0.6) + 600 x (1-0.85) + 550 x (1-0.6) + 40 x
(1-0.99)] = 1,380 Mg/yr ,
where
1.2 = emission factor for atmospheric/pressure column vents
(A/P vents) in Mg/103 Mg
0.33 x (450) + 0.67 x (450) x (1-0.6)
x (1-0.85) = A/P vent "uncontrolled capacities" from American
Hoechst, LA, "A" and "B" plants, respectively,
103 Mg/yr (units given as an example for all
plants)
375 x (1-0.99) = A/P vent "uncontrolled capacity" from Amoco, TX
227 x (1.15) = A/P vent "uncontrolled capacity" from ARCO, TX
600 x (1-0.6) = A/P vents "uncontrolled capacity" from Cos-Mar, LA
0.2 x (700) x (1-0.6) + 0.8 x (700) x (1-0.99) = A/P vents "uncon-
trolled capacities"
from Dow Chemical,
TX, "A" and "B"
Plants
100 x (1-0.99) = A/P vent "uncontrolled capacity" from El Paso, TX
275 x (1-0.6) = A/P vent "uncontrolled capacity" from Gulf, LA
600 x (1-0.85) = A/P vent "uncontrolled capacity" froi,; Monsanto, TX
550 x (1-0.6) = A/P vent "uncontrolled capacity" from Oxirane, TX
40 x (1-0.99) = A/P vent "uncontrolled capacity" from Sun Oil, TX.
A-5
-------
i
A.2.1.3 Continuous Emission Estimation—Industry-wide Emissions
From the Hydrogen Separation Vent. The following calculation shows
how industry-wide emissions from the hydrogen separation column vent
were estimated. Because specific pi ant-by-plant data were unavailable
at proposal, the entire industry nameplate capacity was multiplied by
the emission factor for the hydrogen separation vent times (Infraction
of control]) to estimate emissions due to hydrogen separation vents.
The calculation would be as follows:
4,000 x 103 Mg/yr [2.7 Mg/103Mg x (1-0.99)] = 110 Mg/yr,
where
4,000 = industry nameplate capacity
2.7 = emission factor for hydrogen separation vent
(1-0.99) = uncontrolled percent.
A.2.1.4 Excess Emission Estimation. Excess emissions can occur
during (1) plant startup, (2) plant shutdown, and (3) as a result of
either process or air pollution control equipment failure. A
comprehensive discussion of the sources and amounts of excess emissions
is contained in the proposal BID (EPA-450/3-79-035a; p. AA3). In
summary, the proposal BID explains that at proposal, baseline excess
emissions from EB/S facilities totaled about 133 Mg/yr based on 100 per-
cent nameplate production capacity and existing controls.
A.2.2 Revised U.S. Environmental Protection Agency (EPA) Baseline
Benzene Emission Estimates
Since proposal, the Chemical Manufacturer's Association (CMA)
commented that EPA's benzene emission data were overestimated because
the EPA estimates did not consider plant-specific production and
control levels. CMA and individual companies supplied plant-specific
emission data that included emission sources, benzene emission
rates in grams per second for each emission source, control device
efficiency assumptions, and actual operating capacities for continuous
and excess emissions based on actual operating hours per year (IV-D-13,
IV-D-34, IV-D-35, IV-D-36). The grams-per-second benzene emission
flow rate was calculated by CMA from actual annual benzene flow rates
averaged over 8,760 hr/yr.
A-6
-------
EPA used the CMA data to revise benzene emission estimates from
the EB/S industry. EPA converted the benzene emission rate estimates
presented in the industry submittals from grams per second to megagrams
per year assuming 8,760 hr/yr. Revised EPA estimates, in contrast to
original estimates, reflect actual vent-by-vent emission rates,
capacity utilization, and control levels, and assume 8,760 operating
hr/yr. EPA also recalculated emission estimates from flares by
revising the flare destruction efficiency rating of 60 percent, which
CMA used to estimate emissions from flares, to 98 percent (see
Response 2.2.2 of this document). First, EPA calculated uncontrolled
emissions from the CMA emission estimates at 60 percent control.
Second, EPA multiplied the uncontrolled emission rate by 0.02 (=l-.98)
to estimate emissions at 98 percent control. Emission estimates for
each emission source were summed to estimate a plant's total emissions,
and pi ant-by-plant emission estimates were summed to estimate industry-
wide emissions. It should be noted that according to the CMA data
package (IV-D-13), the EB/S industry currently is not operating at
peak capacity and exhibits a 97-percent overall control level (see
Table A-3).
Based on the data supplied by industry and the assumptions previously
noted, EPA reestimated total benzene emissions from the EB/S industry
to be 208 Mg/yr. Of this total, 186 Mg/yr were estimated to result
from continuous process emission sources and about 22 Mg/yr were
estimated to result from excess process emission sources (see Table A-4).
A-7
-------
TABLE A-3. CAPACITY UTILIZATION, PERCENT CONTROL, AND EB/S
PLANTS IN OPERATION AT BASELINE3
Plant/location
American Hoechst, LA
American Hoechst, TX
Amoco, TX
ARCO, PA
Cos-Mar, LA
Dow Chemical , TX
Dow Chemical , MI
El Paso Products, TX
Gulf, LA
Monsanto, TX
Oxirane, TX
Koch, TX
U.S. Steel, TX
Industry-wide
Capacity utilization
(%)
60
100
90
45
100
100
100
100
100
100
100
100
100
90
Overall control
(%)
99
99
99
65
99
95
71
98
99
95
98
98
99
97
Actual production and control levels from CMA data package (IV-D-13) and
subsequent industry submittals (IV-D-34, IV-D-35, IV-D-36) used by EPA
in estimating current pi ant-by-plant benzene emissions. Operating
hours per year assumed to be 8,760.
A-8
-------
TABLE A-4. REVISED EPA BASELINE BENZENE EMISSION ESTIMATES'
Plant/location
American Hoechst, LA
American Hoechst, TX
Amoco, TX
ARCO, PA
Cos-Mar, LA
Dow Chemical , TX
Dow Chemical , MI
El Paso Products, TX
Gulf, LA
Monsanto, TX
Oxirane, TX
Koch, TX
U.S. Steel, TX
SUBTOTAL
TOTAL
Continuous
emissions
(Mg/yr)
0.47
5.89
14.4
37.0
17.2
35.8
57.1
1.34
9.85
4.35
1.20
0.54
0.42
186
208
Excess
emissions
(Mg/yr)
0.25
0.36
0.59
2.45
0.27
1.64
0.20
0.51
0.05
15.5
0
0.04
0.02
22
Revised EPA baseline benzene emissions estimates after proposal. Calcu-
lated from CMA data package (IV-D-13) and subsequent industry submittals
(IV-D-34, IV-D-35, IV-D-36).
A-9
-------
APPENDIX B
METHODOLOGY FOR ESTIMATING LEUKEMIA INCIDENCE AND MAXIMUM
LIFETIME RISK FROM EXPOSURE TO BENZENE EMISSIONS FROM
ETHYLBENZENE/STYRENE PROCESS VENTS
-------
APPENDIX 8
METHODOLOGY FOR ESTIMATING LEUKEMIA INCIDENCE AND MAXIMUM
LIFETIME RISK FROM EXPOSURE TO BENZENE EMISSIONS FROM
ETHYLBENZENE/STYRENE PROCESS VENTS
B.I INTRODUCTION
The purpose of this appendix is to describe the methodology and to provide
the information used to estimate leukemia incidence and maximum lifetime risk
from population exposure to benzene emissions from ethyl benzene/styrene (EB/S)
process vents. The methodology consists of four major components: estimating
annual average concentration patterns of benzene in the region surrounding each
plant, estimating the population associated with each computed concentration,
computing exposure by summing the products of the concentrations and associated
populations, and estimating annual leukemia incidence and maximum lifetime risk
from exposure and concentration estimates. Due to the assumptions made in each
of these four steps of the methodology, there is considerable uncertainty
associated with the lifetime individual risk and leukemia incidence numbers
calculated in this appendix. These uncertainties are explained in Section B.6
of this appendix.
B.2 ATMOSPHERIC DISPERSION MODELING
The long-term version of the Industrial Source Complex (ISCLT) dispersion
model1 was used to estimate annual average benzene concentrations in the vicinity
of 14 EB/S plants. (For the purpose of this analysis the "A" and "B" units of
the Dow, Texas, plant are assumed to be two plants.)
STAR summaries are the principal meteorological input to the ISCLT Model.
A STAR summary is a tabulation of the joint frequency of occurrence of wind-speed
and wind-direction categories, classified according to the Pasquill stability
categories. For this modeling analysis, 5 consecutive years of wind and
stability data were used to develop the STAR summaries. Hourly data were used
where available, otherwise data taken once every 3 hours were used.
The ISCLT Model also requires the input of ambient temperatures and
mixing heights by stability category. National Weather Service stations were
used to determine the average annual mean, maximum and minimum daily temperature
B-l
-------
was assigned to the A, B, and C stability categories; the average annual mean
daily temperature was assigned to the D stability category; and the average
annual minimum daily temperature was assigned to the E and F stability categories.
Mixing height input data to the ISCLT Model was calculated by averaging hourly
CRSTER Model meteorological preprocessor mixing height data by stability category.
Rural mixing heights and stability categories were used for all plants, except
the USS plant in Houston, Texas where urban mixing heights and stability categories
were used.
The receptor grid consists of ten downwind distances located along 16 radials,
The radials are separated by 22.5 degree intervals beginning with 0 degrees and
proceeding clockwise to 337.5 degrees. The ten downwind distances for each
radial are 0.2, 0.3, 0.5, 0.7, 1.0, 2.0, 5.0, 10.0, 15.0 and 20.0 kilometers.
The x=0, y=0 location at each plant was placed at the center of the recepter
grid.
The ISCLT output for all plants modeled, consisting of annual concentration
estimates at all 160 receptors, is contained in the docket (Docket item IV-J-10).
ISCLT dispersion model concentration estimates have been found to be within a
factor of two of measured concentrations in most tests.2
B.3 POPULATION AROUND ETHYLBENZENE/STYRENE PLANTS
The human exposure model (HEM)3 was used to estimate the population that
resides in the vicinity of each receptor coordinate surrounding each EB/S plant.
A slightly modified version of the "Master Enumeration District List-- Extended"
(MED-X) data base is contained in the HEM and used for population pattern
estimation. This data base is broken down into enumeration district/block group
(ED/BG) values. MED-X contains the population centroid coordinates (latitude
and longitude) and the 1970 population of each ED/BG in the United States (50
States plus the District of Columbia). For human exposure estimations, MED-X
has been reduced from its complete form (including descriptive and summary data)
to produce a randomly accessible computer file of the data necessary for the
estimation. A separate file of county-level growth factors, based on the 1978
estimates of 1970 to 1980 growth factor at the county level, has also been
created for use in estimating 1980 population figures for each ED/BG. The
population "at risk" to benzene exposure was considered to be persons residing
within 20 km of EB/S plants. The population around each plant was identified by
specifying the geographical coordinates of that plant.
B-2
-------
B.4 POPULATION EXPOSURE METHODOLOGY
B.4.1 Exposure Methodology
The plant's geographical coordinates and the concentration patterns computed
by the ISCLT were used as input to the HEM. (The HEM also has its own atmospheric
dispersion model. However, the HEM dispersion model, still under development,
is not as detailed as the ISCLT and less well suited for multiple point emission
sources.)
For each receptor coordinate, the concentration of benzene estimated by the
ISCLT and the population estimated by the HEM to be exposed to that particular
concentration are identified. The HEM multiplies these two numbers to produce
population exposure estimates and sums these products for each plant. A two-level
scheme has been adopted in order to pair concentrations and populations prior to
the computation of exposure. The two-level approach is used because the concen-
trations are defined on a radius-azimuth (polar) grid pattern with nonuniform
spacing. At small radii, the grid cells are much smaller than ED/BG's; at large
radii, the grid cells are much larger than ED/BG's. The area surrounding the
source is divided into two regions, and each ED/BG is classified by the region
in which its centroid lies. Population exposure is calculated differently for
the ED/BG's located within each region.
For ED/BG centroids located between 0.1 km and 2.8 km from the emission
source, populations are divided between neighboring concentration grid points.
There are 96 (6 x 16) polar grid points within this range. Each grid point has
a polar sector defined by two concentric arcs and two wind direction radials.
Each of these grid points is assigned to the nearest ED/BG centroid identified
from MED-X. The population associated with the ED/BG centroid is then divided
among all concentration grid points assigned to it. The exact land area within
each polar sector is considered in the apportionment.
For population centroids between 2.8 km and 20 km from the source, a
concentration grid cell, the area approximating a rectangular shape bounded by
four receptors, is much larger than the area of a typical ED/BG (usually 1 km in
diameter). Since there is a linear relationship between the logarithm of
concentration and the logarithm of distance for receptors more than 2 km from
the source, the entire population of the ED/BG is assumed to be exposed to the
concentration that is geometrically interpolated radially and arithmetically
interpolated azimuthally from the four receptors bounding the grid cell.
Concentration estimates for 80 (5 x 16) grid cell receptors at 2.0, 5.0, 10.0,
15.0, and 20.0 km from the source along each of 16 wind directions are used as
reference points for this interpolation.
B-3
-------
In summary, two approaches are used to arrive at coincident concentration/
population data points. For the 96 concentration points within 2.8 km of
the source, the pairing occurs at the polar grid points using an apportionment
of ED/BG population by land area. For the remaining portions of the grid,
pairing occurs at the ED/BG centroids themselves, through the use of log-log
and linear interpolation. (For a more detailed discussion of the methodology
used to estimate exposure see Reference 3.)
B.4.2 Total Exposure
Total exposure (persons-pg/m3) is the sum of all multiplied pairs of -
concentration-population computed by the previously discussed methodology:
N
Total exposure = z (PiCj) (1)
i=l
where
P-j = population associated with point i,
C-j = annual average benzene concentration at point i, and
N = total number of polar grid points between 0 and 2.8 km and ED/BG
•centroids between 2.8 and 20 km.
The computed total exposure is then used with the unit risk factor to
estimate leukemia incidence and maximum lifetime individual risk. This
methodology is described in the following sections. (Note. "Exposure" as used
here is the same as "dosage" in the computer printout, docket item IV-J-10).
B.4.3 Unit Risk Factor
The unit risk factor (URF) for benzene is 9.9 x 10~8 (leukemia cases per yea
Ug/m3 -person years), as calculated by EPA's Carcinogen Assessment Group (CAG).
This factor is slightly lower than the factor derived by CAG at proposal.
Arguments have been advanced that, in addition to the conservative nature
of the model used, the assumptions made by EPA (CAG) in the derivation of
a unit leukemia risk factor for benzene represented "serious misinterpretation"
of the underlying epidemiological evidence. Among the specific criticisms
are: CAG (1) inappropriately included in its evaluation of the Infante et
al. study two cases of leukemia from outside the cohort, inappropriately
excluded a population of workers that had been exposed to benzene, and
B-4
-------
improperly assumed that exposure levels were comparable with prevailing
occupational standards; (2) accepted, in the Aksoy et al. studies, an unreasonable
undercount of the background leukemia incidence in rural Turkey, made a false
adjustement of age, and under-estimated the exposure duration; and (3) included
the Ott et al. study in the analysis despite a lack of statistical significance.
EPA has reexamined and reevaluated each of the three studies. In summary,
EPA concluded that one case of leukemia was inappropriately included from the
Infante et al. study in computing the original unit risk factor. Additionally,
EPA reaffirmed its decision to exclude dry-side workers from that study in
developing the risk factor. The Agency agrees that the Aksoy et al. study was
adjusted improperly for age; however, the exposures and durations of exposures
are still considered reasonable estimates. The Ott et al. study was not eliminated
from the risk assessment because the findings meet the test of statistical
significance and because it provides the best documented exposure data available
from the three epidemiological studies.
Based on these findings, the unit risk factor (the probability of an
individual contracting leukemia after a lifetime of exposure to a benzene
concentration of one part benzene per million parts air) was recalculated. The
revised estimate resulted in a reduction of about 7 percent from the original
estimate of the geometric mean, from a probability of leukemia of 0.024/ppm4 to
a probability of leukemia of 0.022/ppm.
B.4.4 Calculation of Estimated Annual Leukemia Incidence
The annual leukemia incidence associated with a given plant is the product
of the total exposure around that plant in yg/m3-persons and the unit risk
factor, 9.9 x 10'8. Thus,
Annual leukemia incidence = (total exposure) x (unit risk factor), (2)
where total exposure is calculated according to Equation 1.
B.4.5 Calculation of Maximum Lifetime Risk
The populations in areas surrounding EB/S plants have various risk levels
of leukemia incidence from exposure to benzene emissions. Using the maximum
annual average concentration of benzene to which any person is exposed, it is
B-5
-------
possible to calculate the maximum lifetime risk of leukemia (lifetime probability
of leukemia to persons exposed to the highest concentration of benzene)
attributable to benzene emissions using the following equation:
Maximum lifetime risk = C-j )max x (URF) x 70, (3)
where
Ci,max = tne maximum annual average concentration at any receptor
location where exposed persons reside,
URF = the unit risk factor, 9.9 x 1Q-8 and
70 years = average individual's life span.
B.5 LEUKEMIA INCIDENCE AND MAXIMUM LIFETIME RISK
B.5.1 Input Data, Assumptions, and Methodology
Population exposures were computed for the current level of control for
each plant in order to estimate the baseline leukemia incidence and maximum
lifetime risk.
The plants and their locations are shown in Table B-l. The emission rates
and other dispersion model inputs are shown in Table B-2. Total exposure and
maximum annual average concentrations for each plant are shown in Table B-3.
After population exposure at baseline was computed, leukemia incidence
estimates were made. Population exposure was multiplied by the CAG leukemia
risk factor of 9.9 x 10"8 to estimate the annual leukemia incidence for each
plant.
Maximum lifetime risk for a given plant is found by multiplying the maximum
annual average concentration by the unit risk factor of 9.9 x 10~8 times 70 years
(to obtain a lifetime estimate). The maximum lifetime risk for the industry is
that due to the plant with the highest maximum lifetime risk in the industry.
B-6
-------
TABLE B-l. EB/S PLANTS AND LOCATIONS
Plant
Location
Latitude
Longitude
American Hoechst
American Hoechst
Amoco
ARCO
Cos Mar
Dow
Dow (A)
Dow (B)
El Paso Products
Gulf Oil
Monsanto
Oxirane
Sun Oil
U.S. Steel
Baton Rouge, LA
Bayport, TX
Texas City, TX
Kobuta, PA
Carville, LA
Midland, MI
Freeport, TX
Freeport, TX
Odessa, TX
Dona!dsonvilie, LA
Texas City, TX
Channel view, TX
Corpus Christi, TX
Houston, TX
30°33'08"
29°36'10"
29°21'58"
40°39'21"
30°14'16"
43°35'42"
28°57'15"
28°59'17"
31°49'27"
30°05'44"
29°22'44"
29049'07"
27049'57"
29°42'18"
91°12'40"
95°01'15"
94°55'45"
80°2T20"
91°04'09"
84°12'18"
95°19'08"
95°24'09"
102°19'29"
90°55'19"
94°53'40"
95006'07"
97°3T38"
95°15'06"
B-7
-------
TABLE H-2. EMISSIONS DATA FOR AIR MODELING: BASELINE
Plant/
Location
Am. Ho. /LA
Am. Ho./TX
Amoco/ TX
ARCO/PA
Cos-Ma r/LA
r ~ "*
Emission
Type
C
C
C
E
C
C
C
C
E
C
C
C
C
E
E
C
C
E
C
C
C
C
C
C
C
C
E
E
Control
Device
Absorber
Superheater
Superheater
Flare
Boiler
Boiler
Boiler
Superheater
Flare
None
None
Superheater
Furnace
Flare
None
None
Superheater
None
Condenser
Condenser
Flare
Flare
Superheater
Superheater
None
None
Flare
Flare
Benzene
Rate
(9/s)
0.002
0.002
0.011
0.008
0.0388
0.0388
0.0388
0.0705
0.0115
0.0016
0.0006
0.2854
0.1674
0.0029
0.0158
1.15
0.0236
0.0778
0.0158
0.0158
0.0111
0.0129
0.225
0.262
0.0015
0.0015
0.004
0.0047
Temp.
(°K)
300
561
561
977
455
455
455
460
1,366
330
330
606
611
823
330
318
530
318
280
280
,144
,144
569
577
318
318
,144
,144
Location (m)
(x) (y)
33 0
(28) 46
(28) 30
(486) (84)
24 (7)
42 (7)
62 (7)
(84) 13
30 (107)
(69) (1)
(103) (6)
0 0
15 (2)
(292) 121
(111) (31)
0 0
23 (21)
14 (30)
(73) (158)
61 (158)
35 11
(55) 29
62 (253)
(27) (217)
(22) (157)
64 (160)
35 11
(55) 29
Stack
Height
(m)
12.8
32.0
50.3
2.13
27.1
27.1
27.1
71.5
82.3
4.88
4.88
53.3
31.1
83.3
4.88
25.9
30.4
18.23
12.2
9.1
53.3
53.3
53.3
54.9
11.8
11.8
53.3
53.3
Stack
Dia.
(m)
0.07
1.67
2.84
2.1
1.8
1.8
1.8
3.4
1.06
0.052
0.052
2.85
1.98
0.59
0.052
0.30
1.22
0.1
0.05
0.05
0.4063
0.4063
2.83
3.25
0.05
0.05
0.4063
0.4063
Stack
Velocity
(m/s)
3.7
3.4
5.5
10.0
10.4
10.4
10.4
4.0
7.9
1.72
0.57
5.9
7.22
64.0
4.80
0.135
3.13
0.1
0.1
0.1
0.0014
0.0015
4.51
4.11
0.1
0.1
22.4
26.1
Emission
Direction
U
u
u
U
u
u
u
u
u
u
u
u
u
u
u
D
u
u
D
0
U
u
|J
U
D
D
u
U
Plant
Coordinates
Lat. Long.
30°33'08" 91°12'40'
29036'10" 95001'15'
29°21'58" 94055'45M
40039'21" 80°21'20"
30°14'16" 91°04'09"
Temp
(°C)
20
21
21
10
20
Pressure
(millibars)
1,015
1,016
1,017
990
1,016
CD
CO
-------
TABLE B-2. EMISSIONS DATA FOR AIR MODELING: BASELINE (Continued)
Plant/
Location
Oow/MI
El Paso/TX
Dow/TX/B
[
Emission
Type
C
C
C
C
C
E
E
C
C
C
E
C
C
C
C
C
C
E
E
E
C
C
C
C
C
C
C
C
I C
c
C
c
c
E
E
E
Control
Device
Condenser
Stripper
Furnace
Furnace
Condenser
None
None
Oil heater
Superheater
Flare
Flare
Furnace
Boiler
Boiler
Boiler
None
Flare
None
None
None
Scrubber
Scrubber
Flare
Flare
Flare
Furnace
Furnace
Furnace
Furnace
Furnace
Furnace
Furnace
Furnace
None
None
None
Benzene
Rate
(9/s)
1.70
0.0633
0.0230
0.0230
0.0003
0.0032
0.0032
0.0012
0.0086
0.0328
0.0161
0.0014
0.0072
0.0072
0.0072
0.0032
0.00001
0.0066
0.021
0.001
0.4603
0.4603
0.0040
0.0040
0.0040
0.0978
0.0374
0.0027
0.0027
0.0027
0.0027
0.0043
0.0043
0.0023
0.0014
0.0001
Temp.
(°K)
358
316
422
422
300
314
314
789
444
823
823
433
433
433
433
373
823
373
294
373
294
294
823
823
823
464
503
673
673
673
673
623
623
311
311
311
Location (m)
(x) (y)
0 0
95 41
107 91
107 143
1 0
106 58
106 108
(90) (50)
50 (20)
200 80
(460) (620)
(1,261) 213
(1,278) 248
(1,283) 238
(1,286) 228
(1,013) 267
(1,089) 109
(1,186) 226
(1,186) 226
(1,101) 128
(50) (110)
80 (40)
(135) (50)
40 (115)
415 (235)
(45) 50
95 (20)
98 (2)
102 (5)
116 (8)
125 (11)
85 2
80 10
(30) 45
90 0
110 (20)
Stack
Height
(m)
14.9
0.3
12.7
12.7
0.3
13.5
13.5
15.24
57.9
20
37
55.0
18.0
18.0
18.0
9.0
27.6
21.0
21.0
21.0
29.0
29.0
27.1
27.1
27.1
29.5
18.3
16.8
16.8
16.8
16.8
20.4
20.4
15.2
18.3
18.3
Stack
Dia.
(m)
0.05
0.055
0.430
0.430
0.001
0.200
0.200
1.37
2.05
0.102
0.760
3.66
1.68
1.68
1.68
0.300
0.08
2.13
2.13
0.15
0.076
0.076
0.482
0.482
0.482
1.5
1.2
1.4
1.4
1.4
1.4
0.900
0.900
1.5
0.6
0.500
Stack
Velocity
(
-------
TABLE B-2. EMISSIONS DATA FOR AIR MODELING: BASELINE (Continued)
Plant/
Location
Oxirane/TX
Gulf/LA
Monsanto/TX
Koch/TX
USS/TX
Dow/TX A
Emission
Type
C
C
C
C
C
C
C
C
E
C
C
C
C
C
C
C
E
E
C
C
E
C
C
C
E
C
E
Control
Device
Flare
Condenser
Condenser
Condenser
Condenser
Flare
Superheater
Reheater
Flare
Scrubber
Condenser
None
None
Boiler
Superheater
Superheater
None
Flare
Boiler
Flare
Flare
Superheater
Reboller
Reboiler
None
Furnace
None
Benzene
Rate
(9/s)
0.0380
0.0003
0.0003
0.0020
0.0003
0.0214
0.208
0.080
0.0017
0.0047
0.0002
0.0018
0.0155
0.0478
0.0339
0.0339
0.4899
0.0027
0.006
0.011
0.0013
0.0016
0.0058
0.0058
0.0005
0.0228
0.0195
Temp.
CK)
823
298
298
298
298
993
611
566
993
311
305
324
294
389
505
505
294
823
467
930
930
519
575
575
327
505
294
Location (m)
(x) (y)
0 0
(137) 164
(82) 128
(68) 120
(53) 97
0 0
(119) 178
(103) 186
0 0
(91) (36)
19 (58)
(461) (125)
(60) 57
325) (180)
108) 72
108) 25
42 (43)
(25) 137
0 0
(49) 425
(49) 425
(22) (13)
(5) (1)
5 1
(13) 41
43 51
0 0
Stack
Height
(M)
33.6
10.52
11.79
11.79
10.29
45.72
40.54
46.94
45.72
4.08
28.65
0.76
5.18
15.2
51.8
51.8
41.1
41.4
30.0
43.0
43.0
38.1
59.44
59.44
62.38
39.33
12.2
Stack
Dia.
(m)
0.36
0.05
0.05
0.05
0.05
0.41
2.83
1.52
0.41
0.152
0.152
0.102
0.102
1.60
3.05
3.05
0.050
0.41
1.8
0.5
0.5
1.22
2.35
2.35
0.203
2.44
0.076
Stack
Velocity
(m/s)
0.556
0.1
0.1
0.1
0.1
0.17
3.4
4.3
24.0
0.1
0.1
0.1
0.1
34.74
5.26
5.26
10.76
0.417
12.0
0.2
25.0
9.15
7.80
7.80
7.20
3.32
3.65
Emission
Di rection
U
D
0
D
D
U
U
U
U
H
D
0
D
U
U
U
U
U
U
U
U
U
U
U
U
U
Plant
Coordinates
Lat. Long.
29°49'07" 95°06'07"
30°05'44" 90°55'19"
29°22'44" 94°53'40"
27049.57" 97031>38«
29°42'18" 95°15'06"
28°57'15" 95°19'08"
Temp.
CO
21
20
21
22
21
21
Pressure
(millibars)
1,016
1,016
1,017
1,014
1,016
1,016
CO
I—*
o
-------
TABLE B-3. EXPOSURE AND MAXIMUM ANNUAL AVERAGE CONCENTRATIONS
Plant
Total Exposure Maximum Annual
(ug/m3 - Average Concentrations
person) (yg/m^)
American Hoechst, LA
American Hoechst, TX
Amoco, TX
ARCO, PA
Cos Mar, LA
Dow, MI
Dow, TX (A)
Dow, TX (B)
EL Paso Products, TX
Gulf Oil, LA
Monsanto, TX
Oxirane, TX
Sun Oil, TX
U.S. Steel , TX
94
622
1,390
14,700
449
24,500
188
8,420
328
245
6,370
362
30
190
1.59x10-2
1.69x10-2
9. 54X10-1
2.09
3.96x10-1
1.96xlQl
2.1x10-1
1.11
7.83x10-2
3.56x10-2
1.37
3.02x10-2
5.07xlO-3
8.21x10-4
B-ll
-------
B.5.2 Example Calculations
B.5.2.1 Leukemia Incidence. As an example for calculating leukemia incidenc
the ARCO plant is used. Under the current level of control, the annual leukemia
incidence is computed according to Equation 2 as follows:
Annual leukemia incidence per year = 14,700 x 9.9 x 10~8
Annual leukemia incidence per year = 1.5 x 10"3
B.5.2.2 Maximum Lifetime Risk. Again, ARCO is used to illustrate the
calculation. Under the current level of control, the maximum lifetime risk is
computed according to Equation 3 as follows:
Maximum lifetime risk = 2.09 x 9.9 x 10~8 x 70
Maximum lifetime risk = 1.4 x 1Q~5
B.5.3 Summary of Impacts
The methodology for calculating baseline leukemia incidence and maximum
lifetime risk (described in Section B.5.1) was extended to each plant. The
baseline scenario is the current level of control or the level of control to
be implemented in the near future that would have occurred regardless of this
standard. The estimated leukemia incidence is shown in Table B-4. The
estimated nationwide leukemia cases per year under the assumed baseline level
of control is about 5.7 x 10~3. The estimated maximum lifetime risk is shown
in Table B-5. The estimated maximum lifetime risk under the assumed baseline
level of control is about 1.4 x 10-4.
B.6 UNCERTAINTIES
Estimates of both leukemia incidence and maximum lifetime risk are primarily
functions of estimated benzene concentrations, populations, the unit risk factor,
and the exposure model. The calculations of these variables are subject to a
number of uncertainties of various degrees. Some of the major uncertainties are
identified below.
B-12
-------
TABLE B-4. ESTIMATED ANNUAL LEUKEMIA INCIDENCE (xlO'4)
Plant Baseline
American Hoechst, LA 0.093
American Hoechst, TX 0.62
Amoco, TX 1.4
ARCO, PA 15
Cos Mar, LA 0.44
Dow, MI 24
Dow, TX (A) 0.19
Dow, TX (B) 8.3
El Paso Products, TX 0.32
Gulf Oil, LA 0.24
Monsanto, TX 6.3
Oxirane, TX 0.36
Sun Oil, TX 0.030
U.S. Steel, TX 0.19
TOTAL 57
B-13
-------
TABLE B-5. ESTIMATED MAXIMUM LIFETIME RISK (XI0-6)
Plant Baseline
American Hoechst, LA 0.11
American Hoechst, TX 0.12
Amoco, TX 6.6
ARCO, PA 14
Cos Mar, LA 2.7
Dow, MI 140
Dow, TX (A) 1.5
Dow, TX (B) 7.7
El Paso Products, TX 0.54
Gulf Oil, LA 0.25
Monsanto, TX 9.5
Oxirane, TX 0.21
Sun Oil, TX 0.035
U.S. Steel, TX 0.0057
B-14
-------
B.6.1 Benzene Concentrations
Modeled ambient benzene concentrations depend upon: (1) plant configuration;
(2) emission point characteristics, which can be different from plant to plant;
(3) emission rates which may vary over time, and from plant to plant; and
(4) rneterology, which is seldom available for a specific plant. The particular
dispersion model used can also influence the numbers. Using a different
dispersion model, even with the same emission data input, can produce different
results. The dispersion modeling also assumes that the terrain in the vicinity
of the source is flat. For sources located in complex terrain, the maximum
annual concentration could be underestimated by several fold due to this
assumption. Assuming the inputs to the dispersion model are accurate, the
predicted benzene concentrations are considered to be accurate to within a
factor of 2.
B.6.2 Exposed Populations
Several simplifying assumptions were made with respect to the assumed
exposed population. The location of the exposed population depends on the
accuracy of the census data in the HEM. In addition, the exposed population is
assumed to be immobile, remaining at the same location 24 hours per day, 365 days
per year, for a lifetime (70 years). This assumption may be counterbalanced to
some extent (at least in the calculation of incidence) by the assumption that
no one moves into the exposure area either permanently as a resident or temporarily
as a transient. The population "at risk" was assumed to reside within 20 km of
each plant, regardless of the estimated concentration at that point. The
selection of 20 km is considered to be a practical modeling stop-point. The
results of dispersion modeling are felt to be reasonably accurate within that
distance. The dispersion coefficients used in modeling are based on empirical
measurements made within 10 kilometers of sources. These coefficients become
less applicable at long distances from the source, and the modeling results
become more uncertain. A numerical estimate of the accuracy of these assumptions
regarding the exposed population is not available.
B.6.3 Unit Risk Factor
The unit risk factor contains uncertainties associated with the occupational
studies of Infante, Aksoy, and Ott, and the variations in the dose/response
relationships among the studies. Other uncertainties regarding the occupational
B-15
-------
studies and the workers exposed that may affect the unit risk factor were raised
during the public comment period and focus on assumptions and inconclusive data
contained in the studies. However, those uncertainties have not been quantified
B.6.4 Other Uncertainties
There are several uncertainties associated with estimating health impacts.
Maximum lifetime risk and annual incidence were calculated based on a no-threshol
linear extrapolation of leukemia risk associated with a presumably healthy white
male cohort of workers exposed to benzene concentrations in the parts per million
range compared to the risk associated with the general population, which includes
men, women, children, nonwhites, the aged, and the unhealthy, who are exposed to
concentrations in the parts per billion range. It is uncertain whether these
widely diverse segments of the population have susceptabilities to leukemia that
differ from that of workers in the studies. Furthermore, while leukemia is the
only benzene health effect considered in these calculations, it is not the only
possible health effect. Other health effects, such as aplastic anemia and
chromosomal aberrations, are not as easily quantifiable and are not reflected in
the risk estimates. Although these other health effects have been observed at
occupational levels, it is not clear if they can result from ambient benzene
exposure levels. Additionally, benefits that would affect the general population
as the result of indirect control of other organic emissions in the process of
controlling benzene emissions from ethylbenzene/styrene plants are not quantified
Possible benzene exposures from other sources also are not included in the
estimate. For example, an individual living near an ethyl benzene/ styrene plant
is also exposed to benzene emissions from automobiles. Finally, these estimates
do not include possible cumulative or synergistic effects of concurrent
exposure to benzene and other substances.
B-16
-------
B.7 REFERENCES
1. U.S. Environmental Protection Agency. Industrial Source Complex (ISC)
Dispersion Model User's Guide, Volume I. Research Triangle Park,
North Carolina. EPA-450/4-79-031. 1979
2. U.S. Environmental Protection Agency. An Evaluation Study for the
Industrial Source Complex (ISC) Dispersion Model. Research Triangle
Park, North Carolina. EPA-450/4-81-002. 1981.
3. Systems Applications, Inc. Human Exposure to Atmospheric Concentrations
of Selected Chemicals. (Prepared for the U.S. Environmental Protection
Agency, Research Triangle Park, North Carolina). Volume I (NTIS
No. PB 81 193252) and Volume II (NTIS No. PB 81 193260). May 1980.
4. Albert , R. E. Carcinogen Assessment Group's Final Report on Popula-
tion Risk to Ambient Benzene Exposures. U.S. Environmental Protection
Agency. Publication No. EPA-450/5-80-004. January 1979.
B-17
-------
TECHNICAL REPORT DATA
(Please read Instructions on tne reverse before cumpietingl
. REPORT NO
EPA-450/3-84-003
13. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
Benzene Emissions from Ethylbenzene/styrene Plants--
Background Information for Proposal to Withdraw
Prnnnspd
AUTHOR(S)
7 AUT
5. REPORT DATE
March 1984
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NO.
9 PERFORMING ORGANIZATION NAME AND ADDRESS
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
68-02-3056
12. SPONSORING AGENCY NAME AND ADDRESS
DAA for Air Quality Planning and Standards
Office of Air and Radiation
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final
14. SPONSORING AGENCY CODE
EPA/200/04
15. SUPPLEMENTARY NOTES
JTRACT ' ~ ' ~~~ —— —
This document contains information that formed the basis for the decision to
withdraw standards for the ethylbenzene/styrene industry proposed December 18, 1980
(45 FR 83448). The report includes a summary of industry changes since proposal,
a summary of public comments relevant to the withdrawal decision, and the rationale
for the decision to withdraw proposed standards.
7.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
Air pollution
Pollution control
National emission standards for hazardous
air pollutants
Benzene
Ethyl benzene
Styrene Hazardous air pollutants
b IDENTIFIERS/OPEN ENDED TERMS
Air pollution control
c. COSATI 1 ield/0roup
13B
UTION STATEMEN1
Unlimi ted
; 19 SECURITY CLASS < Thil Report I
\ Unclassified
I 20 SECUR1TV CLASS -This page,
! Unclassified
i 21 .NIC. OF PAGES
L_J.3_
•22 PRICE
~j
HPA Form 2220-1
------- |