&EPA
United States
Environmental Protection
Agency
Office of Air Quality
Planning and Standards
Research Triangle Park NC 27711
EPA-450/3-84-004
March 1984
Air
Benzene Emissions
From Benzene
Storage Tanks —
Background
Information for
Proposal to Withdraw
Proposed Standards
-------
EPA-450/3-84-004
Benzene Emissions from
Benzene Storage Tanks —
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 221 61.
-------
ENVIRONMENTAL PROTECTION AGENCY
Background Information
Benzene Storage Tanks
Prepared by:
Ja£k 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 83952; December 19, 1980) for benzene
emissions from all existing and new Benzene Storage Tanks.
2. Copies of this document have been sent to the following Federal
Departments; Labor, Health and Human Services, Defense, Transportation,
Agriculture, Commerce, Interior, and Energy; the National Science
Foundation; the Council on Environmental Quality; members of the
State and Territorial Air Pollution Program Administrators; the
Association of Local Air Pollution Control Officials; EPA Regional
Administrators; and other interested parties.
3. The comment period for review of this document is 30 days.
Mr. Gilbert H. Wood may be contacted 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
5. Copies of this document may be obtained from:
U.S. EPA Library (MD-35)
Research Triangle Park, NC 27711
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
-------
TABLE OF CONTENTS
Section page
LIST OF FIGURES v
LIST OF TABLES vi
1 SUMMARY i-i
1.1 Summary of Changes Since Proposal 1-1
1.2 Summary of Proposal to Withdraw the
Proposed Standards 1-1
2 SUMMARY OF PUBLIC COMMENTS 2-1
2.1 Selection of Benzene Storage Tanks for Regulation . . 2-1
2.2 Health and Environmental Impacts 2-14
APPENDICES
A EMISSIONS SOURCE TEST DATA AND ANALYSIS A-l
B METHODOLOGY FOR ESTIMATING LEUKEMIA INCIDENCE
AND MAXIMUM LIFETIME RISK FROM EXPOSURE TO
BENZENE STORAGE TANKS B-l
IV
-------
LIST OF FIGURES
Figure page
A-l Process and instrumentation schematic A-3
A-2 Plan view of noncontact bolted IFR A-5
A-3 Elevation view of noncontact bolted IFR in test tank . . A-6
A-4 Plan view of contact welded IFR A-7
A-5 Elevation view of contact welded IFR in test tank . . . . A-8
A-6 Plan view of contact bolted IFR A-9
A-7 Elevation view of contact bolted IFR in test tank .... A-10
A-8 Example of fitting emission bench test apparatus .... A-19
-------
LIST OF TABLES
Tab1e Page
1-1 Changes in Nationwide Impacts 1-2
2-1 List of Commenters on the Proposed National
Emissions Standards for Hazardous Air Pollutants
for Benzene Storage Vessels 2-2
2-2 Vessels Containing Mixtures That May be More Than
10 Percent Benzene 2-13
2-3 Comparison of Emissions as Calculated from the EPA
Series and the 2519/2517 Series 2-20
2-4 Internal Floating Roof Tank Emissions by Source 2-22
2-5 Emissions from a Typical Benzene Storage Vessel 2-23
2-6 Comparison of Convective and Permeability Losses from
Internal Floating Roof Seal Systems in the Model Tank. . . 2-28
2-7 Model Tank Emissions (Mg/yr) from a Fixed Roof Tank
and a Typical Internal Floating Roof Tank 2-29
2-8 Emissions from New and Existing Model Plants 2-30
2-9 Nationwide Emissions from New and Existing
Benzene Storage Tanks . 2-31
A-l Summary of Test Conditions for Phase 1 and 1R A-ll
A-2 Summary of Test Conditions for Phase 2 and 2R A-13
A-3 Summary of Test Conditions for Phase 3 and 3R A-15
A-4 Summary of Test Results for All
Potentially Relevant Tests A-16
A-5 Summary of IFR Deck Fitting Emission Tests A-20
A-6 Permeability of Polyurethane Coated Nylon Fabric A-21
A-7 Comparison of Wiper Seals to Foam-Filled
Vapor-Mounted Seals A-23
A-8 Comparison of Liquid-Mounted Seal to
Vapor-Mounted Seal A-24
A-9 Bolted Deck Seam Emissions A-26
A-10 Comparison of Emissions as a Function of Liquid Type . . . A-27
(continued)
vi
-------
LIST OF TABLES (Concluded)
Page
Plants and Locations for Benzene Storage Tanks B-4
Model Inputs for Each Type of Model Plant B-10
Estimated Maximum Concentration and Exposure
for Benzene Storage Tanks B-16
B-4 Estimated Nationwide Health Impacts
for Benzene Storage Tanks B-21
VII
-------
1. SUMMARY
On December 19, 1980, the U.S. Environmental Protection Agency
(EPA) proposed National Emissions Standards for Hazardous Air Pollutants
(NESHAP) for benzene storage vessels under the authority of Section 112
of the Clean Air Act. The proposed standards were published in the
Federa1 Register (45 FR 83952) with a request for public comment. A
public hearing was held on June 9, 1981. Six individuals representing
three organizations made presentations. A total of 22 comments from
industry, two trade associations, and an environmental group were submitted
during the comment period. Comments submitted relevant to the withdrawal
decision and EPA's responses 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 CHANGES SINCE PROPOSAL
Since the standards for benzene emissions from benzene storage
vessels were proposed (December 19, 1980; 45 FR 83952), estimated benzene
emissions from this source category have declined considerably. This
estimated reduction is due to revised emission factors based on new test
data acquired since proposal. The basis for the revised emission factors
is discussed in more detail in Section 2.2.2.1 of this document. Table 1-1
compares the estimated nationwide baseline benzene emission and health
impacts due to benzene storage vessels at proposal with current estimated
impacts.
1.2 SUMMARY OF PROPOSAL TO WITHDRAW THE PROPOSED STANDARDS
The Administrator is proposing to withdraw the proposal of the
benzene standards for benzene storage vessels. This decision is based
on several factors, including the broad amount of control currently
-------
Table 1-1. CHANGES IN NATIONWIDE IMPACTS
Impact
At proposal
Current
Benzene emissions (Mg/yr)
Leukemia incidence (cases/yr)
Maximum lifetime risk
2,200
0.12 to 0.82
1.5 x 10"4
to
1.0 x 10
-3
620
0.043
3.6 x 10
-5
1-2
-------
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 inability to reduce health risks
significantly with additional controls. This decision is discussed in
greater detail in Section 2.1.2.
1-3
-------
2. SUMMARY OF PUBLIC COMMENTS
A list of commenters, their affiliations, and the EPA docket entry
number assigned to each comment are shown in Table 2-1. Twenty-two
letters commenting on the proposed standards and the Background Information
Document (BID) for the proposed standards were received. Because the
proposed standards are being proposed for withdrawal, only comments and
responses relevant to that decision are addressed in this document.
Significant comments have been combined into the following two categories:
2.1 Selection of Benzene Storage Tanks for Regulation
2.2 Health and Environmental Impacts
2.1 SELECTION OF BENZENE STORAGE TANKS FOR REGULATION
2-1.1 Selection of Source Category
Several commenters contended that the proposed benzene storage
emissions standard is not needed and, therefore, should be withdrawn.
These comments address the following: (1) significance and relative
proportion of risk associated with benzene storage emissions; (2) dupli-
cation of federal and State regulations and guidelines; (3) information
indicating that risks are smaller than estimated in the preamble to the
proposed standards; (4) acceptable residual risk; and (5) lack of data
to demonstrate risk.
Comment: Tnree commenters stated that the EPA has not demonstrated
that benzene storage emissions, relative to other benzene source
categories, pose a significant risk that merits the adoption of a benzene
storage standard (IV-D-lOa, IV-D-16, IV-F-1). One of these commenters
•(IV-D-10a) stated that Section 112 requires that a NESHAP be established
at the level that in the Administrator's judgement provides "an ample
margin of safety to protect the public health from such hazardous air
pollutant." According to the commenter the Supreme Court has held that,
absent a "clear mandate" from Congress to eliminate all risk, the statutory
-------
Table 2-1. LIST OF COMMENTERS ON THE PROPOSED
NATIONAL EMISSIONS STANDARDS FOR HAZARDOUS AIR POLLUTANTS
FOR BENZENE STORAGE VESSELS
Docket entry number3 Commenter/affillation
IV-D-1 Edward W. Warren
Kirkland and Ellis
1776 K Street, Northwest
Washington, D.C. 20006
Counsel for the American Petroleum
Institute
IV-D-2 R. W. Bogan
GATX Terminals Corporation
120 South Riverside Plaza
Chicago, Illinois 60606
IV-D-3 Edward W. Warren
Kirkland and Ellis
1776 K Street, Northwest
Washington, D.C. 20006
Counsel for the American Petroleum
Institute
IV-D-4 John T. Barr
Air Products and Chemicals, Inc.
Box 538
Allentown, Pennsylvania 18105
IV-D-5 John Heinz
Unites States Senate
Committee on Energy and Natural
Resources
Washington, D.C. 20510
With attachment from Sun Petroleum
Production Company
IV-D-6 J. C. Pullen
Celanese Fibers Company
Box 32414
Charlotte, North Carolina 28232
IV-D-7 Herman A. Fritscher
Cities Service Company
Box 300
Tulsa, Oklahoma 74102
(continued)
2-2
-------
Table 2-1. Continued
Docket entry number3 Commenter/affiliation
IV'D-8 E. M. Vancura
Union Oil Company of California
Box 7600
Los Angeles, California 90051
IV-D'9 D. P. Martin
Gulf Oil Company
Post Office Box 2001
Houston, Texas 77001
IV"D'10 Geraldine V. Cox
Chemical Manufacturers Association
2501 M Street, Northwest
Washington, D.C. 20037
IV-D-10a Lance S. Granger
Chemical Manufacturers Association
2501 M Street, Northwest
Washington, D.C. 20037
Attachment to docket entry IV-D-10
IV"D"n Paul J. Sienknecht
The Dow Chemical Company
Midland, Michigan 48640
IV"D'12 Alfred G. Hoerrner
Merck Chemical Manufacturing Division
Post Office Box 2000
Rahway, New Jersey 07065
IV"D"13 Richard K. Meyers
Texaco, Incorporated
Post Office Box 509
Beacon, New York 12308
IV-°-14 F. M. Parker
Chevron U.S.A., Incorporated
575 Market Street
San Francisco, California 94105
IV-D-15 R. J. Samelson
PPG Industries, Incorporated
One Gateway Center
Pittsburgh, Pennsylvania 15222
(continued)
2-3
-------
Table 2-1. Concluded
Docket entry number9 Commenter/affiliation
IV-D-16 Daniel B. Rathbun
American Petroleum Institute
2101 L Street, Northwest
Washington, D.C. 20037
IV-D-17 John J. Moon
Phillips Petroleum Company
Bartlesville, Oklahoma 74004
IV-D-18 Dennis L. Gehlhausen
Eli Lilly and Company
307 East McCarty Street
Indianapolis, Indiana 46285
IV-D-19 David D. Doniger
Natural Resources Defense Council,
Incorporated
1725 I Street, Northwest
Suite 600
Washington, D.C. 20006
IV-D-20 Wells Eddleman
General Energy Consulting
Route 1, Box 183
Durham, North Carolina 27705
IV-D-21 C. D. Mallach
Monsanto Company
800 North Lindbergh Boulevard
St. Louis, Missouri 63166
IV-H-1 T. L. Hurst
Kerr-McGee Corporation
Kerr-McGee Center
Oklahoma City, Oklahoma 73125
IV-F-1 National Air Pollution Control
Techniques Advisory Committee
Transcript of Meeting for National
Emission Standards for Hazardous
Air Pollutants from Benzene
Storage Vessels
U.S. Environmental Protection Agency
Office of Administration
Research Triangle Park, NC 27711
aThese designations represent docket entry numbers for Docket
No. A-80-14. These documents are available for public inspection at:
U.S. Environmental Protection Agency, Central Docket Section, West
Tower Lobby, Gallery 1, Waterside Mall, 401 M Street, Washington, D.C.
20460.
2-4
-------
term "safe" (regarding exposure levels), rather than meaning "absolutely
risk-free," means a level that protects against a "significant risk of
harm." The commenter noted that risk levels that the EPA has calculated
are not "significant" as that term has been used by the court.
Two commenters (IV-D-16, IV-D-21) felt that the risk from exposure
to benzene emissions is insignificant compared to other commonly accepted
societal risks. Two commenters (IV-D-13, IV-D-16) noted that the risk
from benzene storage emissions is insignificant in comparison to the
background leukemia incidence risk.
Two commenters (IV-D-4, IV-D-13) further compared the risk from
benzene storage emissions to other government determinations of risk
acceptability and noted that, under these determinations, the risk from
exposure to benzene storage emission sources would have been considered
not worthy of regulation.
Response: The commenters are judging the significance of benzene
storage vessels based on quantitative risk estimates. In general,
quantitative risk estimates at ambient concentrations involve an analysis
of the effects of a substance in high-dose epidemiological or animal
studies, and extrapolation of these high-dose results to relevant human
exposure routes at low doses. In the case of benzene, the effects
observed were the result of high-dose epidemiological studies. 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. Quantitative
risks to public health from emissions of an airborne carcinogen may be
estimated by combining the dose-response relationship obtained from this
carcinogenicity strength determination with an analysis of the extent of
population exposure to a substance through ambient air.
Most exposure analyses are based on air quality models, available
estimates of emissions from sources of a substance, and approximations
of population distributions near these sources. EPA considers this the
best practicable approach. Even though ambient monitoring data might be
used to estimate quantitative risks to public health, these data are
available only for a few locations near these sources. Thus, use of
ambient monitoring data is not practicable. However, EPA has data to
confirm that the public is exposed to benzene. For example, concentrations
2-5
-------
up to 51 micrograms per cubic meter (on a 24-hour average) were found
around a petrochemical plant in Philadelphia, Pennsylvania.
The air quality models used in exposure analyses generally estimate
exposures out to 20 kilometers from the source. During exposure analyses,
population and growth statistics are examined in conjunction with ambient
concentrations. Using these factors and existing carcinogenicity strength
determinations, estimates are then made of the degree of risk to
individuals and the range of increased cancer incidence expected from
ambient air exposures associated with a substance at various possible
emission levels.
The assumptions and procedures discussed above for extrapolation
and for exposure estimates for benzene emissions are subject to
considerable uncertainty. A small portion of that uncertainty has been
considered by calculating ranges at proposal. The ranges presented at
proposal represent uncertainty in estimates of benzene concentrations to
which workers were exposed in occupational studies of Infante, Aksoy,
and Ott that serves as the basis for developing the benzene unit risk
factor (Part I Docket Item II-A-31). The ranges presented represent
95 percent confidence limits on two sources of uncertainty in the benzene
risk estimates. 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 the source specific data are correct).
Other uncertainties associated with estimating health impacts were
not quantified at proposal. EPA has extrapolated 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.
2-6
-------
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 occupational
studies, often by several orders of magnitude. In relating the occupa-
tional experience to the general population, EPA has applied a linear,
non-threshold model that assumes that the leukemia response is 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 non-linear
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 non-linear alternatives.
This choice may result in an overestimate 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 parameters,
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 overestimates
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
2-7
-------
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.
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 human cancer
risk posed by benzene emitted from the source categories 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.
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 decision making. 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.
When the standard for benzene storage vessels was proposed on
'December 19, 1980, the Administrator made the judgment that "benzene
emissions from benzene storage vessels create a significant risk of
cancer and require the establishment of a national emissions standard
under Section 112" (45 FR 83954).
2-8
-------
The data base used to calculate emissions from storage vessels has
changed since the standard was proposed. This change is based on new
test data acquired since proposal. This data base and the reasons for
using it are described in Section 2.2.2.1. Based on these new data, the
emission estimates for fixed roof tanks (totally uncontrolled tanks)
remains unchanged. The emission estimates for internal floating roof
tanks and external floating roof tanks are lower than at proposal.
Since a substantial proportion of existing tanks have internal or external
floating roofs, this change resulted in a substantial reduction in the
estimate of nationwide emissions from these tanks. For this reason, the
Administrator reevaluated the need to establish Section 112 standards
for benzene storage vessels.
Using the new emission data and a new exposure modeling approach
adopted since proposal, the EPA estimated current leukemia cases and
maximum lifetime risks that occur due to exposure from storage vessels,
and the potential reductions that could be achieved to determine whether
this source category continues to pose significant risk and whether a
standard is warranted under Section 112.
Benzene storage vessels are currently estimated to emit about
620 Mg of benzene per year from about 126 storage facilities. This
amount is about 1 percent of total benzene emissions from stationary
sources. Estimated lifetime risk due to these emissions is about
3.6 x 10 for the most exposed individuals, and over the total exposed
population (within 20 km of each facility) about 0.043 cases per year
are estimated to occur.
For comparison, at proposal, the 126 facilities were estimated to
emit about 2,200 Mg benzene per year. These benzene emissions were
estimated to result in a range of 0.12 to 0.82 leukemia cases per year
and a range of maximum lifetime risk of about 1.5 x 10 to 1.0 x 10"^.
Thus, since proposal, estimated benzene emissions have been revised
downward by over 70 percent, estimated annual leukemia incidence by over
85 percent, and estimated maximum lifetime risk by over 90 percent.
Control measures that can be used to reduce benzene emissions
include the use of certain types of equipment (much of which is already .
in place on many tanks in the industry), such as internal floating
roofs, primary seals, and secondary seals, or enclosure of the storage
2-9
-------
tanks and routing emissions to a combustion device (discussed at proposal).
These control techniques could reduce nationwide emissions over baseline
by about 18 to 98 percent, depending on the technique applied.
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 categories
for which standards are being developed. Using the control techniques
mentioned above, leukemia incidence could be reduced to roughly 0.036 to
0.0009 cases per year (about 16 to 98 percent reduction), and maximum
lifetime risk to roughly 2.9 x 10 to 7 x 10"7, (about 20 to 98 percent
reduction). Although a large percentage reduction could be achieved in
the health risks by enclosing, routing, and combustion, the absolute
amount is small.
Because of the extent of control now exhibited by the industry, the
small amount of benzene emissions from these sources and the small
portion (about 1 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 available
control techniques, the Administrator has concluded that benzene emissions
from benzene storage vessels do not warrant Federal regulatory action
under Section 112.
One commenter (IV-F-1) stated that the "risk levels that EPA has
calculated are not 'significant1 as that term has been used by the
Court." EPA assumes that the commenter refers to the court interpretation
in Industrial Union Department, AFL-CIO v. American Petroleum Institute,
65 L. Ed. 2d 1010, 100 S. Ct. 2844 (1980). This interpretation of the
significance of risk was made in the context of The Occupational Safety
and Health Act of 1970, not the Clean Air Act. It is not necessarily
appropriate to transfer interpretations from one to the other. In any
case, the Court in fact never indicated what actually constitutes a
"significant" risk except to give obvious examples of what constitutes
plainly acceptable and plainly unacceptable risks. The Court stated:
"If, for example, the odds are one in a billion that a person will die
from cancer by taking a drink of chlorinated water, the risk clearly
2-10
-------
could not be considered significant. On the other hand, if the odds are
one in a thousand that regular inhalation of gasoline vapors that are
two percent benzene will be fatal, a reasonable person might well consider
the risk significant and take appropriate steps to decrease or eliminate
it" (48 LW 5034). The Court then stated that it was the duty of the
OSHA Administrator to determine, using rational judgment, the relative
significance of the risks associated with exposure to a particular
carcinogen.
2.1.2 Storage Vessels Attached to Moving Vehicles
Comment: One commenter (IV-D-7) stated that definition of storage
vessels should exclude storage vessels attached to mobile vehicles, such
as tankers, barges and tank trucks.
Response: The control technologies that would be necessary to
control benzene emissions from storage vessels attached to mobile vehicles,
such as tankers, barges, or tank trucks, are completely different from
those that are appropriate for other storage vessels. Additionally,
data collection on tankers, barges, and tank trucks was not part of the
survey performed by the EPA to develop a data base to support the Benzene
Storage Vessel NESHAP. For these reasons, it was never the intent of
the EPA to consider these types of benzene storage vessels as designated
sources under this NESHAP.
2.1.3 Coke Oven Byproduct Vessels
Comment: One commenter (IV-D-15) stated that the regulations
should not exempt tanks at coke oven by-product facilities. He felt
that if benzene is hazardous, all facilities should be required to meet
uniform control requirements.
Response: A separate NESHAP is currently being developed for coke
oven by-product facilities. Vessels at coke oven by-product facilities
were not incorporated into the benzene storage vessels source category
because the applicable control techniques are different than the ones
considered for this source category. This is a function of the nature
of the coking and byproduct processes. For this reason, the Agency has
decided that a separate standard for vessels at coke oven by-product
facilities is appropriate and that such vessels should not be incorporated
into these standards.
2-11
-------
2.1.4 Benzene Mixtures
Comment: One commenter (IV-D-19) pointed out that the proposed
standards would apply only to vessels that store pure benzene. He asked
if vessels that store mixtures of benzene and other substances existed;
and if so, why such vessels were not affected by these standards.
Response: It is true that vessels storing mixtures of benzene and
other chemicals exist, but such vessels were never intended to be part
of this source category. In part, this is because many vessels storing
mixtures, such as those associated with coke oven byproduct processes,
have different control options than those identified for this source
category. The controls and impacts of control strategies for vessels
storing mixtures would have to be examined as part of a separate source
category. For this reason, the Agency decided not to extend the
applicability of this source category to vessels storing benzene mixtures.
However, some information is currently available on vessels storing
benzene mixtures, which for completeness will be presented here. There
are three general classes of stored liquids that are composed of benzene
that would not have been affected by the proposed standards. These are:
1. Liquids such as gasoline, which are stored in large quantities,
but do not, on a fractional basis, contain more than 10 percent
benzene;
2. Mixtures in which benzene may be more than 10 percent; and
3. Benzene that does not meet the specific gravity specification
for industrial grade benzene (crude benzene).
New vessels storing gasoline (^2 percent benzene) are affected
facilities under Standards of Performance for Petroleum Liquid Storage
Vessels (40CFR60: Subpart K(a)). These standards discussed above
require controls that are almost identical in effectiveness to those
that were selected as BAT for new benzene storage vessels in the proposed
NESHAP rule requirements. Many state implementation plans (SIPs) require
that existing gasoline storage vessels be controlled to almost the same
extent as the proposed BAT for existing benzene storage vessels.
Data were gathered on vessels storing liquids of the second class
(Table 2-2). This data was obtained from a data base of 4,054 vessels
associated with the Synthetic Organic Chemical Manufacturing
Industry (SOCMI). Fifteen (15) were thought to possibly contain more
2-12
-------
Table 2-2. VESSELS CONTAINING MIXTURES THAT MAY BE
MORE THAN 10 PERCENT BENZENE3
Vessel contents
Volume (1000's of gallons)
Benzene Caprolactum
Benzene Lactum
Benzene/Toluene
EA, Benzene, Water
EA, Benzene
Light Aromatic Distillate
4.4
7
37.8, 237, 42, 8.8, 17.0, 1272.7l
2 2 2b
f. , f. , £.
4.75
515, 515, 63.5b
Including crude benzene.
""Multiple vessels with same contents.
2-13
-------
than 10 percent of benzene. The total volume of these vessels is about
2.7 million gallons. This can be compared to the estimated 500 vessels
with a total volume of about 308 million gallons that stored industrial
grade benzene in 1979. The total tank volume (tankage) devoted to the
storage of this type of benzene mixtures is less than one percent of the
tankage devoted to benzene. Because vessels storing mixtures will have
reduced amounts of benzene in them, the true amount of benzene stored
may be significantly reduced from the above two million gallons.
The last class of liquids consists of unfinished (crude) benzene or
off specification benzene. Most such liquids are petroleum liquids and
many are affected facilities under Subpart K(a) or the SIPs and as such,
would be controlled to some extent. There was only one such tank in the
data base.
2.2 HEALTH AND ENVIRONMENTAL IMPACTS
2.2.1 Background
The proposed standards, which were based on Best Available
Technology (BAT), would have required the use of a fixed roof in
combination with an internal floating roof. The proposed standards also
would have required that the internal floating roof be in contact with
the liquid surface and be equipped with a liquid-mounted primary seal
and a continuous secondary seal.
Many commenters suggested that the EPA delay the development of the
final standards until the effectiveness of BAT equipment relative to
other equipment types could be reevaluated using data from the American
Petroleum Institute (API) 2519 Task Group testing program. The results
of this testing program have been received and evaluated by the EPA.
Comments were also received on other aspects of BAT, such as control
equipment costs.
2.2.2 Selection of the Level of the Standard
2.2.2.1 Emission Data Base. Seven commenters suggested that the
emissions data base used in selection of the BAT at proposal was erroneous
and that the Agency should await the completion of a new API testing
program before selecting BAT prior to promulgation (IV-D-1, IV-D-2,
IV-D-3, IV-D-8, IV-D-10, IV-D-10a, IV-D-14).
Response: There are four potential emission data bases from which
emission calculations could be developed. These are:
2-14
-------
1. A test series done by Chicago Bridge and Iron (CBI) for an
internal floating roof vendor. This series measured emissions
from a bolted, noncontact internal floating roof equipped with
wiper-type, vapor-mounted primary seals; and a welded contact
internal floating roof. The welded roof was equipped with a
liquid-mounted primary seal and in some instances a secondary
seal. All the tests were performed in a propane/octane binary
mixture. This data base is hereafter referred to as the
Vendor report or series.
2. A large number of tests done on various external floating
roofs with propane/octane as the stored liquid. These tests
were also performed by CBI. The primary emphasis of this work
was to categorize emissions from various types of primary and
secondary seals and was used to update API bulletin 2517,
which is used in estimating emissions from external floating
roof tanks. It was also used in the 1981 revision of EPA
publication AP-42. This work is referred to as the 2517 series
or report.
3. A test series done by CBI for the EPA using benzene as the
test liquid. This program tested a bolted noncontact internal
floating roof with vapor-mounted, shingled, primary and secondary
seals; a welded contact type internal floating roof equipped
with a liquid-mounted primary seal and in some instances, a
secondary seal; and an external floating roof equipped with a
mechanical shoe primary seal and in some instances a secondary
seal. This data base will be referred to as the EPA report or
series.
4. A test program done by CBI for API on emissions from internal
floating roofs. This program tested three roof types (non-
contact, bolted contact, welded contact), three primary seal
types (vapor-mounted wiper; vapor-mounted, foam-filled resilient
seal; liquid-mounted seal) with and without secondary seals,
in three different liquids (propane/octane, hexane, and octane).
Additional work was done on emissions from the components of
2-15
-------
an internal floating roof. This consisted of deck fitting
emission tests, laboratory evaporation tests, laboratory
permeability tests, and bench permeability tests. This data
base will be referred to as the 2519 report or series.
Each of the above test series was performed in the CBI 20 foot
diameter pilot test tank. The first three were completed prior to the
development of the proposed standards, but the 2519 series was completed
after the date of proposal.
In evaluating the emissions data for internal and external floating
roofs prior to proposal it was noted that emissions from the EPA series
were significantly higher than those measured in either the Vendor or
the 2517 series when tests on similar equipment were normalized to the
same vapor pressure and molecular-weight. The Agency believed that the
difference in emissions resulted from a difference in liquids, namely
multicomponent liquids such as propane/octane and single component
liquids such as benzene. The reason for this difference was believed to
be due to that fact that in a mixed product (e.g. the propane/octane
mixture) the emission rate depends upon the ability of the component
with the highest partial pressure (e.g. propane) to migrate through the
liquid to the liquid surface and replenish the component that is lost
through evaporation at the liquid surface. In a single component product
(e.g. benzene), however, the liquid surface does not tend to become
depleted of light ends at the liquid surface during the evaporation
process. Thus, a mixed product of the same vapor pressure as a single
component product was expected to have a lower evaporation rate due to
this phenomenon. Therefore, in selecting BAT at proposal only the EPA
test series was used because it was believed that the previous
propane/octane test work was not representative of single component
emissions such as a vessel storing benzene.
Industry representatives commented that the higher emissions were a
result of the test procedures and did not necessarily result from a
difference in evaporative properties. The 2519 test series shows that
when normalized to a common vapor pressure and molecular weight, there
is no significant emission difference between hexane, octane, and the
propane/octane binary mixture. Based on these results the Agency now
agrees that there is no evidence of evaporative difference between
2-16
-------
single and multicomponent liquids stored in floating roof tanks, and
this is not a reason for the higher emissions measured in the EPA series.
One cause of at least a portion of the higher emissions from the
EPA series is that during certain internal roof tests done for the EPA
series, the roof fittings had openings that would not normally exist in
the field and were sealed with polyurethane film, which, as previously
discussed, is permeable to benzene. This would lead to artificially
higher emissions being measured during the EPA series than would normally
be expected from a typical field tank. During tests done on the same
roof for the 2519 series the roof fitting openings that would not normally
exist in the field were sealed by welded metal seals not permeable to
benzene. This procedure would yield measured emissions more representative
of emissions from a typical field-located tank.
Also during the EPA series, the bolted noncontact internal floating
roof was tested with shingled (i.e., noncontiguous) primary and secondary
seals, which are not as effective in reducing emissions as the more
typical continuous wiper or foam-filled resilient seals. This again
would lead to higher emissions being measured during the EPA series.
Either wiper or foam-filled resilient seals were tested during the other
test series.
Other physical mechanisms that could explain the higher emissions
in the EPA series were sought. The permeability results in the 2519 series
were examined to ascertain if permeation of the seal system could be
responsible for the higher benzene emission. As detailed in Appendix A,
the permeation rate of benzene through a typical seal fabric (polyurethane)
was significantly higher than the rates at which hexane or propane/octane
permeate. Because there are no direct measurements of benzene permeation
rates through an entire seal system, theoretical models were developed.
The most reasonable model of permeation through a liquid-mounted seal
predicts emissions of 0.0102 pound moles per day in the test tank (see
Appendix A). While permeation and equipment differences may explain
•some of the emission differences between the benzene test work and the
other test work, it is not sufficient to account for the total difference.
Another explanation of the higher emissions from floating roof
tanks shown by the EPA series, has to do with the test procedures used.
The vendor series and the 2517 series used the same test procedure as
2-17
-------
the EPA series, that is, a floating roof and seal system is installed in
the pilot tank, and air is blown over the floating roof. The air is
collected and analyzed for hydrocarbon content. In the vendor, 2517,
and EPA series test work, the temperature of the air being blown across
the roof was uncontrolled. During periods when the air is cold (such as
during the winter), the benzene vapor being emitted will condense during
periods when actual tests are not being run. When a test is then begun,
the benzene vapor that condensed will be measured during the test when
it was actually emitted before the test run began. In the case of
benzene (EPA series) this could lead to artificially higher results. In
the case of the propane/octane mixture, the uncontrolled air temperature
is not as important to the results since this mixture is less likely to
condense in the cold air. In the 2519 series, the air temperature was
controlled, and no emissions differences were observed between the three
tested liquids.
As just explained, because the 2519 series test conditions were
more controlled than during the EPA series and because of the equipment
tested (continuous versus shingled seals), this test series resulted in
more representative emission measurements. The 2519 series was also
structured to make it possible to ascertain more accurately the relative
contributions to emissions of the various emission points (e.g., seals,
roof seams and roof fittings). Also, the data obtained from the 2519
series are similar to the vendor series that tested similar roofs and
seals and used a propane/octane binary mixture.
The higher permeability of benzene, the difference in equipment
tested and the differences in test procedures explain most of, but not
all, the higher emissions from floating roof tanks measured during the
EPA series. Currently, however, there is no explanation beyond what has
already been discussed as to why benzene emissions would be any higher
than the hexane and octane emissions measured during the tests done in
the 2519 series.
Since there is no reason (other than possibly permeability, which
is addressed later) for benzene emissions (normalized for vapor pressure
and molecular weight) to be higher than hexane and octane emissions
during the 2519 tests, and since the 2519 series was conducted with more
refined procedures and more thoroughly evaluated the emission sources
2-18
-------
and control techniques for each source, the Agency has decided to use
the data from this series to evaluate the emission reduction potential
for various control technologies applied to fixed roof and internal
floating roof tanks. For similar reasons, the Agency has selected the
2517 series as the data base for evaluating controls for external floating
roof vessels. The 2517 tests are more extensive in terms of equipment
tested and, for the same reasons as the 2519 series, have measured
emissions more representative of emissions from a typical external
floating roof.
Table 2-3 compares emissions from selected floating roof tank types
as calculated using data from the EPA series and as calculated using
data from the 2519 and 2517 series. It should be noted that because of
differences in tested equipment and test procedures, the emissions are
not strictly comparable. However, it can be seen that the sharp difference
in emissions (particularly in terms of mass rather than percentage)
between the equipment configurations vanished in the 2519 and 2517 test
series. Making the decision that the 2517 and 2519 test series are
superior to the EPA test series meant that it was then necessary to
reexamine baseline impacts and effectiveness of control techniques for
benzene storage vessels.
The Agency examined the emission points from possible baseline tank
types and possible control technologies. As explained in the Volume I
BID there are four types of tanks that could be used to store benzene.
These are:
1. Fixed roof tanks;
2. Noncontact internal floating roof tanks;
3. Contact internal floating roof tanks; and
4. External floating roof tanks.
Based on the 2519 test series, there is no inherent difference between
contact and noncontact deck types. Analysis of the data concluded that
deck seams emit at the same rate if they are in contact with the liquid
or saturated vapor. Contact decks may be welded (i.e., no deck seams)
or bolted (e.g., mechanically connected panels or sections that have
seams). A bolted contact deck would have deck seam emissions at the
same rate per foot of deck seam as a noncontact deck. Because of this,
for the purpose of evaluating control efficiencies the two types of
2-19
-------
Table 2-3. COMPARISON OF EMISSIONS AS CALCULATED FROM THE EPA
SERIES AND THE 2519/2517 SERIES
Tank type/equipment
Test series
EPA 2517/2519
emissions emissions
(Mg/yr) (Mg/yr)
I. Internal Floating Roof
A. Bolted deck with vapor-mounted
primary and secondary seals
B. Welded deck with liquid-mounted
primary seal
C. Welded deck with liquid-mounted
primary and secondary seals
II. External Floating Roof with
Mechanical Shoe Primary Seal
3.56J
1.15
0.67
Both primary and secondary seals were shingle design.
"All seals were continuous.
0.42'
0.38
0.34
A. Primary seal only
B. With rim-mounted secondary
6.99
2.63
1.11
0.087
2-20
-------
internal floating roofs were merged into the general classification of
internal floating roof. This procedure reduced the basic starting cases
to three tank types: fixed roof, internal floating roof (bolted deck
assumed), and external floating roof.
The mechanisms of fixed roof tank and external floating roof tank
emissions have been fully discussed in the Volume I BID. Although the
external floating roof tank emission factors have changed based on the
2517 series, the emission mechanisms are still the same. Fixed roof
tank emissions have not changed since proposal.
The 2519 series allows for a more detailed breakdown of internal
floating roof tank emissions into:
1. Standing storage losses, consisting of:
a. Rim seal emissions;
b. Fitting losses; and
c. Deck seam emissions
2. Working losses.
Table 2-4 presents losses from a model benzene storage vessel by point
of loss, and Table 2-5 compares emissions from various selected tank
configurations. The model tank, used in these calculations and all
subsequent calculations in this section, has a volume of 606 m3
(160,000 gallons), a diameter of 9.1 m (30 feet), and undergoes
50 turnovers per year.
Internal floating roofs are typically bolted decks equipped with
vapor-mounted seals and Case A fittings (defined below). In the model
tank, emissions from the vapor-mounted seal are about 35 percent of
total emissions. Emissions from the vapor-mounted seal could be reduced
through the use of a liquid-mounted primary seal, a secondary seal, or
both. A liquid-mounted seal reduces emissions from the vapor-mounted
primary seal by about 55 percent. The addition of a secondary seal to
the vapor-mounted primary seal would reduce emissions by about 63 percent.
The addition of a secondary seal to a liquid-mounted primary seal reduces
emissions by about 46 percent over the liquid-mounted primary seal
alone. Converting a vapor-mounted primary seal system to a liquid-mounted
primary seal with a secondary seal reduces emissions from the seal area
by about 76 percent over the vapor-mounted primary seal alone.
2-21
-------
Table 2-4. INTERNAL FLOATING ROOF TANK EMISSIONS BY SOURCE"
ro
ro
ro
Seal
Type
Vapor-mounted
Liquid-mounted
Vapor-mounted
with secondary
Liquid-mounted
with secondary
losses
Emission
(Mg/yr)
0.19
0.085
0.071
0.046
Fitting losses
Emission
Case (Mg/yr)
A2 0.26
B3 0.16
C4 0.19
Deck losses
Emission
Roof type (Mg/yr)
Bolted 0.06
Welded 0.0
Working losses
Emission
(Mg/yr)
0.03
Tank Parameters:
Volume = 160,000 gallons
Diameter = 30 feet
Turnovers = 50 turnovers per year
"Case A assumes: (1) access hatch, with ungasketed, unbolted cover; (2) automatic gauge float well,
with ungasketed, unbolted cover; (3) built-up column wells, with ungasketed sliding cover; (4) ladder
well, with ungasketed sliding cover; (5) adjustable roof legs; (6) sample well with slit fabric (10%
open area); (7) 1-inch diameter stub drains; and (8) vacuum breaker with, gasketed weighted mechanical
actuation.
Case B assumes: (1) access hatch, with gasketed, bolted cover; (2) automatic gauge float well, with
gasketed, bolted cover; (3) pipe column with flexible fabric sleeve seal; (4) ladder well, with
gasketed sliding cover; (5) adjustable roof legs; (6) sample well with slit fabric (10% open area);
(7) 1-inch diameter stub drains; and (8) vacuum breaker, with gasketed weighted mechanical actuation.
Case C is identical to Case B except that built-up columns with gasketed sliding covers are assumed
instead of pipe columns.
-------
Table 2-5. EMISSIONS FROM A TYPICAL BENZENE STORAGE VESSEL
Tank type/equipment Emissions (Mg/yr)
I. Fixed Roof 9.2
II. Internal Floating Roof
A. Bolted deck, vapor-mounted 0.54
seal, Case A fittings
B. Bolted deck, liquid-mounted 0.44
Case A fittings
C. Bolted deck, liquid-mounted 0.34
seal, Case B fittings
D. Bolted deck, liquid-mounted 0.30
primary with secondary, Case B
fittings
E. Welded deck, liquid-mounted 0.24
primary with secondary, Case B
fittings
III. External Floating Roof
A. Mechanical Shoe Primary Seal 1.11
1. Primary seal only
2. With rim-mounted secondary 0.087
B. Vapor-mounted Primary Seal 6.9
1. Primary seal only
2. With rim-mounted secondary 2.31
C. Liquid-mounted Primary Seal 0.36
1. Primary seal only
2. With rim-mounted secondary 0.080
2-23
-------
The next major source of internal floating roof tank emissions are
losses from fittings. Fittings in general are ancillary equipment such
as hatches or column wells that penetrate the deck. Such penetrations
will emit benzene. Typical fittings are: (1) access hatch, with
ungasketed, unbolted cover; (2) automatic gauge float well, with
ungasketed, unbolted cover; (3) built-up column wells, with ungasketed
sliding cover; (4) ladder well, with ungasketed sliding cover;
(5) adjustable roof legs; (6) sample well with slit fabric (10% open
area); (7) 1-inch diameter stub drains; and (8) vacuum breaker with,
gasketed weighted mechanical actuation. This equipment is referred to
as "Case A". In the model tank, emissions from Case A fittings account
for about 48 percent of total emissions. Emissions from Case A type
fittings could be reduced through the use of gaskets, bolting covers,
and constructing pipe columns with flexible fabric sleeve seals on the
column well in place of built-up columns equipped with ungasketed sliding
covers in the column wells. This configuration of fittings is referred
to as Case B and is the level of control that could be obtained in new
benzene storage vessels equipped with internal floating roofs.
Specifically, "Case B" is defined as: (1) access hatch, with gasketed,
bolted cover; (2) automatic gauge float well, with gasketed, bolted
cover; (3) pipe column wells with flexible fabric sleeve seal; (4) ladder
well, with gasketed sliding cover; (5) adjustable roof legs; (6) sample
well with slit fabric (10% open area); (7) 1-inch diameter stub drains;
and (8) vacuum breaker, with gasketed weighted mechanical actuation.
Case B fittings would reduce emissions from the typical fittings Case A
by about 38 percent.
Existing internal floating roof benzene tanks typically use built-up
columns to support the fixed roof. Such vessels could not be equipped
with pipe columns without replacing the columns. In most instances,
this would be equivalent to requiring the construction of a new tank.
Therefore, an intermediate control strategy was sought. Emissions from
built-up column wells could be controlled by gasketing the sliding
cover. This strategy is referred to as "Case C" and represents the
level of fitting control available in existing internal floating roof
tanks.
2-24
-------
Specifically, Case C is identical to Case B except that built-up
columns with gasketed sliding covers are assumed instead of pipe columns.
Case C fittings would provide about a 27 percent emission reduction over
Case A fittings. Because most existing fixed roof benzene tanks are
equipped with built-up columns, Case C represents the level of control
of fitting emissions generally available for existing tanks.
The next source of internal floating roof tank emissions are deck
seams. Decks that are constructed of sections bolted together have
emissions along the seam. As discussed previously, seams emit at the
same rate if they are in contact with the liquid surface (contact deck)
or contain a saturated vapor on one side (noncontact deck). Because of
this fact, the distinction between contact and noncontact decks has been
dropped, and these decks are now referred to as "bolted" for emission
purposes. Emissions from the deck seams in the model tank are about
11 percent of total emissions.
Deck seam emissions could be controlled by installing decks that
have no seams. Such decks are generally made out of steel sections
welded together. These decks are generally in contact with the liquid
surface, and are referred to as "welded" for emission purposes.
The last emission type in an internal floating roof tank is the
working loss. These losses are fully discussed in the Volume I BID and
account for about 6 percent of typical losses. No controls for working
losses are available.
As Table 2-5 shows, fixed roof tank emissions could be reduced by
about 94 percent by the installation of internal floating roofs. Emissions
could be further reduced through the use of additional controls on
seals, fittings, and deck seams.
Emissions from external floating roofs could be reduced by the
addition of a secondary seal over the primary seal. In the case of the
mechanical shoe primary seal, this would reduce emissions by about
92 percent. Emissions from vapor-mounted primary seals could be reduced
"by replacing these seals with mechanical shoe seals or liquid-mounted
primary seals alone or further reduced with secondary seals.
In examining the effectiveness of the control techniques based on
the 2519 and 2517 test series, it was noted that the emission reductions
for these techniques based on the 2519 and 2517 test series are quite
2-25
-------
different than those calculated at proposal. For example, the 2519 series
showed that the control effectiveness of an internal floating roof (of
any type) in a fixed roof tank is much more effective in reducing emissions
than was believed at proposal, based on the EPA test series. On the
other hand, the 2519 test series showed that a secondary seal in an
internal floating roof tank is much less effective in reducing emissions
than believed at proposal based on the EPA test series. This is to be
expected because the internal floating roof is more effective than
believed previously and as a result, there are less residual emissions
to be controlled by the secondary seal. The 2519 test series showed
that contact and noncontact roofs are equally effective in reducing
emissions. The 2519 test series also showed that liquid-mounted seals
are more effective in reducing emissions than vapor-mounted seals. This
type of seal can be used with both contact and noncontact roofs and was
considered as a control technique. Furthermore, the 2519 series showed
that control of roof fittings, column wells, and roof deck seams does
reduce emissions. Using the 2517 and 2519 data in combination shows
that external floating roofs can, when used with effective seals, reduce
emissions as effectively as internal floating roofs.
As briefly mentioned previously, there is an additional source of
emissions that has not been fully considered up to this point. This is
the permeability of seal systems and gaskets to benzene. The 2519
series and the open literature point to the fact that aromatics such as
benzene have higher permeability rates through polymers than some other
types of compounds. Because no direct measurements of seal permeability
are available, the Agency examined this emission source by developing
theoretical models.
These models represented:
1. A foam-filled liquid-mounted seal;
2. A wiper type, vapor-mounted primary seal; and
3. Each of the above with a wiper type secondary seal.
Each seal consists of two parts:
1. Two layers (top and bottom) of seal fabric; and
2. Open cell foam situated between the fabric layers.
In selecting the fabric layers for modeling, it was discovered that
there was little data on what fabrics are actually in use, and little
2-26
-------
data on measured fabric permeability rates. Because the Agency had
permeability measurements on 0.037 inch thick polyurethane-coated nylon
fabric from the 2519 tests, and because this material is currently in
use in field tanks, the Agency decided to use this material as the
fabric in the models.
In modeling the open cell foam it was assumed that the foam presented
no permeability barrier. Transport between the fabric layers was assumed
to be diffusion (it was assumed that the foam did not allow convective
transport). These models done on a 20 foot diameter tank are contained
in Docket Item IV-A-1. For the purpose of comparability to the model
tank (30 foot diameter) emissions the results have been extrapolated to
the model tank.
Table 2-6 compares the convective losses presented in Table 2-4
with the calculated permeability losses. It is seen that permeation may
account for more than 50 percent of seal losses if:
1. The permeation rates are correct; and
2. The models realistically represent actual systems.
Table 2-7 examines how consideration of permeability affects the
overall effectiveness of controls compared to a fixed roof tank. The
reduction in overall effectiveness when permeability is considered is
less than 3 percent.
However, the Agency examined how permeability emissions may be
controlled. These emissions could be controlled by a seal permeability
specification. Such a specification would limit permeability emissions
to a specified limit per unit area of seal. However, the variation in
measured values in open literature indicate that such measurements would
be difficult to make reliably. Seal materials must withstand abrasion
and flexing as the floating roof moves. At this point in time, the
Agency is aware of no materials or laminar composites that would have
both the necessary characteristics of material strength and permeation
rates lower than the modeled fabric. Such a specification could be made
with additional research on materials.
Table 2-8 shows revised baseline emissions based on the revised
emission equations for each of the four model plants developed during
proposal. Table 2-9 shows revised baseline nationwide emission estimates
based on the revised emission equations.
2-27
-------
Table 2-6. COMPARISON OF CONVECTIVE AND PERMEABILITY LOSSES
FROM INTERNAL FLOATING ROOF SEAL SYSTEMS IN THE MODEL TANK
Seal type
Emissions (Mg/yr)
Convective
Modeled
permeation
Possible
total
losses
Vapor-mounted
Liquid-mounted
Vapor-mounted
with secondary
Liquid-mounted
with secondary
0.19
0.085
0.071
0.046
0.21
0.20
0.11
0.10
0.40
0.285
0.181
0.146
2-28
-------
Table 2-7. MODEL TANK EMISSIONS (Mg/yr) FROM A FIXED ROOF TANK
AND A TYPICAL INTERNAL FLOATING ROOF TANK
Tank type
Emission
Percent control
Fixed roof
Internal floating roof with
9.2
0.54
94.1
bolted deck, Case A fittings,
vapor-mounted primary seal
only, no permeability
Internal floating roof with
bolted deck, Case A fittings,
vapor-mounted primary seal,
permeability
0.75
91.8
2-29
-------
Table 2-8. EMISSIONS FROM NEW AND EXISTING
MODEL PLANTS
Tank dimensions
(meters x meters)
Large benzene producer
12 x 9
18 x 12
8x5
9x9
13 x 13
24 x 9
27 x 15
Total
Small benzene producer
3 x 11
13 x 13
8 x 11
32 x 7
Total
Benzene consumer
12 x 11
18 x 15
Total
Bulk storage terminal
12 x 11
18 x 15
Total
Emissions
Existing
0.72
2.19
0.48
0.59
0.68
1.36
1.82
7.84
1.27
0.68
0.50
2.17
4.61
0.64
0.97
1.61
0.64
0.97
1.61
(Mg/y)
New
0.72
0.13
0.48
0.59
0.68
1.36
1.82
5.78
1.27
0.68
0.50
2.17
4.61
0.64
0.97
1.61
0.64
0.97
1.61
aDiameter x height.
2-30
-------
Table 2-9. NATIONWIDE EMISSIONS FROM NEW AND
EXISTING BENZENE STORAGE TANKS
Emissions (Mg/y)
Model plant Existing New3
Large benzene producer 269 55
Small benzene producer 192 53
Benzene consumer 152 42
Bulk storage terminal 8 2
Total 621 152
aFifth-year (1988).
2-31
-------
2.2.3 Exposure Assessment
A number of commenters (IV-D-1, IV-D-3, IV-D-4, IV-D-10, IV-D-lOa,
IV-D-13, IV-D-21, IV-F-1) stated that the model plant methodology used
by the EPA overestimates risk from benzene exposure. The commenters
suggested that a more realistic and accurate risk estimate would be
obtained using actual plant emission data, actual population data, and
available plant-specific emission data.
One commenter (IV-D-lOa) maintained that the Agency's benzene
emissions exposure analysis relied upon incomplete and inaccurate
meteorologic data. Rather than use site-specific climatological data as
required by the Agency guidelines, the commenter remarked that the
analysis relies entirely on conditions at the Gulf Coast to apply to
storage vessels throughout the nation.
According to the commenter, the EPA concedes that this assumption
causes an overstatement of estimated exposure, noting its data were
"representative of poor dispersion conditions in the area in order to
develop a potential worst-case situation". He concluded that since
climatological data for approximately 300 sites throughout the U.S. are
available in the Agency archives, the EPA's total reliance on Gulf Coast
meteorology was not justified. The commenter also stated that the EPA
arbitrarily oriented the benzene storage vessels of a hypothetical
facility in order to maximize the ambient concentrations at the plant
boundary. According to the commenter, this was done despite the fact
that the actual storage vessels are not usually arranged in a straight
line configuration.
The commenter further asserted that the EPA failed to validate the
results of its air quality modeling as a check on its accuracy, as
required by Agency guidelines. According to the commenter, in this
exposure analysis, the EPA repeatedly has relied upon unsupported
assumptions about emissions, meteorology, population distribution, and
other factors, even though accurate data were readily available. The
'approach taken results in an unacceptably high degrees of uncertainty in
the Agency's exposure estimates; in some instances, the exposure estimate
may be off by a factor of 100 or more.
Response: The commenter is correct in noting that the benzene
storage risk assessment did not make use of plant-specific data relating
2-32
-------
to emissions, meteorology, or plant configurations. However, as explained
below, the plant-specific approach probably would not improve the precision
or accuracy of the results enough to justify the level of effort to use
more specific data. EPA has concluded that a plant-specific approach
would be too costly and not necessary for benzene storage emission
sources. In response to this comment, the EPA has revised its original
risk assessment for benzene storage vessels: the unit risk factor has
been recalculated; new emission estimates have been developed based on
the new API data; the meteorology of the area where each plant is located
has been used; and an improved population model (Human Exposure Model)
has been used.
The EPA considered the option of using plant-specific data for all
parameters in order to run an exposure model for each plant. The EPA
compared the uncertainty that would result using the plant-specific data
approach with the uncertainty that would result using the model plant
and extrapolation approach. The EPA also compared the level of effort
that would be required to complete the two options.
The plant-specific approach probably would not improve the precision
or accuracy of the results enough to justify the level of effort required
to gather the input data. A plant-specific approach would entail using
"Section 114" letters to gather plant information on emissions, meteorology,
and plant configuration from about 130 plants. This would require
substantial effort from plant owners as well as from the EPA. The
dispersion and exposure models would then have to be run about 130 times,
at least once for each plant. The resultant increase in precision and
accuracy would probably be small compared to the uncertainty still
remaining that is inherent in the dispersion and exposure models and in
the input data used. Both the Industrial Source Complex Long Term
computer model (ISCLT) and the Human Exposure Model, even with perfect
input data, are subject to substantial uncertainty. (The ISCLT model,
even with state-of-the-art input data, is estimated to have a 95 percent
'confidence interval of plus or minus a factor of two.) The plant-specific
input data would also exhibit wide variability and thus introduce
uncertainty in the results of the study.
The Agency has not exaggerated the precision of the results of the
model plant extrapolation method, nor has the EPA attempted to refine
2-33
-------
the results of the model plant extrapolation method any more than is
warranted by the quality of the data and the modeling technique.
Uncertainties are clearly delineated. The results are presented in
highly aggregate, nonspecific terms, in a fashion that exhibits much
less uncertainty than if the EPA tried to obtain more detailed, refined
results from the extrapolation. Using the model plant extrapolation
method, inaccurate deviations in the results for specific plants tend to
average out when the total national incidence is computed. Attempting
to validate the results of the air quality modeling would require an
extremely detailed, burdensome, and costly plant-specific approach.
Because a plant-specific approach would be very costly and would not
substantially improve upon the precision and accuracy already achieved
by the model plant extrapolation approach, the EPA has elected to use
the model plant extrapolation approach.
Comment: One commenter (IV-D-lOa) added that deficiencies exist in
the population concentration estimates contained in the exposure analysis.
According to the commenter, the EPA assumed that population is distributed
uniformly in all directions at each site, which introduces an uncertainty
factor of 10 to 100 into the overall exposure estimate.
Response: EPA's revised risk estimate (see Appendix B) was based
upon a more sophisticated population exposure model, which utilized a
population data base characterized as having a high level of resolution.
The Human Exposure Model (HEM) was used to estimate the population that
resides in the vicinity of each receptor coordinate surrounding each
plant. The HEM does not assume population is distributed evenly around
each plant. The population "at risk" to benzene exposure was considered
to be persons residing within 20 km of the plants. The population
around each plant was determined by specifying the geographical coordinates
of that plant.
A slightly modified version of the "Master Enumeration District
List—Extended (MED-X)" data base, a Census Bureau 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/BF) 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
2-34
-------
has been used to produce a randomly accessible computer file of only the
data necessary for the exposure estimation. A separate file of county-
level growth factors, based on the 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 plant's geographical coordinates and the concentration patterns
computed by the model plant extrapolation method were used as input to
the HEM. For each receptor coordinate, the concentration of benzene 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 concentrations 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 exposures are calculated differently for
the ED/BG1s 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 the 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
2-35
-------
that is geometrically interpolated radially and 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.
In summary, two approaches were 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 linear interpolation.
Comment: A commenter (IV-D-lOa) stated the analysis failed to
account for population activity patterns and population mobility, thereby
overestimating exposure levels for persons residing in the affected area
surrounding these plants. He further criticized the EPA's estimate of
"maximum individual lifetime risk" by noting that the Agency has no
evidence that any individual ever lives an entire lifetime 0.1 kilometers
from the plant at a point of maximum benzene concentration.
Response: The maximum individual lifetime risk, as the commenter
understood, is the risk associated with exposure to the maximum
concentration. Maximum concentrations are only modeled estimates and
may overestimate or underestimate the actual concentrations. As discussed
in Docket Item IV-B-4, the maximum concentrations and, consequently, the
maximum individual lifetime risks (which were estimated and used to
make, to the limited extent they were used, decisions) appear to be
underestimates. Provided the air at 0.1 kilometer from plant is located
in a neighborhood, the opportunity for exposure exists. Using the HEM,
exposures to maximum concentrations are generally limited to distances
greater than 0.2 kilometer and to locations where people reside. In the
absence of perfect information regarding the magnitude and duration of
exposure, it is prudent to assume that, as a "maximum", an individual
could face continuous exposure to a maximum concentration.
Comment: One commenter (IV-D-19) felt that the EPA had understated
the risk of exposure to benzene storage emissions. According to one
commenter, the scientific knowledge necessary for reasonably reliable
and precise estimates of human cancer risks simply is not available.
2-36
-------
The commenter felt that, given interactions and synergisms, it is much
more likely that exposure to multiple chemicals will have an additive or
multiplicative effect than that such chemicals will cancel each other
out. This commenter cited many sources of uncertainty in the risk
assessment and concluded that the EPA may have drastically understated
the real leukemia risk associated with benzene. According to the commenter,
the estimates given by the EPA may well underestimate the health benefits
of the increment between the proposed requirements and use of vapor
recovery or thermal destruction of emissions. He added that it is
unacceptable that the noncarcinogenic effects of benzene exposure have
virtually dropped out of the EPA's analysis due to the fact that they
cannot be readily quantified. According to the commenter, the proposal
makes no efforts to see that these effects get appropriate weight in the
decision to stop short of more stringent regulatory alternatives.
Response: While the commenter may be correct that interactions and
synergisms (resulting from exposures to multiple chemicals) may be
additive or multiplicative (or antagonistic) and therefore result in
truly greater (or smaller) risks to persons exposed to benzene, EPA is
unable to estimate these effects and, therefore, has not considered
them. It should be noted that many of the factors used in making the
exposure assessment have uncertainties associated with them and that
these uncertainties can result in underestimation as well as overestimation.
These uncertainties are described in a previous response (2.1.2) and
have been considered as much as is practicable by EPA in the decision to
withdraw the proposed standards.
Comment: A commenter (IV-D-19) noted that the EPA assumed that
many benzene-emitting facilities have a life expectancy of 50 years or
more. Yet the quantifications of risk used to compare the proposed
approach with a more protective one, assume a 20-year lifetime. According
to the commenter, this understates the number of benzene victims for
such facilities by two and one-half times or more, and reconsideration
'of the decision not to adopt Alternatives IV or V (at proposal) with the
appropriate health effects timeframe may lead to a different decision
(IV-D-31).
Response: Twenty years is an average figure for the lifetime of a
plant. Some plants have a life expectancy of 50 years; some have lifetimes
2-37
-------
shorter than 20 years. Roughly, a plant will have a lifetime of 20 years.
Since there are little data available that estimate plant lifetimes, the
EPA considers 20 years to be a reasonable estimate. However, the EPA
agrees there is uncertainty associated with this number.
2.2.4 Risk Methodology Consistency
Comment: A number of commenters (IV-D-lOa, IV-D-4, IV-F-1) stated
that there should be some consistency in risk assessment methodologies
between the four current benzene proposals. One commenter stated that
if benzene is to be regulated by a NESHAP standard, the emission concerns
and risk/benefit analysis should be completed for all types of emissions
(e.g., process emissions, storage tank emissions, fugitive emissions,
etc.) simultaneously. This integrated analysis, the commenter maintained,
would prevent duplication of effort, errors, or inconsistencies and
result in an overall analysis of the risk/benefit of a product. According
to another commenter1s (IV-D-21) review of the four current benzene
proposals, a great deal of duplication has occurred with little or no
health benefit to the public.
Response: The risk assessment methodologies used in evaluating the
four source categories for which benzene standards have been proposed
have been made more consistent. The only area in which they are different
is that the affected industries voluntarily submitted detailed plant-
specific information on the maleic anhydride and ethylbenzene process
vents. Obtaining this kind of information for the 126 or more plants
that have benzene fugitive and storage sources would be too costly
(considering the uncertainty of the final results either way) for the
industry or the EPA to obtain. Because of the detailed information
available on maleic anhydride and ethylbenzene/styrene process vents and
the relatively small number of these plants, the more precise ISC
dispersion model was used for all those plants and the SAI model was
used for the benzene storage and benzene fugitive plants. However, the
ISC model was used for a few plants with benzene storage and fugitive
emissions to compare the results of that model with the SAI model. For
plants containing multiple sources, the same meteorological and population
data were used for each plant each time the risks were calculated for
one of the sources in that plant.
2-38
-------
APPENDIX A
EMISSIONS SOURCE TEST DATA AND ANALYSIS
-------
APPENDIX A - EMISSIONS SOURCE TEST DATA AND ANALYSIS
This appendix provides a summary description of the emission tests
conducted on internal floating roof (IFR) tanks and the major results.
For additional and complete information, refer to the referenced reports.
A.I TEST PROCEDURES
All emissions test measurements were obtained by Chicago Bridge and
Iron Company (CBI) under contract to the American Petroleum Institute.1
The test program was divided into two broad components: pilot tank test
measurements and internal floating roof tank component measurements.
The primary goal of the pilot tank tests was to determine emissions from
IFR seal systems and deck seams; while the purpose of the IFR component
tests was to determine emissions from IFR fittings (hatches, ladder
wells, etc.) and to investigate other issues such as the permeability of
seal systems to the stored hydrocarbon.
A.1-1 Pilot Test Tank Emission Measurements
A. 1.1.2 Description of Test Facility. The tests were performed in
a test IFR tank at CBI's research facility in Plainfield, Illinois. The
test tank was 20 feet in diameter and had a 9-foot shell height (see
Figure A-l). The lower 5'3" of the tank shell was provided with a
heating/cooling jacket through which a heated or cooled water/ethylene
glycol mixture was continuously circulated to control the product
temperature. The effect of air blowing through the shell vents was
simulated by means of a blower connected to the tank by a 12-inch diameter
duct. This air exited from the tank through a similar duct.
Based on wind tunnel tests, it has been possible to determine the
pressure coefficient, C , variation over the exterior surface of the
tank. The air flow rate through the vents over the internal floating
roof was then related to C by means of a mathematical model.1 Thus,
A-2
-------
.Inlet
Concentration
Outlet
Concentration
Air
Heater
.Outlet
T) Damper
Mixing
Section
Shell
Heating
Panel
Air Blower
Propane
P - Pressure
T - Temperature
F - Flow
S - Sample
Glycol
Pump
Product
Circulation
Pump
Stripper
Tower
Figure A-l. Process and instrumentation schematic.
1
-------
internal air flow could be related to ambient wind speed emissions.
During each test, emissions were measured at several equivalent ambient
wind speeds. The recorded data included the inlet and outlet total
hydrocarbon content, system temperatures, and the inlet air flow rate.
A.1.1.3 Pilot Test Tank Internal Floating Roofs and Liquids.
Tests were conducted in three IFR types, and three seal systems. The
first IFR tested (Phase 1, 1R) was a bolted noncontact IFR, equipped
with a wiper type primary seal, and on some tests a secondary seal
(Figures A-2 and A-3). In some tests gaps were intentionally placed
between the seal and the tank shell. Seal gaps were either of 1 or
3 square inches of gap per-foot-of-tank-diameter. In some instances,
0.020 inch thick polyurethane-coated nylon fabric, which was taped in
place using aluminum-backed duct tape, was used to seal off certain
emission sources.
The second IFR tested (Phase 2, 2R) was a welded contact IFR equipped
with a liquid-mounted, foam filled seal (Figures A-4 and A-5). As in
Phase 1, a secondary seal was in place during some tests; the effects of
seal gaps on emissions were investigated; and emission areas were sealed
during some tests.
The final IFR (Phase 3, 3R) was a bolted contact type deck, equipped
with a vapor-mounted, foam-filled primary seal, and (during some tests)
a foam-filled secondary seal (Figures A-6 and A-7).
In each phase, three different test liquids were employed. The
test liquids were a propane/octane mixture, hexane, and octane.
During Phase 1, the primary seal was replaced after Test No. 13.
The primary seal was again replaced at the beginning of Phase 1R (Test
API 73). Each of the primary seals had the same construction.
The initial Phase 1 tests indicated that emissions might vary as a
function of the inlet air-product temperature difference. To control
for this, a heater was installed in the inlet air duct after Test API 19.
Table A-l displays the test conditions for all Phase 1, 1R tests.
Table A-2 displays the test conditions for the Phase 2, 2R tests.
There was a problem with product seepage through a thermocouple during
Tests API 35 through API 44. However, it was possible to correct the
results to account for this problem. Additionally Test API 51 was
performed at the much higher air flow rates that simulate an external
floating roof tank.
A-4
-------
Deck Scam
Clamping Bar-*
Location
Fittings
1 Access Hatch
2 Column Well
3 Vacuum Breaker
0
Air Outlet
Figure A-2. Plan view of noncontact bolted IFR.
A-5
-------
Figure A-3. Elevation view of noncontact bolted IFR in test tank.
A-6
-------
Thermocouple
Locations
D=Deck
L = Liquid
SV* Sec. Vapor
Fittings
1 s Guide Pole
2 * Bleeder Vent
3sColumn Well
4 = Bolted Access
Hatch
Seal
N
O
Rim Brace
Figure A-4. Plan view of contact welded IFR.
A-7
-------
00
(Q
c
-5
(D
I
tn
a>
<
0)
O
3
o
n
o
3
r+
0)
n
CL
a>
a.
3
r+
n>
01
3
63"
Shell Heating Panel
46
II
Product Level
l"
Rim
Heat
Coil
34"
Air Plenum
Roof Elcv.
108'
-------
Thermocouple
Locations
D = Deck
L = Liquid
PV= Pri. Vapor
SV= Sec.Vapor
Air Inlet
Roof Support
Lugs
N
O
Deck Panels
Seal
Column
Well ~\
Bolted Deck
Seams
Figure A-6. Plan view of contact bolted IFR.
A-9
-------
(Q
C
I
M
o
m
(D
<
0)
C*
o
n
o
0)
n
CT
O
(D
Q.
Z3
tt>
OJ
a
63"
Shell Heating Panel
Air Plenum
Rim
Heat
Coil
Product Level
36
Roof EI<2V.
-------
Table A-l. SUMMARY OF TEST CONDITIONS FOR PHASE 1 AND 1R
fcl_._l_A t
Test
number
Phase 1:
API 1
API 2
API 3
API 4
API 5
API 6
API 7
API 8
API 9
API 10
API 11
API 12
New Primary
API 13
API 14
API 15
API 16
API 17
API 18
API 19
API 19A
nuw HUM
Product vapor pressure
type (psla)
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
Seal Installed
nC8
nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
5.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
0.5
0.5
0.5
0.5
0.5
0.5
5.0
5.0
5.0
5.0
5.0
5.0
Gap area
Ort2/ffr rfl«Matav>^
n /ii. cjiameierj
Primary Secondary
0 —
0 —
0 —
0 —
0 —
0 —
1 -•- —
3 —
*
i —
0 —
0 (1) -
0 —
1 —
0 —
0 —
0 0
3 0
3 1
3 1 (2)
Roof components
Column
well
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Sealed
Sealed
Sealed
Sealed
Sealed
Sealed
Sealed
Sealed
Sealed
Unsealed
Unsealed
Sealed
Sealed
Sealed
Sealed
Deck
fittings
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Sealed
Sealed
Sealed
Sealed
Sealed
Sealed
Sealed
Sealed
Sealed
Unsealed
Unsealed
Sealed
Sealed
Sealed
Sealed
Nominal
(mi r-nrnHurt^
Deck temperature difference
seams (°F) Notes
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Variable
Variable
Variable
Variable
Variable
Variable
Variable
Variable
Variable
Variable
Variable
Variable
Variable
Variable
Variable
Variable
Variable
Variable
Variable
Variable
Air product
temperature
difference
was
uncontrolled
Air product
temperature
difference
was
uncontrolled
Air Duct Heater Installed
API 20
API 21A
API 218
API 21C
API 210
API 21E
API 22A
API 228
API 2?C
API 220
API 23
•API 24
API 25
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
Variable
5.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
0 —
0 —
0 —
0 —
0 —
0 —
1
I
1 —
•I ,
Sealed —
Sealed —
Sealed —
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Sealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Sealed
Sealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
0
-15
0
+15
0
Variable
-15
0
+15
Variable
0
0
0
(3)
(4)
(continued)
-------
Table A-l. Concluded
ro
Test
number
API
API
API
API
API
API
API
API
API
API
API
API
API
API
API
API
API
26A
26B
27A
27B
27C
28
29
29R
30
30R
31
31A
32
33
33A
34
34A
Nominal
Product vapor pressu
type
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
nC8
nCfl
nC8
nC8
nC8
nC8
nC6
nC6
nC6
nC6
nC6
(psla)
3.5
2.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
2.5
2.5
2.5
2.5
2.5
Gap area
(InVft diameter
||~g '
Primary
1
1
1
1
1
0
1
1
0
0
1
1
0
1
1
1
1
Roof components
.\
/
— — Pnlimn
^^^ liUIUfflil
Secondary well
_
—
—
—
—
—
_
—
—
—
0
0
_«.
—
—
—
—
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Deck
fittings
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Deck
seams
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Nominal
(air-product)
temperature difference
(°F)
0
0
-15
0
+15
0
0
0
0
0
0
+15
0
0
+15
0
+15
Notes
(5)
(5)
Phase 1R:
API
API
API
API
API
API
API
73
73A
74
75
76
76R
77
Notes: (I)
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
Seal closure
5.0
5.0
5.0
5.0
5.0
5.0
5.0
devices
(?). Gaps fn the secondary
(3)
(4)
(5)
(6)
(7)
Emission test
. Emission test
. Emission test
A column well
Emission test
data Is
data Is
data Is
gasket
data Is
0
0
0
3
Sealed
Sealed
Sealed
were Installed to
seal were rotated
questionable due
questionable due
questionable due
_
—
—
—
—
—
—
Unsealed
Unsealed(6)
Unsealed(6)
Unsealed(6)
Sealed
Sealed
Sealed
eliminate all unintentional
45° to
Unsealed
Unsealed
Unsealed
Unsealed
Sealed
Sealed
Sealed
gaps.
position them directly above the
to variable product temperature causing
to nonequi 1 ibritim condition
to air
inlet heater control
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Sealed
0
0
0
0
0
0
0
(7)
primary seal gaps.
nonequilibrlum
in the rim vapor space due
problems.
conditions.
to prior air purge.
was used during this test.
questionable due
to nonequilibrlum condition
of product caused by Insufficient mixing.
-------
Table A-2. SUMMARY OF TEST CONDITIONS FOR PHASE 2 AND 2R (1)
Test
number
Product
type
Nominal
vapor pressure
(psla)
Gap area
(inVft diami
ter)
Primary
Secondary
Roof components
Column
well
Deck
fittings
Notes
Phase 2
API 35
API 36
API 37
API 38
API 39
API 40
API 41
API 42
API 43
API 44
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
5.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
0
1
0
0.5
3
1
0
1
3
Sealed
0
0
1
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Sealed
Sealed
Sealed
Sealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Sealed
Sealed
Sealed
Sealed
(2)
(2)
(2)
(2)
(2)
(2)
(2)
(2)
(2)
(2)
Repaired Product Seepage Through Thermocouple Fitting
API 45
API 46
API 47
API 48
API 49
API 50
API 51
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
nC8
nC8
5.0
5.0
5.0
5.0
5.0
0.5
0.5
Sealed
Sealed
Sealed
0
0
1
1
Sealed
Unsealed
Unsealed
Unsealed
Sealed
Unsealed
Unsealed
Sealed
Sealed
Unsealed
Unsealed
Sealed
Unsealed
Unsealed
(3)
(3), (4)
Phase 28
API
API
API
AD?
nri
API
API
API
API
67A
67
68
CO
oy
70
71
71A
72
Notes:
(1).
(2).
(3).
(4).
(5).
nC8
nC8
nC6
nC6
C3/nC8
C3/nC8
C3/nC8
C3/nC8
During
Product
Product
0.5
0.5
2.5
2.5
5.0
5.0
5.0
2.5
1
1
1
1
1
1
1
1
;
-
0
0
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
both Phases 2 and 2R, nominal (air-product) temperature difference
seepage through a thermocouple fitting occurred during this test.
contained trace amount of orooane.
During this test
During this test
kept constant at
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
was kept at
the air flow rate was increased to simulate an external floating roof
the inlet air and product heaters were turned off, and the wind speed
about 10 ni/hr.
(3)
(5)
zero.
was
A-13
-------
Table A-3 displays the test conditions of Phase 3 and 3R. During
some tests product penetrated the primary seal. The problem was repaired,
and the tests were repeated.
Table A-4 presents the results of all relevant tests. In summary,
it was found that an air product temperature differential of up to 15F°
had no significant effect on emissions. Small gaps (1 inchVfeet diameter)
did not appear to affect emissions significantly. Also, the tests
demonstrate that ambient wind (particularly at speeds less than 20 miles
per hour) has little or no effect on emissions.
A.1.1.4 IFR Component Tests.
A.1.1.4.1 Deck fitting emission tests. To quantify emissions from
various types of fittings, a series of bench scale tests were performed.
These fittings were placed through the top cover of a liquid-filled
drum, and the drum was then placed on a scale. The weight change and
other data were recorded over a 30 day period. Figure A-8 displays a
sample bench test, and Table A-5 summarizes the results.
A.1.1.4.2 Permeability tests. A series of bench permeability
tests were performed to determine the permeability of the 0.020 inch-
thick polyurethane-coated nylon fabric to various hydrocarbon liquids.
One laboratory test was also performed. Also included was a test on the
same fabric of 0.037 inch thickness with benzene as a test liquid. This
material had been used as the seal envelop material in Phase 2 and 2R,
2
and in earlier test work. The results are shown in Table A-6.
A.2 MAJOR RESULTS
This section discusses the major results of the analysis of test
work. Although the relationship of emission factors to the test results
is discussed, the actual development of emission factors is presented
elsewhere.
A.2.1 Seal Losses
Total measured emissions in a given tank test are the sum of all of
the emission sources in that test. Therefore, to develop an emission
factor the results must be reduced. For example, the permeation emissions
through any sealing material, fittings, and any other source that is not
of interest must be accounted for, and subtracted out before the emissions
from the component of interest are known. Because of this reduction
process, component emissions factors cannot be read directly from Table A-4.
A-14
-------
Table A-3. SUMMARY OF TEST CONDITIONS FOR PHASE 3 AND 3R (1)
Test
number
Product
type
Nominal
vapor pressure
(psia)
Gap area
(in'/ft diameter)
Primary Secondary
Roof components
Column
well
Deck
seams
Rim
plate
Notes
Phase 3
API 52A
API 528
API 52C
API 520
API 52E
API 53A
API 53B
API 53C
API 54A
API 54B
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
5.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
0
0
0
0
0
1
1
1
3
3
0
0
0
0
0
0
0
0
1
1
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
(2)
(2)
(2)
(2)
(2), (3)
(2)
(2)
(2)
(2)
(2)
Product Liquid Removed From Primary Seal
API 52
API 52R
API 53
,API 54
^PI 55A
API 55
API 56
UP I 57
«PI 58
WPI 59
API 60
API 61
API 62
API 63
API 64
API 65
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
C3/nC8
nC8
nC6
5.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
2.5
0.5
0.5
2.5
0
0
1
3
Sealed
Sealed
Sealed
Sealed
Sealed
0
j
3
1
1
1
1
0
0
0
1
Sealed
Sealed
Sealed
Sealed
Sealed
-
-
-
-
-
"
Unsealed
Unsealed
Unsealed
Unsealed
Sealed
Sealed
Sealed
Sealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Sealed
Sealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Sealed
Sealed
Sealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
(4)
(4)
(4)
(4)
Phase 3R
API 65R
API 65A
API 66
API 66R
nC6
nC6
nC6
nC6
2.5
2.5
2.5
2.5
1
1
1
1
0
0
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
Unsealed
(5)
(2)
Notes: (1).
(2).
(3).
(4).
(5).
During both Phases 3 and 3R, Type 1 air flow distribution was used, the nominal (air-product)
temperature difference was kept at zero, and the roof elevation was kept at 63 inches below
the air inlet.
Emission test data is of questionable value since liquid product was present in the primary
Column well cover intentionally positioned off center with a gap.
All taped joints w«r« also caulked during this test.
During this test the primary seal gap plates were intentionally extended down into the
product.
A-15
-------
Table A-4. SUMMARY OF TEST RESULTS FOR
ALL POTENTIALLY RELEVANT TESTS
CBI test number
API-1
API-2
API-3
API-4
API-5
API-7
API-8
API-12
API-13
API-14
API-13R
API-13, 13R
API-14R
API-14, 14R
API-16
API-17
API-18
API-19
API -19 A
API-21A
API-21B
API-21C
API-21AR
API-21A, AR
API-21BR
API-21B, BR
API-21CR
API-21C, CR
API-22A
API-22BI
API-22D
API-22B
API-22BI, B
API-22C
API-21E
API-23
API-24
API-25
API-26A
API-26B
API-27A
Nominal .
true vapor pressure
(psia)
5.00
5.00
5.00
5.00
5.00
5.00
5.00
0.50
0.50
0.50
0.50
0.50
0.50
0.50
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
0.50
(continued)
A- 16
Average
emissions
Ib-mole/day
0.283
0.423
0.309
0.449
1.33
0.224
0.439
0.0181
0.0605
0.0668
0.0567
0.059
0.196
0.159
0.926
0.0698
0.110
0.134
0.147
0.101
0.0891
0.0909
0.171
0.129
0.140
0.102
0.133
0.108
0.142
0.165
0.124
0.176
0.173
0.211
0.128
0.0714
0.120
0.108
0.117
0.128
0.030
-------
Table A-4. Continued
CBI test number
API-27B
API-27C
API-28
API-30
API-29R
API-31
API-31A
API-32
API-33
API-33A
API-34
API-34A
API-35
API-36
API-37
API-38
API-39
API-39R
API-40
API-41
API-42
API-43
API-44
API-45
API-46
API-47
API-48
API-49
API-50
API-51
API-52
API-53P
API-54
API-53
API-53P, 53
API-55
API-56
API-57
API-58
API-52R
API-52, 52R
API-59
Nominal ,
true vapor pressure
(psia)
0.50
0.50
0.50
0.50
0.50
0.50
0.50
2.50
2.50
2.50
2.50
2.50
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
0.50
0.50
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
Average
emissions
Ib-mole/day
0.0196
0.0553
0.0167
0.0316
0.143
0.0357
0.0256
0.0232
0.0306
0.0251
0.0317
0.0347
0.0366
0.0359
0.0297
0.0334
0.0492
0.0387
0.0301
0.0154
0.0176
0.0269
0.0149
0.00693
0.00928
0.0170
0.0246
0.0188
0.00426
0.0390
0.0376
0.0407
0. 0400
0.0372
0.0399
0.0156
0.0338
0.0345
0.0433
0.0435
0.0400
0.0536
(continued)
A-17
-------
Table A-4. Concluded
CBI test number
API-60
API-61
API-62
API-63R
API-64
API-65
API-66
API-66R
API-65R
API-65A
API-67A
API-67
API-68
API-69
API-70
API-71
API-72
API-73
API-73A
API-74
API-75
API-76
API-76R
API-76, 76R
API-77
Nominal ..
true vapor pressure
(psia)
5.00
5.00
5.00
0.50
0.50
2.50
2.50
2.50
2.50
2.50
0.50
0.50
2.50
2.50
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
5.00
Average
emissions
Ib-mole/day
0.0574
0.0690
0.0649
0.00930
0.00867
0.0242
0.0378
0.0322
0.0407
0.0417
0.00779
0.00500
0.0105
0.00715
0.0202
0.0247
0.040
0.0466
0.0628
0.0627
0.0730
0.0509
0.0433
0.0484
0.0417
Nominal average true vapor pressure (TVP) is the TVP at which the
emissions were calculated by using the vapor pressure function to
normalize the measured hydrocarbon concentration to the concentration
expected at the nominal TVP.
A-18
-------
ft
Y
\
22.5* I,D,
X TEST DRUF1
GASKET
1 1/2" mi DIA, ALUfl,
PIPE, SCH, MO
COVER TENSIONING
CABLE
WOODEN PALLET
PLACED ON SCALE
Figure A-8. Example of fitting emission bench test apparatus.
A-19
-------
Table A-5. SUMMARY OF IFR DECK FITTING EMISSION TESTS
ro
o
Test
number
1
2
3
4
5
6
7
8
9A
9B
10
11
12
13
14
15
Description
Access hatch cover, ungasketed
Access hatch cover, gasketed and clamped
lh inch diameter adjustable roof leg
8 inch diameter slotted pipe sample well
8 inch diameter pipe column well
1 inch diameter stub drain
Phase 1 column well, ungasketed
^ inch gap around built-up column
Phase 1 column well, gasketed
Phase 1 column well, ungasketed
Phase 2 column well
Phase 3 column well (1)
1/8 inch gap around built-up column
Access hatch cover with 1/8 inch gap
Sample well with 10% gap area
1/8 inch gap around built-up column (1)
Product
type
nC6
nC6
nC6
nC6
nC6
nC6
nC6
nC6
C3/nC8
C3/nC8
C3/nC8
C3/nC8
nC6
nC6
nC6
nC6
Correlation
coefficient
(-)
0.681
0.689
0.914
0.996
0.989
0.902
0.998
0.998
0.977
0.959
0.964
0.986
0.983
0.997
0.985
0.983
Average
emission
rate (2)
(Ib mole/yr)
0.204
0.158
0.977
4.69
2.11
0.279
4.32
5.42
3.38
5.07
1.22
2.25
2.44
5.61
1.45
2.81
Notes: (1). Test drum was 30 in. diameter.
(2). Average emission rate normalized to a nominal vapor pressure of 5.00 psia.
-------
•£•
ro
Table A-6. PERMEABILITY OF POLYURETHANE COATED NYLON FABRIC
Test
number
16
17
18
19
20
21
22
23
Laboratory
Fabric
thickness
(in)
0.020
0.037
0.037
0.020
0.020
0.020
0.020
1/16" thk
aluminum
permeability
0.020
Fabric
area
(ft2)
2.75
2.75
2.75
2.75
2.75
2.75
2.75
Jest
0.467
Length
of taped
seams Product
(in) type
C3/nC8
C6H6
nC6
C3/nC8
C3/nC8
48 C3/nC8
C3/nC8
60 C3/nC8
nC6
Average
product
temperature
(°F)
59.2
60.5
60.1
53.8
48.1
50.9
43.2
44.2
74.8
Average
vapor
pressure
(psia)
7.13
1.22
1.98
3.86
3.56
4.68
3.59
3.38
1.85
Vapor
mole weight
(Ibm/lbmole)
45.8
78.1
86.2
46.6
46.3
45.9
46.0
46.3
86.2
Average
emission
rate
(Ibm/day)
0.0612
0.159
0.0158
0. 0652
0.0808
0. 0650
0.0344
0.00273
0.0244
Correlation
coefficient
(-)
0.838
0.996
0.663
0.783
0.806
0.863
0.805
0.096
--
Average
rate
(Ibm/ft2 day)
0.0222
0.0578
0.00578
0.0237
0.0294
0.0236
0.0125
0. 0522
Notes
(1)
(1)
Notes: (1). Aluminum backed duct tape was used on all taped seams.
-------
For seal systems, it was found that
Es = Kp Mw D P* (C-l)
Where:
ES = Emissions from the seal area in Ibs/day
Kr = Seal factor
Mw = Molecular weight of vapor
D = Tank diameter
P* = Vapor pressure function
The reduced emissions from seals of similar construction and gap condition
are averaged together. A seal emission factor is the weighted average
of the averaged reduced emissions. Weights are selected according to
field survey data that relate seal gap area to frequency of occurrence.
The emission factor which results from this procedure of repeated
subtraction and averaging does not represent any given tank, but is
rather an expected value.
The analysis shows that for emission purposes seals may be divided
into two types: liquid-mounted and vapor-mounted. An emission comparison
of reduced results between the foam-filled vapor-mounted seal tested
during Phase 3 and 3R and the vapor-mounted wipers tested in Phase 1
and 1R, shows that emissions from the foam-filled seal were lower than
the Phase 1 wiper but higher than the Phase 1R wiper (Table A-7). On
this basis, the results from Phases 1, 1R, 3 and 3R were merged into the
general category of vapor-mounted seal.
The analysis shows that emissions from the liquid-mounted seal
tested in Phase 2 and 2R are lower than both the average of the merged
vapor-mounted seal tests and the individual vapor-mounted seal systems
that were actually tested (Table A-8).
Another finding was the presence of the secondary seal reduced
emissions whether or not the primary seal was gapped. Emissions reductions
obtained by a secondary seal average 47 percent for a liquid-mounted
primary seal and 63 percent for a vapor-mounted primary seal.
A.2.2 Deck Seam Losses
The welded IFR tested in Phase 2 and 2R was assumed to have no deck
seam emissions. The IFR's tested in Phases 1, 1R, 3 and 3R have bolted .
deck seams. The seams in the contact deck (3 and 3R) had a different
construction than those in the noncontact deck (1 and 1R). However, the
A-22
-------
Table A-7. COMPARISON OF WIPER SEALS TO FOAM-FILLED
VAPOR-MOUNTED SEALS
Seal gaps
Seal emissions
(Ib mole/day)
(inVft diameter) Phase 1 wiper Foam-filled Phase 1R wiper
0.0566
0.0248
0.0217
0.0978
0.0324
0.0402
0.0319
No test available.
A-23
-------
Table A-8. COMPARISON OF LIQUID-MOUNTED SEAL TO VAPOR-MOUNTED SEAL
Seal emissions
(Ib mole/day)
Seal gap
(inVft diameter) Liquid-mounted Vapor-mounted1
0 0.0052 0.0217
1 0.0176 —2
3 0.030 0.0319
Based on the best performing vapor-mounted seal (Phase 1R wiper).
No test available.
A-24
-------
test data show that there is no significant difference in emissions from
the seams in the two decks (on a per-foot-of-seam-basis) despite
differences in construction and position relative to the stored liquid
(Table A-9). It should be noted that Test API 76 was not used in making
the comparison. API representatives have stated that due to slight
problems in the test, Test API 76 is not comparable with API 76R.4
The per-foot-of-seam results that appear in Table A-9 were averaged
together and divided by the value of the vapor pressure function to
develop the deck seam emission factor Kd. Further minor mathematical
procedures are needed to develop Krf as it appears in Chapter 3. These
procedures relate seam length to deck diameter.
A.2.3 Effect of Liquid Type on Emissions
Comparisons between previous test programs had indicated that
emissions for single component (pure) liquids (e.g., benzene), could be
significantly higher than emissions from multicomponent liquids
(e.g., gasoline) when normalized for both molecular weight and vapor
pressure. Tests performed in the API program show that between the
tested liquids (hexane, propane/octane, and octane) there were no
significant emissions differences after normalizing for molecular weight
and vapor pressure (Table A-10).
A.2.4 The Effect of Vapor Pressure on Emissions
Several emissions tests (from Phase 2 and 2R) were conducted to
determine the effect of the product vapor pressure, P, on the emissions
rate. This relationship was evaluated during these tests by varying the
product vapor pressure in the pilot test tank which had been fitted with
a contact-type internal floating roof and a liquid-mounted primary seal.
Based on these tests, the emissions are directly related to the vapor
pressure function, P*:
P* =
P
14.7
A.2.5 Fitting Emissions
The fitting emission factors are developed by a procedure similar
to that used for seal factors. A particular fitting design is analyzed
to determine emission points and the results of the bench tests are
A-25
-------
ro
en
Table A-9. BOLTED DECK SEAM EMISSIONS1
Product
Test number type
Bolted, Contact IFR
API 55 C3/nC8
API 56 C3/nC8
Bolted, Noncontact IFR
API 76R C3/nC8
API 77 C3/nC8
Nominal
vapor pressure
(psia)
5.00
5.00
5.00
5.00
Vapor
•ole weight
(Ibm/lbmole)
48.1
48.2
46.8
47.1
Deck
seams
Sealed
Unsealed
Unsealed
Sealed
Total deck
seam length
(ft)
89
89
36
36
Emissions at
nominal vapor pressure
(Ibmole/day)
0.0156
0.0338
0.0433
0.0417
Emissions per foot
of deck seam
(Ibmole/day)
0.0002
0. 00004
Other test conditions:
Primary seal - sealed
Secondary seal - none
Deck fittings - sealed
-------
Table A-10. COMPARISON OF EMISSIONS AS A FUNCTION OF LIQUID TYPE
Test number
Phase 2, 2R
API 50
API 67
API 67A
API 68
API 69
API 71
API 72
API 36
Phase 3, 3R
API 64
API 65
API 65R
API 65A
API 60
Product type
nC8
nC8
nC8
nC6
nC6
C3/nC8
C3/nC8
C3/nC8
nC8
nC6
nC6
nC6
C3/nC8
p
Emissions
(Ib mole/day)
0.0510
0.0599
0.0932
0.0233
0.0159
0.0247
0.040
0.0359
0.103
0.0537
0.0905
0.0927
0.0574
All tests had identical conditions as follows:
a. 1 inVft. diameter of gap on primary seal
b. No secondary seal.
c. All roof components unsealed.
2
Emissions are normalized to 5.0 psia.
A-27
-------
added and subtracted to account for each emission source in the design.
The individual emission sources are summed, and the resulting sum is
made independent of molecular weight and vapor pressure to form the
fitting factor.
The test results show that the addition of gaskets and the bolting
of covers will reduce emissions from fittings. Also demonstrated is the
fact that small fitting design differences can lead to significant
differences in emissions.
A-28
-------
A.3 REFERENCES
1. Laverman, Royce J. et. al. Testing Program to Measure Hydrocarbon
Emissions from a Controlled Internal Floating Roof Tank;
(Unpublished), Chicago Bridge and Iron Co. Chicago, Illinois
March 1982. 304 pp.
2. U.S. Environmental Protection Agency. Measurements of Benzene
Emissions from a Floating Roof Test TanEReport No. EPA-450/3-
79-020. Research Triangle Park, N.C. June 1979.
3. Letter and attachments, from O'Keefe, William, F., American Petroleum
Institute, to Wyatt, Susan R., EPA. January 25, 1983.
4. Moody, W.T., TRW, Meeting on September 2, 1982, Durham, N.C.
between API, EPA, and TRW.
A-29
-------
APPENDIX B
METHODOLOGY FOR ESTIMATING LEUKEMIA INCIDENCE AND
MAXIMUM LIFETIME RISK FROM EXPOSURE TO
BENZENE STORAGE TANKS
-------
APPENDIX B
METHODOLOGY FOR ESTIMATING LEUKEMIA INCIDENCE AND MAXIMUM
LIFETIME RISK FROM EXPOSURE TO BENZENE STORAGE TANKS
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 benzene
storage tanks. The methodology consists of four major components:
estimation of annual average concentration patterns of benzene in the
region surrounding each plant, estimation of the population exposed to
each computed concentration, calculation of exposure by summing the
products of the concentrations and associated populations, and calcu-
lation of annual leukemia incidence and maximum lifetime risk from the
concentration and exposure estimates and a health effects estimate
represented by a unit risk factor. 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 AND PLANT EMISSION RATES
The Human Exposure Model (HEM) was used to estimate concentrations
of benzene around approximately 126 plants that contain benzene storage
tanks. The HEM estimates the annual average ground-level concentrations
resulting from emissions from point and area sources. For point sources,
the dispersion model within HEM is a Gaussian model that uses the same
basic dispersion algorithm as the climatological form of EPA's Climato-
2
logical Dispersion Model. Gaussian concentration files are used in
conjunction with multi-year STAR data and annual emissions data to
estimate annual average concentrations. Details on this aspect of the
HEM can be found in Reference 1.
B-2
-------
Seasonal or annual stability array (STAR) summaries are principal
meteorological input to the HEM dispersion model. STAR data are standard
climatological frequence-of-occurrence summaries formulated for use in
EPA models and available for major U.S. sites from the National Climatic
Center, Asheville, N.C. A STAR summary is a joint frequency of occurrence
of wind speed stability and wind direction categories, classified according
to the Pasquill stability categories. For this modeling analysis,
annual STAR summaries were used.
The model receptor grid consists of 10 downwind distances located
along 16 radials. The radials are separated by 22.5-degree intervals
beginning with 0.0 degrees and proceeding clockwise to 337.5 degrees.
The 10 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 center of the receptor
grid for each plant was assumed to be plant center as determined by
review of maps.
Inputs to the dispersion model include the geographical coordinates
for each plant, and the emission rates, dimensions and plume character-
istics for each storage tank in each plant. The latitudes and longitude
for each plant, used in selecting the STAR site, are listed in Table B-l.
Four model units representing the different types of plants that would
have benzene storage tanks were developed: large producers of benzene,
small producers of benzene, benzene consumers, and bulk storage terminals.
The model units were assigned to each plant according to the uses of
benzene within the plant. Where a plant had two model units assigned to
it (e.g., a plant may be both a producer and consumer of benzene),
emissions from both model units were used in calculating the concentration
pattern around the plant. The model units assigned to each plant are
listed in Table B-l.
Each model unit consists of a set of benzene storage tanks with
specified dimensions, roof types, turnovers, and emission rates. The
tank parameters used in the dispersion model are the same for benzene
consumers and bulk storage terminals; therefore, no differentiation was
made between them for modeling purposes. Table B-2 shows, for each
model unit, the height and vertical cross-sectional area (used in downwash
calculations) of each tank. The table also shows the emissions from
B-3
-------
Table B-l. PLANTS AND LOCATIONS FOR BENZENE STORAGE TANKS
Plant
Region II
1. American Cyanamid
2. DuPont
3. Exxon
4. Standard Chlorine
5. Texaco
6. Ashland Oil
7. ICC Industries
8. Commonwealth Oil
9. Phillips Puerto Rico
10. Puerto Rico Olefins
11. Union Carbide
12. Amerada Hess
Region III
13. Getty
14. Standard Chlorine
15. Sun-Olin
16. Continental Oil
17. Atlantic Richfield
18. Gordon Terminals
19. Gulf Oil
20. Standard Oil
(Ohio)/BP Oil
21. Sun Oil
22. U.S. Steel
23. Allied Chemical
24. American Cyanamid
25. Mobay Chemical
26. PPG
27. Union Carbide
Coordinates
Location
Longitude
Latitude
Boundbrook, NJ
Gibbstown, NJ
Linden, NJ
Kearny, NJ
Westville, NJ
North Tonawanda, NY
Niagara Falls, NY
Penuelas, PR
Guyama, PR
Penuelas, PR
Penuelas, PR
St. Croix, VI
Delaware City, DE
Delaware City, DE
Claymont, DE
Baltimore, MD
Beaver Valley, PA
McKees, PA
Philadelphia, PA
Marcus Hook, PA
Marcus Hook, PA
Neville Island, PA
Moundsville, WV
Willow Island, WV
New Martinsvilie, WV
Natrium, WV
Institute, WV
74°06'04"
75°17'50"
74°12'49"
74°06'39"
75°08'42"
78°55'27"
79000'55"
66°42'00"
66°07'00"
66°42'00"
66042'00"
64°44'00"
75°37'45"
75038'47"
76°25'40"
77°34'02"
80021'20"
80°03'10"
75°12'31"
75°37'45"
75024'51"
80°05'00"
80°48'04"
81°19'08"
80°49'50"
80°52'14"
81047'05"
40°33'25"
39°50'25"
40°38'10"
40045'03"
39°52'05"
42°59'45"
43°03'33"
18004'00"
17°59'00"
18°04'00"
18°04'00"
17°45'00"
39°35'15"
39°33'54"
39°48'20"
39°14'19"
40°39'21"
40°28'22"
39°54'18"
39°35'15"
39°48'45"
40°30'00"
39°55'00"
39°21'45"
39°43'30"
39044'46"
38°22'40"
Model Plant
Type9
C/T
C/T
SP.C/T
C/T
LP.C/T
SP
C/T
LP.C/T
LP.C/T
C/T
C/T
LP
SP
C/T
C/T
C/T
C/T
C/T
LP.C/T
SP
SP
C/T
C/T
C/T
C/T
C/T
C/T
(continued)
B-4
-------
Table B-l. Continued
Plant
Region IV
28. Jim Walter Resources
29. Reichhold Chemicals
30. Ashland Oil
31. B.F. Goodrich
32. GAP
33. 01 in Corporation
34. Chevron
35. First Chemical
Region V
36. Clark Oil
37. Core-Lube
38. Monsanto
39. National Distillers
(U.S.I.)
40. Northern
Petrochemicals
41. Shell Oil
42. Union Oil
(California)
43. Dow Chemical
44. Dow Chemical
45. Sun Oil
46. Vertac/Transvaal
47. Allied Chemical
48. American Hoechst
49. Cities- Service
50. Continental Oil
51. Cos-Mar, Inc.
52. Dow Chemical
Location
Birmingham, AL
Tuscaloosa, AL
Ashland, KY
Calvert City, KY
Calvert City, KY
Brandenburg, KY
Pascagoula, MS
Pascagoula, MS
Blue Island, IL
Danville, IL
Sauget, IL
Tuscola, IL
Morris, IL
Wood River, IL
Lemont, IL
Bay City, MI
Midland, MI
Toledo, OH
Jacksonville, AR
Geismar, LA
Baton Rouge, LA
Lake Charles, LA
Lake Charles, LA
Carrville, LA
Plaquemine, LA
Coordi
Longitude
86°47'30"
87°28I21"
82°36'32"
88°19'51"
88°24'48"
86°07'15"
88°28'37"
88°29'45"
87°42'07"
87°32'30"
90010'11"
88°21'00"
88°25'42"
90004'24"
88°00'10"
89°52I22"
84°12'18"
83°31'40"
92°04'56"
91°03'12"
91°12'40"
93°19'01"
93°16'35"
91°04'09"
91014'30"
nates
Latitude
33°35'30"
33°15'06"
38°22I30"
37°03'19"
37°02'51"
38°00'30"
30°19'04"
30°20'57"
41°39'19"
40°07'10"
38°36'06"
39°47'53"
41°21'28"
38°50'26"
41°40'20"
43°37'21"
43°35'42"
41°36'52"
34°55'36"
30°12'55"
30033'03"
30°10'58"
30°14'30"
30°14'16"
30°19'50"
Model Plant
Type3
C/T
C/T
LP
C/T
C/T
C/T
SP
C/T
C/T
C/T
C/T
C/T
C/T
LP
SP
LP.C/T
C/T
LP
C/T
C/T
C/T
SP
C/T
C/T
LP
(continued)
B-5
-------
Table B-l. Continued
Coordinates
Plant
Region VI (continued)
53. Exxon
54. Gulf Coast Olefins
55. Gulf Oil
56. Gulf Oil
57. Pennzoil United
(Atlas Processing)
58. Rubicon
59. Shell Oil
60. Tenneco
61. Union Carbide
62. Sun Oil
63. Amerada Hess
64. American Hoechst
65. American Petrofina
of Texas
66. American Petrofina
(Cosden Oil)
67. American Petrofina/
Union Oil of
California
68. Atlantic Richfield
69. Atlantic Richfield
(ARCO/Polymers)
70. Atlantic Richfield
(ARCO/Polymers)
71. Celanese
72. Charter
International
73. Coastal States Gas
74. Corpus Christi
Petrochemicals
75. Cosden Oil
Location
Baton Rouge, LA
Taft, LA
Alliance, LA
Donaldsonville, LA
Shreveport, LA
Geismar, LA
Norco, LA
Chalmette, LA
Taft, LA
Tulsa, OK
Houston, TX
Bayport, TX
Port Arthur, TX
Big Spring, TX
Beaumont, TX
Channel view, TX
Houston, TX
Port Arthur, TX
Pampa, TX
Houston, TX
Corpus Christi , TX
Corpus Christi, TX
Groves, TX
(continued)
B-6
Longitude
91°10'17"
90°26'23"
89°58'26"
90°55'19"
93°46'13"
91°00'37"
90°27'35"
89°58'19"
90°27'15"
96°01'15"
95°14'15"
95°01'15"
93°53'20"
101024'55"
93°58'45"
95°07'30"
95°13'54"
93°58'15"
100°57'47"
95°15'09"
97°26'44M
97°31'2r'
93°52'58"
Latitude
30°29'14"
29°59'16"
29°41'00"
30°05'44"
32°28'12"
30°11'06"
29059.42"
29055'56"
29059'17"
36°08'25"
29°41'39"
29°36'10"
29°57'30"
32°16'11"
30°00'00"
29°50'00"
29°43'10"
29051-24"
35032'07"
29°40'17"
27°48'42"
27°50'02"
29°57'46"
Model Plant
Type3
LP
C/T
LP
C/T
LP
C/T
C/T
SP
LP
C/T,SP
C/T
C/T
SP
LP.C/T
SP,C/T
LP
LP
C/T
C/T
SP
LP.C/T
SP,C/T
C/T
-------
Table B-l. Continued
Coordinates
Plant
Region VI (continued)
76. Crown Central
77. Dow Chemical (A)
78. Dow Chemical (B)
79. Dow Chemical
80. DuPont
81. DuPont
82. Eastman Kodak
83. El Paso Natural Gas
84. El Paso Products/
(Rexene Polyolefins)
85. Exxon
86. GATX Terminal Group
87. Georgia-Pacific Corp.
88. Goodyear Tire and
Rubber
89. Gulf Oil Chemicals
90. Gulf Oil Chemicals
91. Hercules
92. Howell
93. Independent Refining
Corp.
94. Kerr-McGee Corp.
(Southwestern)
95. Marathon Oil
96. Mobil Oil
97. Monsanto
98. Monsanto
99. Oxirane
Location
Pasadena, TX
Freeport, TX
Freeport, TX
Orange, TX
Beaumont, TX
Orange, TX
Longview, TX
Odessa, TX
Odessa, TX
Baytown, TX
Houston, TX
Houston, TX
Bayport, TX
Cedar Bayou, TX
Port Arthur, TX
McGregor, TX
San Antonio, TX
Winnie, TX
Corpus Christi, TX
Texas City, TX
Beaumont, TX
Alvin (Choco-
late Bayou)
Texas City, TX
Channel view, TX
Longitude
95°10'30"
95°19'55"
95°24'09"
93°45'14"
94°01'40"
93044 '44"
94°41 '24"
102°19'29"
102°20'00"
95°01'04"
95013'29M
95°03'00"
95°02'44"
94°55'10"
93°58'30"
97016'30"
98°27'36"
94°20'28"
97°25'24"
94054' 47"
94003'30"
95°12'44"
94°53'40"
95°06'29"
Latitude
29044'40"
28°57'23"
28059'17"
30°03'20"
SOW 51"
30°03'24"
32°26'17"
31°49'27"
31°49'22"
29044.50-
29°43'17"
29°37'20"
29039 .43,,
29°49'29"
29°51'30"
31030'15"
29°20'51"
29050'04"
27048'16"
29°22'2r'
30°04'00"
29°15'09"
29°22'44"
29°50'00"
Model Plant
Type3
SP
LP,C/T
LP.C/T
C/T
C/T
C/T
C/T
C/T
C/T
LP.C/T
C/T
C/T
C/T
C/T
LP,C/T
C/T
SP
SP
SP
SP.C/T
LP.C/T
LP,C/T
LP.C/T
C/T
(continued)
B-7
-------
Table B-l. Continued
Plant
Region VI (concluded)
100. Petrounited Terminal
Services
101. Phillips Petroleum
102. Phillips Petroleum
103. Phillips Petroleum
104. Quintana-Howell
105. Shell Chemical
106. Shell Oil
107. Shell Oil
108. Standard Oil
(Indiana)
109. Standard Oil
(Indiana)/Amoco
110. Sun Oil
111. Texaco
112. Texaco/Jefferson
Chemical
113. Union Carbide
114. Union Carbide
115. USS Chemicals
Region VII
116. Chemplex
117. Getty Oil
118. Monsanto
Region IX
119. Atlantic Richfield
120. Chevron
121. Specialty Organics
Location
Houston, TX
Borger, TX
Pasadena, TX
Sweeny, TX
Corpus Christi, TX
Houston, TX
Deer Park, TX
Odessa, TX
Alvin, TX
Texas City, TX
Corpus Christi, TX
Port Arthur, TX
Port Neches, TX
Seadrift, TX
Texas City, TX
Houston, TX
Clinton, 10
El Dorado, KA
St. Louis, MO
Wilmington, CA
Richmond, CA
Irwindale, CA
Coordi
Longitude
95°01'23"
101°22'05"
95°10I53"
95°45'10"
97°27'30"
95°01'45M
95°07'33"
102°19'20"
95°11'55"
94055'45"
97031'38"
93°54'43"
93°56'00"
96°45'59"
94056'33"
95°15'06"
96°17'29"
96052'00"
90°12'00"
118°14'30
122°23'36"
117°55'56M
nates
Latitude
29°33'51"
35°42'05"
29°43'59"
29°04'24"
27°48'30"
29°38'15"
29°42'55"
31°49'05"
29°13'06"
29°21'58"
27049.57..
29°52'00"
29°57'50
28°30'38"
29°22'27"
29°42'18"
41°48'24"
37°47'10"
38°35'00"
33°48'49"
37°56'12"
34006'18"
Model Plant
Type3
C/T
SP
C/T
SP,C/T
SP
C/T
LP
SP
C/T
LP,C/T
LP.C/T
LP,C/T
C/T
C/T
C/T
C/T
C/T
SP.C/T
C/T
SP
SP.C/T
C/T
(continued)
B-8
-------
Table B-l. Concluded
Plant
Location
Coordi
Longitude
nates
Latitude
Model Plant
Type3
Region IX (continued)
122.
123.
124.
125.
126.
Standard Oil of
California (Chevron
Chemical)
Union Carbide
Witco Chemical
Montrose Chemical
Stauffer Chemical
El Segundo, CA
Torrance, CA
Carson, CA
Henderson, NV
Henderson, NV
118°24'41"
118°20'50"
118°14'13"
115°00'40"
115°00'40"
33°54'39"
33°51'11"
33°49'18"
36°02'28"
36°02'28"
SP.C/T
C/T
C/T
C/T
C/T
C/T represents a benzene consumer or bulk storage terminal; LP represents a large producer of
benzene; SP represents a small producer of benzene.
B-9
-------
Table B-2. MODEL INPUTS FOR EACH TYPE OF MODEL PLANT
Tank dimensions
Type of model
plant and tank number
Benzene Producer: Large
Facility (throughput of
224.6 x 106 liters/year)
1
2
3
4
5
6
7
Benzene Producer: Small
Facility (throughput of
46.3 x 106 liters/yr)
1
2
3
4
Benzene Consumer or
Bulk Storage Terminal
1
2
Vertical
cross-sectional
Height area
(m) (m2)
9
12
5
9
13
9
15
11
13
11
7
11
15
108
216
40
81
169
216
405
33
169
88
224
132
270
Baseline
Roof
type
ncIFR
EFRps
cIFRps
cIFRps
ncIFR
ncIFR
ncIFR
FR
ncIFR
ncIFR
cIFRps
ncIFR
cIFRps
Emissions
(kg/yr)
720
2,190
480
590
680
1,360
1,820
1,270
680
500
2,170
640
970
FR - Fixed-roof tank, IFR - internal floating-roof tank, ERF - external
floating-roof tank, c - contact roof, nc - noncontact roof, ps - primary seal,
ss - secondary seal, 1m - liquid-mounted seal.
B-10
-------
each tank for the baseline (current level) level of control. Emissions
from all the tanks were assumed to be at ambient temperature, which the
model assigns as 293°Kelvin. Because the gas exit velocity is negligible,
it was assumed to be 0 m/s. The model was run in the rural mode. More
information on the development of model plants and emission rates can be
found in Chapter 2 of this document.
B.3 POPULATION AROUND PLANTS CONTAINING BENZENE STORAGE TANKS
The HEM was used to estimate the population that resides in the
vicinity of each receptor coordinate surrounding each plant containing
benzene storage tanks. 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 1978 estimates of the 1970 to 1980 growth factor at
the county level, has been used to estimate 1980 population figures for
each ED/BG. The population "at risk" to benzene exposure was considered
to be persons residing within 20 km of plants containing benzene storage
tanks. The population around each plant was identified by specifying
the geographical coordinates of that plant. The geographical coordinates
are shown for each plant in Table B-l.
B.4 POPULATION EXPOSURE METHODOLOGY
B.4.1 Exposure Methodology
The HEM uses benzene atmospheric concentration patterns (see
Section B.2) together with population information (see Section B.3) to
calculate population exposure. For each receptor coordinate, the
concentration of benzene 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
B-ll
-------
computation of exposure. The two-level approach is used because the
concentrations 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 generally 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.
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 exposures, see Reference 1.)
B-12
-------
B.4.2 Total Exposure
Total exposure (persons-pg/m3) is the sum of the products of
concentration and population, computed as illustrated by the following
equation:
N
Total exposure = I (P-C-) (1)
i=l n n
where
P. = population associated with point i,
C. = 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. This methodology and the derivation of
maximum lifetime risk are described in the following sections.
B.5 LEUKEMIA INCIDENCE AND MAXIMUM LIFETIME RISK
B.5.1 Unit Risk Factor
_Q
The unit risk factor (URF) for benzene is 9.9 x 10 (cases per
year)/ (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 the assumptions made by EPA
(Carcinogen Assessment Group [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
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 adjustment of age, and underestimated the exposure
duration; and (3) included the Ott et al. study in the analysis despite
a lack of statistical significance.
B-13
-------
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/ppm to a probability of leukemia of
0.022/ppm.
B.5.2 Annual Leukemia Incidence
Annual leukemia incidence (the number of leukemia cases per year)
associated with a given plant is the product of the total exposure
around that plant (in persons - ug/m3) and the unit risk factor,
9.9 x 10"8. Thus,
Cases per year = (total exposure) x (unit risk factor), (2)
where total exposure is calculated according to Equation 1 and the unit
risk factor equals 9.9 x 10"8.
B.5.3 Maximum Lifetime Risk
The populations in areas surrounding plants containing benzene
storage tanks have various risk levels of contracting leukemia from
exposure to benzene emissions. Using the maximum annual average concen-
tration of benzene to which any person is exposed, it is possible to
calculate the maximum lifetime risk of leukemia (lifetime probability of
leukemia to any person exposed to the highest concentration of benzene)
attributable to benzene emissions using the following equation:
B-14
-------
Maximum lifetime risk = C. x (URF) x 70 years (3)
I j Hid A
where
C. max = the maximum concentration among all plants at any receptor
location where exposed persons reside,
URF = the unit risk factor, 9.9 x 10"8, and
70 years = the average individual's life span.
B.5.4 Example Calculations
The following calculations illustrate how annual leukemia incidence
and maximum lifetime risk were calculated for specific plants listed in
Table B-l. Table B-3 presents the maximum annual average concentration
and the total exposure for each plant under the baseline (current level)
control level.
B.5.4.1 Annual Leukemia Incidence. As an example for calculating
annual leukemia incidence the Gulf Oil plant in Philadelphia, Pennsylvania,
is used. As shown in Table B-3, the total exposure under the current
(baseline) level of emission control is 3.30 x 10 persons-ug/nv*.
Therefore, under the baseline, the cases per year are computed according
to Equation 2 as follows:
Cases per year = 3.30 x 104 x 9.9 x 10"8
Cases per year = 0.003
B.5.4.2 Maximum Lifetime Risk. Plant numbers 73 (Coastal States
and Gas) and 117 (Sun Oil) had the highest maximum annual average benzene
concentration of 5.22 ug/nr1. Using this maximum concentration and
Equation 3, maximum lifetime risk under the current (baseline) level of
control is calculated as follows:
Maximum lifetime risk = 5.22 x 9.9 x 10"8 x 70
Maximum lifetime risk = 3.62 x 10
B.5.5 Summary of Impacts
Table B-4 summarizes the estimated nationwide impacts for the
baseline (current level) level of emission control. The nationwide
annual leukemia incidence was calculated by summing the total exposure
over all the plants and multiplying by the unit risk factor. The maximum
lifetime risk was calculated as shonw in Section B.5.4.2.
B-15
-------
Table B-3. ESTIMATED MAXIMUM CONCENTRATION
AND EXPOSURE FOR BENZENE STORAGE TANKS
Plant
number
Region II
1
2
3
4
5
6
7
8
9
10
11
12
Region III
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
Baseline
Maximum annual
average benzene
concentration
(ug/m3)
4.45X10"1
2.50X10"1
2.60x10°
l.OOxlO"1
3.03x10°
3.19x10°
7.44X10'1
b
b
b
b
b
1.77x10°
2.50X10"1
8.67xlO"3
2. 50x10" l
2.50X10"1
4. 50x10" l
3.03x10°
1.77x10°
1.77x10°
4.50X10"1
l.SOxlO"2
2.50X10"1
9.62xlO"3
S.OOxlO"1
4.87X10"1
(continued)
B-16
Total exposure
(person pg/m3)
5.26xl03
2.27xl03
3.05xl04
1.44xl04
2.32xl04
3.95xl03
7.93xl02
b
b
b
b
b
1.21xl03
3.20xl02
S.SOxlO1
1.31xl02
5.20xl02
3.75xl03
3.30xl04
1. 21xl03
4.50xl03
2.32xl03
5.39xl02
1.21X102
7.24X101
1.03xl02
1.07xl03
-------
Table B-3. Continued
Plant
number
Region IV
28
29
30
31
32
33
34
35
Region V
36
37
38
39
40
41
42
43
44
45
Region VI
46
47
48
49
50
51
52
Baseline
Maximum annual
average benzene
concentration
(ug/m3)
3.47X10"1
l.OOxlO"1
1.00x10°
1.04xlO"2
1.92xlO"2
1.53xlO"2
9. 78x10" 3
S.OlxlO"1
3.70X10"1
2.50X10"1
4.11X10"1
6.05xlO"3
3.70X10"1
2.51x10°
1.64x10°
1. 00x10°
3. 88x10" l
2.87x10°
l.OOxlO"1
2.50X10"1
2.50X10"1
5. 00x10" l
2.50X10"1
2.50X10"1
2.66x10°
(continued)
B-17
Total exposure
(person ug/m3)
1.70xl03
4.04xl02
2.57xl03
6.45X101
1.19xl02
1.41xl02
3.69xl02
3.77xl02
5.30xl03
1.96xl02
2.67xl03
4.75X101
4.91xl02
3.42xl03
3.43xl03
2.51xl02
4.81xl02
1.22xl04
1.63xl02
1.42xl02
6.57xl02
8.88xl02
3.65xl02
1.59.X102
1.49xl03
-------
Table B-3. Continued
Plant
number
Region VI
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
Baseline
Maximum annual
average benzene
concentration
(|jg/m3)
(continued)
1.00x10°
5.75X10"1
2.61xlO"2
2.50X10"1
3.72x10°
2.50X10"1
5.00X10"1
1.00x10°
2.50x10°
3.19x10°
4.39X10"1
8.13X10"1
1.00x10°
2.50x10°
1.00x10°
5.19x10°
1.00x10°
2. 50x10" l
3.25X10"1
2.01x10°
5.22x10°
3.78x10°
l.OOxlO"1
S.OOxlO"1
3.10x10°
1.00x10°
l.OOxlO"1
(continued)
B-18
Total exposure
(person ug/m3)
1.05xl04
3.51xl02
2.44xl02
2.01xl02
1.55xl04
1.35xl02
2.29xl02
1. OSxlO4
1.14xl03
5.29xl03
4.61xl03
7.39xl02
1.60xl03
7.50xl02
2.03xl03
4. 34xl03
1. 95xl04
3.43xl02
4. 55xl02
1.54xl04
4.89xl03
1.24xl03
3.59xl02
7.30xl03
l.OSxlO3
1.21xl03
3.42xl02
-------
Table B-3. Continued
Plant
number
Region VI
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
Baseline
Maximum annual
average benzene
concentration
(fjg/m3)
(continued)
2. 50x10" 1
l.OOxlO"1
l.OOxlO"1
l.OOxlO"1
4.12X10"1
1. 00x10°
4.39X10"1
8. 13x10" 1
2.50X10"1
"2.50X10"1
1.00x10°
2.50X10"1
4.72x10°
1.56xlO"2
3.07x10°
2.26x10°
2.50x10°
1.00x10°
3.10x10°
5.00X10"1
8. 13x10" l
1.00x10°
8.13X10"1
1.00x10°
3.07x10°
8.13X10"1
2.50x10°
(continued)
B-19
Total exposure
(person ug/m3)
3.57xl02
3.42X102
3.40xl02
5.94xl02
7.13xl02
4.70xl03
3.62xl03
7.71xl02
S.OOxlO2
2.07xl02
1.95xl03
1.32xl02
1. 36xl04
8.62X101
5.13X103
2.16xl03
3.24xl03
1.71xl02
3.81xl03
6.24xl02
7.33xl02
2.72xl02
2.60xl03
2.79xl02
1.73xl03
8.31xl02
7.55xl03
-------
Table B-3. Concluded
Plant
number
Region VI
107
108
109
110
111
112
113
114
115
Region VII
116
117
118
Region IX
119
120
121
122
123
124
125
126
Baseline
Maximum annual
average benzene
concentration
(pg/m3)
(concluded)
1.85x10°
3. 36x10" 3
3.10x10°
5.22x10°
2.50x10°
l.OOxlO"1
-T
5.00x10 L
2. 50x10" l
4.39X10"1
6.11xlO"3
S.OOxlO"1
2.50X10"1
2.85x10°
4. 50x10°
5. 61x10" l
3.85x10°
7.33X10"1
5. 61x10" l
2. 50x10" l
2.50X10"1
Total exposure
(person pg/m3)
1.76xl03
9.02x10°
2.66xl03
2.13xl03
3.60xl03
6.22xl02
i
2.47X101
4.88xl02
4.55xl03
1.58X101
6.08xl02
2.82xl03
2.44xl04
1.07xl04
6.29xl03
2.40xl04
6.93xl03
8.47xl03
5.18xl02
5.18xl02
This table lists the maximum annual average benzene
concentration to which at least one person is exposed.
DPopulation estimate is not included in the HEM for this
plant.
B-20
-------
Table B-4. ESTIMATED NATIONWIDE HEALTH IMPACTS
FOR BENZENE STORAGE TANKS
Baseline
Max. Annual Average 5.22
Concentration (ug/m5)
Maximum Lifetime Risk 3.6 x 10"5
Total Exposure. 4.37 x 10^
(persons-|jg/m3)
Incidence (cases/yr) 0.043
B-21
-------
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.6.1 Benzene Concentrations
Modeled ambient benzene concentrations depend upon: (1) plant
configuration, which is difficult to determine for more than a few
plants; (2) emission point characteristics, which can be different from
plant to plant and are difficult to obtain for more than a few plants;
(3) emission rates, which may vary over time and from plant to plant;
and (4) meteorology, which is seldom available for a specific plant.
The particular dispersion modeling used can also influence the numbers.
The dispersion models also assume 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. 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. Assuming the inputs to
the dispersion model are accurate, the predicted benzene concentrations
are considered to be accurate to within a factor of 2. This uncertainty
factor was not included in the calculations in this analysis.
The Industrial Source Complex - Long Term (ISC-LT) dispersion model
is considered to be a more complex and accurate dispersion model than
the dispersion model subprogram of the HEM. However, it is too resource-
intensive for modeling a large number of sources, such as benzene storage
vessels. To evaluate the effect of using the HEM dispersion model, the
ISC-LT was run on the model plants for several geographic sites and the
results were compared with those from the HEM dispersion model. The
results of the analysis can be found in Docket A-80-14, Item IV-B-4.
For three sites (New Orleans, Houston, and Chicago) the maximum and
mean ring concentrations predicted by each model were compared. In all
cases, the ISC-LT resulted in higher estimates than the dispersion model
B-22
-------
of the HEM. For the same three sites and two additional sites (Los Angeles
and Philadelphia), the concentration at each receptor point times the
corresponding area around the receptor point was summed over all receptors
at each plant. (NOTE: Docket Item IV-B-4 calls this sum "total exposure."
The usage in the docket item is different from that defined in Section B.4
of this appendix.) The ISC-LT results in a higher estimate of this sum
(ranging from about 20 to 60 percent) than the HEM dispersion model for
New Orleans, Houston, and Philadelphia. For Chicago and Los Angeles,
the HEM and ISC-LT give very similar results for this sum, within
10 percent of one another.
This analysis shows that the ISC-LT and the HEM dispersion model
may give different results. In many cases, the ISC-LT predicts higher
concentrations than the HEM. However, because of the degree of uncertainty
in the basic data available for the model and in dispersion analysis,
the degree of effort to model all the plants specifically using the more
sophisticated dispersion model (ISC-LT) is not warranted.
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 was 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
considering the uncertainty of dispersion estimates beyond 10 km. The
results of dispersion modeling are felt to be reasonably accurate within
that distance (see above). The uncertainty of these assumptions has not
'been quantified.
B-23
-------
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 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 leukemia incidence were
calculated using the unit risk factor, which is based on a no-threshold
linear extrapolation of leukemia risk and applies to a presumably healthy
white male cohort of workers exposed to benzene concentrations in the
parts per million range. It is uncertain whether the unit risk factor
can be accurately applied to 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 may have
susceptabilities to leukemia that differ from those 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 benzene
storage tanks are not quantified. Possible benzene exposures from other
sources also are not included in the estimate. For example, an individual
living near a benzene storage tank 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-24
-------
B.8 REFERENCES
1. Systems Applications, Inc. Human Exposure to Atmospheric Concentra-
tions of Selected Chemicals. (Prepared for the U.S. Environmental
Protection Agency, Research Triangle Park, North Carolina).
Volume I, Publication Number EPA-2/250-1, and Volume II, Publication
Number EPA-2/250-2. May 1980.
2. Busse, A.D. and J.R. Zimmerman. User's Guide for the Climatological
Dispersion Model. (Prepared for the U.S. Environmental Protection
Agency, Research Triangle Park, North Carolina.) Publication
Number EPA-R4-73-024. December 1973.
3. Albert, R. E. Carcinogen Assessment Group's Final Report on
Population Risk to Ambient Benzene Exposures. U.S. Environmental
Protection Agency. Publication No. EPA-450/5-80-004 Docket
Number A-79-27-II-A-28. January 1979.
B-25
-------
TECHNICAL REPORT DATA
fflease read Instructions on the reverse before completing)
EPA-450/3-84-004
4. TITLE AND SUBTITLE
Benzene Storage Tanks - Background Information for
Proposal to Withdraw Proposed Standards
7. AUTHOR(S)
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Office of Air Quality Planning and Standards
Environmental Protection Agency
Research Triangle Park, North Carolina 27711
12. SPONSORING AGENCY NAME AND ADD
DAA for Air Quality Planning
Office of Air and Radiation
U.S. Environmental Protectior
Research Triangle Park, North
RESS
and Standards
i Agency
i Carolina 27711
3. RECIPIENT'S ACCESSION NO.
5. REPORT DATE
March 1984
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NO.
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
68-02-3063
13- JYPEpOF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
EPA/200/04
15. SUPPLEMtN IAHY NOTES —
It is proposed to withdraw the proposed National Emission Standards for
Hazardous Air Pollutants for the control of Benzene emissions from
Benzene Storage Tanks. Previously, standards had been proposed under
Section 112 of the Clean Air Act. This document contains background
information considered in the proposed withdrawal of those previously
proposed standards.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
COSATI Field/Or
Air pollution
'ol lution control
storage tanks
Floating roof and seal systems
Chemical manufacturing plants
Benzene
Emissions standards for Hazardous
Air Pollution Control
13 B
19 SECURITY CLASS (This Report,
;21. NO. OF PAGES
Unlimited
I 20 SECURITY CLASS /THispaget
i Unclassified
I22 PPICE
ing
EPA Form 2220-1 (Rev. 4-77) PREVIOUS EDITION is OBSOLETE
------- |