EPA-450/3-33-011b
Inorganic Arsenic Emissions from
Glass Manufacturing Plants —
Background Information for
Promulgated Standards
Emission Standards and Engineering Division
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Air and Radiation
Office of Air Quality Planning and Standards
Research Triangle Park, North Carolina 27711
May 1986
-------
-------
ENVIRONMENTAL PROTECTION AGENCY
Background Information
Final Environmental Impact Statement
for Inorganic Arsenic Emissions from Glass Manufacturing Plants
i fi2
Prepared by:
jack R. Farmer
Director, Emission Standards and Engineering Division
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
1. The promulgated standards will limit emissions of inorganic arsenic
from existing and new glass manufacturing plants. The promulgated
standards implement Section 112 of the Clean Air Act and are based on
the Administrator's determination of June 5, 1980 (48 FR 37886) that
arsenic presents a significant risk to human health as a result of air
emissions from one or more stationary:source categories, and is
therefore a hazardous air pollutant.
2. Copies of this document have been sent to the following Federal
Departments: Office of Management and Budget; 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. For additional information contact:
Mr. Robert L. Ajax
Standards Development Branch (MD-13)
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
Telephone: (919) 541-5624
4. Copies of this document may be obtained from:
U.S. EPA Library (MD-35)
Research Triangle Park, North Carolina 27711
National Technical Information Service
5285 Port Royal Road
Springfield, Virginia 22161
m
-------
-------
TABLE OF CONTENTS
Page
Chapter 1 - Summary 1-1
1.1 Summary of Changes Since Proposal 1-1
1.2 Summary of Impacts of Promulgated Action 1-3
Chapter 2 - Summary of Public Comments 2-1
2.1 Affected Facilities 2-1
2.2 Stringency of the Proposed Emission Limit 2-11
2.3 Selection of Control Technology 2-17
2.4 Cost Effectiveness of the Proposed Standard 2-22
2.5 Economic Impact 2-22
2.6 Exemptions and Allowances 2-25
2.7 Monitoring and Measurement Methods 2-29
2.8 Determination of Compliance 2-37
2.9 Reporting Requirements 2-41
2.10 Arsenic Emissions from Soda-Lime Furnaces 2-42
2.11 Zero Production Offsets 2-48
2.12 Risk Assessement and Risk Management 2-49
Appendix A - Summary of Results of Emission Testing
Performed After Proposal A-l
Appendix B - Updated Cost and Economic Analysis B-l
Appendix C - Available Arsenic Emissions Data for
Existing Arsenic-Using Glass Plants C-l
Appendix D - Inorganic Arsenic Risk Assessment for
Glass Manufacturing Plants D-l
-------
-------
1.0 SUMMARY
On July 20, 1983, the Environmental Protection Agency (EPA) proposed
standards for inorganic arsenic (48 FR 33112) pursuant to Section 112 of the
Clean Air Act which requires national emission standards for hazardous air
pollutants (NESHAP). The proposed standard covered the following categories
of sources of inorganic arsenic: high-arsenic primary copper smelters,
low-arsenic primary copper smelters, and glass manufacturing plants. EPA
also identified six additional source categories, but determined that
standards are not warranted at this time.
Public comments were requested on the proposal in the Federal Register.
This document addresses the comments pertaining to glass manufacturing
plants. After proposal, the public comment period for the proposed
standards for glass manufacturing plants was reopened on March 20, 1984
(49 FR 10278). Additional comments were solicited for one month concerning
inorganic arsenic emissions from soda-lime glass manufacturing and
concerning zero production rate offsets.
There were 24 comment letters submitted concerning the standards for
glass manufacturing. Eleven letters were from industry,, seven from state
agencies, four from environmental groups, one from the Office of Management
and Budget, and one from two private individuals. In addition, two industry
and one environmental group spokesman testified about the arsenic NESHAP for
glass manufacturing plants at the public hearing.
The comments submitted, along with responses to these comments, are
summarized in this document. The summary of comments and responses serves
as the basis for the revisions made to the standard between proposal and
promulgation.
1.1 SUMMARY OF CHANGES SINCE PROPOSAL
In response to public comments received on the proposed rulemaking and
as a result of EPA re-evaluation, five major changes were made to the
proposed standard. These changes involve (1) revising the cutoff for
existing furnaces, (2) revising the format of the emission limits,
1-1
-------
(3) allowing the control device to be by-passed for periods of maintenance,
(4) eliminating the exemption to 40 CFR Part 60, Subpart CC for sources that
comply with the NESHAP, and (5) establishing a provision to exempt certain
sources from testing procedures.
Existing Furnace Cutoff
Further examination of risks associated with arsenic emissions from
specific existing glass manufacturing plants led the Agency to change the
regulation by establishing the cutoff on uncontrolled arsenic emissions for
existing glass melting furnaces at 2.5 Mg (2.75 tons) per year. The
proposed cutoff on uncontrolled arsenic emissions of 0.4 Mg (0.44 ton) per
year is retained for new or modified furnaces.
Format of the Standard
The second major change in the regulation since proposal involves a
change in the format of the emission limits. The promulgated standard
requires owners or operators of existing glass furnaces to ensure either
that uncontrolled total arsenic emissions are less than 2.5 Mg (2.75 tons)
per year or that arsenic emissions are reduced by 85 percent.
By-pass of the Control Device
The third major change in the regulation allows glass furnace owners or
operators to petition the Administrator for permission to by-pass the
control device for a limited period for control device maintenance purposes.
After receiving a comment requesting this provision, the Agency investigated
the cost and environmental impacts associated with performing routine
maintenance on emission control devices installed on affected glass
furnaces. EPA concluded that control devices could be by-passed during
maintenance because the cost of requiring a temporary shutdown during
maintenance would be excessive without offsetting environmental benefits and
because the use of well maintained control devices is essential in
effectively controlling arsenic emissions on a continuing basis. However,
the Agency has included provisions to requie plants to minimize arsenic
emissions during maintenance periods and will allow it only upon
demonstration of its necessity.
1-2
-------
Elimination of Exemption from NSPS ' ;
In the proposed standard, participate emission limits were identical to
those in the glass manufacturing NSPS (40 CFR Part 60, Subpart CC) and no
furnace was allowed to operate with uncontrolled arsenic emissions in excess
of 0.4 Mg (0.44 ton) per year. The promulgated standard has been revised
such that the emission limits are no longer identical and the exemption from
the NSPS is no longer appropriate.
Compliance Testing
In the final major change, EPA created a provision to exempt certain
sources from performing emission tests to demonstrate compliance. Existing
sources that add less than 8.0 Mg arsenic per year to the batch and new or
modified sources that add less than 1.0 Mg arsenic per year to the batch
will be exempted from performing emission tests. Analysis has shown that if
an existing or new source adds less arsenic than these amounts to the raw
materials, it would emit less than 2.5 Mg (2.75 tons) or 0.4 Mg (0.44 ton),
respectively. In these cases, the Administrator has concluded that emission
testing is not necessary.
1.2 SUMMARY OF IMPACTS OF PROMULGATED ACTION
1.2.1 Alternatives to Promulgated Action
Regulatory alternatives considered during development of the standard
are discussed in Chapter 4 of "Inorganic Arsenic Emissions from Glass
Manufacturing Plants - Background Information for Proposed Standards",
EPA-450/3-83-011a, April 1983. This document is also referred to as the
Background Information Document (BID). These regulatory alternatives
reflect the different levels of emission control. After proposal, the
Agency revised the emission limits as discussed in Section 1.1. These
revisions made it necessary to recalculate1 environmental, energy, and
economic impacts because fewer furnaces are affected by the revised emission
limits.
i
1.2.2 Environmental Impacts of Promulgated Action
Under the promulgated standard, eight existing glass melting furnaces
are subject to the control requirement and are required 10 reduce emissions
1-3
-------
of arsenic by 85 percent. Six of these furnaces are presently equipped with
controls and will not be required to install any additional controls as a
result of this standard. The reduction in arsenic emissions currently
achieved on these six furnaces totals 30.4 Mg (33.4 tons) per year. The
remaining two furnaces are presently uncontrolled and are required to
install control devices capable of reducing arsenic emissions by at least
85 percent. However, one of these furnaces is expected to change production
to a non-arsenic containing glass type while the other will install a
control device. The total reduction in arsenic emissions expected to be
achieved from these two furnaces is 14.6 Mg (16.1 tons) per year.
Because only one furnace is anticipated to install controls as a result
of this standard, the total water and solid waste impact associated with the
standard are those arising from this furnace. Application of controls to
this furnace will not result in any emissions to water. The solid waste
collected in the control device would amount to about 20 Mg (22 tons) per
year. However, because reuse of this waste would result in cost savings for
raw materials and because of the high costs of disposing of this waste, it
is anticipated that- about 90 percent of the total waste will be recycled to
the melting furnace. Therefore, the total solid waste impact of the final
standard is estimated to amount to about 2 Mg (2.2 tons) per year.
The analysis of environmental impacts presented in Volume I of this
document, as modified by the changes described above, is the final
environmental impact statement for this action.
1.2.3 Energy and Economic Impacts of Promulgated Action
Revisions in the emission limits since proposal have resulted in
changes in the energy and economic impacts presented in the BID for the
proposed standard.
Because only one presently uncontrolled furnace is anticipated to
install controls as a result of the promulgated standard, the total energy
impact of the standard is that associated with controlling this furnace. It
is estimated that the application of controls will increase energy use at
the plant by about 185 MW-hr per year. This represents about 0.1 percent of
the total energy use at the plant.
1-4
-------
The revised economic impacts are summarized and compared to the impacts
of the proposed standard in Appendix B of this BID. Although at proposal
information was available on the financial characteristics of the affected
glass companies, sufficient data were not available to estimate the cost
impacts of controlling all arsenic-using glass furnaces operated by those
companies. Detailed information gathered after proposal was used to
reassess economic impacts of control options. The revised analysis in
Appendix B indicates moderate profit declines for most potentially affected
furnaces with severe profit declines for a few furnaces. However, no
adverse economic impacts are estimated for the single plant anticipated to
install controls as a result of this action. There was little impact on
capital availability.
1.2.4 Other Considerations :
1.2.4.1 Irreversible and Irretrievable Committment of Resources.
Other than the fuels required for power generation and the materials
required for the construction of the control system, there is no apparent
irreversible or irretrievable commitment of resources associated with this
regulation.
1.2.4.2 Environmental and Energy Impacts of Delayed Standards.
Delay in implementing the standard would result in a delay in emission
reduction of arsenic from affected furnaces and delay in realization of
other estimated impacts.
1.2.4.3 Urban and Community Impacts.
Neither plant closures nor impacts on small business are forecasted.
The issue of plant closure is discussed in more detail in Section 2.5.
1-5
-------
-------
2.0 SUMMARY OF PUBLIC COMMENTS
A total of 24 letters commenting on the proposed standard for the glass
industry were received. In addition, three speakers commented on the
proposed standard at the public hearing. The transcript from the public
hearing, comments made at the public hearing, and the 24 letters have been
recorded and placed in the docket. The list of commenters, their
affiliation, and the docket number assigned to each comment by the
Environmental Protection Agency (EPA) are shown in Table 2-1. The docket
reference number is indicated in parentheses in each comment summary. All
docket references are part of Docket Number A-83-08, Category IV. The
comments have been summarized according to the following 11 categories:
2.1 Affected Facility
2.2 Stringency of the Proposed Emission Limit
2.3 Selection of Control Technology
2.4 Cost Effectiveness of the Proposed Standard
2.5 Economic Impact
2.6 Exemptions and Allowances
2.7 Monitoring and Measurement Methods
2.8 Determination of Compliance '
2.9 Reporting Requirements ;
2.10 Arsenic Emissions from Soda-lime Furnaces
2.11 Zero Production Offsets
2.12 Risk Assessment and Risk Management
2.1 AFFECTED FACILITIES
j
2.1.1 Comment: Three commenters (F-l, D-5, D-10) addressed the issue
of whether the presence of arsenic as an impurity in the raw materials used
to manufacture glass should be considered in determining the applicability
of the standard. Each of these commenters expressed concern that it would
be burdensome and costly to require facilities which do not use arsenic as
2-1
-------
TABLE 2-1. LIST OF COMMENTERS ON THE PROPOSED STANDARDS FOR ARSENIC
EMISSIONS FROM GLASS MANUFACTURING PLANTS
Docket Number A-83-08, IV
Public Hearing
Commenter
Dr. Robert A. Drake
Vice President, Technical
Glass Packaging Institute
6845 Elm St., Suite 209
McLean, Virginia 22101
Mr. H. Neal Troy
Manager of Environmental Control
Owens-Illinois, Incorporated
One Seagate
Toledo, Ohio 43666
Mr. David D. Doniger
Senior Staff Attorney
Natural Resources Defense Council, Inc.
1725 I St., N.W. '
Suite 600
Washington, D.C. 20006
Letters
Mr. Thomas J. Koralewski
Senior Environmental Engineer
Libbey-Owens-Ford Company
1701 East Broadway
Toledo, Ohio 43605
Dr. George and Adriana Hess
4437 West Grandview Place
Tacoma, Washington 96466
Mr. Ben A. Brodovicz, Chief
Division of Technical Services and Monitoring
Bureau of Air Quality Control
Department of Environmental Resources
Post Office Box 2063
Harrisburg, Pennsylvania 17120
Docket Reference
F-l
F-2
F-3
D-l
D-2
D-3
2-2
-------
TABLE 2-1 CONTINUED
Commenter
Mr. John T. Barr
Regulatory Response
Air Products and Chemical, Inc.
Box 538
Allentown, Pennsylvania 18105
Mr. F. P. Partee !
Principal Staff Engineer ;
Stationary Source Environmental Control Office
Ford Motor Company
One Parklane Blvd.
Dearborn, Michigan 48126 :
Mr. Paul F. Munn
Chief Engineer
Department of Public Utilities
Environmental Services Agency
26 Main Street
Toledo, Ohio 43605
Mr. Bill Stewart
Executive Director
Texas Air Control Board
6330 Hwy 290 East
Austin, Texas 78723
Anita Fries
A-95 Coordinator
State Clearinghouse
30 East Broad Street
Colombus, Ohio 43215 :
Mr. Michael Gregory
Grand Canyon Chapter, Arizona
Sierra Club
Rt. 1, Box 25A
McNeal, Arizona 85617 '
Mr. Lowell D. Turnbull
Leva, Hayes, Symington & Martin
8l5 Connecticut Ave., N.W.
Washington, D.C. 20006 ;
Executive Office of the President
Office of Management and Budget
Washington, D.C. 20503
Docket Reference
D-4
D-5
D-6
D-7
D-8
D-9
D-10
D-ll
2-3
-------
TABLE 2-1 CONTINUED
Commenter
Mr. David F. Zoll
Vice President and General Counsel
Chemical Manufacturers Association
2501 M Street, N.W.
Washington, D.C. 20037
Mr. John L. Cherill
Energy and Environmental Control
Corning Glass Works
Corning, New York 14830
Mr. Richard A. Kamp
Smelter Crisis Eduction Project
Post Office Box 5
Naco, Arizona 85620
Mr. H. M. Howe
Chief Siting Engineer
Pacific Gas and Electric Company
77 Beale Street
San Francisco, California 64160
Robert Abrams
Attorney General of the State of New York
Mary Lyndon
Assistant Attorney General of the State of New York
Two World Trade Center
New York, New York 10047
David D. Doniger
Senior Staff Attorney
Natural Resources Defense Council
1725 I Street, N.W.
Suite 600
Washington, D.C. 20006
Mr. H. Neal Troy
Manager of Environmental Control
Owens-Illinois, Incorporated
One Seagate
Toledo, Ohio 43666
David D. Doniger
Senior Staff Attorney
Natural Resources Defense Council
1725 I Street, N.W.
Suite 600
Washington, D.C. 20006
2-4
Docket Reference
D-12
D-13
D-14
D-15
D-16
D-17
D-18
D-23
D-24
-------
TABLE 2-1 CONTINUED
Commenter
Mr. John L. Cherill
Energy and Environmental Control
Corning Glass Works
Corning, N.Y. 14830
Steven 6. Kuhrtz
Director
New Jersey Department of Environmental Protection
Division of Environmental Quality
Trenton, N.J. 08625
Mr. Herbert Engel
Decor, Incorporated
60 Cedar Lane
Englewood, N.J. 07631
Mr. John L. Cherill
Energy and Environmental Control
Corning Glass Works
Corning, N.Y. 14830
Docket Reference
0-25
D-26
D-27
D-28
Docket entries D-19, D-20, D-21, and D-22 are not included in this list
because they did not include comments pertaining to the proposed standard.
1-5
-------
a raw material to demonstrate that emissions arising from trace arsenic
contamination of other raw materials would not result in exceedance of the
proposed emissions cutoff. The commenters requested that EPA explicitly
exclude from the promulgated regulation all furnaces that do not
intentionally use arsenic as a raw material. One commenter (D-10) noted
that the arsenic content of raw materials is not routinely specified by
purchasers since arsenic is not known to impair glass quality. However, the
commenter indicated that in a telephone survey of glass manufacturers and
raw material suppliers, no evidence was found that arsenic exists in
significant quantities as an impurity of raw material components. The only
detectable quantity of arsenic was found in Green River soda-ash at
concentrations ranging from 0.03 to 0.05 parts per million (ppm). These
concentrations would result in maximum uncontrolled arsenic emissions of
about 1.1 kg (2.5 pounds) per year from a typical 227 Mg/day (250 tons/day)
glass container furnace. This commenter concluded that EPA should give no
consideration to the arsenic content of raw materials since there is no
reason to believe that the arsenic content of raw materials used for glass
manufacture is any higher than it is in raw materials used in other process
industries. Another commenter (D-5) pointed out that for the size of
furnace typically used to produce flat glass (500 to 600 tons/day), trace
amounts of arsenic in the raw materials on the order of 2 to 3 ppm by weight
could result in uncontrolled arsenic emissions approaching the proposed
emission cutoff of 0.4 Mg/year (0.44 ton/year). The commenter is aware of
only one conventional raw material that contains arsenic as an impurity.
That one exception, an additive used in small amounts in producing
body-colored glass, would result in arsenic emissions of less than one pound
per year.
2.1.1 Response: EPA has examined the problems posed by the presence
of arsenic as an impurity in various raw materials used in the production of
glass, and has concluded, based on available information, that this source
of arsenic is not expected to affect significantly the emissions of arsenic
from glass manufacturing furnaces. The specific comment which appears to
indicate that the presence of arsenic impurities may result in emissions
2-6
-------
approaching 0.4 Mg (0.44 ton) per year was closely examined. It was
determined that the calculations present an unrealistic situation in
presuming that all of the raw materials entering the furnace contain 2 to 3
ppm arsenic by weight, and that all of the arsenic entering the furnace is
emitted. Because it would be uncommon for all raw materials to contain
arsenic at that level, and because most of the arsenic is known to be
retained in the product, EPA has concluded that the emissions calculated in
the example given in the comment are substantially overstated and not
indicative of an actual condition that might occur.
EPA has also independently investigated the concentration of arsenic
found in the bulk raw materials commQnly used in the glass industry.
(Docket reference IV-B-12). During an emissions test of an arsenic-using
furnace, samples of the bulk raw materials were taken and analyzed for
arsenic content. With the single exception of barium carbonate, the
concentrations of arsenic in the raw materials from this plant were below
the detection limits of the analytical method used. The measured
concentration of arsenic in the barium carbonate sample was 2.32 ppm.
However, barium carbonate is not widely used in large quantities within :he
glass industry. Even assuming that the concentration of arsenic in bulk raw
materials is equal to the detection limit of the analytical methods used on
the test samples, the maximum uncontrolled emissions of arsenic arising from
raw material impurities would be about 0.19 Mg/year (0.21 ton/year) from a
furnace producing 500 Mg/day (550 tons/day) of glass.
Based on all of the information available to the Agency, glass melting
furnaces which do not use commercial arsenic as an ingredient of their batch
composition would not emit enough arsenic to be affected by the promulgated
cutoff of 0.4 Mg (0.44 ton) per year for new and modified furnaces and
2.5 Mg (2.75 tons) per year for existing furnaces. EPA agrees it would be
unreasonable to require demonstration of this and, therefore, the
applicability section of the regulation has been revised to exclude from the
promulgated standard all furnaces which do not use commercial arsenic as a
raw material. Commercial arsenic is any variety of arsenic or arsenic
compounds which is produced by extracting arsenic from minerals.
2-7
-------
2.1.2 Comment: Two commenters (D-13; F-2) stated that the proposed
regulation should be applicable only to furnaces that are at present
uncontrolled. One commenter (D-13) added that all existing glass melting
furnaces equipped with fabric filters or electrostatic precipitators should
be excluded from further regulation under this NESHAP. The commenter stated
that although uncontrolled furnaces are relatively few in number, the
largest reductions in arsenic emissions could be obtained from these
furnaces.
2.1.2 Response: Under the promulgated cutoff on arsenic emissions for
existing sources, six currently controlled furnaces would be affected by the
emission limits. However, all six of the furnaces are at present achieving
the required reduction in arsenic emissions and no additional controls are
required on these furnaces. The Agency wants to ensure that existing
control devices will be properly operated and maintained because the health
risks associated with emissions above the cutoff could be significantly
reduced at reasonable cost. Therefore, it would not be in keeping with
Section 112 of the Clean Air Act to exempt existing furnaces and to fail to
ensure that control systems on these furnaces remain effective in reducing
emissions.
2.1.3 Comment: One commenter (D-26) stated that EPA has relied
totally on cost and economic factors in establishing 0.4 Mg (0.44 ton) per
year as the cut-off between affected and unaffected furnaces in the proposed
standard, and that no analysis of the impact of this cutoff on exposure or
risk has been presented. The commenter urged that the data on exposure and
risk should be presented so that the impact of the cutoff can be fully
evaluated.
Another commenter (F-2) stated that the proposed emission cutoff of 0.4
Mg (0.44 ton) arsenic per year is overly stringent given the insignificant
environmental benefits that would be gained from controlling emissions at
this level. According to the commenter1s calculations, the emissions from
one of the commenter's plants, which emits more than the proposed cutoff,
would contribute only 0.17 micrograms/cubic meter to the ambient
concentration over an 8 hour period. This is only 2 percent of the OSHA
2-8
-------
standard of 10 micrograms/cubic meter. The commenter also referred to the
previous testimony of Dr. Lamm (November 8, 1983 public hearing transcript,
Docket No. A-83-08-IV-F), that ambient concentrations of 300
micrograms/cubic meter over a lifetime would be necessary to cause
significant health impacts. If the furnace were controlled to the proposed
emission limit of 0.2 Ib particulate/ton of glass produced, the
corresponding reduction in the ambient concentration of arsenic would be
0.13 micrograms/cubic meter. The commenter concluded that such a reduction
is not justified given the costs required to achieve it and that the
proposed emission cutoff should be increased by a factor of two or three and
apply only to uncontrolled furnaces. i
2.1.3 Response: At the time of proposal, the Agency's standard
setting approach involved first selecting a standard that was achievable
through the application of best available.technology (BAT). Determination
of BAT was based on the capability of existing technologies to reduce
emissions, as well as on the costs of emission controls and on the economic
impact of applying the controls at specific facilities. The residual risks
remaining after application of BAT to furnaces which would have been
affected by the proposed cutoff (0.4 Mg/year of arsenic prior to control)
were then considered to determine if a more stringent standard would be
necessary to protect public health. However, Agency policy has since
evolved to place greater emphasis on risk and risk reduction in determining
which specific sources within a source category shall be subject to an
emission limit under Section 112. Costs and economic impact are still
considered in relation to the reductions in risk achievable through the use
of selected control technologies.
Because of various site-specific factors, the degree of risk associated
with emissions of inorganic arsenic from glass manufacturing plants does
not, in all cases, directly correlate with the absolute magnitude of those
emissions. Moreover, risks to the population in the vicinity of a plant
must be assessed in terms of emissions of arsenic from the entire plant,
rather than emissions from individual furnaces within the plant. Therefore,
in establishing an emission cutoff, the emphasis has shifted from
2-9
-------
consideration of the magnitude of the emissions arising from individual
furnaces and the costs of controlling those emissions, to consideration of
the magnitude of the risks associated with specific plants and the degree to
which those risks can be reduced at a reasonable cost.
The Agency estimated the risks associated with arsenic emissions from
existing glass manufacturing plants and determined that most of the
emissions and risks were associated with 11 uncontrolled furnaces emitting
more than 0.4 Mg per year each. These furnaces are located at five
different plants. A sixth plant, having nine furnaces, all of which emitted
arsenic at a level below the proposed cutoff of 0.4 Mg per year, also had
relatively high aggregate risks. Because of the proposed cutoff of
0.4 Mg/yr, however, none of these furnaces would have been subject to the
proposed control requirements. Nevertheless, EPA considered whether the
cutoff should be changed to require control of the emissions from these nine
furnaces by estimating and analyzing the emission and risk reductions
achievable and the associated costs and economic impacts. The analysis
indicated that the cost of controlling emissions would be disproportionately
high at this plant compared to the reduction of risk achievable.
Of the other five plants, three posed relatively high risks. The
analysis indicated that the reduction in risk achievable from the first two
plants is about three to four times greater than the reduction in risk that
could be achieved from the third plant. Also, the costs of controlling
emissions from the third plant are disproportionately high when compared to
the reduction in risk which would be gained. For these reasons, the
promulgated standard establishes an emission cutoff for existing glass
melting furnaces that would require add-on control of emissions for only the
first two plants. Emissions from both of these plants arise from arsenic-
using glass furnaces which individually emit more than 2.5 Mg/yr (2.75
tons/year) of uncontrolled arsenic. Therefore, the cutoff on uncontrolled
emissions of inorganic arsenic for existing glass melting furnaces is
established at 2.5 Mg/yr (2.75 tons/yr). A more detailed discussion of the
information on which this decision is based is included in Appendix D.
Agency estimates indicate that there will be eight existing furnaces subject
to the NESHAP.
2-10
-------
The Agency does not evaluate risks to the general public based on OSHA
occupational standards. The OSHA standards are set based on exposure to
healthy workers during the 40 hour work week. The general public includes
individuals at a greater risk than healthy workers (e.g., children, the
elderly) and may be exposed continuously for a very long period of time.
Dr. Lamm testified at the public hearing concerning risks from arsenic
emissions from the copper smelter in Ajo, Arizona. A response to this
comment is found in the background information document for promulgated
standards on primary copper smelters (EPA 450/3-83-010b).
2.2 STRINGENCY OF THE PROPOSED EMISSION LIMIT
2.2.1 Comment; One commenter (D-13) expressed support for EPA's
'"•!!'
emissions control proposal and emissions reduction targets, given that EPA
concludes that arsenic emissions from the glass industry must be reduced.
2.2.1 Response: .No response necessary.
2.2.2 Comment: One commenter (D-18) stated that the standards for
emissions of arsenic from glass manufacturing plants should be more
stringent than the 90 percent reduction on which the proposed standards are
based. The commenter referred to data in the Background Information
Document on the performance of electrostatic precipitators and fabric
filters in controlling particulate matter emissions. These data, the
commenter asserted/show that with only one exception fabric filters and
electrostatic precipitators have control efficiencies greater than
90 percent. Setting the level of emissions reductions achievable by these
technologies at the level of the least effective system does not satisfy the
requirements of.Section 112 of the Clean Air Act.
2.2.2 Response: It is the intent of the Agency to require the most
effective possible control of arsenic emissions where these emissions may
reasonably be expecfed to endanger public health. The proposed standards
were not based ori the achievability of 90 percent reduction of arsenic
emissions from glass melting furnaces. Rather, the proposed standards
required particulaty emission from affected furnaces to be controlled to the
2-11
-------
established under the new source performance standards (NSPS) for glass
manufacturing plants. These standards represent the level of control
achievable through the application of the best demonstrated technology (BDT)
for controlling particulate emissions from glass melting furnaces. Because
the data available to the Agency at the time of proposal indicated that at
least 90 percent of the arsenic emitted from glass melting furnaces would be
emitted as particulate matter, it was believed that in meeting the NSPS
particulate emission limits at least 90 percent of the emitted arsenic would
be collected as well. As a result, the best available technology (BAT) for
controlling arsenic emissions was determined to be identical to BDT for
control of total particulate emissions. Therefore, the proposed standards
were not based on the reduction of arsenic emissions achievable by the least
effective system, but on the reductions in arsenic emissions achieved by the
application of the best available technology. The 90 percent reduction
cited in the background information document (BID) only reflects the minimum
level of control for arsenic that was expected to be achieved through the
application of BAT.
Data gathered by EPA after proposal, however, have indicated that in
some instances significantly less than 90 percent of the arsenic emitted
from glass melting furnaces is in particulate form. No correlations were
identified between the proportion of arsenic emitted as particulate matter
and the type of arsenic used as a raw material, the type of glass produced,
or other process parameters.
In addition, EPA found that some furnaces, including the largest
emitting furnace, could meet the proposed emission limits by reducing
particulate emissions by as little as 45 percent. In this case, the
corresponding reduction achieved in arsenic emissions would be no greater
than 45 percent. For these reasons, it was concluded that the proposed
emission limits would not guarantee that the most effective control of
arsenic emissions would be achieved on all affected furnaces. In
considering all of the available data, the Agency determined that a standard
requiring arsenic emissions to be reduced by a specific percentage would be
2-12
-------
necessary to ensure that emissions control devices are installed and
operated in a manner which best reflects their effectiveness in controlling
arsenic emissions.
In setting the level of the final standard, consideration was given to
the performance of existing control devices in reducing arsenic emissions,
to the factors affecting control device performance, to the level of control
achievable from uncontrolled furnaces, and to the costs of control. The
data obtained on arsenic-using furnaces with existing controls indicated
that the efficiency of these devices is variable, ranging from 92 percent to
99 percent. The data further indicate that the variability observed in
control device performance does not correlate to the design or operating
characteristics of either the glass melting furnace or the control device.
Rather, the efficiency of these systems in reducing arsenic emissions is
primarily a function of the amount of arsenic emitted as particulate matter.
Although the available data do indicate that exhaust gas cooling may be
effective in increasing the proportion of arsenic emitted as particulate
matter, sufficient data are not available to predict quantitatively the
extent to which cooling will increase the effectiveness of control.
As a result of these uncertainties, no technical basis exists for
concluding that the efficiency of any existing control device could be
improved by modifying either the design or the operation of the existing
control system. In addition, the costs of modifying any existing control
device would be disproportionate to the incremental reduction in arsenic
emissions that might be achieved, even if there were reason to believe that
these modifications would increase the effectiveness of control. Therefore,
the Agency concluded that the standard should be set at a level which would
not require any additional control of furnaces equipped with existing
control devices.
Because of the variability observed in control device performance, and
because no correlations exist between control device performance and process
or operating characteristics, the efficiency of existing control devices in
reducing arsenic emissions could not be generalized to furnaces that are not
presently controlled. Therefore, emission testing was performed on the
2-13
-------
uncontrolled furnace that will be required to install controls under the
promulgated standard. These tests (summarized in Appendix A) indicated that
a maximum reduction in arsenic emissions of 85 percent could be achieved
within the range of anticipated furnace production rates if the exhaust gas
was first cooled to less than 121°C (250°F) prior to entering the control
device. Because no methods for reducing uncontrolled emissions from this
furnace by more than 85 percent (including further cooling) could be
identified, and because all affected furnaces with existing controls have
demonstrated removal efficiencies of 85 percent or more, the level of the
final standard was established at a 85 percent reduction of uncontrolled
arsenic emissions.
2.2.3 Comment: One commenter (D-18) stated that there is a
discrepancy between the data concerning arsenic trioxide removal and the
theoretical considerations relied on by EPA in setting the proposed
standards. This commenter asserted that EPA found much more arsenic is
removed from glass furnace flue gas by electrostatic precipitators than
would be predicted by theory. Since the data support more stringent control
requirements, EPA should not develop an emission standard that relies on
this current theory.
2.2.3 Response: The standards were developed based on data collected
by EPA which reflect the level of arsenic emissions reductions achieved by
particulate matter control devices. The standards were not developed based
on theoretical calculations of the anticipated emissions of arsenic.
Indeed, if theory were relied upon, it would have been concluded that
particulate matter control devices are ineffective in reducing arsenic
emissions.
2.2.4 Comment: One commenter (D-17) stated that EPA has not proposed
standards which address arsenic in the vapor phase or in fine particulate
form. The commenter points out that arsenic trioxide is appreciably
volatile at temperatures of 100°C (212°F), a temperature common to many
industrial processes. Because arsenic trioxide sublimes from the solid to
the vapor phase more rapidly than it condenses from vapor to liquid phase,
flue gas may become supersaturated by arsenic trioxide, resulting in more
vapor-phase arsenic trioxide than expected. Also, condensation of the
2-14
-------
vapor-phase arsenic will not occur as expected. Unless the process vapor is
substantially saturated with arsenic trioxide, lowering the temperature will
not force the vapor to condense. Therefore, control methods based on the
condensation of arsenic trioxide will not be effective. Further, the
commenter points out that arsenic trioxide adsorbs most readily onto small
particulates, and that the removal efficiencies of particulate matter
control devices are lower on fine particulates than on total particulates.
The commenter concludes that fabric filters are the most effective
technology in removing particulate matter emissions over a range of flue gas
temperatures and particle sizes. Therefore, the proposed standards should be
based on the performance of a well designed and maintained fabric filter
system.
2.2.4 Response: Arsenic emissions from most glass manufacturing have
been shown to be predominantly in the solid phase. Although this apparently
contradicts theoretical considerations which predict that all of the arsenic
would be emitted in the vapor phase, theoretical considerations do not take
into account the presence of other chemical species in the stack gases.
Therefore, EPA has not relied on theoretical data in developing this NESHAP.
Rather, EPA has conducted a series of tests on glass furnaces to provide an
indication of the physical form in which arsenic emissions occur, and to
assess the effect of temperature on the proportion of arsenic found in the
solid phase. In most tests, more than 90 percent of the arsenic has been
shown to exist in the solid phase at typical stack gas temperatures of 288'°C
(550°F).
One test conducted on a furnace melting soda-lime glass indicated that
a relatively small percentage (-x-40%) of total arsenic exists in the solid
phase at 288°C (550°F). When the filtered gases were cooled to 121°C
(250°F), however, up to 80 percent of the total arsenic was found in the
solid phase. In this instance, condensation did appear to occur even though
the vapor-phase concentration at 288°C (550°F) was less than 1.0 x 10"6 of
the theoretical vapor-phase concentration at saturation at that temperature.
The results of this test were inconclusive, however. Although the amount of
solid-phase arsenic increased as the filtered gas temperature in the
sampling train was cooled from 288°C (550°F) to 121°C (250°F), the amount of
2-15
-------
vapor-phase arsenic did not decrease proportionally. Therefore, it is not
certain that the increase observed in the amount of solid-phase arsenic
captured in the sampling train occurred as a result of condensation of
vapor-phase arsenic. Another emission test performed by EPA on a soda-lime
glass furnace showed that all of the arsenic emitted from the furnace was in
the solid phase regardless of temperature.
In another test on a glass melting furnace, additional information was
obtained by EPA on the factors affecting condensation of gaseous arsenic.
Both Reference Method 108 samples at 121°C (250°F) and single-point samples
at elevated temperatures (204°C and 288°C) were collected during this test.
In addition, backup filters maintained at 121°C (250°F) were placed prior to
the impingers in some high temperature sampling trains. The fraction of
arsenic detected in the Reference Method 108 sampling trains averaged
39 percent. The fraction of arsenic detected on the filters at 204°C
(400°F) averaged 34 percent and an average of 36 percent of the total
arsenic was collected on the filters at 288°C (550°F). These data indicated
that condensation of gaseous arsenic was not occurring, even though the
concentration of gaseous arsenic was relatively high. However, substantial
amounts of arsenic were collected on the backup filters maintained at 121°C
(250°F) in the high temperature sampling trains, which indicated that
condensation did occur between the high temperature filters and the low
temperature backup filters. These data suggest that the rate of
condensation in the sampling train may be sensitive to residence time,
temperature gradient, and/or probe location.
Although EPA agrees that the removal efficiencies of particulate matter
control devices are lower for the smallest particle sizes, EPA is not aware
of any data to support the contention that fabric filters are universally
more effective in controlling arsenic emissions from glass melting furnaces
than electrostatic precipitators (ESP). EPA has conducted three separate
tests on controlled arsenic glass furnaces. Two of these were equipped with
ESPs and the third with a fabric filter. The average control efficiency of
the fabric filter in removing particulate arsenic was 99.7 percent. One of
the ESPs demonstrated an average control efficiency of 99.6 percent, and the
2-16
-------
other achieved an average participate arsenic removal efficiency of
99.0 percent. Therefore, the EPA has concluded that ESPs are as effective
as fabric filters in controlling emissions of arsenic from glass manufacturing
plants.
2.3 SELECTION OF CONTROL TECHNOLOGY
2.3.1 Comment: One commenter (D-12) stated that EPA should not assume
that because a particular control device has been adequately demonstrated on
one type of source, that it should be considered BAT for a different source
category. EPA should consider only those control technologies that have
been demonstrated at a commercial scale on a plant of the same type for
which the control requirement is to be imposed. The commenter also stated
that the definition of BAT should also vary among existing sources within
the same general source category on the basis of the age and remaining
useful life of the facilities. While a particular technology may be
considered BAT for new plants, it may not necessarily represent BAT for
older plants. The commenter endorsed EPA's approach of not regulating
facilities where a level of control equivalent to BAT is already in place.
Finally, the commenter stated that controls corresponding to BAT should not
be imposed if they are not needed to eliminate a significant health risk,
particularly in cases where a less costly technology will effectively
eliminate all risks of mortality or serious illnesses.
2.3.1 Response: The selection of ESPs and fabric filters as the
basis for the standard for the control of arsenic emissions from glass
manufacturing furnaces is not based, on assumptions, but on the actual
experience of the glass industry in using these technologies to control
arsenic emissions. The promulgated standard is set at a level that will
require installation of emission control devices on only those existing
glass furnaces where emissions of inorganic arsenic have been determined to
present a significant risk to public health.
2.3.2 Comment: One commenter (D-18) objected to the EPA's reliance on
compliance with OSHA standards for fugitive emissions of arsenic in the
workplace in deciding not to propose standards for these emissions. The
2-17
-------
commenter stated that (1) this reliance was based solely on statements made
by company representatives and had not been independently verified by the
Agency; (2) although OSHA standards, if implemented, may provide protection
to workers in glass manufacturing plants, they do not give persons living
around the plants the enforcement power to compel compliance with the
standards that would be available under the Clean Air Act; (3) the Agency
should, at the least, incorporate into a Section 112 standard the equipment
and work practice requirements needed to comply with the OSHA standards.
2.3.2 Response: The Administrator believes that where standards
established under separate authorities are effective in reducing emissions,
redundant standards need not-be established by EPA. The Agency establishes
separate standards when there is evidence that either the control measures
are not likely to remain in place or are unlikely to be properly operated
and maintained. EPA has again reviewed the emission sources at glass
manufacturing plants to determine any need for controls beyond those
required by OSHA.
Information gathered after proposal during visits to glass plants that
use arsenic indicated that fugitive emissions from some plants may not be
controlled. As a result of this finding, EPA has estimated the magnitude of
the emissions of arsenic which could arise from fugitive sources within
glass manufacturing plants. (Docket reference IV-B-11) These estimates were
based on published fugitive emissions factors for various material handling
operations, as well as on data gathered during visits to glass plants which
use arsenic. To be conservative, "worst case" conditions were assumed in
estimating potential fugitive arsenic emissions. For example, in this
analysis it was assumed that the plant uses unusually high concentrations of
arsenic (14 Ib/ton) in the batch raw materials. The major potential source
of fugitive particulate emissions at glass manufacturing plants are the
materials handling operations associated with the unloading, storage and
weighing of the bulk raw materials. However, arsenic is not present during
these operations. Arsenic is added later, just prior to mixing of the
batch. Fugitive emissions of arsenic could occur during mixing of the batch
materials, during the transfer of these materials to the furnaces, when the
2-18
-------
materials are charged into the furnace, and when control devices (if used)
are emptied and the waste products removed for disposal or recycled back
into the melting furnace. In considering all of the possible sources of
fugitive emissions from glass manufacturing plants, and employing the best
information currently available to the Agency, EPA estimated that the
maximum,fugitive emissions of arsenic from a large, 545 Mg/year (600 tons
per day) plant would amount to 0.21 Mg/year (0.23 ton/year) if emission
control devices were not used. For a plant of this size, uncontrolled stack
emissions would be about 145 Mg/year (160 tons/year). The same plant, if
controlled, would emit about 7 Mg/year (8 tons/year) out of the stack(s);
fugitive arsenic emissions from a 545 Mg/day (600 tons/day) controlled plant
were estimated to be 0.33 Mg/year (0.36 ton/year) under worst case
conditions. Because all of the plants known to use arsenic have capacities
less than 545 Mg/day (600 tons/day), and because the estimates
summarized above are based on "worst case" assumptions, EPA has concluded
that fugitive emissions of arsenic from glass manufacturing plants are
negligible and do not endanger public health. Therefore, the promulgated
standard does not require controls for fugitive arsenic emissions at glass
manufacturing plants nor have OSHA requirements been incorporated into the
promulgated standard as suggested by the commenter,,
2.3.3 Comment: Four commenters (F-l, F-2, D-10, D-18) discussed the
elimination of arsenic as a raw material in the manufacture of glass. Two
commenters (F-l, D-10) stated that the use of arsenic in the manufacture of
glass containers has been completely eliminated, and that there is no
technical reason to use arsenic in the manufacture of glass container
products. These two commenters posed no objection to a requirement that
arsenic be eliminated from glass container manufacturing, as long as no
additional administrative burdens were placed upon container glass
manufacturers. One commenter (F-2) stated that the use of arsenic in the
manufacture of pressed and blown glassware is essential and that no
acceptable substitutes are currently available. Without arsenic, tableware
glass tends to have an objectionable green tint. Another commenter (D-18)
objected to the contention that the elimination of arsenic in pressed and
2-19
-------
blown glass manufacturing would have serious consequences for this sector of
the glass manufacturing industry. This commenter stated that the only
benefit to the glass industry stemming from the use of arsenic is that it
improves the cosmetic qualities of the glass by making it clearer. This
commenter asserted that cosmetic benefits are insufficient to justify public
exposure to arsenic emissions and urged that the standards be amended to
eliminate arsenic from the manufacture of pressed and blown glass. The
commenter also stated that if there are specialized, non-substitutable uses
for arsenic which rise above the level of cosmetics, then EPA should set a
standard requiring extremely stringent controls for a small number of
furnaces dedicated to such uses.
2.3.3 Response: Based on the public comments received and the
information available before and after proposal of the standards, EPA has
concluded that the container glass, flat glass, and wool fiberglass segments
of the glass manufacturing industry do not use arsenic as a raw material in
the manufacturing process. Because the promulgated standard applies only to
furnaces that use arsenic as a raw material, no furnaces in the container,
flat, or wool fiberglass segments of the glass industry would be affected.
Owners or operators of furnaces that do not melt a glass in which arsenic is
added as a raw material are not subject to the requirements of this NESHAP,
including those for reporting and recordkeeping. If an owner or operator of
a furnace in any of these segments of the industry were to begin using
arsenic, the furnace would be subject to the standard.
Arsenic is used in the manufacture of some products in the pressed and
blown segment of the glass industry, however. A case-by-case assessment of
the potential to eliminate arsenic use was conducted by contacting all six
major manufacturers of pressed and blown soda-lime glassware, (docket
reference IV-B-13) Although some companies have been successful in removing
arsenic entirely from their raw batch materials, other companies producing
similar types of glass have been unable to obtain a product of acceptable
quality when arsenic is removed. The qualities achieved by the inclusion of
arsenic (clarity, elimination of unwanted color, etc.) are not just
"cosmetic" in the sense that they have no economic value, but reflect
2-20
-------
certain physical attributes of the final product which are required by the
consumer. Demand for these products is inherently connected to their
physical appearance which, therefore, has a tangible economic value. The
EPA expects that producers of pressed and blown glasswares will continue to
try to eliminate arsenic from their batch recipes to avoid being subject to
the requirements of this NESHAP. It is not clear, however, when (and if)
these efforts will be successful. Because a requirement to eliminate the
use of arsenic in the pressed and blown glass noncontainer segment of the
industry could have severe economic impacts for some producers, it is not
included in the final standard but will be evaluated as part of the 5-year
review of the standard.
EPA considered banning the use of arsenic in those segments of the
glass industry where arsenic is known to be no longer used by establishing a
zero emission limit for certain segments of the glass industry. However,
the promulgated standard does not ban the use of arsenic within any segment
of the glass manufacturing industry. The decision not to ban the use of
arsenic in the flat, container and wool fiberglass industry segments was
made for two main reasons. First, the only available system for
differentiating among the various segments of the glass industry are the
Standard Industrial Classifications (SIC) devised and administered by the
Department of Commerce. EPA has no assurance that all plants at which
arsenic is used at present have been classified within the pressed and blown
glass segment of the industry (SIC 3229). Therefore, placing a ban on the
use of arsenic according to SIC categories might place an unfair and
unintended burden on some individual plants.
Second, the decision to establish an arsenic NESHAP for glass
manufacturing plants was made on the basis of the finding that atmospheric
emissions of arsenic pose a potentially significant risk to human health.
The health risks arising from individual glass plants were carefully
evaluated in establishing the promulgated standard. Emissions of arsenic
from plants producing container, flat, or wool fiberglass would pose the
same risks as those from plants in the pressed and blown glass segment of
the industry. Therefore, EPA sees no reason to establish a separate
2-21
-------
emission limit for plants producing container, flat, or wool fiberglass that
is any different than that for pressed and blown glass.
Glass melting furnaces which do not at present use arsenic, but which
will use arsenic in the future, would be considered to be modified furnaces
under the general provisions of 40 CFR Part 61. These furnaces would then
be required under the promulgated standard to install a control device if
emissions of arsenic exceed 0.4 Mg/year (0.44 ton/year).
2.4 COST EFFECTIVENESS OF THE PROPOSED STANDARD
2.4.1 Comment: One commenter (F-2) stated that the costs of applying
add-on control technology to furnaces producing tableware glasses are
disproportionate to the resulting reductions in emissions.
2.4.1 Response: The EPA's analysis of the costs of applying the
promulgated standards to the production of tableware glass indicates that
the costs of the final standard do not have significant adverse impacts on
the industry and are reasonable in light of the reductions in arsenic
emissions and health risks achieved.
2.4.2 Comment; One commenter (D-13) stated that only glass furnaces
which are at present uncontrolled should be subject to the proposed
regulations because the most cost-effective emissions reductions can occur
from these furnaces.
2.4.2 Response; The response to comment 2.1.2 explains why the Agency
decided to establish emission cut-offs based on emissions prior to any
control device for existing and new or modified furnaces rather than making
the NESHAP applicable only to uncontrolled furnaces as the commenter
suggests.
2.5 ECONOMIC IMPACT
2.5.1 Comment: One commenter (F-2) stated that the monetary costs
required to comply with the standard would severely affect an already
depressed market which is facing significant and increasing competition from
2-22
-------
foreign producers of glass tableware. Between 1979 and 1982, the compound
growth in imports has been 6.8 percent, while growth in the domestic share
of the market has declined by 0.4 percent. In addition, over the past ten
years there has been a decline in real total dollar market value for the
U.S. tableware industry. Two tableware manufacturers have recently closed
plants. The strong U.S. dollar will continue to favor imports of glass
tableware. The commenter stated that reducing emissions to the level
proposed by the standard is estimated to cost $14.20/ton of glass. These
costs would increase operating costs by over 2 million dollars/year. This
represents an increase of 2.1 percent in production costs over 1982 levels,
which would have decreased 1982 profits by! 25 percent.
2.5.1 Response: The EPA recognizes that machine-made glass tableware
manufacturers are facing competition from foreign producers of glass
tableware; and in the economic analysis conducted after proposal, it was
assumed that prices cannot be raised and that companies must absorb the
control costs as decreased profits (See Appendix B). The costs cited by the
commenter were for a specific plant owned and operated by the commenter.
The cost and economic impacts of the promulgated standard were analyzed for
this plant, and EPA concluded that they would be disproportionately high
compared to the risk reduction that would be achieved through application of
control technology. Therefore, the plant is one of several that would have
been required to install controls under the proposed standard but, due to
the revised emission cutoff for existing furnaces, would not be required to
install emissions control devices under the promulgated standard.
The economic analysis indicated a potential closure for only one
furnace currently using arsenic and emitting arsenic at a rate above the
revised cutoff. Company representatives have informed EPA, however, that
they plan to eliminate the use of arsenic at this furnace (Docket reference
IV-E-58); therefore, it would not be affected by the standard. The EPA's
analysis indicates that no other furnace closures would result from this
NESHAP.
2.5.2 Comment: One commenter (D-13) stated the belief that some
plants would close down if the proposed standard were promulgated.
2-23
-------
2.5.2 Response; As mentioned above, EPA has reanalyzed the economic
impacts and has raised the uncontrolled emissions cutoff for existing
furnaces. Within the narrowed range of affected furnaces, EPA believes that
there will be no furnace closures due to the promulgated regulation.
2.5.3 Comment: One commenter (D-18) stated that the "worst case"
economic analysis conducted by EPA has been grossly exaggerated in reaching
a conclusion that under certain conditions the proposed regulations could
cause some furnaces to close. Further, the commenter stated that the
assertion that the elimination of arsenic from pressed and blown glass would
make U.S. manufactured glassware uncompetitive with glassware imported from
countries which do not restrict arsenic use has not been supported by hard
data or analysis. The commenter stated that if the regulations do impose a
competitive disadvantage on U.S. glass manufacturers, other steps should be
taken to protect their position, such as the imposition of duties on imports
of arsenic containing glass.
2.5.3 Response: The revised economic analysis (Appendix B) of the
promulgated standard explains that cost absorption (profit reduction) by
producers, rather than cost pass-through to consumers, is more likely to
result because of the competitive role of imports. Using this assumption,
all control costs were analyzed as additions to baseline operating expenses.
No closures are anticipated as a result of the promulgated standard.
The EPA's assertion that U.S. manufacturers of pressed and blown glass
would be at a competetive disadvantage with foreign manufacturers if arsenic
were eliminated is based on the fact that the properties that arsenic
provides in the glass products have an economic value. Such properties as
clarity and sparkle are desired by the consumer and, thus, are considered
necessary for certain products to be competitive in the market. The
economic value of these properties has not been quantified but is,
nevertheless, real. The commenter's suggestion that duties be imposed on
imports of pressed and blown glass that contain arsenic cannot be
implemented because EPA does not have legislative authority to impose such
duties or to take any similar measure to reduce possible competition to U.S.
glassware manufacturers by foreign glass.
2-24
-------
2.6 EXEMPTIONS AND ALLOWANCES
2.6.1 Comment: One commenter (D-13) requested that EPA include
provisions for conducting normal maintenance on control devices. Most glass
furnaces operate continuously for a period of years, while emission control
devices require frequent maintenance. The commenter stated that the
maintenance requirement on an electrostatic precipitator is about 144
hours/year and that a provision should be made for by-pass of the control
device while maintenance is being conducted.
2.6.1 Response: EPA has investigated the cost and environmental
impacts associated with performing routine maintenance on emission control
devices installed on affected glass furnaces. (Docket reference IV-B-10)
Two alternatives were considered. The first alternative would be to require
the glass furnace to be shut down during these maintenance periods in order
to avoid uncontrolled emissions of arsenic. The second alternative would
allow furnace operators to by-pass the control device for a limited period
of time for maintenance purposes. Emissions of arsenic during these periods
would not be controlled. EPA analysis compared the increase in the costs
incurred by a model manufacturing plant which would result from the first
alternative to the increase in emissions which would follow from the by-pass
alternative. In this analysis both large and small furnaces and high and
low glass production costs were considered. In total, the cost and
environmental impacts associated with the alternative requirements were
evaluated for eight different cases.
In the first four cases the impacts were calculated for two furnace
sizes (50 tons/day and 150 tons/day) and for two levels of specific arsenic
emissions (0.05 Ib arsenic/ton of glass produced and 1.00 Ib arsenic/ton of
glass). In the first four cases, relatively low glass production costs were
assumed, on the order of $0.34/lb of product. The second four cases assumed
the same furnace sizes and specific arsenic emission rates, but were based
on the assumption of a glass with higher production costs of $1.90/lb.
These values represent the low and high end of the ranges for actual glass
furnaces which use arsenic. In all cases, it was assumed that the time
required for maintenance of control devices is 144 hours/year.
2-25
-------
The results of this analysis showed that a large furnace with a high
arsenic emission rate could emit up to 0.41 Mg (0.45 ton) of arsenic during
the 144 hours that the control device is by-passed. Small furnaces with low
arsenic emission rates would emit 0.01 Mg (0.01 ton) of arsenic during this
maintenance period. The costs of furnace shutdown were estimated to range
from a low of $63,000 for a small furnace producing a low cost glass, to a
high of $1,000,000 for a large furnace producing a high cost glass. Thus,
the cost effectiveness of requiring all arsenic-using furnaces to be
shut down while maintenance is carried out on emissions control devices
would range from about $420,000 per ton of arsenic removed to over
$47,000,000 per ton of arsenic removed.
Because the impacts of requiring furnaces to be temporarily shut down
while maintenance is performed on emissions control devices would be
excessive in some cases, and because the use of well maintained control
devices is essential in effectively controlling arsenic emissions on a
continuing basis, the promulgated standard allows emission control devices
installed on furnaces affected by the standard to be by-passed for purposes
of conducting necessary maintenance. EPA has also determined, however, that
arsenic emissions from glass melting furnaces can be reduced by implementing
certain work practices during maintenance periods. Therefore, the
regulation requires owners or operators of affected furnaces to minimize
arsenic emissions to the maximum extent practicable during periods when the
control device is shut down for maintenance purposes. Each owner or
operator of an affected furnace that wishes to by-pass the control device
for maintenance purposes is required to submit a plan to the Administrator
which details (1) the length of time it will be necessary to by-pass the
control device, (2) the emissions of arsenic which would occur during
maintenance periods if no steps were taken to reduce them, (3) the
procedures and work practices which will be implemented to minimize arsenic
emissions during maintenance periods, and (4) the expected reduction in
emissions of arsenic achieved by the implementation of these procedures and
work practices. Only after approval by the Administrator of this plan will
the by-pass of an emission control device be allowed.
2-26
-------
In some cases, emissions of arsenic can be prevented entirely while
control devices are undergoing maintenance. For example, control device
maintenance could be scheduled during periods of normal furnace shutdown
whenever possible. For some plants, it may be feasible to switch production
temporarily during periods of control device maintenance to glasses which do
not contain arsenic. All facilities affected by the regulation should make
maximum use of control devices which are divided into two or more
independently operated sections. Use of so called "sectionalized" control
devices enables maintenance to be performed on one section of the device
without affecting the operation of the other(s). Other steps that can be
taken to minimize emissions of arsenic during maintenance of control devices
are the maximum use of oil let, the temporary reduction in arsenic feed, or
the temporary reduction of furnace output.
2.6.2 Comment: One commenter (D-13) requested that furnaces melting
arsenic-containing glasses over limited time periods (1 to 10 weeks) for
research and development purposes not be subject to the standard. Only
after a glass is commercially proven and melted for production purposes
should the glass/furnace become subject to the regulation.
2.6.2 Response: New or modified glass manufacturing furnaces emitting
more than 0.4 Mg (0.44 ton) of arsenic per year or existing furnaces
emitting more than 2.5 Mg (2.75 tons) per year are subject to regulation
without regard to the purposes of the particular glass production. It is
not anticipated that research and development furnaces will exceed the
cutoffs. However, should emissions from a research and development furnace
exceed these levels, the furnace would be subject to the standard. Whether
the glass is produced as a research project or for sale, the arsenic
emissions pose the same hazard and may be controlled through the application
of the same control devices. Therefore, no exemption is provided in the
final standard for research and development glass production.
2.6.3 Comment: One commenter (D-13) requested that, in those cases
where production is shifted from a controlled furnace to an uncontrolled
furnace, sufficient time (24 months) be allowed for installation of a
control device. The commenter considered it unrealistic for the industry to
2-27
-------
anticipate the need to switch furnaces 2 years in advance in order to have a
control device installed at start-up.
2.6.3 Response: The promulgated standards do not provide for an
interim period following the switching of an existing furnace to the
production of arsenic-containing glass to allow for the installation of
control equipment. Basically, the switching of a furnace is viewed as
similar to the construction of a new furnace and the same degree of planning
prior to the change should be undertaken. Section 112 provides for the
possibility of a 2-year waiver period only for existing sources of arsenic
emissions.
2.6.4 Comment: One commenter (D-13) stated that because the grain
loading from electrically boosted furnaces is on the order of 0.05
grains/dscf, an add-on control device may not be able to meet the standard
for these furnaces at all times. It was noted that there are no add-on
emissions control devices currently in use on electrically boosted glass
furnaces. The commenter suggested that EPA may need to establish a separate
category to reflect this.
2.6.4 Response: The comment is assumed to imply that the efficiency
of particulate matter control devices will be adversely affected by low
concentrations of particulates in the stack gases. As the commentor noted,
there are no electrically boosted furnaces currently in operation which are
equipped with an add-on control device. Therefore, EPA has no means to
determine whether an electrically boosted furnace would be able to meet the
proposed particulate limit at all times. No information has been submitted
to EPA, however, that would indicate that electrically boosted furnaces
would be unable to do so if equipped with an appropriate control device.
EPA did test one electrically boosted furnace in the development of this
NESHAP. The average particulate grain loading from this furnace was
0.11 gr/dscf, more than double the concentration cited by the commenter.
Although electric boosting decreases the amount of particulate matter
emitted, it also decreases the flow rate of the furnace exhaust. As a
result, electric boosting is expected to have only a minor impact, if any,
on the concentration of particulate matter at the inlet of the control
device or on the emissions reduction efficiency of the control device.
2-28
-------
Results of tests conducted by EPA in the development of this NESHAP
have demonstrated that ESPs and fabric filters can achieve high arsenic
collection efficiencies even when the concentration of arsenic in the flue
gas is low. Average particulate arsenic removal efficiencies of 97.8, 99.7,
and 99.7 percent were achieved by control devices on particulate arsenic
inlet concentrations of 0.024, 0.006, and 0.01 gr/dscf, respectively.
Because no data have been presented which Indicate that emissions from
electrically boosted furnaces cannot be controlled as efficiently as
emissions from conventional furnaces and because ESPs and fabric filters are
capable of high pollutant removal efficiencies even at low pollutant inlet
concentrations, EPA sees no reason to establish a separate category for
electrically boosted glass melting furnaces.
2.6.5 Comment: One commenter (D-13) recommended that the non-arsenic
,portion of the particulate emissions reduced in complying with the proposed
standard should be available for future bubbling purposes allowed by State
regulations. For example, the commenter stated that for a furnace emitting
7 Ib/hour of total particulate, including 1 Ib/hour of arsenic, that is
required to reduce its arsenic emission to 0.1 Ib/hour, the corresponding
reduction in total particulates of 5.4 Ib/hour should remain available for
future bubbling purposes. The commenter stated that the state agencies
should have this option. ',
2.6.5 Response: The availability of particulate matter emission
reductions achieved under the NESHAP for "bubbling" is an issue outside the
scope of this rulemaking, and is therefore,not being addressed here.
2.7 MONITORING AND MEASUREMENT METHODS
2.7.1 Comment: One commenter (D-6) supported the EPA's position that
a material balance or other non-stack test data be used to establish whether
a facility is affected by the proposed regulation and to monitor compliance.
However, the commenter requested clarification on two points. First, what
confidence level does EPA have in the estimates of arsenic retention in
glass? Specifically, should the low end of the estimate, 70 percent
2-29
-------
retention, be used in estimating uncontrolled arsenic emissions? Second,
how should the arsenic content of the cullet be determined? Is it accurate
to assume that all of the arsenic entering with the cullet remains in the
glass, and thus has no impact on arsenic emissions?
2.7.1 Response: The estimates of the amount of arsenic retained in
the glass product were provided by the glass industry. Data obtained from
tests conducted by EPA have been found to be reasonably consistent with data
supplied by industry representatives. It should be noted, however, that the
amount of arsenic retained in the glass can vary significantly according to
the specific recipe used in making the glass.
The 70 percent retention value published in the proposal BID
(EPA-450/3-83-011a) was supplied by industry representatives as a typical
retention rate for lead silicate type glass. Data gathered by the EPA after
proposal have demonstrated that at least 70 percent of the arsenic is
retained in the glass product regardless of its composition. However, the
amount of arsenic retained in the glass product is not strictly a function
of the type of glass produced. For any given type of glass, the percentage
of arsenic retained in the product can vary widely. For example, data
collected by EPA show that the percent of arsenic retained in soda-lime
glass can range from about 70 percent to about 90 percent. Therefore, in
estimating uncontrolled arsenic emissions, the arsenic retention value
should be based on actual laboratory analysis of the glass produced in a
specific melting furnace. If analytical data are not available, an assumed
retention value of 70 percent would provide an estiamte of the maximum rate
of uncontrolled arsenic emissions from a glass melting furnace. In
developing a material balance for monitoring compliance, it is the
responsibility of the furnace owner or operator to provide a theoretical
emission factor which accurately takes into account the amount of arsenic
retained in the glass. Retention values should be based on actual
analytical data for the specific type(s) of glass produced by the furnace.
The amount of arsenic entering the furnace in the cullet should be
explicitly accounted for. Some furnaces may add mixed cullet that is not
exactly similar in chemical composition to the type of glass being melted.
2-30
-------
When the cullet added is identical to the glass being produced, the
percentage of arsenic in the cullet can be assumed to be identical to the
percentage retained in the glass. Thus, the arsenic entering with the
cullet would not have any impact on arsenic emissions. When this assumption
is made, however, care must be taken to calculate the amount of arsenic
retained in the glass on the basis of the percent of product weight that is
derived from fresh raw materials rather than on the basis of the total
product weight.
For example, consider a hypothetical furnace which produces 2000 pounds
of glass per hour. Each 2000 pounds of product is made from a batch
composition which consists of 1700 pounds of fresh batch materials,
including 0.3 pound of arsenic, and 500 pounds of cullet. The Gullet is
identical in composition to the product and contains 0.012 percent arsenic.
Thus, the total amount of arsenic entering the furnace per hour is 0.36
pound (0.3 lb/hr + [0.00012 x 500 Ib/hr]). The amount of arsenic retained
in the product per hour is 0.24 pound (2000 Ib/hr x 0.00012). On this basis
the expected emissions of arsenic from the furnace would be 0.12 Ib/hour
(0.36 lb/hr - 0.24 lb/hr). However, if the amount of arsenic entering with
the cullet is ignored, then the portion of the total product that is derived
from the cullet must also be ignored in determining the expected emission
rate. Thus, if the total arsenic input is assumed to be 0.30 Ib/hr (the
amount of arsenic added in the fresh batch) then the amount of arsenic
retained in the product per hour would be 0.18 Ib/hour ([2000 Ib/hour - 500
lb/hr] x 0.00012). This procedure yields the same expected emission rate as
that obtained when the cullet arsenic is explicitly accounted for. It would
be erroneous, on the other hand, to assume that of the 0.30 Ib/hour of
arsenic added to the fresh batch, 0.24 Ib/hour is retained in the glass.
2.7.2 Comment: Two commenters (F-2, D-13) stated that the proposed
requirement for opacity monitoring is unnecessary and inconsistent with the
NSPS for glass manufacturing, which does not require opacity monitoring.
One commenter (F-2) indicated that opacity monitoring would represent an
unjustifiable cost burden. The other commenter (D-13) stated that opacity
monitoring is administratively burdensome, and readings cannut be correlated
2-31
-------
with emissions of either arsenic or participates, especially when multiple
furnaces are exhausted to a common stack. The commenter noted that
excessive stack opacity occurs in one of the commenter's furnaces as a
result of gaseous fluoride emissions from melting one type of glass, and
that this opacity is unrelated to arsenic or total particulate emissions.
2.7.2 Response: The requirement for opacity monitoring was proposed
as a means to ensure that emissions control devices installed on arsenic-
using glass furnaces are continuously operated and maintained in a manner
consistent with the procedures followed to comply with the standard
initially. These requirements have been retained in the promulgated
standard. The fact that opacity monitoring is not required as part of the
NSPS for glass manufacturing plants has no bearing on this action because
the intent of this regulation is to control a hazardous air pollutant which
is not specifically regulated under the NSPS. With respect to the costs of
opacity monitoring, EPA has determined that the costs involved are
reasonable in light of the improved effectiveness in enforcement that will
be gained as a result of this requirement. No information has been
presented to the Agency which indicates that continuous monitoring of
opacity represents an unjustifiable cost burden.
The promulgated standard does not set any specific limit on stack gas
opacity based on correlations between opacity and emissions of either
particulate matter or of arsenic. Rather, the promulgated standard requires
that a 6-minute average opacity value for a given furnace be determined
during compliance testing. Any subsequent exceedance of the average opacity
value demonstrated in a compliance test must be reported semiannually. If
excess opacity occurs as a result of a change in the composition of the
glass being melted in a furnace, this cause should be cited in the report.
Alternatively, if multiple types of glass are typically melted in a single
furnace, and stack gas opacity is expected to be significantly higher for
one type of glass, the initial compliance test may be performed while this
glass is being melted. Finally, paragraph 61.163(i) of the promulgated
standard allows owners or operators of affected furnaces to petition the
2-32
-------
Administrator for approval of any alternative continuous monitoring system
that can be demonstrated to provide accurate and representative monitoring
of a properly operating control device.
2.7.3 Comment: Two commenters (D-l, D-3) stated that the inclusion of
a procedure for measuring sulfur dioxide in the flue gases from glass
manufacturing is unnecessary. One commenter (D-l) further indicated that
the procedure specified for sampling sulfur dioxide may give inaccurate
results when applied to glass melting furnaces because of the formation of
nitric oxide and sulfur trioxide in the manufacture of glass. The nitric
oxide will be measured as sulfur dioxide because the analytical procedure is
non-specific; it is a general test for all acidic species formed by passage
through the hydrogen peroxide impingers. In conclusion, the commenter
recommended that all references to the measurement of sulfur dioxide be
eliminated in the description of EPA Method 108 procedures. The other
commenter (D-3) indicated that the use of three hydrogen peroxide impingers
is unnecessary, and that much of the currently available source sampling
equipment cannot accommodate a six impinger sampling train. The commenter
recommended that sulfur dioxide sampling be eliminated from the testing
procedure for glass manufacturing plants.
2.7.3 Response: The three hydrogen peroxide impingers at the back of
the sampling train were included only to protect the metering system from
corrosion when sampling gas streams with high sulfur dioxide concentrations.
The analysis steps for measuring sulfur dioxide were included to allow the
tester to determine the approximate sulfur dioxide concentration so that the
final sample volume could be corrected for the sulfur dioxide collected by
the sampling train. However, neither of these steps would be necessary in
collecting samples from glass furnaces so all references to sulfur dioxide
collection and analysis have been deleted from the method.
2.7.4 Comment: One commenter (D-3) recommended that EPA establish
greater consistency in specifying sampling probe and filter temperatures
for EPA Reference Method 5 particularly for source categories that do not
emit significant concentrations of sulfur dioxide. The commenter pointed
out that a probe and filter temperature of 320°F was recommended for boilers
2-33
-------
(NSPS), while a probe and filter temperature of 350°F was specified for
glass melting furnaces (NESHAP). The commenter saw no reason to
differentiate between sources firing more than, or less than, 0.5 percent by
weight sulfur content. The commenter recommended that a set probe
temperature be specified rather than a range, since the ratio of the
front-half to back-half catch can vary significantly according to sample
collection temperature. Finally, the commenter stated that comments
previously made in response to Document No. A-81-19 relating to proposed
Quality Assurance revisions to Methods 6 and 7 of 40 CFR, Part 60, Appendix
A, as published in the Federal Register on March 30, 1983, are applicable
here as well, and should be considered.
2.7.4 Response: Because the proposed particulate matter emission
limits have not been retained in the promulgated standard, EPA Reference
Method 5 sampling procedures are no longer applicable. Therefore, the
temperature requirements referred to by the commenter are no longer relevant
to this action. The Agency agrees that there is no reason to differentiate
between sources firing fuel with greater than 0.5 percent by weight sulfur
and those with less than 0.5 percent, and has revised the standard
accordingly.
The commenter referred to comments made on the quality assurance and
quality control revisions' proposed for Reference Methods 6 and 7 in the
March 30, 1983 Federal Register (48 FR 13388). He felt comments made on the
proposed revisions to Reference Methods 6 and 7 were applicable to the
requirement for an audit sample in Reference Method 108. The EPA
promulgated the quality assurance and quality control revisions to Reference
Methods 6 and 7 (49 FR 26522) on June 27, 1984. Comments and Agency
responses are summarized in the promulgation notice.
This commenter (D-3) referred to the comments on Reference Methods 6
and 7, applying them to Reference Method 108, noting that an audit per set
of compliance samples is over-regulation. The proposed quarterly (Reference
Method 6 and 7) exemption covers repetitive testing at a given source only.
Most control agencies will have a central laboratory and will use different
personnel and analytical systems. The arsenic analyses are very time
2-34
-------
consuming, and a team of chemists may be routinely assigned to analyze the
samples so they can be completed in a timely manner. The requirement that
only one person analyze arsenic samples (paragraph 4.4) could delay the
preparation of the compliance test report past deadlines now set by the
States and EPA. The proposal should be amended to allow more than one
person to analyze the samples. Any analytical variations would be checked
by quality control procedures currently incorporated in EPA methods (e.g.,
equipment calibrations, duplicate testing and analysis, and inter-
laboratory testing). Furthermore, since the "oneness" rationale is not and
should not apply to the sample collection phase, it should not be mandated
for the analyst and analytical equipment phase.
The comments on Reference Method 6 and 7 went on to say that the
rulemaking ignores the logistical problems that will be created with a
one-on-one audit requirement, especially for control agencies conducting
numerous tests.
1. How will EPA guarantee delivery of the audit samples?
2. Will EPA ensure that the concentrations are within the analytical
range of the samples?
3. Can it ensure a different audit concentration for each request?
4. Can EPA guarantee fast delivery of audit samples? Our regulations
require analysis within 48 hours of sample Collection.
In response to these comments on Reference Method 6 and 7 and their
applicability to Reference Method 108, the quarterly auditing allowance for
analyses performed on a frequent basis has been changed. The revised
requirement (Reference Method 6, 7 and 108) is to analyze audit samples at
least once per month. This discourages the use of 3 month-old analytical
reagents as allowed in the quarterly audit and also eliminates the need to
perform an audit with each set of compliance samples if samples are analyzed
on a frequent basis.
The intention of the rule is not to restrict the analysis to une
person but rather to ensure that all parties involved in the analysis of the
compliance samples likewise take part in the analysis of audits. The names
of all persons participating in the audit analysis are to be included in the
report that is submitted to the appropriate enforcement agency.
2-35
-------
EPA will be able to respond if notified soon after a decision has been
made to perform a compliance test. The quality assurance revisions will
have instructions to notify the EPA Quality Assurance Management Office
30 days prior to the actual test.
Since all sample concentrations must be diluted to fall within the
calibration range of the standards for Method 108, the audit sample
concentrations will be within these analytical ranges. Coordination within
the auditing program will ensure that sources receive different audits on
different requests.
2.7.5 Comment: One commenter (D-7) stated that the acid digestion
bombs required by the Method 108 protocols are a potential explosion hazard,
and recommends that alternative methods be developed. The commenter also
noted that since a five hour digestion period is specified by the methods,
it would be desirable to have multiple-bombs in order to digest several
samples simultaneously. However, since these bombs are expensive, small air
pollution laboratories could incur significant expense in obtaining multiple
bombs.
2.7.5 Response: A review of data collected at glass furnaces shows
that there is little or no arsenic in the insoluble paniculate matter which
must be digested by the bomb digestion procedure. Although significant
amounts of arsenic can be found in the insoluble particulate from copper
smelters, the digestion procedure will be eliminated from Method 108 when it
is applied to glass furnaces.
2.7.6 Comment: One commenter (D-13) stated that the time allowances
for testing under the proposed Section 61.163 were inflexible and
inadequate, and that the specified testing procedures were inflexible and
unnecessary. In support of this view, the commenter provided data showing
that other analytical methods can provide similar results to those obtained
when using the specified EPA Method 108 procedures.
The major difference between the procedure proposed by the commenter
and the EPA Method 108 procedure was in the method used in determining
arsenic concentration of the samples. The procedure proposed by the
commenter employed the colorimetric molybdenum blue method instead of atomic
2-36
-------
absorption. There were also slight differences in the types of reagents
employed, and the procedures followed, in leaching the materials collected
by the probe, filter, and impingers. In the example provided, the amount of
arsenic detected when using the molybdenum blue method was 21 mg, 5 mg, and
0.2 mg in the filter, probe, and impingers, respectively. These results
compared to detected arsenic levels when using EPA Method 108 procedures of
21 mg in the filter, 1 mg in the probe, and 0.4 mg in the impingers.
2.7.6 Response: Under Section 40 CFR 61.14 in Subpart A - General
Provisions, the Administrator may allow the use of any alternative method
which he has determined to be adequate for indicating whether a source is in
compliance. Anyone wishing to have a method approved as an alternative may
submit comparative data between the candidate method and the reference
method for evaluation by the Administrator. EPA has notified the commenter
of these provisions (docket reference).
2.8 DETERMINATION OF COMPLIANCE
2.8.1 Comment: One commenter (D-13) stated that not all furnaces that
are at present controlled can be assumed to meet the proposed standard. For
example, a controlled furnace at one of the commenter's plants is in
compliance with the proposed emission limits at low and medium production
rates. The furnace has not been tested at high rates of production,
however, and it is not known if the furnace would achieve compliance at high
rates of production. The commenter stated that it would be costly and
burdensome to test this furnace at artificially high production rates, and
unnecessary in light of the fact that the control device installed on this
furnace represents the best system available at present.
2.8.1 Response: Under the promulgated standards, it would not be
necessary to test glass manufacturing furnaces at artificially high
production rates. Rather, for purposes of, demonstrating compliance, testing
will be performed at the typical expected production rate of a given
furnace. Because the proposed limits on particulate emissions have not been
retained in the promulgation standards, determinations of compliance for
2-37
-------
furnaces which are required to install control devices will not be affected
by the production rate of the furnace. These furnaces must only demonstrate
that emissions of arsenic are being reduced by the percentages specified in
Section,61.162.
2.8.2 Comment: One commenter (D-13) provided an example of the
difficulties that may arise in demonstrating compliance with the proposed
emission limits for furnaces that are already equipped with control devices.
The commenter provided a hypothetical situation in which a baghouse controls
emissions from two furnaces which melt four different types of glasses, and
three of these glass compositions contain arsenic. Each module of the
baghouse has its own "goose neck" exhaust stack. The commenter states that
these goose neck stacks cannot be tested using EPA Method 5 procedures, and
cannot be equipped with opacity monitors. The commenter suggested that EPA
should consider accepting the efficiency rating of the control device as
specified by the manufacturer, in lieu of requiring emissions testing.
2.8.2 Response: Method 108 has replaced Method 5 as the required test
method in the promulgated standard, but the sampling procedures in
Method 108 would not be appropriate for testing the specific source cited by
the commenter either. However, there is another test method, Method 5D,
whcih was developed specifically for testing baghouses and whose sampling
procedures would be applicable to this source. For sources equipped with
positive pressure fabric filters, Section 4 of Method 5D should be used in
conjunction with Method 108 to determine a suitable sampling location and
procedure. Although it would be possible to equip a source such as this
with opacity monitoring equipment, it would not be very cost effective since
multiple opacity monitors would be required. A source could request an
alternative procedure such as the one allowed at electric arc furnaces where
daily observation of visible emissions (in accordance with Method 9) by a
certified visible emissions observer is used to monitor opacity.
2.8.3 Comment: One commenter (D-13) raised the question of the
applicability of the standard in cases where multiple furnaces are exhausted
to a common stack. An example was given of a group of five furnaces which
melt more than 30 different glass compositions, 9 of which contain arsenic.
2-38
-------
These furnaces are controlled by two electrostatic precipitators which
exhaust to a common stack. The commenter stated that more than 100,000
combinations of furnaces/glasses could occur on any given day. For this
reason, it would be virtually impossible to determine if each combination is
in compliance with the applicable proposed emission limit. Finally, the
commenter indicated that the control devices installed on these furnaces
reflect the best technology currently available, yet do not always comply
with the proposed standards. The commenter provided another example in
which a furnace melting fluoride opal glass is known not to meet the
proposed emission limits at all times even through the furnace is controlled
with the Best Available Technology (BAT). The commenter stated that this
situation is attributable to EPA's classification of fluoride opal glass in
the "all other" category during NSPS development. In conclusion, the
commenter indicated that furnaces that are;already equipped with BAT should
not be subject to further regulation under the proposed standard.
2.8.3 Response: EPA has investigate^ the problems associated with
demonstrating compliance with the proposed;emission limits on total
particulates in cases where multiple furnaces, producing different types of
glass, are exhausted to a common stack. Because the proposed particulate
emission limits varied according to glass type, the emission limit
applicable to a given furnace would change with a corresponding change in
the type of glass produced. No satisfactory approach could be developed for
determining compliance with a particulate emission limit under these
circumstances, or for prorating emissions from multiple furnaces which
exhaust to a common stack when these furnaces melt various types of glass.
Partly for this reason, the proposed emission limits on total particulates
have been abandoned and replaced by a requirement that arsenic emissions be
reduced by a specified percentage. (See response to comment 2.2.2.)
EPA also investigated the issue raised by the commenter in regard to
the compliance status of furnaces which are equipped with existing control
devices. In reviewing all available data on controlled arsenic-using
glass furnace, the Agency has found that many of these control devices
achieve more than 95 percent reduction of total arsenic emissions, although
2-39
-------
some of them are not capable of reducing emissions of total participates to
the levels prescribed by the proposed standard. The cost of upgrading these
control devices to meet the proposed particulate emission limits was
investigated and found to be excessive given the minimal incremental
reduction in arsenic emissions that would be achieved. Therefore, basing
the control requirement on the NSPS emission limits originally proposed
would result in the failure of some furnaces with existing control devices
to achieve the standard, even though these control devices represent an
effective control for emissions of arsenic. For this reason also, the
proposed emission limits have not been retained in the promulgated standard.
Data made available to EPA from industry representatives indicate that all
existing controlled furnaces with uncontrolled arsenic emissions of more
than 2.5 Mg/year (2.75 tons/year) will be able to demonstrate compliance
with the promulgated emission limits without additional control.
The EPA has investigated the emission reductions currently achieved
from the furnace producing a fluoride-opal glass composition referred to by
the commenter. The control device installed on this furnace was found to
reduce arsenic emissions by more than 98 percent, although particulate
emissions from the furnace exceed the applicable NSPS emission limits. As
discussed above, the proposed emissions limits were not adopted in the final
regulation, and no additional control is required on this furnace under the
promulgated standard.
2.8.4 Comment: One commenter (D-26) stated that the provision in the
proposed standards requiring emissions to be monitored only at the stack
outlet and reported every 6 months is unenforceable. The commenter
suggested that an efficiency standard for arsenic removal, which would
require the measurement of arsenic levels at both the inlet and the outlet
to the control device, or a standard limiting the amount of arsenic in each
batch would be enforceable and would reduce arsenic emissions to the
atmosphere.
2.8.4 Response: Although the commenter provided no information to
support the contention that the proposed standard would have been
unenforceable, the proposed emission limits on total particulates have been
replaced in the promulgated standard by a requirement that emissions of
2-40
-------
arsenic be reduced by a specified percentage. In analyzing data made
available after proposal, EPA found that particulate emission rates from
some uncontrolled arsenic-using furnaces are significantly less than would
normally be expected. Thus, these furnaces could conceivably meet the
proposed limit dVi particulate emissions by reducing particulate emissions by
as little as 45 percent. In this case, the corresponding reduction achieved
in arsenic emissions would be only 40 to 45 percent even if all of the
arsenic were emitted as particulate matter. In considering these data, EPA
has found that emission control equipment that would meet the proposed
particulate emission limits may not, in all instances, represent the most
effective control technology for arsenic emissions. Therefore, the
promulgated standard requires emissions of arsenic to be reduced by a
specified percentage which reflects the level of control achievable by the
most effective control technologies.
Setting a limit on the amount of arsenic added to the raw materials
would constitute a work practice standard as defined in Section 112 of the
Clean Air Act. Section 112 allows work practice standards to be set only in
those instances in which it is not feasible to prescribe or enhance an
emission standard. Because emission standards for arsenic from glass
manufacturing plants can be established and enforced, EPA has no authority
to establish a work practice standard. In any case, the amount of arsenic
added to the raw materials varies widely within the industry as a function
of the type of glass being produced and the market for the final product.
Therefore, it would be impossible to set any specific limit on arsenic usage
which could be applied to all glass melting furnaces, or to categories of
glass melting furnaces, that would not fundamentally interfere with process
operations.
2.9 REPORTING REQUIREMENTS
2.9.1 Comment: One commenter (F-2) stated that it is unreasonable and
irrational to require 12-month projections of arsenic emissions from glass
plants. A semiannual reporting of past emissions should be sufficient for
enforcement purposes.
2-41
-------
2.9.1 Response: The requirement that arsenic emissions be projected
over a 12-month period is necessary in order for the operator of the glass
manufacturing furnace to anticipate the level of control that will be
required for each facility. Only in this way can possible instances of
noncompliance with the proposed standards be prevented. The calculation of
past emissions may reveal actual instances of noncompliance, but only after
unacceptable levels of arsenic have been emitted to the atmosphere. This
result would be inconsistent with the objectives of Section 112 of the Clean
Air Act.
2.9.2 Comment: One commenter (D-13) stated that many administrative
problems could result with EPA's semiannual reporting requirements under the
proposed Section 61.163.
2.9.2 Response: The administrative problems referred to in this
comment have not been specified. However, it is the EPA's conclusion that
the reporting, recordkeeping, and other requirements contained in the
standards are both necessary to the implementation of the regulation and
reasonable in their impact on the glass manufacturing industry and
individual furnace owners and operators.
2.10 ARSENIC EMISSIONS FROM SODA-LIME FURNACES
2.10.1 Comment: Three commenters (D-23, D-24, D-25) addressed the
issue of whether the temperature of the gas exhausted from furnaces melting
soda-lime glass affects the percentage of arsenic emitted in the solid
phase, and the three options proposed by EPA for controlling arsenic
emissions from soda-lime glass furnaces.
One commenter (D-25) stated that data from one of the commenter's
soda-lime furnaces indicate that the percentage of arsenic in the solid
phase increases with increasing stack gas temperature. He noted that the
proportion of arsenic found in particulate from this furnace varied widely
from a low of about 50 percent to a high of 99 percent. The commenter
offered no explanation of why these data conflicted with the data gathered
by EPA, but felt that there may be some complex chemical reactions taking
2-42
-------
place in the melting furnaces which are not at present understood. Data
obtained from another of the commenter's plants (melting an "all other"
glass type) have also shown a trend of increasing solid-phase arsenic with
increasing stack gas temperature. For 23 representative samples collected
on this furnace, from 30 to 100 percent of the total arsenic was emitted as
particulate. The commenter concluded that temperature alone is not the only
factor affecting the level of arsenic control achievable from glass melting
furnaces.
The commenter (D-25) also responded to the options proposed by EPA for
controlling arsenic emissions from soda-lime glass manufacturing furnaces.
According to the commenter, a stack gas temperature of 250°F is at, or
below, the acid dew point for most exhaust streams and would lead to severe
corrosion of control equipment. Corrosion; problems would decrease the
on-line availability of control devices leading to higher emissions of
particulate arsenic than would occur if the temperature of the exhaust gas
was maintained at a higher temperature. The commenter stated that dry
scrubbing is not, in practice, a reliable technology. In the commenter's
opinion, EPA should not require cooling of exhaust gas to 250°F at the
present time without at least first demonstrating the technology on a pilot
scale. The commenter also noted that the dew point of exhaust gases from
glass furnaces is constantly changing, and cannot be accurately monitored on
a continuing basis. Also, temperatures profiles across an emission control
device are not uniform. Therefore, maintaining the temperature of the
exhaust gases at the inlet of the control device at 10° to 20°C above the
dew point is not practical, and would not ensure that the temperatures
within the control devices would not fall below the dew point. According to
the commenter, a control option based on maintaining exhaust gas
temperatures at 10° to 20°C above the acid dew point demands more from
technology than current technology can provide. The commenter concluded
that the reduction in arsenic emissions from soda-lime furnaces achievable
with ESPs or fabric filters would, in practice, be much better than theory
would suggest, and recommended that EPA test this hypothesis on a pilot
scale before selecting any other regulatory option.
2-43
-------
Another commenter (D-23) challenged the validity of the data presented
by EPA in support of the options being considered for control of inorganic
arsenic emissions from soda-lime glass furnaces. The commenter noted that,
based on a concern over the data and information presented in the March 20,
1984 Federal Register notice, additional information was requested from, and
provided by EPA. After receiving and thoroughly reviewing this additional
information, the commenter concluded that the test data used by EPA were
faulty, and also misrepresented in the March 20, 1984 notice in the Federal
Register. Specifically, the commenter stated that Figure 1 was not a graph
showing the percentage of arsenic in the solid-phase, but rather a graph
showing normalized data produced from inaccurate test data. The commenter
also noted that a meeting was held with EPA on April 12, 1984 to review the
information provided by EPA. On the basis of that meeting, it was concluded
by the commenter that the data were sufficiently flawed such that the
conclusions presented by EPA could not be drawn. The commenter stated that
the total measurements of arsenic emissions in each of the three tests
showed a decreasing amount of total particulate, and that this could only be
the case only if the sampling train failed to collect material. [The
Administrator assumes that this commenter meant to refer to "total arsenic"
rather than "total particulate".] The commenter found this position
difficult to accept based on EPA's contention that the sampling train is 99
percent efficient in capturing gaseous and particulate emissions.
The commenter stated that if an arsenic emission limitation is to be
placed on soda-lime glass melting operations, then it should be based on the
effects of emissions on ambient air quality and on risks to public health.
The method for meeting an established emission limitation should be left up
to glass manufacturers. In conclusion the commenter urged EPA to review its
earlier position that at least 90 percent control of arsenic emissions can
be achieved by control equipment, a position that the commenter believes to
be correct. The commenter also contended that sufficient time and resources
have already been committed to the study of soda-lime glass melting
operations and that further activities are unnecessary.
2-44
-------
A third commenter (D-24) stated that the data presented by EPA
demonstrate that emissions of participate arsenic increase sharply as the
temperature of the furnace exhaust gases decreases. This shows that
substantial additional control of arsenic emissions can be achieved by
cooling the gas stream prior to the particulate control device. Therefore,
in reviewing the three options considered by EPA for controlling arsenic
emissions from soda-lime glass furnaces, the commenter stated the belief
that the option of not requiring any cooling should be rejected. The
commenter noted that the drawback to the second option (require cooling to
10° to 20°C above the dew point of the furnace exhaust gases) is that the
cooling requirement would need to be tailored to each facility which would
increase the task of implementation and enforcement. This option is still
preferable to requiring no cooling, however. The commenter expressed the
belief that the best option is to require cooling of the furnace exhaust
gases to 121°C, or to an even lower temperature. However, the commenter
challenged EPA's conclusion that cooling to this temperature would mean that
"most" operators would need to install dry scrubbers upstream of the
particulate control device. The commenter noted that EPA neither presented
nor referred to, any data to substantiate this claim and that the Background
Information Documents (BIDs) for smelters suggested that temperatures of
121°C or lower are tolerable. Nonetheless, the commenter supported the
option of requiring operators to reduce exhaust temperatures to 121°C, even
if scrubbers are necessary, because of the substantial reductions in
emissions of arsenic achievable through gas cooling.
2.10.1 Response: Since the three options proposed by EPA for
controlling arsenic emissions from soda-lime glass melting furnaces were
published in the Federal Register (49 FR 10278), the Agency has performed
two additional emission tests and has reviewed additional data provided by
industry representatives. In considering all of the available data, EPA
concludes that there is insufficient evidence to support a requirement that
the exhaust gas from soda-lime glass furnaces, or from furnaces producing
any other type of glass, be cooled to any specific temperature prior to
entering a particulate control device.
2-45
-------
The available data do indicate that arsenic emissions from some glass
melting furnace may occur less predominantly as particulate matter than was
previously believed, and that cooling can be effective in increasing the
proportion of total arsenic emitted as particualte matter. However, no
correlations have been identified between the proportion of arsenic emitted
as particulate matter and the type of glass produced, the type of melting
furnace used, the type of arsenic added to the raw materials, or any other
source characteristics. In addition, EPA does not have sufficient data to
conclude that cooling of furnace exhaust gases would be effective in
increasing the efficiency of a control device in all cases. Therefore, a
requirement that the exhaust gas from all affected furnaces be cooled to
some specific level prior to entering a cotnrol device could result in
increased costs with no guarantee that additional control would be achieved.
The Agency does believe, nonetheless, that both the format and the level of
the final standard are sufficient to ensure that furnace exhaust gases are
cooled in those instances where the effectiveness of control is dependent on
the operating temperature of the control device.
The results of the first test on a furnace melting soda-lime glass
showed that less of the total arsenic emitted from the furnace was in
particulate form compared to the previous tests (about 74 percent compared
to more than 90 percent) at the standard EPA Method 108 sampling temperature
of 121°C (250°F). In addition, samples taken simultaneously at three
different temperatures (121°C, 204°C, and 288°C) showed that the amount of
arsenic in the particulate matter generally increased as the filtered gas
was cooled from 288°C (550°F) to 121°C (250°F). However, the amount of
vapor-phase arsenic detected in these samples did not decrease in proportion
to the increase observed in particulate arsenic, and the total amount of
arsenic collected at 288°C (550°F) was uniformly less than the total amount
collected at a filtered gas temperature of 121°C (250°F). The results of
this test were also complicated by the fact that some of the filters used
during the test were later found to be torn. Because there was not a
decrease in vapor-phase arsenic emissions in proportion to the apparent
increase in particulate arsenic, no basis exists for concluding that cooling
2-46
-------
of the exhaust gases caused vapor-phase arsenic to condense and form
particulate arsenic. The Agency agrees with the commenter that the data
obtained from the first test on a soda-lime furnace are inconclusive, and
are insufficient to support a limit on the temperature of the gases at the
inlet of particulate control devices.
After publication of the notice in the Federal Register on March 20,
1984, a second arsenic emission test was performed on a soda-lime glass
melting furnace. No significant amounts of vapor-phase arsenic were found
in the emissions from this furnace regardless of the temperature of the
filtered gas. In all test runs, more than 99 percent of the total arsenic
was captured as particulate matter. Therefore, even if the results of the
first test on a soda-lime furnace had demonstrated a relationship between
temperature and the amount of arsenic emitted as particulate matter, this
relationship could not be generalized to all furnaces producing soda-lime
glass.
The EPA also performed emission tests on a glass melting furnace
producing an alumino-silicate glass. Although the furnace is not presently
equipped with a permanent control device, a pilot-scale fabric filter system
had been recently installed on the furnace. The test program included both
EPA Reference Method 108 and single-point sampling, as well as a series of
performance tests on the pilot-scale fabric filter. The results of these
tests did conclusively demonstrate that cooling of the furnace exhaust gases
caused gaseous arsenic to condense, and thereby increased the effectiveness
of the fabric filter in reducing arsenic emissions. When the temperature of
the exhaust gas was cooled to below 121°C (250°F), control efficiencies
ranged from about 75 percent to 97 percent and averaged about 87 percent.
When the temperature of the exhaust gas was maintained above 121°C (250°F),
control efficiencies ranged from about 58 percent to 82 percent and averaged
about 71 percent. The data also indicated that the effectiveness of cooling
is sensitive to the concentration of gaseous arsenic in the exhaust gas and
to the residence time of the gas stream at lower temperatures. However, the
data collected during these tests are not sufficient to correlate specific
temperatures to specific removal efficiencies.
2-47
-------
Although the promulgated standard does not require exhaust gas cooling,
EPA does believe that dry scrubbing is a demonstrated technology for control
of emissions from glass melting furnaces. At least two manufacturers offer
dry scrubbing systems for glass furnaces and eight dry scrubbing systems are
known to be installed on glass melting furnaces. Both manufacturers and all
eight plants were contacted for information on the performance, reliability,
and costs of these systems. The information obtained from these contacts
(Docket reference IV-B-14) indicates that dry scrubbing systems are a
demonstrated technology for removing potentially corrosive pollutants from
the exhaust gases of glass melting furnaces, and would enable cooling of the
exhaust gases to a temperature of 121°C (250°F).
2.11 ZERO PRODUCTION OFFSETS
2.11.1 Comment: Two commenters (D-24, D-25) responded to EPA's
proposal for revising the method of applying zero production offsets in
determining compliance with the proposed particulate emission limits. One
commenter (D-24) discouraged EPA from developing either tighter offsets,
which would place existing control equipment in jeopardy of being out of
compliance, or more complex offsets which would make it more difficult to
obtain a meaningful guarantee from vendors of new control equipment. The
commenter expressed the belief that the existing offset provisions are
sufficient to adequately protect the environment, and that reducing or
changing them will only result in miniscule reductions in emissions. At
worst, changing the zero production offset provisions will result in trivial
non-compliance situations, and uncertainty in engineering new control
devices. The commenter concluded that EPA should have more important
matters to pursue than worrying about a few tenths of a pound per hour of
particulate emissions from small glass melting furnaces with an existing, or
new, ESP or fabric filter control device.
One commenter (D-24) expressed support for EPA's analysis of the zero
production offset issue. The commenter stated that these offsets were
developed in an unrelated proceeding to suit a purpose which is neither
2-48
-------
controlling nor relevant to the development of a NESHAP for glass
manufacturing plants. According to the commenter, retention of the offsets
would allow higher emissions from some facilities than would occur if the
offset provisions were dropped. Therefore, the commenter stated support for
dropping the use of zero production offsets.
2.11.1 Response: The issue of zero production offsets would apply
only to a standard which set a limit on emissions from glass melting
furnaces specific to the amount of glass produced in the furnaces, however,
the format of the standard has been revised to require affected furnaces to
reduce arsenic emissions by a certain percentage. Therefore, the zero
production offset provisions of the NSPS for glass manufacturing plants are
not applicable to the requirements set forth in the final NESHAP.
2.12 RISK ASSESSMENT AND RISK MANAGEMENT
2.12.1 Comment: One commenter agreed with EPA's judgment that an
arsenic emission standard is needed for glass manufacturing plants. The
commenter characterized glass plants as sources of substantial arsenic
emissions. They opposed the idea that glass manufacturing plants emit less
than smelters EPA has proposed to exempt and should therefore not be
regulated. The commenter offered three reasons for not exempting glass
manufacturing plants from the regulation: (1) the commenter believed that
the proposed smelter exemptions were unlawful and unwise; (2) the commenter
contended that section 112 of the Clean Air Act requires EPA to set
standards for all arsenic source categories; and (3) the commenter saw the
opportunity to reduce or eliminate 37 Mg/yr of arsenic as substantial and
not to be ignored.
A contrary view was offered by two other commenters. They called
arsenic emissions from the entire glass manufacturing industry insignificant
because glass manufacturing contributed only 3 percent of the total arsenic
emissions. One commenter asked how EPA could conclude that 37 Mg/yr from
the entire glass industry represents an "unreasonable risk" while
2-49
-------
simultaneously finding that emission rates of 627, 172 and 140 Mg/yr from
individual copper smelters after control by the proposed standards are
reflective of an ample margin of safety.
2.12.1 Response: The Agency noted that the above commenters focused
their reasoning to regulate or not regulate glass manufacturing plants on
the arsenic emissions from the various plants and source categories. In an
approach that is sometimes used by the Agency, the commenters are using
emissions estimates as a surrogate for ambient concentrations and public
exposure on which risk assessment is based. However, surrogates must be
carefully applied and the underlying assumptions that make the use of
surrogates reasonable techniques must be carefully considered. For example,
when comparing emissions from different sources as a surrogate or substitute
to comparing public exposure, the commenter is assuming that similar
emissions cause similar ambient concentrations, that is, emission parameters
for and population distributions near each source or source category are
similar. The commenters' comparison of the arsenic emissions from copper
smelters to glass manufacturing plant emissions is a case of where the
surrogate approach breaks down. In this case, primary copper smelters emit
a large portion of their emissions from tall stacks and thereby create
relatively small concentrations around the plant. Unlike the smelting
industry, glass manufacturing plant emissions are coming from relatively low
stacks and are causing larger ambient concentrations per ton of emissions
than the primary copper smelters. And, indeed, EPA air dispersion modeling
has borne this out. So, EPA cannot determine the need to regulate or not
regulate the glass manufacturing plants by comparing glass plant emissions
to the primary copper smelters emissions; EPA analysis has shown that this
approach would lead to faulty decisions when protecting public health.
2.12.2 Comment: One commenter estimated that one of his glass plants
•3
(at Shreveport) contributes 0.16 ug/m arsenic averaged over 8 hours, and he
called this an insignificant environmental impact.
2.12.2 Response: Based on the Agency's calculations, the maximum
o
concentration to which people near this plant are exposed is 0.018 ug/m
averaged over several years. Although EPA has not calculated an 8-hour
2-50
-------
average concentration, it appears that the two estimates are reasonably
consistent when considering the difference in averaging times. The
Administrator believes that the risk associated with Shreveport plant's
arsenic emissions is low.
2-51
-------
-------
APPENDIX A
SUMMARY OF RESULTS OF EMISSIONS TESTING PERFORMED AFTER PROPOSAL
A.I SUMMARY OF EMISSION TESTING ON A SODA-LIME FURNACE
A.1.1 Introduction
Emission tests were conducted on a soda-lime glass melting furnace in
July 1983. Arsenic was being added to the batch raw materials in the form
of powdered arsenic trioxide. Emissions from the furnace were not
controlled. Two different types of emission tests were performed. First,
in order to evaluate whether the temperature of the furnace exhaust gases
has an effect on the proportion of arsenic emitted in the solid phase, a
series of three single point tests were conducted. In the single point
testing, three sampling trains were operated simultaneously. In each train,
the filtered gas was maintained at a different temperature: one at 121°C
(250°F), one at 204°C (400°F), and one at 288°C (550°F). Second, four
modified EPA Method 108 tests were performed. The results of these tests
are presented below.
A.1.2 Summary of Single Point Tests
Table A-l summarizes the results of the single point sampling. The
values given for the S-l-A and S-2-A results are suspect because small tears
were identified in the filters used when these samples were taken.
Nonetheless, the proportion of solid-phase arsenic appears to increase as
the filtered gas temperature is cooled from 288°C (550°F) to 121°C (250°F).
In run S-3, a three-fold increase in solid-phase arsenic occurred as a
result of cooling the gas stream. The amount of vapor-phase arsenic present
did not, however, decrease in proportion to the increase in solid-phase
arsenic. In run S-3, vapor-phase emissions decreased by about half when the
filtered gas temperature was reduced from 288°C (550°F) to 121°C (250°F).
In runs S-l and S-2, the amount of vapor-phase arsenic detected remained
roughly constant at all temperature levels. Because the amount of
A-l
-------
TABLE A-l. SUMMARY OF SINGLE POINT TESTS ON A SODA-LIME FURNACE
Run ID
S-l-A
S-l-B
S-l-C
S-2-A
S-2-B
S-2-C
S-3-A
S-3-B
S-3-C
MEAN (S-A
MEAN (S-B
MEAN (S-C
Sample
Temp (°F)
250
400
550
250
400
550
250
400
550.
Arsenic
Solid
(gr/dscf)
0.00141
0.00144
0.00097
0.00217
0.00161
0.00069
0.00343
0.00230
0.00102
0.00234
0.00178
• 0.00089
Emissions
Vapor
(gr/dscf)
0.00153
0.00133
0.00155
0.00109
0.00090
0.00123
0.00074
0.00085
- 0.00155
0.00112
0.00103
0.00144
Total
(gr/dscf)
0.00294
0.00277
0.00252
0.00326
0.00251
0.00192
0.00417
0.00315
0.00257
0.00346
0.00281
0.00234
Solid
(%)
47.96
51.99
38.49
66.56
64.14
35.94
82.25
73.02
39.69
65.59
63.05
38.04
A-;
-------
vapor-phase arsenic detected did not decrease proportionally to the increase
in the amount of solid-phase arsenic detected as the filtered gas
temperature was decreased from 288°C (550°F) to 121°C (250°F), the total
amount of arsenic detected in the high temperature samples was uniformly
less than the total amount detected in the low temperature samples.
The percentage of the arsenic detected in the solid phase at a filtered
gas temperature of 121°C (250°F) ranged from a low of 48 percent to a high
of about 82 percent. The validity of the S-l-A and S-2-A values is
questionable, however, because of the tears identified in these filters. At
a filtered gas temperature of 288°C (550°F), from about 36 to 40 percent of
the total arsenic catch was collected in the front half of the sampling
train as particulate arsenic. Once again, the validity of these results is
questionable because the total amount of arsenic detected declined with
increasing filtered gas temperature.
Because of the inconsistencies in the data, and because the measured
concentrations of arsenic were so uniformly low, it is difficult to draw any
general conclusions from the results of this test. The data did appear to
indicate that a relationship exists between the temperature of the filtered
gas and the amount of arsenic captured in the probe and filter (particulate
arsenic). However, the data did not consistently indicate that the increase
in the amount of particulate arsenic captured as the temperature of the
filtered gas was cooled resulted from condensation of vapor-phase arsenic.
Therefore, other explanations for the trends observed in the data cannot be
completely discounted. For example, the collection efficiency of the entire
sampling train may have been somehow impaired during these tests when the
filtered gas was maintained at elevated temperatures.
A.1.3 Summary of EPA Method 108 Tests
Table A-2 gives the results obtained from the four EPA Method 108 tests
on the soda-lime furnace. The amount of total arsenic detected in the
solid-phase ranged from 66.9 percent to 83.6 percent. On the average, about
74 percent of total arsenic was captured in the front half of the sampling
trains. This was considerably less than the percentage of solid-phase
arsenic found in other tests. The other tests had indicated that more than
A-3
-------
TABLE A-2. SUMMARY OF EPA METHOD 108 TESTS ON A SODA-LIME FURNACE
Run ID
Vapor
(gr/dscf)
Arsenic Emissions
Solid Total
(gr/dscf) (gr/dscf)
Solid
p-1
P-2
P-3
P-4
0.00064
0.00105
0.00045
0.00116
0.00249
0.00218
0.00230
0.00234
0.00313
0.00323
0.00275
0.00350
79.55
67.49
83.64
66.86
MEAN
0.00083
0.00233
0.00315
74.38
A-4
-------
90 percent of the arsenic emitted from glass melting furnaces is in the
solid phase and collected by conventional participate control devices.
A. 2 SUMMARY OF EMISSION TESTING AT FOSTORIA GLASS COMPANY
A.2.1 Introduction
Emission tests were conducted at Fostoria Glass Company from
October 11-13, 1983, and from November 1-3, 1983. During the October tests,
arsenic was being added to the batch raw materials in the form of powdered
arsenic trioxide. For the November tests, liquid arsenic acid was
substituted. Both EPA Method 5 (particulate emissions) and EPA Method 108
(arsenic emissions) were employed. In addition, three single point samples
were taken during both the October and November tests. During each single
point run, two sampling trains were used simultaneously. The temperature of
the filtered gas was maintained at 121°C (250°F) in one, and at 288°C
(550°F) in the other. The objective of the single point tests was to test
the effect that temperature has on the phase composition of the emitted
arsenic.
Fostoria Glass Company produces a lead-crystal glass in a single
melting furnace. Arsenic is added to the batch as a refining agent in the
form of powdered arsenic trioxide under normal operating conditions.
Emissions from the furnace are controlled by an electrostatic precipitator
(ESP).
A.2.2 Summary of Single Point Tests
Table A-3 provides a summary of the results from the single point
tests. Runs S-l, S-2, and S-3 were performed in October when arsenic
trioxide was being used; runs S-4, S-5, and S-6 were conducted while arsenic
acid was being used during November.
In each run, at both 121°C (250°F) and 288°C (550°F), over 99 percent
of the arsenic collected was in the solid phase. The highest concentration
of vapor-phase arsenic detected was 0.00023 gr/dscf in run S-3 at 121°C
A-5
-------
TABLE A-3. SUMMARY OF SINGLE POINT TESTS AT FOSTORIA GLASS COMPANY
Run ID
S-l-A
S-l-B
S-2-A
S-2-B
S-3-A
S-3-B
MEAN (S-A)
MEAN (S-B)
S-4-A
S-4-B
S-5-A
S-5-B
S-6-A
S-6-B
MEAN (S-A)
MEAN (S-B)
Sample
Temp (°F)
550
250
550
250
550
250
550'
250
550
250
550
250
Arsenic
Solid
(gr/dscf)
0.01890
0.02490
0.02580
0.02610
0.02090
0.02610
0.02187
0.02570
0.02620
0.02150
0.01930
0.01840
0.01510
0.01640
0.02020
0.01877
Emissions
Vapor
(gr/dscf)
0.00006
0.00004
0.00007
0.00008
0.00006
0.00023
0.00007
0.00012
0.00015
0.00010
0.00008
0.00007
0.00009
0.00007
0.00011
0.00008
Total
(gr/dscf)
0.01896
0.02494
0.02587
0.02618
0.02096
0.02633
0.02193
0.02582
0.02635
0.02160
0.01938
0.01847
0.01519
0.01647
0.02031
0.01885
Solid
(%)
99.67
99.86
99.72
99.68
99.72
99.12
99.70
99.55
99.43
99.52
99.59
99.65
99.39
99.56
99.47
99.58
A-6
-------
(250°F). This corresponds to 0.9 percent of the total arsenic detected in
this run. On average, 99.6 percent of the total arsenic catch was in the
participates.
Little or no difference was apparent in the total amount of arsenic
captured in the sampling train as a function of temperature. In runs S-l
and S-3 the total amount of arsenic detected was about 20 percent less at a
filtered gas temperature of 288°C (550°F) than it was at 121°C (250°F).
However, this difference is probably not significant given the small
concentrations of arsenic present. Similarly, the data did not indicate
that the temperature of the filtered gas had any significant effect on the
proportion of arsenic captured in the particulate matter. In all runs, more
than 99 percent of the arsenic was found in the solid phase regardless of
temperature.
There was no significant difference between the overall results
obtained when arsenic acid was used and when arsenic trioxide was used.
A.2.3 Summary EPA Method 108 Tests
Table A-4 summarizes the results of the EPA Method 108 tests. On
average, 0.021 gr/dscf of arsenic were emitted from the furnace and about
0.00025 gr/dscf were finally emitted to the atmosphere.
The overall efficiency of the ESP in controlling emissions of arsenic
ranged from 97.5 percent to 99 percent. An average efficiency of
98.7 percent was demonstrated. Control levels for solid-phase arsenic
ranged from a low of 97.5 percent in run P-l, to a high of 99.2 percent in
run P-4. The data also suggest that vapor-phase emissions were reduced
across the control device. For example, in run P-2 vapor-phase emissions
decreased by about 86 percent. However, because the outlet samples were
collected at the stack rather than at the outlet duct from the control
device, some condensation of arsenic could have occurred in the ducting or
in the heat exchanger which immediately preceded the stack.
As in the results from the single point sampling, there were no
indications that the form in which arsenic is added to the batch materials
has any effect on either controlled or uncontrolled arsenic emissions.
A-7
-------
TABLE A-4. SUMMARY OF ARSENIC EMISSIONS DATA FROM FOSTORIA GLASS COMPANY
Run ID
P-l
P-2
P-3
MEAN
P-4
P-5
P-6
Uncontrolled Emissions
Solid Vapor Total
(gr/dscf) (gr/dscf) (gr/dscf)
0.018
0.022
0.022
0.021
0.017
0.024
0.024
0.000010
0.000036
0.000054
0.000033
0.000016
0.000068
0.000080
0.0175
0.0224
0.0218
0.0206
0.0165
0.0236
0.0245
Controlled Emissions
Solid Vapor Total
(gr/dscf) (gr/dscf) (gr/dscf)
0.0004
0.0003
0.0002
0.0003
0.0001
0.0003
0.0002
0.000001
0.000005
0.000025
0.000011
0.000004
0.000012
0.000015
0.0004
0.0003
0.0002
0.0003
0.0001
0.0003
0.0002
Control
Efficiency
(%)
97.50
98.53
98.95
98.33
99.24
98.88
99.02
MEAN
0.021
0.000055 0.0215
0.0002
0.000010
0.0002 99.05
A-8
-------
A.3. SUMMARY OF EMISSION TESTING AT INDIANA GLASS COMPANY
A.3.1 Introduction
Emission tests for inorganic arsenic were conducted at Indiana Glass
Company over the period May 14 to 19, 1984. The furnace on which the tests
were performed produces a crystal glass from a soda-lime recipe. Arsenic
trioxide powder is added to the recipe to improve the clarity of the glass
and to eliminate undesirable coloration. Emissions from the furnace are not
controlled.
The purpose of the test was two-fold. One purpose was to gather
further data on the proportion of arsenic emitted from soda-lime furnaces as
particulate matter. As reported in Section A.I, an earlier test on a
soda-lime furnace indicated that emissions from furnaces producing soda-lime
glass might consist of relatively greater amounts of vapor-phase emissions
compared to those from furnaces melting other types of glass. In addition,
there were some indications from this test that cooling of the furnace
exhaust gases might be effective in increasing the amount of arsenic emitted
as particulate matter, and thereby increase the efficiency of conventional
control devices i'n reducing arsenic emissions from soda-lime glass melting
furnaces. However, the results from the previous test were difficult to
interpret because some of the filters used during this test had torn while
sampling was in progress.
The second purpose of the test was to determine the precision of the
proposed EPA Reference Method 108 in measuring emissions of inorganic
arsenic from glass melting furnaces. The precision of the method was
estimated by calculating the relative standard deviations (the standard
deviation expressed as a percentage of the mean value) of samples collected
in a four-train (quad) sampling apparatus. This system allows four trains
to sample simultaneously at essentially a single point in the stack which
reduces the effect of variations in the velocity and particulate profiles on
the sampling results. It also permits a statistically significant number of
samples to be collected in a short amount of time. A total of nine quad
train runs, representing 36 individual samples, were performed using EPA
Method 108 sampling and analytical procedures.
A-9
-------
The effect of temperature on the proportion of arsenic emitted as
participate matter was evaluated by maintaining the temperature of the
filtered gas in the single-point quad sampling train at temperatures of
204°C (400°F) and 288°C (550°F). Altogether, four test runs were conducted
at elevated sampling temperatures. In each run, the temperature of the
filtered gas in two of the sampling trains was maintained at 204°C (400°F)
and at 288°C (550°F) in the other two trains. During each run, a single
Method 108 sampling train was operated at a filtered gas temperature of
121°C (250°F) for reference purposes. At the completion of the single-point
sampling, three multi-point EPA Method 108 runs were conducted with a
standard sampling train. The results of these tests are presented below.
A.3.2 Summary of Single Point Tests
The results of the single-point sampling at elevated filtered gas
temperatures are given in Table A-5. During runs 11D, 12C, and 12D the
filter supports ruptured as a result of thermal decomposition of the silica
rubber gaskets. These ruptures allowed particulate matter to pass directly
into the impingers. However, in run 12C the rupture occurred in the final
minute of the 70 minute test run. Run 12D was terminated immediately after
the rupture occurred. Sampling continued after the gasket failed in run 11D
which invalidated the results from this run.
In every run, virtually all of the arsenic was emitted from the furnace
as particulate matter. The temperature of the filtered gas had no effect on
the proportion of arsenic captured as particulate. In each test run, less
than 0.05 mg of arsenic was detected in the back-half of the sampling train.
Because up to 0.05 mg of arsenic was detected in unused samples of the NaOH
rinse and HgO impinger solutions, after-sampling values of 0.05 mg or less
are considered insignificant. Therefore, no significant amounts of arsenic
were emitted from the Indiana Glass furnace in the vapor phase.
A.3.3 Summary of EPA Method 108 Tests
The results of the standard EPA Method 108 tests are presented in
Table A-6. As was evident from the results of the single-point sampling, no
significant amounts of vapor-phase arsenic were detected in the emissions.
The total arsenic concentrations measured while traversing the stack were
equivalent to the concentrations measured during single-point sampling.
A-10
-------
TABLE A-5. SUMMARY OF SINGLE POINT TESTS AT INDIANA GLASS
Run ID
10A
10B
11A
11B
12A
12B
13A
13B
MEAN
IOC
10D
11C
11D
12C
120
13C
13D
MEAN
10-RT
11-RT
12-RT
13-RT
Sample
Temp (°F)
400
400
400
400
400
400
400
400
400
550
550
550
550
550
550
550
550
550
250
250
250
250
Arsenic Emissions
Solid Vapor
(gr/dscf) (gr/dscf)
0.00343
0.00384
0.00359
0.00384
0.00354
0.00383
0.00358
0.00388
0.00369
0.00378
0.00412
0.00367
0.00372
0.00418
0.00341
0.00400
0.00384
0.00383
0.00387
0.00313
0.00379
0.00001
0.00001
0.00001
0.00002
0.00000
0.00001
0.00001
0.00001
0.00001
0.00001
0.00001
0.00000
0.00000
0.00001
0.00002
0.00002
0.00001
0.00000
0.00001
0.00000
0.00002
Total
(gr/dscf)
0.00344
0.00385
0.00360
0.00386
0.00354
0.00384
0.00359
0.00389
0.00389
0.00379
0.00413
0.00367
0.00372
0.00419
0.00343
0.00402
0.00385 '
0.00383
0.00388
0.00313
0.00381
Solid
99.7
99.7
99.7
99.5
100.0
99.7
99.7
99.7
99.7
99.7
99.8 .
100.0
100.0
99.8
99.4
99.5
99.7
100.0
99.7
100.0
99.5
MEAN
250
0.00366
0.00001
0.00366
99.8
A-ll
-------
TABLE A-6. SUMMARY OF EPA METHOD 108 TESTS AT INDIANA GLASS
Run ID
Arsenic Emissions
Vapor Solid Total
(gr/dscf) (gr/dscf) (gr/dscf)
Solid
CD-I
CD-2
CD-3
MEAN
0.00001
0.00001
0.00001
0.00001
0.00374
0.00359
0.00324
0.00352
0.00375
0.00360
0.00325
0.00353
99.73
99.72
99.69
99.72
A-12
-------
A.4 SUMMARY OF EMISSION TESTING AT CORNING GLASS WORKS IN MARTINSBURG, WEST
VIRGINIA ."...'
A.4.1 Introduction
Emission tests for inorganic arsenic were conducted by EPA at Corning
Glass Works in October 1984. The furnace on which the tests were performed
produces a pyro-ceramic composition which is derived from an aluminosilicate
batch recipe. Liquid arsenic acid (H2As04) is added to the recipe as a
fining agent. Emissions from the furnace were not controlled.
The purpose of the test was two-fold. One purpose was to gather
further data on the proportion of solid phase arsenic emitted from the
aluminosilicate glass furnace. Previous tests conducted by Corning
demonstrated that the proportion of arsenic emitted as particulate matter
varied significantly from test to test and this variability was apparently
independent of process parameters. The second purpose was to determine the
effect of sampling train filter temperature on the collection of particulate
arsenic.
The emission tests consisted of five separate single point runs using a
four train (quad) sampling system, and a single standard EPA Method 108
sampling train. Single point testing consists of simultaneous operation of
the four sampling trains at a single point in the offgas stream. During the
tests, the temperature of the filters in two of the single point sampling
trains was maintained at 204°C (400°F) and at 288°C (550°F) in the other two
trains. In addition, each of the two higher temperature sampling trains
(204°C and 288°C) had a back-up filter before the impingers. These back-up
filters were maintained at 121°C (250°F).
A standard EPA Method 108 sampling train with a filter temperature of
121°C (250°F), was also operated during each of the test runs for reference
purposes. Except for the third run, the standard reference tests and the
single point tests were conducted simultaneously.
During the same period, tests were also conducted by Corning Glass
Works on a pilot scale fabric filter system. The efficiency of the pilot
system in controlling arsenic emissions at varying operating temperatures
was determined.
A-13
-------
A.4.2 Summary of EPA Method 108 Tests
Table A-7 summarizes the results of the EPA Method 108 tests. Total
arsenic emissions ranged from 0.0325 to 0.0536 gr/dscf. On the average,
0.0436 gr/dscf of arsenic were emitted from the furnace. The proportion of
solid phase arsenic ranged from 29 to 45 percent. On the average, the solid
phase arsenic was approximately 39 percent.
A.4.3 Summary of EPA Single Point Tests
The results of the single-point sampling at elevated filtered gas
temperatures are given in Table A-8. The percentage of the arsenic detected
in the solid phase ranged from a low of 26 percent to a high of about
51 percent. Ten samples were collected at a filtered gas temperature of
204°C (400°F) and 9 samples were collected at 288°C (550°F). The average
solid phase arsenic collection using the 204°C (400°F) sample trains was
34 percent, while the average of the 288°C (550°F) trains was 36 percent.
This average was not statistically different from the average (39 percent)
of the Method 108 reference train.
However, while this indicates no effect of temperature on the
proportion of solid phase arsenic emitted from this furnace, substantial
amounts of arsenic were collected in the back-up filters maintained at 121°C
(250°F). As can be seen from Table A-9, the sampling trains with back-up
filters show a significant degree of condensation of arsenic between the
higher temperature filters and the 121°C filters. This discrepancy may
result due to the fact that condensation may be occurring in the sampling
train as a function of residence time in the sampling system, temperature
gradient of the sampling train, and/or the probe location. Therefore, in
instances where the concentration of vapor-phase arsenic is sufficiently
high to be sensitive to condensation effects, the sampling train may not
provide a reliable indication of the actual proportion of solid phase
arsenic in the furnace exhaust or the effectiveness of a control device in
reducing total arsenic emissions.
A-14
-------
TABLE A-7. SUMMARY OF EPA METHOD 108 TESTS ON CORNING/MARTINSBURG
ALUMINOSILICATE FURNACE
Run ID
Solid
(gr/dscf)
Vapor
(gr/dscf)
Total
(gr/dscf)
Solid
RT-1
RT-2
RT-3
RT-4
RT-5
0.01140
0.01043
0.02178
0.02081
0.02116
0.02108
0.02609
0.03181
» ;
0.02587
0.02737
0.03247
0.03652
0.05359
0.04664
0.04853
35.09
28.55
40.64
44.62
43.61
MEAN
0.01712
0.02644
0.04355
38.50
A-15
-------
TABLE A-8. SUMMARY OF SINGLE POINT TESTS ON CORNING/MARTINSBURG
ALUMINOSILICATE FURNACE
Arsenic Emissions
Run ID
S-l-A
S-l-B.
S-l-Ca
S-l-D
S-2-A
S-2-B
S-2-C
S-2-D
S-3-A
S-3-B
S-3-C
S-3-D
S-4-A
S-4-B
S-4-C
S-4-D
S-5-A
S-5-B
S-5-C
S-5-D
Filter
Temp (°F)
361
352
511
351
349
504
518
356
352
514
527
351
347
505
518
358
356
516
527
Solid
(gr/dscf)
0.01250
0.01571
0.01443
0.01632
0.01522
0.01500
0.01228
0.01456
0.01619
0.01390
0.01703
0.01967
0.01654
0.01654
0.01311
0.01518
0.01465
0.01241
0.01421
Vapor
(gr/dscf)
0.02719
0.03806
0.02358
0.01553
0.02825
0.02350
0.01492
0.03802
0.03934
0.03080
0.03159
0.02891
0.02948
0.02666
0.01958
0.03590
0.03560
0.03454
0.03287
Total
(gr/dscf)
0.03964
0.05377
0.03802
0.03190
0.04347
0.03850
0.02719
0.05258
0.05557
0.04074
0.04858
0.04853
0.04598
0.04321
0.03269
0.05108
0.05025
0.04695
0.04721
Solid
(%)
31.52
29.21
37.96
51.17
35.02
38.97
45.15
27.70
29.14
34.13
35.05
40.53
35.98
38.29
40.11
29.72
29.16
26.43
30.10
MEAN S-A
MEAN S-B
MEAN S-C
MEAN S-D
355
351
510
520
0.01565
0.01566
0.01446
0.01421
0.02911
0.03415
0.02889
0.02451
0.04475
0.04981
0.04235
0.03874
36.13
31.70
34.46
37.67
Run S-l-C is void due to excessive post-test leak rate.
A-16
-------
TABLE A-9. PERCENTAGE OF TOTAL ARSENIC COLLECTED ON FRONT HALF, GLASS
CONNECTOR AND BACKUP FILTER AT CORNING/MARTINSBURG FURNACE
Run ID
Filter
Temp (°F)
Solid (%)
S-l-A
S-l-D
361
511
75%
86%
S-2-A
S-2-D
351
518
88%
89%
S-3-A
S-3-D
356
527
91%
91%
S-4-A
S-4-D
351
518
94%
73%
S-5-A
S-5-D
358
527
95%
93%
A-17
-------
A.4.4 Summary of Pilot Scale Fabric Filter Performance Tests
The purpose of the pilot test was to investigate the effect of offgas
temperature and cooling techniques on the arsenic reduction performance of
the pilot scale fabric filter system. Furnace offgases were tapped from the
process stream prior to the main stack control damper. The hot exhaust
gases at 272°C to 318°C (522 to 605°F) were cooled by a liquid spray tower
or by air dilution prior to the fabric filter. Water and a mixture of
sodium hydroxide and water were the liquids used during the spray tower
cooling tests. Thermocouples located at the inlets and outlets of the spray
tower and the fabric filter provided gas stream temperature measurements
during the test runs. A total of 33 test runs were conducted by
simultaneously collecting samples upstream and downstream of the pilot scale
fabric filter system. All samples were collected using EPA Method 108
sampling procedures. Nine test runs were performed with baghouse inlet
temperatures between 129°C and 139°C (265 and 283°F). These nine tests are
referred to as high temperature runs. Twenty-four test runs were conducted
with baghouse inlet temperatures ranging from 92 to 111°C (198 to 232°F) and
these are referred to as low temperature runs.
The performance of the fabric filter was tested under 4 sets of
conditions. These are (1) water cooling with high temperature baghouse, (2)
water cooling with low temperature baghouse, (3) water and sodium hydroxide
cooling with low temperature baghouse, and (4) cooling by air dilution with
low temperature baghouse.
A.4.5 Summary of Fabric Filter Performance Data
The results of the pilot scale fabric filter performance tests also
indicate that the stack gas sampling protocols may not provide a reliable
indication of the proportion of arsenic emitted as particulate matter when
this proportion is sensitive to gas stream temperature. Table A-10 shows
the fraction of solid phase arsenic detected prior to the fabric filter.
These data were collected with EPA Method 108 sampling trains (filter
temperature of 121°C) and flue gas temperatures similar to those for EPA
tests. An average of 77 percent of the arsenic was found to be in the solid
A-18
-------
1—
o
LU
••3 =3:
0 H-
o^ ^^
Q. CD.
1— 00
CD z:
_1 CD
Q- OO
OO
CI5 I-H
rv* y
rD LU
CO
OO CJ
2: t— 1
HH ^-^
t— LU
o; oo
ec o;
;y ^^
-*^
OO Q
CO LU
-
i-H C£
1 <=C
•< ^.
LU ^3
_l CO
CD
i^C
1—
CU
i— re ti-
re r- u
•4-> 3 CO
o o -a
*~ T> "tT
s- en
D-^
o
"O 'f-
•r- E »
i — CU &^
O tO- — •
oo i-
^_^
0 t-
1— •!- O
re E co
OO 4-> CU T3
S O !/J • — .
CD I— i- S_
I-H S- S-
CD «a; en
fV- .
t—
O '-T
C_3 O 4-
z: -o -r- o
ZD T- E to
t— tu ~o
O to *^>,
oo s- s-
•=C en
^_^
E
O
•1—
+J
Q-
•r™
J^
O
O
O
co
CO
to
LO
o
•
CD
J^,
0
o
CD
cr>
'^f'
0
O
^-^
CU
1—
E en
i — i E
•r—
LU i—
Lu 0
O
O.O
E
CU S-
1— CU
•*->
0 3
CM
»— i
1
^"
^^
1*^*
O
CD
CM
OO
«-H
LO
O
CD
CTl
CD
CD
CD
CM
^H^
O
•
CD
en
E
-|J -i—
CU i—
t— O
E O
I-H O
Lu S-
Lu CU
4_>
Q. re
E 2
|— ~[^
O
3 re
0 2=
LO
r— 1
1
CO
i-H
CO
O
CD
*5J~
co
CM
to
CD
•
O
o
i-H
CD
CD
CM
LO
O
•
O
+J
a>
p«.
c
I-H C
o
LL. -r-
LL_ 4->
23
Q.C—
CU Q
1—
S-
S *r™
o
3
.Q.1—
E *>^
CU O
1—
jg •!—•
o <:
t— i
CM
, i
cr>
r- 1
LO
LO
O
CD
CO
to
f^^
CO
CD
CD
CM
i-H
CD
CD
LO
CM
CD
•
CD
•4^
CU
r—
E
I-H E
O
Lu -i-
Lu •}->
3
E *r""
CU ' — *
I—
S ••-
0«=t
CM
1
CM
CM
i— 1
^J-
O
CD
CO
to
[*^
CM
O
CD
O
t-H
O
*
o
r--.
i— i
CD
•
O
•4-3
cu
^— •
c**
i— i co
c-
U_ -r-
LU r—
0
Q. O
E 0
(^ s_
CU
^= 4->
en to
^
CM
1
to
CM
CO
^J-
o
CD
i-H
r-^.
(-H
CO
o
m
O
o^
CD
O
CD
CM
CM
CD
*
O
+-*
O)
1— "
E en
I-H E
•r-
LU i—
Lu 0
O
Q.O
E
CU i-
f— CU
S re
0 3
0
CO
i
CO
CM
^^
"%-
^
1^^
co
0
CO
o
CD
•=*•
CD
O
CD
to
CM
O
CD
4-3
CO
E
I-H E
O
Lu T-
1 1 ) ^
3
CLr—
CU O
t—
s ••-
—1
CO
CO
1
r—4
CO
CO
to
o
CD
f-^»
f"«*v
co
CD
O
^^
1—4
CD
CD
^«
co
0
CD
"Z-
f^
UJ
s:
A-19
-------
phase compared to only about 39 percent for the EPA reference Method 108
sampling trains and about 87.5 percent for the EPA single-point tests with
back-up filters.
Table A-ll presents a summary of the fabric filter performance data.
The initial six test runs were conducted while the temperature of the
control device was maintained at a temperature of about 138°C (280°F).
Control efficiencies for total arsenic during these runs ranged from about
58 percent to 82 percent, and averaged about 71 percent. Significantly
higher removal efficiencies were obtained during the following 15 test runs
(runs 7 through 21) after the operating temperature of the fabric filter was
lowered to about 104°C (220°F). The control efficiencies for total arsenic
emissions averaged about 90 percent during these 12 test runs. Although the
efficiency of the control device decreased during the next three test runs,
these were performed after the production rate of the furnace had been
decreased by about 20 percent.
For test runs 25, 26, and 27, the operating temperature of the control
device was increased to about 138°C (280°F). The data from these runs
indicated that the average amount of arsenic exiting the fabric, filters was
only slightly less than the amount measured at the inlet. The increase in
temperature may have caused the evaporation and subsequent entrainment of
some of the arsenic previously collected in the filters.
The results obtained from the final six test runs (all at low
temperatures) showed wide variability in arsenic removal efficiencies,
possibly as a result of further declines in the production rate of the
furnace.
Table A-12 shows the average control efficiencies achieved under
various operating conditions of both the control device and the melting
furnace. Significantly higher removal efficiencies were obtained when the
temperature of the fabric filter was maintained below 121°C (250°F)
regardless of furnace production rate or cooling method. In considering the
data obtained at low control device temperatures only, the production rate
of the furnace appeared to exert the greatest influence on arsenic removal
efficiencies. During the tests conducted with water spray cooling, average
A-20
-------
co
i—
co
LU
H-
LU
O
2: o
or uj
o --3
U- O
or or
uj a.
o_
i—
UJ _l
I— 1-1
__ i a.
i — t
U. CD
fv"
O ^3
I-H CO
or co
ca :z
- CD
or z:
•a: .-I
2: 2:
2: or
=3 O
oo o
•
r—l
t
•=c
LU
on
«=c
1—
O)
IT}
'(0 'rJ
&^ +•* O
O '1-
E h- +J
O S-
| » O
O
3
-o
S- -r-
•=t o c-~ -
O. CL> S«
(TS t/J
> s_
•*
3)
4_>
03 U
I— T—
^5 d •""""•*
O
o.
E!
o
(/I
(U
Q
d •
3 0
or z
co cn CM r-.
n • # •
O oo CO *d-
co cn cn cn
LO LO CM O
• • • •
O CM CM O
t*^ cn cn cn
co LO co cn
• • • •
i^» co cn CM
CM CO LO >5j-
r»~ CM co co
• • • •
cn LO p-^ co
r**« cn cn cn
co LO o r-.
1 — r— 1 CO CM
CM CM CM CM
cn
0) -4-> +J •!- +J
r— OJ O) t—
E r— r— O r—
l-H CO E CO CO E
E I— I E I— I O I — IE
U_ •!— •!— O
LUr— Ll_t— LUS- Ll_^i-
O U- O U_ Cl) U_ 4-»
Q. O O •»-> 3
E 0 0.0 Q. (0 Q.I—
1— r o
CO CO
o «*
• •
r-) O
LO CO
CO «H-
• •
•3jt" I ' S
cn cn
«>J- LO
CM CM
CM CM
-M -!->
CD U_ 4->
3 3
Q.I— Q.r-
0) C^ QJ O
h- 1—
S- S-
S -I- 3S -r-
O ^C O ^C
_1 _l
CM CM
1 1
cn CM
i-H CM
CM
•
CO
IO
LO
•
LO
CD
•
CM
oo
CM
•
^J-
i-H
1
CO
CO
CM
CD
^—
E
l-H CO
E
U_ -r-
U. r—
0
Q. O
E 0
H^ S-
O)
CO as
•r- 3
CM
1
LO
CM
CM «=C
• ""-"x^
co z:
cn
CM CM
• •
CO CO
i^ cn
O CM
• •
cn CM
co t — .
r*^- o
• •
CM IO
cn cn
cn et:
* — i °^^»
CM z:
+J +J
(U CD
"E co "E
I-H E >— i E
•r- O
U_ r— U_ •!—
i 1 Q i t ^J
O 3
0.0 Q-r—
-------
TABLE A-12. ARSENIC CONTROL EFFICIENCY OF CORNING/MARTINSBURG
PILOT FABRIC FILTER UNDER VARIOUS TEST CONDITIONS
Fabric Filter
Temperature
High
High
Low
Low
Low
Low
Low
Low
Furnace
Production Rate
High
All
High
Low
All
High
Low
All
Method of
Cooling
Water Spray
Water Spray
Water Spray
Water Spray
Water Spray
Air Dilution
Air Dilution
Air Dilution
No. of
Data Points
4
7
9
3
12
6
5
11
Average
Arsenic
Reduction
(*)
70.5
42.6
92.1
78.2
88.6
87.4
80.8
86.5
A-22
-------
removal efficiencies of about 92 percent were achieved at high furnace
production rates, compared to an average of about 78 percent at lower
production rates. Similarly, when cooling was achieved by air dilution the
average percent reduction obtained at high production rates was higher than
those obtained at low production rates. As can be seen in the data provided
in Table A-10, the concentration of particulate arsenic entering the control
device decreased with decreasing production rate while the concentration of
gaseous arsenic remained relatively constant. Therefore, at lower
production rates proportionally less arsenic entered the control system in
particulate form and was captured in the fabric filters. The concentration
of total arsenic in the gas leaving the control device remained constant at
all production rates. Cooling by air dilution also decreased the
concentration of arsenic at the inlet of the control device, although it did
not appear to affect significantly the proportion of particulate arsenic.
As a result, the results obtained during cooling by dilution air are not
significantly different from those obtained during water cooling.
A-23
-------
A.5 SUMMARY OF EMISSIONS DATA SUPPLIED BY INDUSTRY REPRESENTATIVES
A.5.1 Introduction
In response to requests from EPA, industry representatives supplied
additional data on arsenic emissions from glass melting furnaces. These
data are summarized in the following sections. Because EPA was not present
during these tests, and because in some cases the test methods employed do
not fully conform to EPA reference test methods, the results of the tests
cannot be fully evaluated.
For each test, the average temperature of the stack gas is given.
However, the temperature of the filtered gas in the sampling train was
maintained at about 121°C (250°F) in each of these tests. Therefore,
arsenic present in the stack gas as vapor-phase arsenic might condense into
particulate arsenic as the gas is cooled within the sampling train, and
thereby obscure any relationship between stack gas temperature and the
proportion of arsenic emitted as particulate matter.
A.5.2 Summary of Arsenic Emissions Testing on the Corning Soda-Lime Furnace
in Charleroi, Pennsylvania
A soda-lime glass recipe is melted in a furnace located at Corning
Glass Works' facility in Charleroi, PA. Emissions from the furnace are
uncontrolled and exhausted to two separate stacks. Each stack has been
tested five times for emissions of both solid- (probe and filter catch) and
vapor-phase (impinger catch) arsenic. Prior to 1976, powdered arsenic
trioxide was added to the batch. After 1976, the plant switched to liquid
arsenic acid.
The results of these five tests are shown in Table A-13. In most
tests, more than 99 percent of the emitted arsenic was found to be in
particulate matter. However, in the April 1983 test only 50 to 60 percent
of the arsenic emitted from the main stack was detected in the solid phase.
-------
'
LU
t— «
1
1
CO
1— 1
o
1 1 1
—I
Sg
<_>
CD
>— 1
z:
O
o
o
13
z:
i—
co
UJ
r—
CO
o
co
CO
t— 4
sc
UU
u
•^f
LU
CO
err
.
0£
j|
—i
CO
•
CO
T-H
1
•a:
LU
i
CO
•a:
h-T
•o
v— ' — *
^ ^^
o ^-^
CO
i — O J=
ro a.-.—
+•> ro X)
•r- .C
"o XI
CO i —
'!— *— *
r— (3-5
o —
co
, "^"
0 -C
ro ja
.*• ^d-
u
ro
CO -O 1-
•r- -C
i?~o xT
ro CO r—
•r— , — -^
•r-ro
X <1)
3 S-
< 3
W ••— •
S- U-
Ol 0
Q.-
f—
^.
«^
o
fl3 **" "H
4-> XJ S-
CO •!- J=
sr 'o xi
•i- CO i —
(O •---
z:
ro
(U
3
cn cn
cn cn cr»
A A A
r-H ' — I 1 — 1
O O O
0 3 O
V V V
o o --vi
r-H r-H "^
O O O
IO LO i-l
I— « CO T-H
cn en CP>
A A A
o o o
0 O 0
V V V
T-H CM T-H
«*'*«*
O O O
•— I •* O
r-.r-.vo
LO LO LO
J3 -Q -O
ll* ^i* ^^
CO OO 00
**^» **^^ ^1*11*
IO tQ f*«^
"x*»x*^\
oo ro oo
000
CO CO VO
cn cn co
VO VO ^"
O O T-H
o o o
co co cn
o O o
r»» LO co
oo cn cn
CO li-H 1— I
o o o
3 O O
O O r-.
CM CM <^-
O O O
. CM CO O
LO VO O
LO LO IO
VO CO O
en en co
co uo co
O O T-H
0 O O
t-- UO CM
vo vo un
000
o*> *~H r^»>
CO UO CO
.Q t~^ ft
CO OO OO
co oo oo
"""^ "V^ **»•»,, '
CO OO CO
"-N. 'S^. '^^.
r*^ r^* f^^
o o o
r-. co
uft vo
A A
2S
o o
J V
r— r*.
T-H T-H
0 O
vo vo
CO CO
A A
i-H r—4
O O
O O
v
vo vo
o o
0 0
vo O
CM O
CO CO
O r-H
un vo
T-H O
0 0
r-H T-H
r-H i-H
0 0
UO CM
O O
LO un
-O XI
CO CO
CO OO
*^^ "X^
CO OO
CVJ Cvj
*^ *^t*
o o
cn cn cn
cn cn en
A A A
CM CM CM
O 0 O
O O O
V V V
•a "a ~a
r-. co vo
co co co
o o o
cn cn cn
cn cn cn
A A A
T-H r— 1 T— 4
O O O
O O O
V V V
•y x) ~a
cn o o
r-H CM CM
000
CO r-H CO
•* CO -=f
un un in
CTI cn cn
cn cn cn
A A A
000
o o o
V V V
-o -o -o
co cn r-.
vo vo vo
000
un co vo
in un cn
o o o
T-H T-H T-H
u u u
("*•» p"^ I*1^
^•x. "**^ *"""•'*
co •* un
o o o
VO VO VO
o o o
r-» co r^
cn cn cn
A A A
CM CM CM
O O O
O O O
V V V
"X3 T3 "O
CO t~- CO
VO !"-» VO
O O O
cn c^ cn
cn cn cn
A A A
t— 1 r-H r—4
O O O
O O O
V V
-o -o -a •
vo co en
CM CM T-H
o o o
r^ cn cn
^d" CO CM
un un un
cn cn cn
cn cn cn
A A A
o o o .
o o o :
V V V
•a -a -o
CM *a- cn
^f un •*
000>
r-. T-H un
CO CO CO
cn cn cn
F
u u u
i^f »qj» 1^-
r^ i^*» r^«
^^ ^*^» "^^
vo ^o r^
CSJ CXJ Csl
^~ ^" v^*
o o o
o
•r—
C
e:
in
e^
0) O
ro ~O 1-
$- -r- Q-
O) X
Q.T3 O ••>
S(l_ *_• ^Ifc.
• |_ if.. ^|^
C:
o
CO O O
Cn-r- -r- i-
03 C C
-------
During the same test, more than 86 percent of the arsenic in the auxiliary
stack was emitted as particulate matter. In the test conducted in July,
1983, the proportion of arsenic collected as particulate matter ranged from
80 percent to 96 percent in the main stack, and from 87 to 98 percent in the
auxiliary stack. No explanation could be found for the variations in the
solid/vapor split during tests conducted at different times.
A.5.3 Summary of Arsenic Emission Testing on the Corning Aluminosilicate
Furnace in Charleroi, Pennsylvania
The results of an emissions test on Coming's aluminosilicate glass
melting furnace in Charleroi, PA are given in Table A-14. This furnace is
also equipped with two separate stacks. Liquid arsenic acid is used as a
raw material in this furnace. Arsenic emissions from the main stack were
sampled on April 21, 1983; the auxiliary stack was tested the following day.
Tests on the main stack showed that 95 percent of the arsenic is emitted in
the solid phase. A somewhat smaller fraction (about 90 percent) of the
total arsenic was found in the solid phase in the emission from the
auxiliary stack.
A.5.4 Summary of Arsenic Emission Testing on the Corning Aluminosilicate
Furnace in Martinsburg, West Virginia
An aluminosilicate glass composition is produced in the Corning/
Martinsburg furnace and arsenic is added to the batch raw materials as a
liquid acid. This furnace has been tested numerous times since 1978 for
emissions of arsenic. Table A-15 summarizes the results of these tests.
The fraction of arsenic captured in the particulate matter ranges from a low
of 29 percent to a high of nearly 100 percent. Significant variability is
apparent even for tests performed on the same day. During the test
performed on November 4, 1982, the proportion of arsenic captured as
particulate matter ranged from less than 30 percent to more than 60 percent.
Similar variability is apparent in the data obtained from the test on
November 30, 1983. Additional data supplied fay Corning (docket entry
A-83-08, IV-D-28) show that the fraction of arsenic emitted in the solid
phase did not correlate to the production rate of the furnace or the amount
of Gullet added to the fresh raw materials.
-------
TABLE A-14. SUMMARY OF ARSENIC EMISSIONS TESTING ON CORNIN6/CHARLEROI
ALUMINOSILICATE FURNACE
Date
4/21/83^
4/21/83°
4/22/83C
4/22/83c
Temperature9
(°F)
698
699
791
800
Solid
(Ib/hr)
0.82
0.88
0.67
0.80
Vapor
(Ib/hr)
0.04
0.05
0.08
0.09
Total
(Ib/hr)
0.86
0.93
0.75
0.89
Solid
95
95
89
90
. Temperature of stack gas
Main stack
Auxiliary stack
A-2 7
-------
TABLE A-15. SUMMARY OF ARSENIC EMISSIONS TESTING ON CORNING/MARTINSBURG
ALUMINOSILICATE FURNACE
Date
05/03/78
05/08/78
05/09/78
05/11/78
05/11/78
05/15/78
05/16/78
06/12/78
06/29/78
11/04/82
11/04/82
11/04/82
11/30/82
11/30/82
11/30/82
03/29/83
03/29/83
05/06/83
05/06/83
05/06/83
05/24/83
05/24/83
05/24/83
MEAN
Temperature3
(°F)
570
560
500
814
824
820
817
748
722
598
592
594
562
557
560
569
559
559
607
608
613
630
614
635
Solid
(Ib/hr)
10.40
10.90
9.60
6.57
8.33
7.74
8.23
4.18
2.43 .
1.15
1.09
2.23
1.14
0.91
1.17
1.95
1.95
2.43
4.12
4.19
4.85
4.58
2.92
4.48
Vapor
(Ib/hr)
0.19
0.02
0.27
1.10
1.30
1.30
1.20
1.80
1.50
2.88
2.06
1.41
0.80
1.70
1.29
0.58
0.33
0.76
0.56
0.54
0.86
0.55
0.44
1.02
Total
(Ib/hr)
10.59
10.92
9.87
7.67
9.63
9.04
9.43
5.98
3.93
4.03
3.15
3.64
1.94
2.61
2.46
2.53
2.28
3.19
4.68
4.73
5.71
5.13
3.36
5.50
Solid
(%)
98
100
97
86
87
86
87
70
62
29
35
61
59
35
48
77
86
76
88
89
85
89
87
75
Temperature of stack gas
-------
A.5.5 Summary of Data Provided by Owens-Illinois, Inc.
Emissions from four furnaces located at two different Owens-Illinois
plants have been tested for arsenic. All of these furnaces produce a
soda-lime glass. The results of these tests are summarized in Table A-16.
In four of the eight tests, no arsenic was detected in the vapor phase. In
one test, trace amounts of vapor-phase arsenic were detected, but were too
small to be quantified. However, in two other tests, significant amounts of
vapor-phase arsenic were found. The second test run on the Shreveport "C"
furnace indicated that only about 78 percent of the arsenic was emitted as
particulate matter. In the second test run on the Shreveport "A" furnace,
about 85 percent of the emitted arsenic was found in the particulate matter.
-------
TABLE A-16. SUMMARY OF ARSENIC EMISSIONS TESTING ON OWENS-ILLINOIS
SODA-LIME FURNACES
Date
06/27/83^
06/27/83°
06/29/83°!
06/30/83°
08/02/83®
08/03/83e
08/05/831
08/05/831
Temperature3
570
560
500
490
640
630
480
470
Solid
(Ib/hr)
0.15
0.22
0.08
0.25
0.05
0.06
0.02
0.02
Vapor
(Ib/hr)
NDC
0.06
Trace
0.04
NDC
NDC
NDC
NDC
Total
(Ib/hr)
0.15
0.29
0.08
0.29
0.05
0.06
0.02
0.02
Solid
100
78
<100
85
100
100
100
100
. Temperature of stack gas
Shreveport furnace "C"
^ None detected
Shreveport furnace "A"
I Toledo furnace "G"
Toledo furnace "D"
A *"»••*
,-i-OU
-------
APPENDIX B
UPDATED COST AND ECONOMIC ANALYSIS
B.I INTRODUCTION
In the Background Information Document (BID) for the proposed NESHAP
for the glass industry, the costs and the economic impacts associated with
the proposed standard were analyzed. Although at the time of proposal
information was available on the financial characteristics of affected glass
companies, sufficient data were not available to estimate the cost impacts
of controlling all arsenic-using glass furnaces operated by those companies.
There was also some uncertainty as to whether all arsenic-using furnaces had
been identified. Therefore, the capital and annualized costs of arsenic
emissions control were estimated for six model glass melting furnaces
(Chapter 6). Based on these cost estimates, the economic impact on
potentially affected glass companies was evaluated (Chapter 7).
In order to more accurately evaluate the economic impacts associated
with this NESHAP, detailed information was gathered after proposal on
potentially affected furnaces and the plants at which those furnaces are
located. This information enabled the costs associated with alternative
control options to be estimated for specific furnaces and specific plants.
On the basis of the updated cost analysis, the economic impacts of
alternative control options were reassessed.
This Appendix provides updated information on the costs and economic
impacts which were considered in establishing the promulgated standard. In
the following section, the changes made after proposal in the methodology
B-l
-------
for estimating costs and economic impacts are reviewed. Section B.3
summarizes the results of the updated cost analysis, and section B.4 reports
the results of the updated economic analysis.
B.2 CHANGES IN COST AND ECONOMIC ANALYSIS SINCE PROPOSAL
B.2.1 Changes in Cost Analysis
The capital and annualized costs associated with controlling emissions
of arsenic from glass melting furnaces are a function of the exhaust gas
flow rate through the control device and the level of control required.
At proposal, the costs of reducing total particulates to the levels
specified in the proposed NESHAP for glass manufacturing plants were
estimated on the basis of exhaust gas flow rates for model plants. However,
the cost estimates for the promulgated standard are based on actual flow
rates at existing furnaces. The required level of control is assumed to be
a reduction in particulate arsenic emissions of at least 92 percent for
soda-lime glass and at least 95 percent for other types of glass. On the
basis of data provided in the BID for the NSPS for glass manufacturing
plants, the specific collection areas (SCA) of ESPs necessary to achieve
these percent reductions were determined. For soda-lime glass, an SCA of
2
400 ft /l.OOO acfm was determined to be necessary to achieve a reduction in
particulate arsenic emissions of at least 92 percent. For other types of
glass, an SCA of 1,000 ft2/!,000 acfm would be required to achieve a
95 percent reduction in particulate arsenic emissions.
B-2
-------
Because the exhaust gas flow rate is dependent on numerous site-
specific factors including the capacity of the furnace, the type and
amount of fuel used, the type of glass produced, the temperature of the
exhaust gases, and other furnace operating parameters, actual flow rates
were obtained for all arsenic-using furnaces emitting more than 0.4 Mg
(0.44 ton) of arsenic per year. The actual flows were corrected to reflect
a control device temperature of 400°F and then multiplied by the required
SCA to determine the total plate area necessary to achieve the assumed
control level. Capital and annualized costs were then estimated on the
basis of the required plate area for actual furnaces using the same cost
algorithms presented in Chapter 6 of the proposal BID.
In the cost analysis performed prior to proposal, it was assumed that a
separate control device would be installed on each affected furnace.
However, the costs of ESPs and fabric filters do not increase proportionally
with increasing collection area. Therefore, at plants with multiple
furnaces in close proximity to one another, a single control device would
likely be installed to control emissions from multiple furnaces. The
locations of individual furnaces within potentially affected glass
manufacturing plants were evaluated with respect to the feasibility of
controlling multiple furnaces with a single control device. In cases where
it was determined technically feasible to do so, the costs of controlling
multiple furnaces with a single control device were estimated.
B.2.2 Changes in Economic Analysis
One objective of the economic analysis is to estimate the increase in
production costs that would be incurred in controlling emissions of arsenic
B-3
-------
from glass melting furnaces. The increase in production costs can then be
used to estimate either the decline in profits (cost absorption) or
increases in prices (cost pass-through) that would occur as a result. At
the time of proposal, no data were available on the actual costs of
producing specific types of glass within the pressed and blown segment of
the glass industry. Instead, production costs were estimated on the basis
of producer price indexes for various product categories. However, the
producer price indexes related costs of production to the quantity of
products shipped rather than to the weight of the products shipped. As a
result, rough estimates were made of the specific production costs
(i.e. $/unit weight of product) for various product types. In order to gain
a more accurate perspective on the increases in production costs associated
with a NESHAP for arsenic-using glass furnaces, individual companies were
contacted for information on their current production costs for specific
arsenic-containing glasses. This information was used in the updated
economic analysis.
The previous analysis was also based on the assumption that all of the
glass melted in a furnace was formed into a marketable product. In reality,
however, a percentage of the glass produced does not meet quality
specifications and is discarded as cullet. In the updated economic
analysis, it is assumed that 70 percent of furnace production is actually
sold as a final product. The 70 percent value was selected on the basis of
typical cullet ratios for glass melting furnaces which average about
30 percent.
In the previous economic analysis, both impacts on price and profit
were estimated. However, in reviewing the overall economic climate of the
B-4
-------
pressed and blown segment of the glass industry, it was determined that cost
absorption (profit reduction) would be more likely for all types of
products. Therefore, only impacts on profit were calculated in the updated
economic analysis, and profit ratios of indivuidual companies were applied to
furnaces of these companies.
Finally, the updated economic analysis is based on the estimated costs
of controlling specific furnaces rather than model furnaces. Accordingly,
profit impacts were calculated for specific furnaces with known production
rates of arsenic-containing glasses. The estimated decline in profit from
an individual furnace provides an indication of the probability that a
furnace would be shut down rather than equipped with a control device.
B.3 RESULTS OF UPDATED COST ANALYSIS
B.3.1 Number of Furnaces Considered
Control costs were estimated for the 11 uncontrolled furnaces that
currently emit more than 0.4 Mg (0.44 ton) of arsenic per year. In
addition, control costs were estimated for 13 uncontrolled furnaces which
individually emit less than 0.4 Mg/year (0.44 ton/year), but are either
located at the same plants as the higher emitting furnaces or at plants with
relatively high total plant-wide emissions of arsenic. These 24 total
furnaces are located at 6 separate glass manufacturing plants and have
combined arsenic emissions of about 28.8 Mg/year (31.7 tons/year).
B.3.2 Estimated Control Costs
The estimated capital and annual costs of controlling each of the
24 arsenic-using glass furnaces are given in Table B-l. Only costs for ESPs
B-5
-------
TABLE B-l. RESULTS OF COST ANALYSIS FOR UNCONTROLLED GLASS FURNACES
Plant
Code
1
1
2
2
2
3
4
5
6
6
a) At a
Furnace
Code
A
B
C
A,B,C
A
B
C
D
E
F
G
H
I
A,B,C,D,E
F,G,H,I
A
A
A
A
B
C
D
E
F
G
H
I
A,B,C,D
temperature of
b) Assumes^a specific
Glass
Type
Soda-lime
Soda-lime
Soda-lime
Soda-lime
Soda-lime
Soda-lime
Soda-lime
Soda-lime
Soda-lime
Soda-lime
Soda-lime
Soda-lime
Soda-lime
Soda-lime
Soda -lime
Al'-Silicate
Al -Silicate
Al -Silicate-
Soda -lime
Soda-lime
Soda-lime
Soda-lime
Soda-lime
Soda-lime
Soda -lime
Soda-lime
Soda-lime
Soda-lime
400°F.
collection area
Control
Device
Flowrate
(acfm)
18,081
17,727
33,939
69,747
7,842
9,124
6,086
12,965
14,557
17,899
15,960
6,103
10,125
50,574
50,087
25,800
25,696
18,993
12,961
10,296
10,296
8,842
7,025
7,994
6,904
6,420
6,420
42,395
-)
of 400 ft VI ,000
Estimated . ESP Capital
PlateJ\reaD
(ft2)
7,232
7,091
13,576
27,899
3,137
3,650
2,434
5,186
5,823
7,160
6,384
2,441
4,050
20,230
20,035
25,800
25,696
18,993
5,184
4,118
4,118
3,537
2,810
3,198
2,762
2,568
2,568
16,958
acfm for soda-1
Cost
($)
1,342,100
1,328,100
1,873,800
2,746,300
862,000
933,900
753,700
1,125,200
1,196,500
1,335,000
1,256,200
754,700
987,000
2,352,500
2,297,900
2,633,500
2,627,800
2,238,800
1,125,000
995,800
995,800
918,600
813,200
870,900
805 ,800
775,300
775,300
2,108,300
ime glass and
ESP Annua
Cost
($/year)
268,000
265,100
376,500
555,800
170,700
185,200
148,900
223,900
238,400
266,500
250,500
149,100
195,900
474,700
463,500
532,500
531,400
451,300
223,900
197,700
197,700
182,100
160,800
172,500
159,300
153,200
153,200
424,600
1000 ft /I,000 acfm for other glass types.
B-6
-------
were calculated. Except for the very smallest furnace sizes, the capital
and annualized costs for EPSs are somewhat greater than those for fabric
filter systems.
As mentioned above, the costs assume that furnaces producing soda-lime
glass are required to reduce particulate arsenic emissions by 92 percent,
and that furnaces producing other types of glass are required to reduce
particulate arsenic emissions by 95 percent. At plants where it would be
technically feasible to do so, the costs of installing a single device to
control emissions from multiple furnaces have been estimated in addition to
the costs of controlling the furnaces individually. The costs of combined
control systems are, in all instances, lower than the costs for controlling
each furnace separately. For example, the annual cost of controlling the
three furnaces at plant 1 would be about $556,000 if a single control device
were installed, compared to a total annual cost of about $910,000 for
controlling these furnaces separately.
B.4 RESULTS OF UPDATED ECONOMIC ANALYSIS
Since the arsenic NESHAP for glass manufacturing plants was proposed on
July 20, 1983, additional information has been collected on companies that
operate arsenic-using furnaces. Therefore, the purpose of this section is
to supplement the analysis of economic impact that was presented in the BID
for the proposed standard.
Currently, three companies operate six plants having furnaces with
potential arsenic emissions of more than 0.4 Mg (0.44 ton) per year.
Twenty-four arsenic-using furnaces are located at these six plants. The
three companies are Corning Glass Works, Owens-Illinois, Inc., and Lancaster
B-7
-------
Colony Corporation. Background financial data for Corning Glass Works and
Owens-Illinois, Inc. are presented in Tables 7-10 and 7-14, respectively, of
the proposal BID and are not repeated here. Background financial data for
Lancaster Colony Corp. are given in Table B-2.
B.4.1 Maximum Percent Profit Reductions
Estimated maximum profit impacts are based on the assumption that
companies would be unable to increase their prices for products to cover the
costs of installing and operating an emissions control device. If product
prices are not increased and net sales remain constant, profits before taxes
will be reduced. Therefore, the assumption of complete cost absorption
presents a "worst case" from the standpoint of glass manufacturing
companies.
The extent to which profit would be reduced can be estimated by
comparing the bas.eline (i.e. before control) return on sales to the return
on sales after emissions controls are installed. Return on sales is defined
as profit before tax divided by total revenue (profit ratio). Before tax
return on sales is used instead of after tax return on sales to avoid the
complication of varying tax rates.
Financial data for glass companies are generally reported according to
major segments of the business. For the glass industry, these business
segments generally correspond to broad categories of products. However,
within any overall product category, for example machine-made consumer
glassware, a wide range of different individual products may be included.
Therefore, profit ratios derived from financial data for these broad
business segments may not precisely indicate the margin of profit on a
B-8
-------
TABLE B-2. FINANCIAL CHARACTERISTICS OF LANCASTER COLONY CORPORATION'
($105)
Year
Company
Total
Machine-made
Consumer
Glassware
1979
1980
1981
1979
1980
1981
1979
1980
1981
1979
1980
1981
Average
1979
1980
1981
Average
1979
1980
1981
Average
Sales
267.5
291.1
307.7
Profits Before Taxes
20.6
12.9
19.6
Assets
174.6
172.1
182.1
Return on Sales
(Percent)
7.7
4.4
6.4
6.2
Before Tax Return on Investment
(Percent)
10.0
Sales to Assets
(Ratio)
1.53
1.69
1.69
1.63
95.6
111.2
115.9
9.1
8.1
11.1
60.9
65.3
65.5
9.5
7.3
9.6
8.8
14.9
12.4
16.9
14.7
1.57
1.70
1.77
1.68
B-9
-------
particular arsenic-containing glass produced in a particular furnace.
Nonetheless, more disaggregated data on before tax profits and total
revenues are not publicly available, and the available data are considered
sufficient for purposes of this analysis.
The ratios of before tax profits to sales used to estimate profit
declines are shown in Table B-3. These ratios were derived from the
financial data provided in Tables 7-10 to 7-17 of the proposal BID as well
as from the financial data for Lancaster Colony Corporation shown in
Table B-2. With the exception of machine-made consumer ware, the ratios
reflect unweighted industry-wide averages for each product category
(business segment). For furnaces producing machine-made consumer ware,
individual company-wide profit ratios were used.
Table B-4 summarizes the profit declines estimated for the 24 glass
melting furnaces under consideration. For each range of decline in profit,
the number of furnaces within that range are shown, both when it is assumed
that each individual furnace is controlled separately and when it is assumed
that multiple furnaces are controlled by a centralized emissions control
device. Assuming that all furnaces are individually controlled, profits
from 21 of the 24 furnaces would decline by 15 percent or more. This range
of profit decline is considered significant. Profit declines for 2 furnaces
would be moderate, in the range of 5 to 15 percent, and applying controls to
one furnace would have an insignificant impact of profit. However, when the
more realistic situation is considered in which single control devices are
applied, wherever feasible, to multiple furnaces, the overall impact on
profit would be moderate for most furnaces. In this case, significant
impacts on profit are estimated for 7 of the 24 furnaces.
B-10
-------
TABLE B-3. SUMMARY OF FINANCIAL RATIOS FOR PROFIT IMPACTS
Product
Before Tax
Profi ts
to Sales
(Percent)
Machine-made tableware
T.V. envelope tubes
7.5 to 8.3
11.7
The financial ratios are for the business segments in the firms producing
that product
B-ll
-------
TABLE B-4. SUMMARY OF PROFIT IMPACTS
Range of
Maximum Profit
Declines
(Percent)
Number of
Furnaces
(Separate Control)
Number of
Furnaces
(Multiple Control)
0-5
5-15
15-30
30-50
50+
TOTAL
1
2
10
6
5
24
1
16
1
1
5
24
B-12
-------
The estimated impacts on profit for Individual furnaces are given in
Table B-5. :
B.4.2 Capital Availability
The ability of the companies to finance the capital costs of emissions
control devices must also be considered. ; The ratio of long-term debt to
total capitalization (long term debt plus equity) provides a measure of
capital availability. A high ratio (expressed as a percentage) generally
indicates that additional debt financing may not be possible.
Table B-6 provides data on the ratio of long-term debt to total
capitalization for the three firms considered. Data are presented for a
period of four years in order to identify any trends in company debt ratios.
i . '
The ratios of long-term debt to total capitalization for Owens-Illinois,
Inc. and Lancaster Colony Corporation are more than double that for Corning
Glass Works. All. three firms show relatively stable debt ratios.
For purposes of this analysis, it was assumed that the costs of
installing emission controls are financed solely through debt. The impact
on capital availability of installing emissions control devices on arsenic-
using furnaces can be measured by comparing the debt ratio prior to control
to the debt ratio after control for each affected company. These ratios are
shown in Table B-7, along with the percent increase in debt ratio resulting
from installation of emissions control devices. Capital availability does
not appear to be a major constraint. The highest estimated percent increase
in debt ratio is for Lancaster Colony Corporation which increases by about
10 percent. However, for all three companies the ratio of long-term debt to
total capitalization after controls is within the range of debt ratios
experienced by these firms over the past four years.
B-13
-------
CO
UJ
JJ
2=
oi
u.
CD
U.
CO
UJ
1 1
_J
o
UJ
a
t— 4
U.
o
a_
UJ
CO
UJ
*
tn
i
CO
UJ
§
<4- CO
O •*-> C
CO <*- r—
en o o
C S- CU
ctj d. CD
tO CO -Q
in o r—
(O •!
i— S- *fl-
CO O-
>,
+•> i-
0 0
3 en
-0 CO
O •+"*
S- re)
a. o
. »
•a to
CO O
r— •«-> O
~O •r^'io^'
I— +J 3
U) C
UJC
D_
CO +J
UJ in
O
-o o
CO
fO to •%>*)•
•*-> a.
10 re)
UJ O
s.
I—
10
-
vo vo vo vo
co co co co
CD O O CD
CO CO CO CO
t- S— i- &->
(O rcJ (O fO
CO CO CO CO
_Q J/^ ,tf^) t^
(O fl3 (O fO
t-l— J- h-
O O CD O
CD CD O O
CD t-4 in co
co in vo in
vo vo !•••• in
CM CM CO in
O O CD O
o o o o
t— 1 •— 4 CO CO
CM OO CO VO
*J" ^^1 ^^ ^^*
co co co r*»*
i— t rH r-H CXI
03 d) 03 O)
^s _s _E _e
J3HJ3 H
co co co co
C_3
«=C CQ 0 CO
gj"
t— 4 t— 1
ooooooooo m in
cocoininincococoin «— i »-i
'0 ' ' ' ' 0 0 0 'o 0 '
uDinooc^Lninino in in
I— l^HCOCOCO.— l«-4l— 4CO
cococococococococo co co
ooooooooo o o
COCOCOCOCOCOCOCOCO CO CO
COCOCOCOCOCOCOCOCO CO CO
-Q-Q.a-Q-a-O.a-a_a J2 ^3
fO (O (& (d rt3 *O fO fO r~- in
o in co co co vo o en m «* co
t^*» co **J" CM co vo in ^d* en r*^ vo
^-Ir-li— ICMCMCMCMi-l'-l «* •*
OCDOCDCDCDCDCDCD CD CD
OOOOOOOOO O O
oenr~»CMinoc\]r-o in en
CMcocoinvoinvo«a-t~> CM i*^
vocoincMcncoinmco in en
cocnr»«r-<»-4cocMr-^en co CM
i— 1 <— 1 t— < r-4 CM CM
COCOCOCOCOCOCOCOCO CO CO
cococococococococo co co
UJ
*
CD *~4
A A
ca co
A A
-------
o> o o
c i- o>
«3 0. 0
10 a> .a
«/J O i
03 -r-
o
o
oo
LU
O
CD O
f— M-> O
+-> E i— -
O -r- «-
H- •*-> =J
w c
LU C
O
C£
a.
a
Q.
00 4->
LU 10
O
T3 O
0)
iO 03 -fa^-
a.
Q.
ra
5
QJ
o
(O
(O
SI
OOOOC3C3OOO
oororotnooooo
i i i i t—i i—i i—i i—11—i
6-S&9&5&5A A A A A
in LO ir> o
t—I r—I r-> CO
in
co oo oo co co co co co co
o'o'o'o'o'o'ooo.
co
•
o
01(1)
0)0)
0)
O)
S-
oooooooo
oooooooo
CMt— Ir— It— «
O
O
to
CsJ
OOOOC3CDOOO
OOOOOOOOO
OOOOOCMO'>COfOCO
CvJOICTli— li— i
o
o
CO
A
§
evi
(LI
oooooooooooooooooo oo
ca
f\
•=c
B-15
-------
i TABLE B-6. RATIO OF LONG-TERM DEBT TO TOTAL CAPITALIZATION (PRE-CONTROL)'
1978
1979
1980
1981
Average
1978
1979
1980
1981
Average
1978
1979
1980
| 1981
Average
Corning
Glass
163.4
147.1
153.6
200.1
166.0
905.1
975.1
1,066.3
1,165.5
1,028.0
18.1
15.1
14.4
17.2
16.2
Owens-
Illinois
Long-Term Debt
($ 106)
691.
697.
687.
618.
673.3
Total Capitalization
(Long-Term Debt Plus Equity)
1,797.
1,898.
2,003.
2,033.
1,932.8
Ratio of Long-Term Debt to
Total Capitalization
(Percent)
38.5
36.7
34.3
30.4
34.8
Lancaster
Colony
31.4
37.1
44.0
40.9
38.4
93.8
106.1
116.8
122.3
109. 7 1
33.5
35.0
37.7
33.4
34.9
B-16
-------
TABLE B-7. RATIO OF LONG-TERM DEBT TO TOTAL CAPITALIZATION (POST CONTROL)
Company
Ratio of Long-Term
Debt to Total Capitalization
Before After Percent
Controls Controls Change
Corning Glass
Lancaster Colony
Owens-Illinois
16.2
34.9
34.8
16.8
38.4
35.1
3.7
10.0
0.9
B-17
-------
B.4.3 Regulatory Flexibility
The Regulatory Flexibility Act (RFA) of 1980 requires that differential
impacts of Federal regulations upon small business be identified and
analyzed. The RFA stipulates that an analysis is required if a substantial
number of small businesses will experience significant impacts. Both
requirements, substantial numbers of small businesses and significant
impacts, must to met to require an analysis. If either measure is not met,
then no analysis is required.
The Small Business Administration (SBA) definition of small business
for SIC 3229 is a company that employs 750 persons or fewer. Table B-8
shows recent employment levels for the three firms that will potentially be
affected by this regulation. All of the firms have more than 750 employees.
Therefore, none of the firms meets the SBA definition of small business and
no regulatory flexibility analysis is required.
B-18
-------
TABLE B-8. EMPLOYMENT IN AFFECTED COMPANIES, 19813"6
Company
Employment
Corning Glass
Lancaster Colony
Owens-Illinois
30,200
5,900
51,000
B-19
-------
-------
APPENDIX C
AVAILABLE INORGANIC ARSENIC EMISSIONS DATA FOR
EXISTING ARSENIC-USING GLASS PLANTS
Plant Furnace Glass Type
Existing Control
Device Type
Baseline Arsenic Emissions
(kg/hr) (Ifa/hr) (Mg/yr) (ton/yr)
1
2
3
4
5
6
7
8
9
10
11
A
B
C
A
A
B
A
B
C
D
E
F
G
H
I
A
B
A
A
A
A
C
A
B
C
D
E
A
B
Soda -Li me
Soda-Lime
Soda-Lime
Soda -Lime
Lead
Lead
Soda-Lime
Soda-Lime
So da -Li me
Soda -Li me
Soda -Li me
Soda-Lime
Soda -Li me
Soda-Lime
Soda -Li me
Strontium
Lead
Lead
Lead
Aluminosilicate
Aluminosilicate
Fluoride/Opal
Various
Various
Various
Lead
Lead
Aluminosilicate
Lead
None
None ',
None
None
ESP
None
None
None
None ;
None
None
None
None
None
ESP
ESP
ESP
ESP
None
None
MC/ESP
ESP
ESP
ESP
ESP
ESP
None
FF
0.089
0.095
0.102
0.048
0.023
0.024
0.024
0.024
0.010
0.009
0.033
0.026
0.024
0.024
0.003
0.001
0.010
0.005
1.581
0.390
0.001
0.006
0.008
0.095
0.005
0.196
0.209
0.225
0.105
0.050
0.052
0.052
0.052
0.023
0.021
0.073
0.058
0.052
0.052
0.007
0.002
0.021
0.010
3.486
0.860
0.003
0.013
0.018
0.209
0.010
0.75
0.80
0.86
0.40
0.19
0.20
0.20
0.20
0.09
0.08
0.28
0.22
0.22
0.22
0.03
0.01
0.08
0.04
13.28
3.41
0.01
0.05
0.07
0.80
0.04
0.82
0.88
0.95
0.44
0.21
0.22
0.22
0.22
0.10
0.09
0.31
0.24
0.22
0.03
0.01
0.09
0.04
14.64
3.75
0.01
0.06
0.08
0.88
0.04
C-l
-------
Plant Furnace Glass Type
Existing Control
Device Type
Baseline Arsenic Emissions
(kg/hr) (Ib/hr) (Mg/yr) (ton/yr)
12
13
14
15
16
17
fe
19
20
21
22
23
A
A
A
A
B
C
D
E
F
G
H
I
A
B
C
D
A
A
B
A
B
A
A
A
B
C
A
B
Lead
Lead
Borosilicate
Soda -Li me
Soda-Lime
Soda-Lime
Soda-Lime
Soda-Lime
Soda-Lime
Soda -Li me
Soda -Li me
Soda -Li me
Soda-Lime
Soda-Lime
Soda -Li me
Soda -Li me
Soda-Lime
Soda-Lime
Soda-Lime
Soda-Lime
Soda -Li me
Soda-Lime
Soda-Lime
Soda-Lime
Soda-Lime
Soda-Lime
Soda-Lime
Soda -Lime
ESP
ESP
FF
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
0.002
0.023
0.023
0.269
0.162
0.162
0.113
0.046
0.107
0.040
0.006
0.006
0.048
0.048
0.048
0.048
0.000
0.008
0.008
0.000
0.000
0.006
0.019
0.011
0.011
0.011
0.002
0.002
0.005
0.050
0.050
0.593
0.357
0.357
0.249
0.102
0.232
0.089
0.013
0.013
0.105
0.105
0.105
0.105
0.000
0.018
0.018
0.001
0.001
0.014
0.042
0.024
0.024
0.024
0.004
0.004
0.02
0.19
0.19
2.26
1.36
1.36
0.95
0.39
0.90
0.34
0.05
0.05
0.40
0.40
0.40
0.40
0.00
0.07
0.07
0.00
0.00
0.05
0.16
0.09
0.09
0.09
0.01
0.01
0.02
0.21
0.21
2.49
1.50
1.50
1.05
0.43
0.99
0.37
0.06
0.06
0.44
0.44
0.44
0.44
0.00
0.08
0.08
0.00
0.00
0.06
0.18
0.10
0.10
0.10
0.01
0.01
C-2
-------
Plant
Furnace Glass Type
Existing Control
Device Type
Baseline Arsenic Emissions
(kg/hr) (Ib/hr) (Mg/yr) (ton/yr)
24
25
26
27
TOTAL
A
B
C
D
E
F
A
B
C
D
A
B
C
D
E
A
B
C
Soda-Lime
Soda-Lime
Soda-Lime
Soda-Lime
Soda -Li me
Soda -Lime
Soda-Lime
Soda -Li me
Soda-Lime
Soda-Lime
Soda-Lime
Soda-Lime
Soda-Lime
Soda -Li me
Soda -Li me
Various
Various
Various
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
0.003
0.003
0.003
0.003
0.003
0.003
0.002
0.002
0.002
0.002
: o.ooo
0.000
0.000
0.000
0.000
0.018
0.018
0.018
3.828
0.007
0.007
0.007
0.007
0.007
0.007
0.005
0.005
0.005
0.005
0.000
0.000
0.000
0.000
0.000
0.039
0.039
0.039
8.438
0.03
0.03
0.03
0.03
0.03
0.03
0.02
0.02
0.02
0.02
0.00
0.00
0.00
0.00
0.00
0.15
0.15
0.15
32.15
0.03
0.03
0.03
0.03
0.03
0.03
0.02
0.02
0.02
0.02
0.00
0.00
0.00
0.00
0.00
0.17
0.17
0.17
35.44
ESP = Electrostatic Precipitator; MC = Multiple Cyclons; FF = Fabric Filter
C-3
-------
-------
APPENDIX D
INORGANIC ARSENIC RISK ASSESSMENT FOR
GLASS MANUFACTURING PLANTS
-------
-------
QUANTITATIVE EXPRESSIONS OF PUBLIC CANCER RISKS FROM EMISSIONS OF
INORGANIC ARSENIC FROM GLASS MANUFACTURING PLANTS
D.1 INTRODUCTION :
D.1.1 Overview The quantitative expressions of public cancer risks presented
In this appendix are based on (1) a dose-response model that numerically
relates the degree of exposure to airborne inorganic arsenic to the risk of
getting lung cancer, and (2) numerical expressions of public exposure to
ambient air concentrations of inorganic arsenic estimated to be caused by
emissions from stationary sources. Each of these factors is discussed
briefly below and details are provided in the following sections of this
appendix.
D.I.2 The Relationship of Exposure to Cancer Risk
The relationship of exposure to the risk of getting lung cancer is
derived from epidemic!ogical studies in occupational settings rather than
from studies of excess cancer incidence among the public. The epidemic!ogical
methods that have successfully revealed associations between occupational
exposure and cancer for substances such as asbestos, benzene, vinyl chloride,
and ionizing radiation, as well as for inorganic arsenic, are not readily
applied to the public sector, with its increased number of confounding
variables, much more diverse and mobile exposed population, lack of consoli-
dated medical records, and almost total absence of historical exposure data.
Given such uncertainties, EPA considers it improbable that any association,
short of very large increases in cancer, can be verified in the general
population with any reasonable certainty by an epidemic!ogical study.
Furthermore, as noted by the National Academy of Sciences (NAS)1, "...when
there is exposure to a material, we are not starting at an origin of zero
cancers. Nor are we starting at an origin of zero carcinogenic agents in
our environment. Thus, It is likely that any carcinogenic agent added to
the environment will act by a particular mechanism on a particular cell
population that is already being acted on by the same mechanism to Induce
cancers." In discussing experimental dose-response curves, the NAS observed
D-l
-------
that most information on carcinogenesis is derived from studies of ionizing
radiation with experimental animals and with humans which indicate a linear
no-threshold dose-response relationship at low doses. They added that although
some evidence exists for thresholds in some animal tissues, by and large,
thresholds have not been established for most tissues. NAS concluded that
establishing such low-dose thresholds "...would require massive, expensive,
and impractical experiments ..." and recognized that the U.S. population
"...is a large, diverse, and genetically heterogeneous group exposed to a
large variety of toxic agents." This fact, coupled with the known genetic
variability to carcinogenesis and the predisposition of some individuals to
some form of cancer, makes it extremely difficult, if not impossible, to
identify a threshold.
For these reasons, EPA has taken the position, shared by other Federal
regulatory agencies, that in the absence of sound scientific evidence to
the contrary, carcinogens should be considered to pose some cancer risk
at any exposure level. This no-threshold presumption is based on the view
that as little as one molecule of a carcinogenic substance may be sufficient
to transform a normal cell into a cancer cell. Evidence is available from
both the human and animal health literature that cancers may arise from a
single transformed cell. Mutation research with ionizing radiation in cell
cultures indicates that such a transformation can occur as the result of
interaction with as little as a single cluster of ion pairs. In reviewing
the available data regarding carcinogenicity, EPA found no compelling
scientific reason to abandon the no-threshold presumption for inorganic
arsenic.
In developing the exposure-risk relationship for inorganic arsenic, EPA
has assumed that a linear no-threshold relationship exists at and below the
levels of exposure reported in the epidemiological studies of occupational
exposure. This means that any exposure to inorganic arsenic is assumed
to pose some risk of lung cancer and that the linear relationship between
cancer risks and levels t>f public exposure is the same as that between cancer
risks and levels of occupational exposure. EPA believes that this assumption
D-2
-------
is reasonable for public health protection in light of presently available
information. However, it should be recognized that the case for the linear
no-threshold dose-response relationship mbdel for inorganic arsenic is not
quite as strong as that for carcinogens which interact directly or in
metabolic form with DNA. Nevertheless, there is no adequate basis for
dismissing the linear no-threshold model for inorganic arsenic. At low
concentrations it is the Agency's belief that the exposure-risk relationship
used by EPA represents only a plausible upper-limit risk estimate in the
sense that the risk is probably not higher than the calculated level and
could be much lower.
The numerical constant that defines the exposure-risk relationship
used by EPA in its analysis of carcinogens is called the unit risk estimate.
The unit risk estimate for an air pollutant is defined as the lifetime cancer
risk occurring in a hypothetical population in which all individuals are
exposed throughout their lifetimes (about 70 years) to an average concentra-
tion of 1 ug/m3 of the agent in the air which they breathe. Unit risk
estimates are used for two purposes: (IKto compare the carcinogenic potency
of several agents with each other, and (2) to give a crude indication of
the public health risk which might be associated with estimated air exposure
to these agents.
The unit risk estimate for inorganic arsenic that is used in this
appendix was prepared by combining the five different exposure-risk numerical
constants developed from four occupational studies.2 The methodology used
to develop the unit risk estimate from the four studies is described in D.2
below.
D. 1.3 Public Exposure
The unit risk estimate is only one of the factors needed to produce
quantitative expressions of public health risks. Another factor needed
is a numerical expression of public exposure, i.e., the numbers of
people exposed to the various concentrations of inorganic arsenic. The
difficulty of defining public exposure was noted by the National Task Force
on Environmental Cancer and Health and Lung Disease in their 5th Annual
Report to Congress, in 1982. 3 They reported that "...a large proportion of
D-3
-------
the American population works some distance away from their homes and
experience different types of pollution in their homes, on the way to and
from work, and in the workplace. Also, the American population is quite
mobile, and many people move every few years." They also noted the necessity
and difficulty of dealing with long-term exposures because of "...the long
latent period required for the development and expression of neoplasia
[cancer]..." [The reader should note that the unit risk estimate has been
changed from that value used in the inorganic NESHAP proposal as a result
of EPA's analysis of several occupational epidemic!ogical studies that have
recently been completed.]
EPA's numerical expression of public exposure is based on two estimates.
The first is an estimate of the magnitude and location of long-term average
ambient air concentrations of inorganic arsenic in the vicinity of emitting
sources based on air dispersion modeling using long-term estimates of source
emissions and meteorological conditions. The second is an estimate of the
number and distribution of people living in the vicinity of emitting sources
based on 1980 Bureau of Census data which "locates" people by population
centroids in census tract areas. The people and concentrations are combined
to produce numerical expressions of public exposure by an approximating
technique contained in a computerized model. The methodology is described
in D. 3 below.
D.I.4 Public Cancer Risks
By combining numerical expressions of public exposure with the unit
risk estimate, two types of numerical expressions of public cancer risks are
produced. The first, called individual risk, relates to the person or
persons estimated to live in the area of highest concentration as estimated
by the computer model. Individual risk is expressed as "maximum lifetime
risk." As used here, the work "maximum" does not mean the greatest possible
risk of cancer to the public. It is based only on the maximum annual average
exposure estimated by the procedure used. The second, called aggregate risk,
is a summation of all the risks to people estimated to be living within the
vicinity (usually within 50 kilometers) of a source and is customarily summed
for all the sources in a particular category. The aggregate risk is expressed
D-4
-------
as incidences of cancer among all of the exposed population after 70 years
of exposure; for convenience, it is often divided by 70 and expressed as
cancer incidences per year. These calculations are described in more
detail in D.4 below.
There are also risks of nonfatal cancer and other potential health
effects, depending on which organs receive the exposure. Mo numerical
expressions of such risks have been developed; however, EPA considers all
of these risks when making regulatory decisions on limiting emissions of
inorganic arsenic.
D.2 THE UNIT RISK ESTIMATE FOR INORGANIC ARSENIC2
The following discussion is summarized from a more detailed description
of the Agency's derivation of the inorganic arsenic unit risk estimate as
found in EPA's "Health Assessment Document for Inorganic Arsenic (EPA-600/
8-83-021F).
D.2.1 The Linear No-Threshold Model for Estimation of Unit Risk Based on
Human Data (General)
The methodologies used to arrive at quantitative estimates of risk
must be capable of being implemented using the data available in existing
epidemiologic studies of exposure to airborne arsenic. In order to extrap-
olate from the exposure levels and temporal exposure patterns in these
studies to those for which risk estimates are required, it is assumed that
the age-specific mortality rate of respiratory cancer per year per 100,000
persons for a particular 5-year age interval, i, can be represented using
the following linear absolute or additive risk model:
a-j(D)=ai + 100,OOOa'D (1)
With this model, a, is the age-specific mortality rate per year of respira-
tory cancer in a control population not exposed to arsenic, a1 is a parameter
representing the potential of airborne arsenic to cause respiratory cancer,
and D is some measure of the exposure to arsenic up to the ith age interval.
For example, D might be the cumulative dose in years-ug/m3, the cumulative
dose neglecting exposure during the last 10 years prior to the ith age inter-
val, or the average dose in ug/m3 over some time period prior to the ith
age interval. The forms to be used for D are constrained by the manner in
which dose was treated in each individual epidemiologic study. At low
exposures the extra lifetime probability of respiratory cancer mortality
will vary correspondingly (e.g., linearly).
D-5
-------
The dose-response data available in the epidemiologic studies for esti-
mating the parameters in these models consists primarily of a dose measure
Dj for the jth exposure group, the person-years of observation Yj, the
observed number of respiratory cancer. deaths Oj, and the number Ej of these
deaths expected in a control population with the same sex and age distribution
as the exposure group. The expected number Ej is calculated as
zYjiaj/100,000 (2)
Ej =
i
where Yj,- is the number of person-years of observation in the ith age cate-
gory and the jth exposure group (Yj = z Yj,-). This is actually a simplified
i
representation, because the calculation also takes account of the change in
the age-specific incidence rates with absolute time. The expected number
of respiratory cancer deaths for the ith exposure group is
E(0j) = zYj, (a, + 100,OOOa'Dj)/100,000
« EJ + a'Yj°j (3)
under the linear absolute risk model. Consequently, E(0j) can be expressed
in terms of quantities typically available from the published epidemiologic
studies.
Making the reasonable assumption that Oj has a Poisson distribution,
the parameter a' can be estimated from the above equation using the method
of maximum likelihood. Once this parameter is estimated, the age-specific
mortality rates for respiratory cancer can be estimated for any desired ex-
posure pattern.
To estimate the corresponding additional lifetime probability of res-
piratory cancer mortality, let bi,...,bi8 be the mortality rates, in the
absence of exposure, for all cases per year per 100,000 persons for the age
Intervals 0-4, 5-9 80-84, and 85+, respectively; let ai,...,ai8 represent
the corresponding rates for malignant neoplasms of the respiratory system.
The probability of survival to the beginning of the ith 5-year age interval
is estimated as
n [1 - 5bj/100,000]
3=1
D-6
(4)
-------
Given survival to the beginning of age interval i, the probability of dying
of respiratory cancer during this 5-year interval is estimated as
5ai 7100 ,000 (5)
The probability of dying of respiratory cancer given survival to age
85 is estimated as aiQ/biQ. Therefore, the probability of dying of respir-
atory cancer in the absence of exposure to arsenic can be estimated as:
17 i-1
PQ • £ [5ai 7100 ,000) n (l-5bj/100,000)] (6)
i=l j=l
17
+(ai8/bi8) n (1 - 5 bj 7100,000)
j=l
Here the mortality rates a-j apply to the target population for which risk
estimates are desired, and consequently will be different from those in
(l)-(5), which applied to the epidemiologic study cohort. If the 1976 U.S.
mortality rates (male, female, white, and non-white combined) are used in
this expression, then PQ = 0.0451.
To estimate the probability PEP of respiratory cancer mortality when
exposed to a particular exposure pattern EP, the formula (6) is again used,
but a^ and bj are replaced by a-j(Di) and b-j(D.j), where D-j is the exposure
measure calculated for the ith age interval from the exposure pattern EP.
For example, if the dose measure used in (1) is cumulative dose to the be-
ginning of the ith age interval in ug7m3-years, and the exposure pattern
EP is a lifetime exposure to a constant level of 10 ug7m3, then D^ =
(i-1) (5) (10), where the 5 accounts for the fact that each age interval has a
width of 5 years. The additional risk of respiratory cancer mortality is
estimated as
PEP - PO (7)
EP
- P is
If the exposure pattern EP is constant exposure to 1 ug7m3, then P
called the "unit risk."
This approach can easily be modified to estimate the extra probability
of respiratory cancer mortality by a particular age due to any specified
exposure pattern. ,
D-7
-------
D.2.2 Risk Estimates from Epidemlologlc Studies
Prospective studies of the relationship between mortality and exposure
to airborne arsenic have been conducted for the Anaconda, Montana smelter
and the Tacoma, Washington smelter. Table D.I summarizes the fits of both
absolute and relative-risk models, with either k = 1 or k = 2, to dose-
response data from 4 different studies at the two smelters. (See the
"Health Assessment Document for Inorganic Arsenic," chapter 7, EPA-600/!
8-83-021F for detailed description of occupational studies.)
Table D.I also displays the carcinogenic potencies a1. It should be
noted that the potencies estimated from different models are in different
units, and are therefore not comparable.
In every case, a linear model (k = 1) fitted the data better than the
corresponding quadratic model (k = 2). In every case but two, the fits of
the quadratic model could be rejected at the 0.01 level. The two exceptions
involved the two smallest data sets (Higgins et al. absolute risk, and Ott
et al.) and in the former case the fit was very marginal (p = 0.017). On the
other hand, for each data set a linear model provided an adequate fit. Also,
in every case, an absolute-risk linear model fit the data better than the
corresponding relative-risk linear model. The p-values for the fits of the
absolute-risk linear model ranged from 0.025 to 0.75.
The estimated unit risk is presented for each fit for which the chi-
square goodness-of-fit p-value is greater than 0.01. The unit risks derived
from linear models—8 in all—range from 0.0013 to 0.0136. The largest of
these is from the Ott et al. study, which probably is the least reliable for
developing quantitative estimates, and which also involved exposures to penta-
valent arsenic, whereas the other studies involved trivalent arsenic. The
unit risks derived from the linear (k = 1) absolute-risk models are considered
to be the most reliable; although derived from 5 sets of data involving 4 sets
of investigators and 2 distinct exposed populations, these estimates are quite
consistent, ranging from 0.0013 to 0.0076.
To establish a single point estimate, the geometric mean for data sets is
obtained within distinct exposed populations, and the final estimate is taken
to be the geometric mean of those values. This process is illustrated in
Table D.2.
D-8
-------
u
1
oo
UJ
oo
j
^^
-y*
^£
v;
»— i
C£
UJ
P
r-
1—4
t—
O"
U_
O
^_
C£
*^*
j=
±5
OO
T— 1
UJ
_J
§
I
•P
to
CO
t—
4^>
•p—
U.
o
to
CO
c
, "O
§
w
CD
tj-
o
•p
f— •
^g
to
CO
^
v^
to «.
*Jjj
s
•r-
C
3
s
CO
"io
Q.
^^
•
•
3
°X
ro
O *—
p—
C ro
O >>
C O
i!
o a.
jy»
i_
CO
o
*—
CO
•o o
C S-
rO 3
O
^5
3 ro
•P +•>
OO ro
O
c
o
CO "P
to ro
O P-
O. 3
X Q-
UJ O
a.
•n
•p -p -p
•—.**" **~ **""
OO *l <4— ^4—
t O O O
00 U U U
CM i— i— <—
p-4 ^- p-4
O O O
to o o o
CM 0 0 0
o o o o
o o o o
—~
to
r*1* to to to
CM CO 00 O"l
t-4 VO CM l~~
• — '
l-» i-4 «*• 00
1 1 1 1
OO O"t O C71
«* o o o
CM CM OO* CM
i-4 CM i-4 CM
^ ^
to to
•1— »p-
S- S_
CO CO
•p >
3 ••-
O ro
CO P-
ja co
> CO
U. > -P
CO CO -r-
CO JC E
1 ~_~ Q
ro
-O 4- CO
C CO t-
0 4-> CO
O i— J*
lO ^~
t-5 CD CM to
t— 1 t-l
"~-
r-. t-i «* oo
vo <* r~> oo
oo r~ i— 1 1— i
CNJ i-H CO CM
rH CM r-4 CM
^ ^
to to
E si
CO CO
3 f
r- +»
O ro
V) r—
_a co
ra t-
ro
+J
CO
CO
c
*5>
•^
<-_
oo
1
to
CM
«— 1
^H
•Sf
O
«"~
C^
o
•^
•— •
vo
1
to
CTi
t-l
^
to
•r-
S-
co
3
*o
to
i*^
ra
3
J=
CJ
oB
C
o
CO
4^ •
3 -r-
r- +J
o n
10 i—
ja co
ro S-
1
J=
to
S-
^o
s:
OiJJ
<•-•»
CD O)
c 4-> 4->
*» t* »•
*^~ *t*~ *^~
CO H— 4— H—
1 O O O
vo u u o
• (O t3 fO
^** P"" ^^ l~~
00 CM
o o
O VO O
«* o o o
t-l O O O
o o o o
«*«*«*<•
o t-i to cy>
r— vo «* oo
CM r-l CM
*^' — *-.~*
r-- S «d- ?
till
to t-i oo vo
00 O t-l O
CO CM to t-4
t-l CM t-4 CM
.* .*
to to
•1— •!—
s- s-
CO CO
•p >
3 ••-
r- -P
^3 t5
to f—
j=i co
ro S.
J-
to
s-
ro
21 "en
oo ra
p—
CO
c s-
•^
^^
•— 1 —
1 ~^1_J
CM to
cr> t-i
t-4 CM
^
to
^
CO
^
•p-
40
ro
p—
CO
i.
co
S-
3
to
. o
i— 0-
ra X —
CO T5
•P CO
CO J= -P
•p j= 'i
o«— o
to
s_
CO
&-
o
CO 2
•o
•i- CO
0 S.
•P 4<^
to o
CO
P
CO T-
CO P—
to
•o °
CO t|-
to
p— *p-
CO P-
•§5
0
c S
o
CL S-
3 CO
u
cn c:
c
S-3
^3 fl3
S-
"•p-
CO O.
•p to
•p- CO
c J-
3
t-r
•P O
c:
CO *"*
i_ to
CO -P-
t|_ S-
tp.
•p- CO
•a e
c 5
•p- CO
t4—
CO t-
S. r—
ro
1^
to
. 1
VO CL
o -o
cn
1
0
-------
TABLE D.2 COMBINED UNIT RISK ESTIMATES FOR ABSOLUTE-RISK LINEAR MODELS
Exposure Source Study
Unit Risk
Geometric
Mean Unit
Risk
Final Estimated
Unit Risk
Anaconda smelter
ASARCO smelter
Brown & Chu 1.25 x 10~3
Lee-Feldstein 2.80 x 10'3
Hlgglns et al. 4.90 x 10-3
Enter!Ine &
Marsh 6.81 x 10-3
7.60 x ID'3
2.56 x ID"3
4.29 x 10-3
7.19 x 10-3
D.3 QUANTITATIVE EXPRESSIONS OF PUBLIC EXPOSURE TO INORGANIC ARSENIC
EMITTED FROM GLASS MANUFACTURING PLANTS
D. 3.1 EPA's Human Exposure Model (HEM) (General)
EPA's Human Exposure Model is a general model capable of producing
quantitative expressions of public exposure to ambient air concentrations
of pollutants emitted from stationary sources. HEM contains (1) an atmos-
pheric dispersion model, with included meteorological data, and (2) a
population distribution estimate based on Bureau of Census data. The only
input data needed to operate this model are source data, e.g., plant location,
height of the emission release point, and temperature of the off-gases.
Based on the source data, the model estimates the magnitude and distribution
of ambient air concentrations of the pollutant in the vicinity of the source.
The model is programmed to estimate these concentrations within a radial
distance of 50 kilometers from the source. If the user wishes to use a
dispersion model other than the one contained in HEM to estimate ambient
air concentrations in the vicinity of a source, HEM can accept the concen-
trations if they are put into an appropriate format.
Based on the radial distance specified, HEM numerically combines the
distributions of pollutant concentrations and people to produce quantitative
expressions of public exposure to the pollutant.
D-10
-------
D. 3.1.1 Pollutant Concentrations Near a Source
The dispersion model within the HEM is a climatological model which uses
a sector-averaged gaussian dispersion algorithm. The algorithm has been
simplified to improve computational efficiency.5 The algorithm is evaluated
for a representative set of input values as well as actual plant data, and
the concentrations input into the exposure algorithm are arrived at by
interpolation. Stability array (STAR) summaries are the principal meteoro-
logical input to the HEM dispersion model. STAR data are standard climato-
logical frequency-of-occurence summaries formulated for use in EPA models
and available for major U.S. meteorological monitoring sites from the
National Climatic Center, Asheville, N.C. A STAR summary is a joint
frequency-of-occurrence of wind speed, atmospheric stability, and wind direc-
tion, classified according to Pasquill's categories. The STAR summaries in
HEM usually reflect five years of meteorological data for each of 314 sites
nationwide. The model produces polar coordinate receptor grid points
consisting of 10 downwind distances located along each of 16 radials which
represent wind directions. Concentrations are estimated by the dispersion
model for each of the 160 receptors located on this grid. 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, 0.5, 1.0, 2.0, 5.0, 10.0, 20.0, 30.0, 40.0, and 50.0 kilometers. The
center of the receptor grid for each plant is assumed to be the plant center.
D. 3.1.2 Justification for 50 Kilometer Modeling
At proposal, exposure and risk were estimated for people residing
within 20 kilometers of the smelter. Some commenters pointed out that
since people beyond 20 kilometers are exposed to some level of arsenic due
to a source's emissions, EPA's proposal analysis underestimates the total
exposure and risk. EPA agrees with this comment and has expanded its
analysis out to 50 kilometers. There are several reasons for EPA to extend
its analysis out to 50 kilometers. When applying air dispersion models,
the EPA's modeling guidelines recommend that, because of the increasing
uncertainty of estimates with distance from the modeled source and because
of the paucity of validation studies at larger distances, the impact may
0-11
-------
extend out to 50 kilometers, but the analysis should generally be limited
to this distance from the source. ^ Such site-specific factors as terrain
features (complex or flat), the objectives of the modeling exercise, and
distance to which the model has been validated will determine the appropriate
distance (whether greater than or less than the guideline distance) for which
the Agency should apply the model.
Unless the technical or other considerations do not allow it, EPA has
decided to extend their hazardous air pollutant regulatory analysis for
inorganic arsenic out to 50 kilometers. The Administrator believes that
the potential to identify additional significant public exposure outweighs
the increased inaccuracies of applying the models beyond the previously
accepted 20 kilometer radius.
D. 3.1. 3 Methodology for Reviewing Pollutant Concentrations
Before making HEM computer runs, EPA reviewed U.S. Geological Survey
topographical maps (scale 1:24000) to verify locational data for each glass
manufacturing plant. Plants were given exact latitude and longitude values
which were then incorporated into the HEM program. Such information was
critical to the accuracy of the model in utilizing STAR data for each site.
The other major inputs for the HEM, emissions data, were also carefully
screened to insure accurate model predictions.
After completing the HEM runs, nearby monitoring sites with ambient
air quality data were identified by computer from the National Aerometric
Data Bank (NADB) (Table D. 3). In some cases, data for several years at
individual monitors were included in the NADB retrieval and were expressed
as average values with different sample sizes for the years monitored. In
these instances, weighted multi-year averages were calculated to provide an
overall mean for each monitoring site. For purposes of mean calculations,
values measured below minimum detection limits were considered as equal to
one half the detection limit. These ambient arsenic data were then compared
to HEM predicted values in order to gauge the accuracy of the air dispersion
model's estimates. As noted above, HEM predicted values were based on
concentrations calculated to occur at 160 polar coordinate receptor grid
D-12
-------
points consisting of 10 downwind distances located along each of 16 radials
which represented wind directions. Because the actual monitoring site
locations identified in the NADB retrieval usually did not correspond to
exact grid point Vocations, a log-linear interpolation scheme was used to
caluclate an HEM estimated concentration at the site (Figure D-l).
D. 3.1. 3.1 Use of Ambient Data
Certain criteria were considered in review of ambient levels. Mean
concentration values derived from smaple sizes of less than 25 data points
were disregarded. When reviewing the available monitoring data, it appeared
that monitors situated at distances greater than 15 km from the glass plant
were considered too far from the source to accurately measure emissions with-
out interference from other arsenic sources. Furthermore, at distances
greater than 15 km from the source, plant impacts were often predicted to be
significantly lower than minimum detection limits (Table D. 3). These data
were not incorporated in the analyses. A third consideration in reviewing
ambient data concerned the percentage of monitored data which fell below
minimum detection limits. Although some monitoring sites registered data
with over 90 percent of the values above minimum detection levels, many had
about half the data points below such levels. Instances where more than 50
percent of the data were below MDL were disregarded.
Although fifteen glass manufacturing plants were included in the
analyses, review of National Aerometic Data Bank data revealed that only four-
sites had monitors located within a 15 km radium of the source (Table D.I).
Three of these listed only one monitor - Glass Plant Numbers 1, 4 and 14.
Because concentration was expressed as a single weighted mean value for
each monitoring location, the result was a single data point. This was
insufficient information on which to run a regression analysis. The fourth
location, Glass Plant Number 6, cited four monitors inside 15 km. However,
this plant could not be analyzed due to the fact that between 95 and 99 per-
cent of the monitored concentrations were below analytical detection limits.
D-13
-------
flgur* D-l Craap 2 KC/EH lattraolatton
Gives:
A •» Uia a&fla Sa radians aubtaadtd elocfcvlat about tha aourea froa due south to the
•»*
BC/ED eantroid;
Al - the aaflo froa 4m ooutk to th* radial liaa taaodlataly eoontar-clockwis* of A,
ar fault* thraufli A if tfeara ia aa axaet Mtc&;
A2 • lha aagla froa tfoa aoutb to tha radial llaa faa«diataly eloekviaa of Al (42 ia 0
if it ia *•* aontb);
m »»• ilataaea in an froa taa aovrea to taa BC/CD caatraid;
«1 • &a iiataaea from taa aeorea to tka iart»st circular are of radiua iaa* than B;
»2 *iaa «iataaea fraa tha •oarea to tba ia^llaat cirealar are af *adf«a ftraatar than
•r «faal ta B; ._/' - *
CI » Aa aatoral lagaritta «f tba caaeaatratioa valsa at Cal.BDs
C2 • Tk« aataral logaritW »f taa caaeaatratioa valaa at C*l,B2)s
• .
C3 • Iaa aatural logarithm of tba eoeeaatratioa ^alna at (A2.B1);
S
D-14
-------
€4 « The Mtitral logarithm of the coaeeatreftion valve at CA2,ft2) ;
•
then: : • • •- •
VISIT » to(R/tl)/lB(ft2/Ri);
CC2-Ci)s%T!MP);
CZ » CAl •*• (CA2-CM)sAT»SP.
vhsrc CX is the interpolated cenceatratiee at the BC/EO centroid
D-15
-------
D. 3.1.4 The People Living Near A Source
To estimate the number and distribution of people residing within 50
kilometers of the glass manufacturing plant, the HEM model uses the 1980
Master Area Reference File (MARF) from the U.S. Bureau of Census. This
data base consists of enumeration district/block group (ED/BG) values.
MARF contains the population centroid coordinates (latitude and longitude)
and the 1980 population of each ED/BG (approximately 300,000) in the United
States (50 States plus the District of Columbia). HEM Identifies the
population around each plant, by using the geographical coordinates of the
plant, and identifies, selects, and stores for later use those ED/BGs with
coordinates falling within 50 kilometers of plant center.
For each of the 15 glass plant locations, a detailed check was made to
determine whether the exposed population as predicted by the HEM was located
accurately. Based on information obtained from a U.S. Geological Survey
map of the area around Glass Plant 15, a revised maximum exposure con-
centration was used. Plants 1 and 8 were modeled using site-specific
terrain and plant parameters (Section D.4. 2.1). For these plants, exposure
concentrations were taken from the site-specific model predictions. These
results are summarized 1n Table D.4.
D. 3.1.5 Exposure5
The Human Exposure Model (HEM) uses the estimated ground level
concentrations of a pollutant together with population data to calculate
public exposure. For each of 160 receptors located around a plant, the
concentration of the pollutant and the number of people estimated by the
HEM to be exposed to that particular concentration are identified. The
HEM multiplies these two numbers to produce 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 non-uniform spacing. At small radii, the grid
cells are usually smaller than ED/BG's; at large radii, the grid cells are
usually larger than ED/BG's. The area surrounding the source is divided
into two regions, and each ED/BG 1s classified by the region in which its
D-16
-------
centroid lies. Population exposure is calculated differently for the ED/BG's
within each region. For ED/BG centroids located between 0.2 km and 3.5 km
from the emission source, populations are divided between neighboring concen-
tration grid points. There are 64 (4 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 and respective concen-
trations are assigned to the nearest ED/BG centroid identified from 1980 U.S.
Census Bureau data. Each ED/BG can be paired with one or many concentration
points. The population associated with the ED/BG centroid is then divided
among all concentration grid points assigned to it. The land area within
each polar sector is considered in the apportionment.
For population centroids between 3.5 km and 50 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. Since
there is an approximate linear relationship between the logarithm of concen-
tration 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 logarithmically interpolated radially and arithmetically
interpolated azimuthally from the four receptors bounding the grid cell. Con-
centration estimates for 96 (6 x 16) grid cell receptors at 5.0, 10.0, 20.0,
30.0, 40.0 and 50.0 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 64 concentration points within 3.5 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 model used to estimate
exposure, see Reference 5.)
D. 3.2 Public Exposure to Inorganic Arsenic Emissions from Glass Manufacturing
Plants
D. 3.2.1 Source Data
Fifteen glass manufacturing plants are included in the analysis.
Table D.5 lists the names and addresses of the plants considered, and Table
D-17
-------
D.6 lists the plant data used as input to the Human Exposure Model (HEM).
D. 3. 2.2 Exposure Data
Table D.7 lists, on a pi ant-by-plant basis, the total number of people
encompassed by the exposure analysis and the total exposure. Total exposure
is the sum of the products of number of people times the ambient air concen-
tration to which they are exposed, as calculated by HEM. Table D.8 sums,
for the entire source category (15 plants), the numbers of people exposed
to various ambient concentrations, as calculated by HEM.
D-18
-------
Table 0.3
Plant
Arsenic Concentrations Near Select
Glass Manufacturing Plants
ISCLT/
HEM Valley MAOB
# Obs. Distance1 Bearing Predicted2 Predicted Measured
MDL3 Percent!le4
Gl ass PI ant #6
Glass Plant #14
Glass Plant #1
Glass Plant #4
76
77
77
69
41
65
24
84
.5
.6
.6
.7
8.6
3.2
1.0
2.1
73.7
246.1
43.4
23.2
168.7
66.1
189.6
172.7
0.00154
0.00097
0.0017
0.0017
0.000115
0.0025 0.0035
0.013
0.0062
0.025
0.025
0.026
0.026
0.007
0.003
0.003
0.003
0.05
0.05
0.05 95< 3
0.05 95< 3
0.0055 5099*
>99*
t..<99*
I <99*
I <70*
I <95*
>99*
I <90*
* Indicates data point was disregarded; see Section 0.3.1.3.1
1 Distance from source to monitor (km)
2 Concentration predicted by Human Exposure Model (HEM) - see Section 0.3.1
3 Minimum Detection Limit
4 Percentile indicates percentage of data falling below minimum
detectable levels
0-19
-------
.
*
a
S
/2
r—
C
4.
c u
•2 2
"T— *T"
re a
u
0
— J •£
«= *£ CO
O — O
•r- J-
4-> CO 3
re o o
r— • C CO
3 re
CX4J E
O 10 O
0. -r- J-
^— * 1 •
— J IL.
«
t~l
O CD <
t- CO C
4-> f- C
c > o
o co o
s=. o cu
u
•i—
3= COCM
c xj •
O "r- CO 0
*•> i— VI C
CO -I- O
«/> XJ > O
c co re co
O to CO C£
4-> CX
re x
y it i
C CO r- XJ
CO S- O CO CO
u re t_ r— to
C •«->«*- CX O
o co c o o a.
Or- O CO X
re o *tts a. ui
§ "O
j^ ,^
•i— > CO
£ S £ 01 «
«er* c* ,U1, IA
•Cii «•• !••• f^ l/f
*-* CO «*- CX O
"O CO O O D.
co re co x
•M co ^ a. ui
l]*^
tn
CO
ex.
"0^-4
S- •
•M U
c c
0 0
o o
^
Ci-l
•t— •
r— U
CO C
CO O
re o
CO
e\
sf
0
CO
* * * *
Ul Z Ul Z Ul
co co
Ul Ul
* * * * in
CM CM in in CM
o o o o o
1
1
o
I-H
X
CO
CO —1 -4
i-4 CO VO ^- CTi
CO —1 -H
•K -Jc 4c ^c CM
•i" co co co i
1 1 1 1 O
O O O O i-l
i-l rH >-4 i-l
x x x x x
VO
CO p~ VO «3- CM
»* VO CM CO i-H
•1
* * * * ^H
^* co co cO i
1 1 1 t O
O O O O 1-4
x x x x x
CO P** VO ^" Cf»
«4- VO CM CO i-H
in vo ** co co
1— 1 f— 1
UIUIUIUIUIUIZLUZU1
CO Z Z CO Z Z
Ul Ul Ul
****** •}:*
ininooeMCMomcMCM
OO»-»«-IOO-HOOO
CM
1
o
i-H
, x
1
1 C7^
co
CM
O 1
•-4 0
X ""*
<*>
00 CM
i-4 CM
3C2S~*£2CSIOP''>00
como ooincMcrip^.
P«« i-4 -— 1 f~. po
A
i-H
COCOCOO-4COOOCMOO
co in o co in CM P-
* *
ICO^COCOCO ICMCO 1
i-HOOOOOi-HOOi-H
rHi-4i-4^Hi-4 ^ ^H
xxxxxxxxx
•— » o^ in
VOVOi-HCMincrtCMOfHCO
^H«i-VOCOin«3-i-Hr-4CMi-H
* *
>******CMCM*
ICM^-COCOCO 1 ICMCM
O 1 1 1 | | o O 1 1
i-4OOOOO,_4i-lOO
^xxxxx _xx
vocM-i-icoincnp^oo'*
•-ia>vovoin«3-i-H.-i T-
w ••-
c: o
O CO
•i- CX
(13 1
O CO
O -P
• *""
e
0 2?
•r- jQ
fw V
•35
CX 0
O •!-
CXXI
O)
at ex
4J
U (/>
•r* C
•a o
I- +J
as
3: co
u
0 0
u
o c
to o •
'£ TS •
re co CM
CX CO •
es re •%!
u a CM
00 1
t- c o
o -o o
re '-M
co c co re
CS « CO 4-
C/J •— 3
**^ Q™ C^
• 0
t-. o co re
0 4-> •«->
» « S"?
x co re co
• CO C i— i.
i — -r- O CX CO
••- -O
X» CO O -r-
O t- T- 2 to
E > +•> c
C CO O
1- O Q£ CO O
co co
4-> X> O CO
3 CO • JC J-
ex to 4-> re
E re co
O -i- O 4->
in t-
>> re +J o
JQ i— c 3 s:
a. co -o ui
xj -oca:
CO O ••— O 1
+->•»-> CO O 1
re co x>
O C i. CO
•r- O CO O)
XJ -i- i — 10 c
c re >, re
*~ "> *J *« 0
CO CO O C C
-------
TABLE D.5
IDENTIFICATION OF GLASS MANUFACTURING PLANTS
Plant Number Code
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
1
Plant Name and Address
Owens-Ill 1nois
Shreveport, LA
Owens-Illinois
Mt. Pleasant, PA
Owens-Illinois
Pittston, PA
Owens-Illinois
Toledo, OH
£CA
Circleville, OH
GTE
Versailles, KY
Fostoria
Mouridsville, WV
Corni ng
Marti nsburg, WV
Corning
Charleroi, PA
Corning (Fall Brook)
Corning, NY
Corning (Main Plant)
Corning, NY
Corning
State College, PA
North American Phillips
Danville, KY
GTE Sylvania
Central Falls, RI
Indiana Glass
Dunkirk, IN .
D-21
-------
O 4->
T- c
to t-
cu
DU
LU
o o
P
CO CO t/»
O
IB
Jit J*
O O
-P -P
CO CO CO
o
Q.
c:
*r™
1
O
>
O CD -P O
CO 4-> T- O CO
CO C X O --.
»i— "r- LU r— P-
SO CU
LU a. >
* r- — >
C (BCM
O C E
i- -P CO O —
CO C CO -r-
CO -r- O -P IB
f- O t- O CU
E CX. O CU X-
LU CO «C
C J_
o cu — •
•r* -P -P CO
CO C CU S-
CO 'r- £ CU
T- O IB -P
* 1 1 f^\ j^
^--
C
c= o «-»
O -P T- W»
i- C -P t_
«/» f- «B CU
to o > +J
•r- Q. CU CU
invo cn CM o vo co cot->.>-i
cjco cn r— ^- vo oo 4*envo
in ^t- vo «a- «* vo 5- vo vo ^
ocn oo in in «a- CM ovo^-
^"O^ ^" OO ^ CO ^- COCOP-.
**"" f~* CM r— «— 1
gg g g g g g ggg
S£ g g S 3 g ESS
* * * * • • • •••
i— O *H O f— CM f— O O — 1
*4
OO C3 O O O O OOO
CMO VO O r-^ f-I VO CnOOi-4
CM ^ ^ «* «* * 3h CO
r^ in t»» CM
t-» in i^ CM
^J" ^^ ^^ ^O
cn o CM to
CM «3- in CM
CM r— i— •-(
0 00 0
in in in in
o o en CM
CM t— VO 10
• • • •
•— r- O fH
O O O O
• • • •
vo r-» r^ cn
CM CM CM CM
vo
co
vo
CO
o
in
"
<— i
vo
•
O
o
co
f— 1
C3 *5t" cn
co »* vo
in in in
CM CM <-4
CM CM CM
^H --I CM
OOO
in in in
"^
o o in
OOO
• • •
OOO
ro co cn
in in CM
w
o
•r—
CO
Vt
•i—
CU
•P
«•-
o
IB
£
IB
IB
c
•r—
O
CU
CO
CO
CO
e
fH*
IB
O
1_
CU
•
fit ^
VW *lto
•P IB
2
.^ =
•r-
—>
c
o
•1—
CO
CO
iS
cu s-
** >>
IB "^
oc cn
^
vo cn o
r~ in in
CM r>.
in
o
cn
o
o
cn
t-4
cS
CM
CO
•*
co
CM
0
oo
rv.
r—l
O OO
cn CM
CM r*.
vo
<— i
0
•
o
CM
i— i
0 r-
in co
cn cn
in
ooo
o oo r*. in ao
o co «t cn in
«* r*. r^ co
«*s
o
cu
1^
1 Longitude
1 Latitude
i i
CO
cu
cu
cn
cu
o
to
t_
cn
s
1 Mi nutes
CO
cu
•P
c
Seconds)
i Seconds)!
1
CU
o
IB
0825718
1-1
1-1
co
cn
co
0844504
VO
CM
O
CO
CO
0712320
VD
«d-
CO
m
0844948
CM
in
r^
co
0780103
CM
co
CM
CO
0774728
VO
O
O
in
o
0795403
in
IB
in CM vo
O —t O
!-•• in oo
•HI in CM
r-c cn CM
«* co co
•P
IB
a.
t— CM
in
CM CO
CM
CM CO
VO
CO
00
CTi
CO
r-. t-i
f— O
c— CU
IB J-
•r"
i. -O
O
«*- TJ
CM ••-
E 3B
o «=
in
-------
cu
•o
o
o
en
c
•1—
i_
5
U C
•o o
o o
S c
i. ••-
en cv
O CO
a.
!e>
•.- c o
«/>••— o.
*" P 4^»
•i— OL f—
5
CO
c= ia
o en •
co +J E a*
CO C 9) o
E o
uj a.
CO
C >
O 13 •»-> O
•i- 4-> T-
LU a. =»
* »— ^~
C flSCM
0 C E
.,- 4-> CO O —
CO C CO •!-
CO T- O 4-» <8
E a. o +J CO
co c a> s-
co •!* E O
•i- O
J^5 §
c
c o —•
O +•> t- CO
co T- fa a>
•r- a. a> a>
e r- s:
us ui *-*
c
o — .
CO 4-» >,
CO (O • —
•i- Q£ O>
E ^
UJ —
C
•r- S- =3 O
O> O> C O
c
o a s to
_j ^-
d; co co co
•o a» a» ~o
3 C
4-> S_ =3 O
•i- 01 C 0
•«-> 0) T- inioir>vo
COI^COP~«*'-4lf>P-.CO
csjcMi-H«d-cviir>
•-< <-l CM
ooooooooo
mmininininminin
ooooooooo
tnir><4-aocooooocMCM
ooooooooo
^-Ir-l^- lOOOOO<-«'^
( — f*»P««^-tOr-ltOP»-P*-
OOOOOOr^CrtCOOOOOO
OICTiCTiOOl — ^f»~fMCftCT>
,-1 ^-i ,— i CM CM i— 1 •— 1
CM
O
^H
CO
CO
OO
o
t— (
1— 1
o
<*
— 1
«*
•-ieMco^-ir>ior^.oocr>
«*
^Jl£^£^^^^£^^
OOOOOOOOO
fljj t/)tOt/>OOun
C7) CTk O> O^ C7) CTt C7) O> CT)
OOOOOOOOO
OOOOOOOOO
CMCMCMCMCMCMCMCMCM
^ooaioa^ooo
o«d-oaO'*co«d-«*^-
i— O
r-
-------
O)
•f"
3
4*}
u
1—
3
C
CO
SE
t/j
(d
3
r- •
0) .-.
•o to
Or-
CO 4->
t- {=
3 O
*A / ^
MO
M. « __
xfe
UJ CO
o cn
fl3 fe*
4-> 3
to to
C J vi
4-> "I— ^
3
C
>— l
JQ
0
S
0 •«->
f- C CO
in .— pi
UJ
in
C A3
0 CO •
tn B co o
•t— »f— ^— *•-*
E 0
UJ O.
in
c to >,
O CD 4-> O
f- 4-> f- CO
10 C X O — .
^- -t- UJ r— £
S O CO
UJ Q- >•
* r- •"—
C tO O —
t/) f* I/) *r~
v) *f* O 4-* CO
•r- O S» O CO
E O. O CO i-
UJ CO «C
__
C i-
O CO — •
•r- 4-» 4-> IO
to C QJ t.
c/t i** E co
•t— o to 4^
E Cx. T— CO
uj Q z:
c= o *—
O 4-> i- «/)
w> «r- re co
tO O > 4->
•r* CU CO CO
E r— E
UJ UJ -^
c
o *—
*f— (D (_
^1^ ^
•r- a: en
UJ «—
CO ^-»
3 CO CO T3
4^> CO 4-> C
•r- L. 3 O
cn cn c o
C CO ••- CO
O O S CO
I ^ -~
co to to to
•o co co -o
3 CO 4-> C=
4J fc. 3 O
f- cn c: o
4-> CO -r; CO
to a E co
"co
o
03
c
3
xj It
c «^
(O
Q-
u
to
CO
5
•^j
t-H
CM
o
in
CM
-1
o
in
CO
o
0
vo
co
0934753
vo
0
CO
CM
CM
co
~*
to
co
CO
in
vo
o
CM
o
in
vo
0
o
CO
o
§
0793201
f--.
CO
CM
.1C
u
<0
CO
CM
CM
in
vo
CM
0
in
CM
in
t-H
o
co
CM
o
o
o
0754619
in
o
t-4
CO
u u
-------
>nduded)
o
o>
•^
+J
CL tO
X CO
C3>
0 C
4-> •!-
W)
0 <
*>'"'
CL
I-H
«3
•
O
"i
UJ
Emission*
C
o
•^
•7-
ut
o
«/»
•f
o>
o
_l
Latitude
+•»
C
tt.
«/»
C
•r-
O
O-
tn
(O
ID
+J
C
•f—
o
Q-
4-»
C
•*-
O
o.
+J
C
o
Q.
4J
*f«.
O
o.
4->
It)
0£
V
ta-
ll'
•
t
O>
UJ
(Degrees
Mi nutes
(Degrees
Minutes
Q
o
'o
C/)
^f
5
«3
o>
s:
^
o>
^
Seconds)
1
Seconds )
(Furnace)
^
o
^"
0
U3
>*
•*
CO
o
CM
CO
CM
ff—
e^
0
3
in
5
«*
t— i
.*£
u
W
§8
CM CM
CM CM
O O
•* «*
0 0
O O
in in
^ t-4
0 O
in in
•* «*
CO CT>
in
CM
i
o
-------
TABLE D.7 TOTAL EXPOSURE AND NUMBER OF PEOPLE EXPOSED
(GLASS MANUFACTURING PLANTS)*
Plant
Total
Number of
People Exposed
Total
Exposure
(People - M9/m3)
5
6
14
13
8
12
9
10
11
3
7
1
2
4
15
835,000
434,000
2,530,000
428,000
346,000
257,000
2,080,000
270,000
268,000
697,000
372,000
369,000
1,450,000
834,000
408,000
5
10
148
7
1620
1
1830
17
82
87
21
690
163
1070
627
A 50-kilometer radius was used for the analysis of glass manufacturing
plants.
D-26
-------
TABLE D.8
PUBLIC EXPOSURE FOR GLASS MANUFACTURING PLANTS
•AS PRODUCED BY THE HUMAN EXPOSURE MODEL
Concentration
Level (M9/nr)
Population
Exposed
(Persons)*
Exposure
(Persons - ug/m3)**
0.216
0.1
0.05
0.025
0.01
0.005
0.0025
0.001
0.0005
0.00025
0.0001
0.00005
0.000025
0.00001
0.000005
0.0000025
0.000001
0.000000593
0
486
4740
20800
79800
189000
523000
1210000
2380000
3900000
5100000
6770000
7570000
9580000
10800000
11400000
11500000
11600000
0
64
346
907
1830
2580
3740
4820
5610
6180
6370
6500
6530
6560
6570
6570
6570
6570
*Column 2 displays the computed value, rounded to the nearest whole number, of the
cumulative number of people exposed to the matching and higher concentration levels
found in column 1. For example, 0.5 people would be rounded to 0 and 0.51 people
would be rounded to 1. ,
**Column 3 displays the computed value of the cumulative exposure to the matching
and higher concentation levels found in column 1.
D-27
-------
D.4 QUANTITATIVE EXPRESSIONS OF PUBLIC CANCER RISKS FROM INORGANIC ARSENIC
EMITTED FROM GLASS MANUFACTURING PLANTS
D.4.1 Methodology (General)
D.4. 1.1 The Two Basic Types of Risk
Two basic types of risk are dealt with in the analysis. "Aggregate
risk" applies to all of the people encompassed by the particular analysis.
Aggregate risk can be related to a single source, to all of the sources in
a source category, or to all of the source categories analyzed. Aggregate
risk is expressed as incidences of cancer among all of the people included
in the analysis, after 70 years of exposure. For statistical convenience,
it is often divided by 70 and expressed as cancer incidences per year.
"Individual risk" applies to the person or persons estimated to live in the
area of the highest ambient air concentrations and it applies to the single
source associated with this estimate as estimated by the dispersion model.
Individual risk is expressed as "maximum lifetime risk" and reflects the
probability of getting cancer if one were continuously exposed to the esti-
mated maximum ambient air concentration for 70 years.
D.4.1 .2 The Calculation of Aggregate Risk
Aggregate risk is calculated by multiplying the total exposure produced
by HEM (for a single source, a category of sources, or all categories of
sources) by the unit risk estimate. The product is cancer incidences among
the included population after 70 years of exposure. The total exposure, as
calculated by HEM, is illustrated by the following equation:
N
Total Exposure = I
where
I - summation over all grid points where exposure is calculated
Pf - population associated with grid point i,
C-j = long-term average inorganic arsenic concentration at grid point i,
N * number of grid points to 2.8 kilometers and number of ED/BG
centroids between 2.8 and 50 kilometers of each source.
D-28
-------
To more clearly represent the concept of calculating aggregate risk, a
simplified example illustrating the concept follows:
EXAMPLE
This example uses assumptions rather than actual data and uses only
three levels of exposure rather than the large number produced by HEM. The
assumed unit risk estimate is 3 x 10~3 at 1 M9/m3 and the assumed exposures
are:
ambient air : number of people exposed
concentrations to given concentration
2 Mg/ro3 1,000
1 (jg/m3 10,000
0.5 M9/m3 100,000
The probability of getting cancer if continuously exposed to the assumed
concentrations for 70 years is given by:
concentration
unit risk
probability of cancer
6 x 10-3
3 x 10-3
1.5 x 10-3
2 ug/m3 x 3 x TO'3!
1 M9/m3 x 3 x 10~3
0.5 ug/m3 x 3 x 10~3
The 70 year cancer incidence among the people exposed to these concentrations
is given by:
cancer incidences
number of people at after 70 years
each exposure level
1,000
10,000
100,000
probability of cancer
at each exposure level
6 x 10-3
3 x 10-3
1.5 x 10-3
of exposure
x
x
x
6
30
150
TOTAL = 186
The aggregate risk, or total cancer incidence, is 186 and, expressed as
cancer incidence per year, is 1864- 70, or 2.7 cancers per year. The total
cancer incidence and cancers per year apply to the total of 111,000 people
assumed to be exposed to the given concentrations.
D-29
-------
D.4.1.3 The Calculation of Individual Risk
Individual risk, expressed as "maximum lifetime risk," is calculated by
multiplying the highest concentration to which the public is exposed, as
reported by HEM, by the unit risk estimate. The product, a probability of
getting cancer, applies to the number of people which HEM reports as being
exposed to the highest listed concentration. The concept involved is a
simple proportioning from the 1 |jg/m3 on which the unit risk estimate is
based to the highest listed concentration. In other words:
maximum lifetime risk the unit risk estimate
highest concentration to = l jjg/m3
which people are exposed
D.4.2 Risks Calculated for Emissions of Inorganic Arsenic from Glass
Manufacturing Plants
The explained methodologies for calculating maximum lifetime risk and
cancer incidences were applied to each glass manufacturing plant assuming
a baseline level of emissions. A baseline level of emissions means the
level of emissions after the application of controls either currently in
place or required to be in place to comply with current state or Federal
regulations but before application of controls that would be required by a
NESHAP.
Table D.9 summarizes the calculated risks. To understand the relevance
of these numbers, one should refer to the analytical uncertainties discussed
in section D.5 below.
D.4.2.1 Site-Specific Analysis
In its original risk assessment, EPA did not consider terrain effects
or the full effect of building downwash on stack emissions from glass manu-
facturing plants. These plants oftentimes have short stacks whose effluents
can be entrained in the building wake on the leeward side of the furnace
buildings or other adjacent structures. As a consequence, it was regarded
as likely that airborne arsenic concentrations to which persons might be
exposed near or adjacent to these plants could be underestimated. In addi-
tion, it was felt that the extent of building downwash could be expected to
be different depending on the degree of control applied. If so, the relative
D-30
-------
reduction in risk achieved by any particular control option might differ
from that estimated previously.
For these reasons, additional dispersion analyses were carried out to
examine these concerns further. These analyses were done for two glass
plant locations, Martinsburg, Virginia and Shreveport, Louisiana. For the
Shreveport analysis, the Industrial Source Complex Long Term (ISCLT) model
was used in conjunction with a joint frequency distribution of wind speed,
stability class, and wind direction derived from surface weather observa-
tions at Greater Shreveport Municipal Airport. In order to better assess
.the impact of building downwash on the rise of the hot furnace gases, a
modified plume rise treatment similar to that in the Buoyant Line and Point
Source (BLP) model was used in addition to the standard ISCLT plume rise
treatment. Both the building downwash and standard ISCLT analyses included
an enhancement to the dispersion of the plume as appropriate, an enhance-
ment which is a part of the ISCLT model. The two sets of analyses were
intended to bracket the expected effects of building downwash on airborne
inorganic arsenic concentrations.
Results obtained from HEM and the two site-specific analyses for
Shreveport can be seen in Tables D.10 - D.12. Table D.10 outlines arsenic
concentrations estimated by the Human Exposure Model (HEM) to occur at 16
wind directions and eight distances downwind from the Shreveport plant
center. Table D.ll shows corresponding values based on the standard ISCLT
model run in conjunction with a joint frequency distribution of wind speed,
stability class and wind direction. Table D.12 shows values based on the
ISCLT model used in conjunction with modified plume rise treatment. Agree-
ment between the HEM and ISCLT estimates is fairly good with differences
rarely exceeding a factor of 1 or 2. The HEM tends to underpredict slightly
in regions of higher and lower concentration.
For the Martinsburg analysis, the same approach was taken, except that
the effects of terrain were also considered. Martinsburg is located on a
broad plateau at the foot of a long ridge. ISCLT was used to estimate air-
borne inorganic arsenic concentrations at any receptor located at or below
D-31
-------
the top of the furnace stack. However, for receptors above the top of the
furnace stack, the Valley model was used. The Valley model allows the plume
to intersect terrain features under stable conditions, resulting in high
concentrations. For receptors well above the plume center!ine, the impact
of the plume is gradually reduced. Both models were used in conjunction
with a joint frequency distribution of wind speed, stability class, and wind
direction derived from surface weather observations at Martinsburg Airport.
The modified plume rise treatment described for Shreveport was also included
as part of the Martinsburg analysis in addition to the standard treatment of
plume rise in ISCLT and Valley. This again resulted in two sets of disper-
sion analyses, which, as before, were intended to bracket the expected
effects of building downwash.
Results obtained from HEM and the two site-specific analyses for
Martinsburg can be seen in Tables D.13 - D.15. Table D.13 outlines arsenic
concentrations estimated by the HEM to occur at 128 points (see above)
around the Martinburg plant. Table D.14 shows corresponding values based
on the standard ISCLT and Valley models run in conjunction with a joint
frequency distribution of wind speed, stability class, and wind direction.
Table D.15 shows values based on the ISCLT/Valley models with modified
plume rise treatment. Agreement between the HEM and ISCLT runs is fairly
good with differences rarely exceeding a factor of 2. The HEM tends to
overpredict slightly at regions of higher concentration with more varied
results at regions of lower concentration.
D-32
-------
TABLE D.9 MAXIMUM LIFETIME RISK AND CANCER
INCIDENCE FOR GLASS MANUFACTURING PLANTS
Plant
Maximum Lifetime Risk
Baseline Control
Cancer Incidences Per Year
Baseline Control
Owens-Illinois
Shreveport, LA
Owens-Illinois
Mt. Pleasant, PA
Owens-Illinois
Pittston, PA
Owens-Illinois
Toledo, OH
RCA
Circleville, OH
GTE
Versailles, KY
Fostori a
Moundsville, WV
Corning
Marti nsberg, WV
Corning
Charleroi, PA
Fall
Corning - Brook
Cornint, NY
Corning - Main
Corning, NY
Corning
State College, PA
N. Amer. Philips
Danville, KY
GTE Sylvania
Central Falls, RI
Indiana Glass
Dunkirk, IN
7
8
4
2
3
1.8
3
2
8
1.6
4
3
3
7
1.4
1.1
9
x 10-5
x 10-5*
x 10-5
x 10-5
x 10-^
x 10-6
x 10-5
x 10-5
x 10-4
x 10-3*
x 10-4
x 10-6
x 10-5
x 10-7
x 10-5
x 10-5
x 10-4
6 x 10-6
4 x 10-5
2 x 10-5
9 x 10-6
1.8 x 10-6
3 x 10-5
2 x 10-5
5 x 10-5
2 x 10-5
3 x 10-6
1.4 x 10-5
7 x 10-7
1.4 x 10-5
1.1 x 10-5
1.7 x 10-4
0.035
0.042*
0.010
0.0053
0.066
0.0003
0.0006
0.0013
0.12
0.099*
0.11
0.0006
0.0050
0.0001
0.0011
0.0091
0.038
0.0037
0.010
0.0053 .
0.0066
0.0003
0.0006
0.0013
0.013
0.012
0.0006
0.0016
0.0001
0.0011
0.0091
0.0085
* Represents risk estimates calculated from site-specific analyses using
ISCLT model.
D-33
-------
C3
.O
17E-Q
i/i
n
Ul
c
o
*•»
«j
i
o
o
u.
Xt
Ul
V>J
re-
i
r.
0
Ul
Ul
id
IT
Ul
Ul
C
Cf
U.'
in
ro
o
ro
Ul
l
c
o
Xt
to
o
*+
c,
CO
I
C."
o
>J\
m
Xt
,_,
Xt
LU
*•
1
(_>
•_>
Ul
vn
03
vn
Ul
1
cs
Ul
~
ro
o-
o»
vn
1
c.
o
Ul
-J
o
•J
U
i
o
o
to
Jj
c.
•0
o
o
m
m
ro
•o
o-
JCt
^»
1
Vw
U
UJ
*
UI
o
OC
1
u-
Ul
o
ro
ro
vn
c
o
•J
«o
-J
Ul
1
c
a
x-
Ul
J*
u.
ro
Ul
i
0
Ci
m
UJ
Xt
-o
0
tr
1
0
O
u
Xr
vn
o-
1
Vr.
0
UJ
Ul
o
n
z
m
UJ
^j
M
^
«fi
i
0
o
Ul
Xf
o
•o
»-»
1
Vr.
Ul
u
o
m
Xt
Ul
*»J
•»
1
u<
t_>
UJ
vn
Ul
H*
O
1
UJ
JS
O
z
rn
Xt
O
0-
JB
o.
1
l^
0
Ul
vn
00 i
t—
X:
1
b
UJ
ft
CO
z
z z
i-* vn
i* Id
C vn
*\) f
rv «-
1 i
O */
1
UJ
0
Ul
•o
ro
o
X:
1
UJ
-O
ro
«j
-o
"i4
c.
c o u o
ui ro ui
«j -o ro f*
co
•o
CO
1
Xt
•J O-
vn ui i"»
vn ro ro
UJ
NO e-
CD
0» 0 Ut *
Xr
•
^i U
h- Ul
Xt Ul
1 1
o 6 o
O c U
»• • . ««
oo
o
Ul
1
o
o
-ff
C
c
Xt
Ul
Ul
'-4
er
1
C
Q
UJ
O
^
ro
a:
1
fe-.
0
UJ
o-
ro
Ul
o?
ro
C
o
UJ
M
Ul
Ul
CD
ro
i
o
o
Ul
Ul
UJ
Cf
o-
**
1
o
o
•*.
•V.
M
|d
ro
Cf
i
o
o
JBt
^,
UJ
CO
ro
Ul
i
o
o
Ul
Ul
UJ
ro
1
o
o
Ul
••••••i
**
Ul
•o
{y
Ss
•0
•o
4*
4*
Ul
|
O
a
Ul
.g
£.
0s
00
•o
0
0
Ul
•••
tj
o-
•o
1
o
2
ro
o
Ui
ro
i
o
o
Xt
M
00
ro
e
i
o
o
Xf
(M
t«J
Xt
oo
1
ro
u
•j
Eh UJ
O UI
h* 0-
•*
«j
o
OO 00 0 >*
i
1 1
1
w GJ o a o
2
£
2 2
id id H» >• >* ui ro
Q«
o»
1
o
o
CO
Xf
ro
00 " '
?
o
o
Ul
UJ
•o
Ul
Oo
1
o
o
Xt
«o
j»
-j
ro
1
o
o
Ul
ro
e
t*
•
o
o
00
-J
Ul
K
1
€
Ul
2
UJ
»d UJ «O UJ I-
Ul
M
Ul
1
O
o
*
00
Ul
ro
o
l
o
o
Ul
••J
•O Ul
!«• O OB
UI
1
o
o
4»
0
•0
o
00
Ul
1
a
o
Ul
-j to
1 1
o o
o o
ro »•»
03 -J
0- O
-J *
-J ao
1 1
0 0
o o
«*
Ul
•4
1
O
a
*
ro
ro
•0
00
ro
l
o
9
UJ
o
ro
0
i
0
0
Xt
ro
o
0
UJ
o
1
o
o
*
M
X-
Xt
60
€
XT
•E:
Xt
c
•c
lid
o
1
o
o
Ul
*
o
-J
c
1
cr
O
UJ
vn
M
to
o
1
o
e
Ul
o
oo
H*
oo
|
C
o
Xr
Ul
oo
C
OD
OD
I
O
o
1*.
ro
td
Ul
5
i
0
o
Xt
~
Xr
O
00
Ui
1
o
0
£
~
0
^*
a-
i
*
c
c
to
x:
0
«*J
Co
i
w
Vj
UJ
UJ
t>
o>
OC
1
H
Ul
ro
•o
^
|d
Xt
1
C
o
Ul
Ul
ro
X!
00
Ul
1
0
o
Xt
ro
o
0*
Ul
1
c
a
-»•
**
»*
ro
Ul
i
&
2
^
*
M
O
ro
i
o
o
Ul
Ul
UJ
Ul
o
1
o
o
Ul
c:
ro
Ul
Co
-J
U-'
1
O
Ul
»•
OD
x-
0
0-
1
o
o
U!
**
ro
vn
Ul
ro
l
o
o
UJ
*
00
X
c
Xt
1
o
0
ro
«j
(d
00
1
0
0
Xf
M
0>
o-
Ul
1
o
o
*
*-
ro
CD
1
o
e
*
o
rn ^3
o «-*
1- C"
o
z
•
v<-
CJ
^/
.
t.
*,
^
rv>
• o
o o
c. r
o z
c
1— <
0
o
C. I/*
C- J»
o z
o n
m
Jt
ro
c
b
o
o
UJ
o
o
a
a
«
o
•
o
o
o
Ul
o
o
o
o
• — -oo «c rn
m -s o ri-
3 ro ->•->•
•—< 3 3
,-,-ffi ;iS.
OT3 c-h r+
fa O << TO
— ' -S Q.
0 rt- O
5i" "** ^
fl> r— 3= (f>
c+ 3=> ro — l
— '•*• CD 3 QJ
0 —a _J. CT
3 oa CD n — >
(/} (/) C™"5
fD o a
Q- IS 3 .
Cu O I—1
o 3 ro o
3 C 3
-b c*-
3: cu -s
3 t-f c-t-
O> C. -J.
3 -5 O
— »• 3
m 3 to
X IQ
•a ->•
O ID 3
tn — i
C OK c+
-$ 3 3-
fD c+ ro
0 3*
rt
ro
-------
O)
.o
10
(O
C -r-
^1 C 3 i—
4-> T3 O)
•— C 13
C D- >-< O
cn-o
«/> C S-'—-
E: -r- CU
S= S T3 r-
O Q)
CJ> t/) CD
«/) "0 £=
o n) co o
CU E O
T3 i_ O -r-
CU >> O O 4->
+J ^J Q. (O
to •!- CU CU i—
E E > O 3
•i- •!- CO S_ O
4-» O S- 3r—
OO *r— J^ O *O
UJ 5> 1/1 OO O
o
c
o
•
a
to
o
o
o
•
a
CM
O
O
o
D
C-
-3
CM
o
Q
1
ar
•O
to
ar
*-»
a»
O
O
1
co
to
o
. 9
ar
o
o
1
CM
CO
in
CO
CM
ar
O
a
i
o
m
ar
o
in
<0
O
f
o»
co
o
CM
-*
to
°D
^5
1
fO
Ul
CM
•H
Ul
to
O
O
1
,\4
7)
-M
fi
a-,
""•
CM
ar
O
.0
sr
ro
0
tn
in
ro
r>
^
i
•o
CM
O
lf>
AJ
ro
O
T
1
-a
CO
in
eo
CM
Ul
O
O
i
-a
CO
CM
CM
Ul
m
o
o
o
a
in
•o
a
ar
o
0
1
in
CM
af
O
-
&
0
o
1
ar
O
ar
CO
r-i
ar
ra
o
r>
r»
•o
•o
ar
ro
••^
O
1
!*%
CM
o*
-*
N
fO
o
-^
1
r-
ar
CM
f"
ru
ro
?3
T3
1
*-*
ffl
0*
ar
3
U".
in
O
•3
1
0
^M
°
•°
0
O
t
o
CM
9
CO
ar
o
o
t
fO
o
CM
^
ar
O
O
i
to
0
CM
1-1
CM
ar
O
' '?
CM
CM
O
ro
in
1
. S4
ar
•"^)
0
*•
•o
o
o
i
o
o
ro
a-
ar
x>
0
3
t
X>
"5
-O
^
2
D-35
O
•-•3 •
1
'^}
•*>,
*- .
O
o
1
to
CM
&
O
o
1
CO
at
O
«-*
".
ar'
O
O
1
I"":
U^
3»
in
tn
JO
O
O. '
PI
in
r»
fO;
"
fO
b1
o
i
o*
ar
in
CO
•ar:
ro
&
O'
1
5T
O
ar
«-«
Ul
ro
O
O
g
P*
r-
*H'
•=•
3
Z3
O
0
1
ar
m
in
CM
ar
O
O
1
CM
ar
CM
ar
a?
0
O
1
o
VI •
=r
o
in
ar
O
O
1
•o
CO
CO
CO
to
n
Q
0
CM
•o
•-«
CM
fO
o
o
1
tM^
cr
00
.en
•°
ro
^1
o
1
'O
in
or
to .
-"
fO
O
•3
1 .
sr
•O
f,
ar
-3
Z
' 0
i
0
0
-*
ar
O
O
1
CM
CM
O
Ul
CM
ar
O
O
1
e«
i>^
G*
P-
to
ar
O
O
1
«_i
^3
fO
r-
«o
tn
>^
o
i
ro
«>»f
O
f*
*••
^
T1
O
1
*o
en
to
ItA
•O
.0
,7^
•^
1
j^
O
ro
i
CM
O
co
ar
—a
. ar
O
i
rH
-O
(O
*•«
^
•ar
O
O
1
1
CM
in
o
0
1
ou
0
-o
CM
CM
ar
O
O
1
to
p*»
CM
o
ar
JO
1
o-
CM
ar
"S
-
.•o
TT*
"^
1
<^
CO
^.^
.1
Ul
ro
•rj
i
^
r»
•o
in
-o
-o
o
1
CM
O
*
jj.
-z.
P
J
o
-"M
0
ar
O
0
m
0
CO
«M
ar
O
O
1
co
o
co
co
•*
^
o
o
1
SO
o
CM
to
to
ar
o
o
o
CM
«-4
<0
00
^
'"
rs
t
o*
ro
iO
!fl
rO
••o
~^\
—i
1
'/)
^
ar
ar
iO
^
•3
1
=r
O
0
CM
LJ
£1
•=r
r^
^
^
•c,
»-i
ar
O
-O
ar
to
O
0
f«.
in
•o
**
o
o
1
•o
CM
ar
to
ar
d>
y
f3
O
to
CO
*o
"5
1
P^
.->
r»-
.0
ro
O
1
"3
ro
•j.
ca
ro
iO
rj
-j
i
r-
0
-
Ul
U)
a-
1
0
co
fO
•-4
-
O
o
CM
ii
in
ar •
O
o
1
f-
CM
CM
CM
CM
ar
o
O
1
CO
5O
co
to
ar
O
f»
o*
CM
m
*
-o
•n.
3
|
-j^
-s..,
<\,
HO
t<\
-2
3
1
ai
CM
-o
M
.0
75
(O
C2,
-o
-
Lu
^-j
'M
O-
Ul
o
JT
O
o
m
r*-
CM
ar
o
0
1
CM
to
co
cc
•-t
• a-
O
O
1
CM
H^
CM
to
to
ar
CD
n»
i>
m
ro
oo
ro
-3
1
CO
fO
-0
.•o
1-5
-5
1
r-
•>.
•o
r^
iO
-5
-5
1
|
—4
•41
1C
ro
JO
<-^
-»
i
fO
-o
•o
CM
Ul
i/:
n
.
^
£
,—.
-o
ar
a-
ff\
O
!»
0
3-
ar
Ul
1/1
ir
-------
f— c-
•r- U) r~-
03 CU
CU 4-> 3 -0
J= C C 0
C O- O E
CD CD d)
l/> C C 1—
C T- •!-
O S- "O CD
•r- 3 r— C
•»-» •*->•!- O
(d U 3 _J
J- (13 CO
OvJ H-> <4~ X
r-l C 3 C CU
* OJ C O i— •
o u >o a,
C S T3 E
tt> O CU O
t— o tn in o
_O (/) eg
rc> o no aa cu
C CD •> $_
CO «=t 3
f> ra _J O
i- CO l/l
r— O
•a s- rcs-r-
OJ >» O «r- 4->
H-> H-> O. J- 03
(TJ •!- CU •*-> r—
E C > CO 3
•i- T- CU 3 U
H-» U S- -O •—
t/> -r— JZ C tf
LL! >» 1/3 r— I CO
D
in
O
O
o
o
*
o
o
o
o
o
o
rM
r»
o
^)
•r-l
"^
o
3
OJ
"3
"3
uj -o
1^
I™"
or, ^
rH **
_ 1
I
a:
r-4
O
sr
-3
1
•o
=r
C
03
sr
-o
o
rH
sr
o
o
1
0
cc
CM
O
CM
sr
0
o
1
CM
p^
CM
in
m
r>
CJ
ft
sr
f*»
•o
rH
2
"
"^
1
•O
CM
ul
O
•^
•T*
t
o-
IX1*
•—4
CM
•Tt
fl
•3
3
1
n
•O
O
CM
irt
ui
t>
1
ut
;n
sr
«-i
O
1
o
•o
CO
sr
o
o
1
CO
0
sr
CM
in
"}
JT *
-J
i
vM
r-l
^
f,
.^
:r*
1
•a
.-*
CM
CM
•"•
fl
O
3
1
*O
•1".
O
•""
*""
to
U)
o
1
o
-o
r—
CM
in
0
o
•o
sr
r-
-r
o
o
1
sr
•o
o
**
if
o
o
1
r-l
f»
O
o
«H
sr
O
7
^u
-o
o
o
sr
.0
-~
TJ
i
r-t
\^>
>>
CM
•n
o
i
'""O
o
p-
(X)
f,
fl
o
^>
1
fl
cr
r^
CM
3
O
O
O
o
1
•o
r-
f»
CO
sr
o
0
1
r-l
O
CM
CM
^
sr
0
0
1
«•<
K*\
Q
CM
CM
sr
o
o
sr
rH
CO
<—
in
f «K
t3
1
|-~
fl
f>
^
fl
•O
1
•M
CM
sr
CM
ar
CO
"3
-^
1
,"™^
in
!M
vO
3
3
sr
i
in
o
-3
I
CM
•H
in
(n
«"'
^
^
o
t
in
fw
0
CM
.j.
O
o
1
*O
f*"%
CM
sr
fl
o
"l
sr
r*
•D
-H
^}
"^
|
*\
3r
T-4
•"
n
r^
i
CM
•J>
•^»
U\
^J
f>
C^
^S
T
^«
—<
c*
fi
0
3
o
i
r-
0
0
1
.0
rH
CM
CM
.f
a
o
I
o
in
o
CM
f>
.^
O
o
1
in
O-
O)
CO
in
f>
1
i^»
CM
U1
rH
'0
3
1
•n
0
»
r-
fi
~5
^
sr
fl
fi
CD
fl
"3
*^
1
rj*
CM
AJ
•>
•-
3
3
D-3
i
o
in
sr
O
i
o
«
.,.
o
o
1
rn
rH
CM
in
sr
O
o
1
CM
|S^
n
o
o
o
r-l
fl
CM
CM
ro
•j
3
1
u^
*°
~
CM
— »
1
sr
in
•~M
^
«
n
r^J
^
1
CM
fl
sr
in
O-
3
16
o
o
1
•c
c>
«-
•sr
0
T*
o
CM
sr
o
o
1
-o
m
r-l
o>
fl
sr
0
o
1
^^
-o
o
(*-
o
o
fl
in
03
30
^
'!?
,.'
3
1
a
CM
r—
*A~<
O
CM
O
^
{••«
j.
D
Og
"•
CM
O
O
i
O1*
CM
CO
;'^
*-*
3
-5T
O
"l
rH
Cv)
M
O
i
r-
G
fi
sr
.,.
O
o
1
CM
rH
O
in
•0
ft
o
o
l
>o
CO
rH
n
"3
JO
JO
sr
«-<
fi
-J
1 .'
-3
I
••H
tO
r-l
-«
^
-I
—1
I
•"1
'*O
cc
-
•M
fT:
_— \
1
.*-)
r»
o
^
r-t
*~
1
rH
~
»
'a
i
rH
in
-
^
O
o
1
CO
o
ft
fl
CM
sr
o
o
1
r*»
sr
CM
sr
rt
a
i
•0
•u
m
-»
—4
•o
~
"^
1
n
C>
T
'-
."M
-7^
t
•r
_ (
rH
rH
•M
"3
— >
1
•O
,r\
-U
•rH
UJ
^
-0
o
sr
3
1
O
in
r-4
0
sr
o
o
1
o
sr
m
o
rH
sr
o
0
1
. sr
CM
in
ft
sr
O
?
CO
O
fO
ft
0*
-I
_7/
*)
1
•J
lv'l
a^
-1
1
o
rH
•>
•il
."1
..-
"V^
T
^
,—•
^
T-*
*
LJ
^~
O
1
JO
.M
-
sr
c^
o
sr
CO
-
sr
o
o
1
r»
«r
o
CM
•sr
o
o
1
r**
O*
o
•o
ft
•sr
b
r-l
O
fN.
•o
0
0
~TJ
"^
1
T'"l
|^-
r»»
»
'M
1
jX,
•fc^
CJ
Hr
-1
.-)
~?
'M
O-
V,
^J
••J
1
o
j.
sr
ro
,-.
3
1
Q
*
u,
'_
JM.
"
'
,'
J
U',
LU
on
•5
t
-3
CX
-------
o
§
o
o
o
I
CM
o
•
ea
i
§
•9
03
CM
O O,
o 3
i t
S S
P. P- «n
a
o
O
CO
o
o
C3
CM
•O
• • •
1=4 •"! CM
co
o
ar
•
CM
Q o
3 o
I i
o "•
CM »•«
O
in
co
tn
-o
a •
m **
o
o
1*1
!*\
•O
CM «*
CS, O
o cs
ft, CM
in o
sr co
o -H
• •
CM •*
O
e
o
o
«r
o
o
i
o
•
CM
O
0
1
1*1 J*
O O
& o
*4 O
f. «=«
w co
m o
m »
« • •
,*«"•«•
O
o
I
o
o
CO
•
CO
o o
o a
i i
co
tn
a
00
(M
CO
O
a
CM
i
in
e
o
i
o
CM
•«*
O
O
O
e
O O
o a
d CM
CO O
«« o
CM *«
ar eg
O
O
1
CO
o
o
1
CM «M
O O
O O
CM *
•O CO
•
CM
•
CM
•0
CM
in
CO 4->
-E C
+-> (t)
r— C
C Q- (O
c T-
o s.
•i- 3
(O O
s- '
(O -r-
-|_J (J
CO -r-
s=
O
T3 CO
ai c
to O
« •«-
ca -t-J
fO
S» U
i —
+->
•••a
O> O
s- s
co S-
C 3
-i- CO
+J O
i. Q.
(O X
s: LU
e
a
o
9
a
it
o
o
o
3
CM
O
O
o
o
a
o
3
o
o
I
CO
o
9
•o
CM
o
o
at
CM
in
O
O
I
CO
«•*
CM
0
co
in
»
»t
o
i
a
CO
I
O
ar
09
CO
O
o
CO
•o
o
a
I
m
S
ca
i
in
ru
CM
i
Sf
CM
O
a
i
co
O
p-
a
CM
o
CM
«
O
o
Q
o
CM
O
O
•O
a
O O
O O
I t
GO f\
O
CM
I
in
in
o
CM «4
t*\ r?i
O O
3 3
ar CO
O I-
P- CM
O
O
I
O
o
CO
o
o
I
o
o
•o
CO
o
t
•o
*v
n
•3
t
o
o
9
CM
o
a
i
o
in
a
O
O
I
m
CM
9
Sf
fO
o
i
•o
o o
o o
I I
p. o>
«• co
CO O
r* co
• ••
in CM
O
o
I
1
;M
O
a
i
o
in
e
CM 'O
O 3
•1 >O
a •
O ar
CM «M
O O
? ?
o
o
I
•a
CO
p»
a
n
o
o
in
-o
ni
•
•o
o
o
I
PV
o
o
I
co
co
•o
o
*
CM
O
O
a"
c
o
cp
f*\ K>
O O
O O
a i
•o o
<*l ro
p» »
a •
tn CM
m
o
o
n
•o
b-
a
CM **>
O O
O -5
rr\ CM
•O !>
O O ft «H rt "4
a • a 9 , a
CM »H «^ i-( ^4 . CO
o
o
^
a
M
CM
O
1
1-t
ft
^4
•O
a
O
.>J
O
1
CM
ar
JO
sr
a
ut
rM
rs
o
i
m
—i
it
a
a?
.M
O
1
S)
ft
r*»
p-
a
••*
--M
O
3
1
O
CO
CO
e
CM
CM
O
0
1
in
a-.
rt
O
a
VM
CM
O
1
CM
ft
P-
p-
a
•O
CM
•3
O
1
CM
v4
O*
a
jy
.^
•3
I
0
.-1
i«4
3
e
i-»
CM
O
1
CO
•3 .
ar
ar
e
0
5
o
1
t3
CO
ar
C\J
a
»"«
3
CS
1
00
vH
v^
p-
a
-O
•:\j
rs
o
i
o
^
/7»
ft
a
f?
0 0
0 O
1 1
ft p*
p« at
«-( 1-1
a a
CO O
CM
rt
i
••>*
•o
•o
a
ut
co
<3,
I
.t
CM
CM
I
ar
-JO
a
CM
?
CM
O
CM
O
Lit
CM
a
CM
I
C3
I i
CM
e
O
O
CM
P.
CO
CM
M
p»
•O
CJ
I
-o
ut
a
I
r—
cvi
n
I
O
rj
I
ar
•JO
a-
-3
I
CVJ
O U
•3 o
I I
-o g
ft at
CO CM
«1 CO
CM
O
CM
sr
I
ut
at
CM
o
•o
CM ^i
c> o
o o
I I
^* '-tl
co ij-
•O «4
-3
9
•O
p-
9
O
o
I
o
-o
CM
Ut
a
CO f^>
I I
rt t>
CO ^
O
I
O
=0
•-J
I
•o
!^\
oO
a
O
3-
JO
•O
OvJ
O •-•
M O
3 fe
JS
I/
in
3
«/•
3
a-5 ^B
3D-37
3
Ul
UJ
UJ
UJ
LJ
l/i
Ul
I/'
i/i
-------
o
n
o
o
o
o
•
.o
o
CO
JO
(O
o
co 4-> -a to
j= c i- HH
4-> ro ro •—
r—-a
c a. c E
cn*j co
C -r-
O S- C C71
•r- 3 O C 10
+J 4J O C
« O-O _1 O
4.) lj_ CO X -f->
C 3 (O CO ra
CO C 00 i—r—
u ro a. 3
c S • E <->
O rO
w> o
O ro 4-> CO
•r- r— V) O i—
E CD CO I- CO
CO S 3T3
(O O O
eC It- 0>
O S- •— >»
T3 3 rO CO
CO >>J3 -r-r—
4J ^J to $-•—
ra -i- c 4J ro
E C T- to >
•r- •!- 4-> 3
4-> O S. T3T3
W -i- rO C C
UJ > S »-H «
O
O
o
o
to
o
o
o
•
o
CM
O
o
o
D
c-
CM
UJ
O
I/! -C
a:
a
rO
O
•x»
CM
.•o
O
•3
0
in
ro
«-i
CM
O
O
|
to
ro
T^
CM
to
fO
O
O
1
o
ar
a-
to
tn
CM
o
o
1
,T
04
r-
~*
**
.,
-3
7>
1
O-
>
ilO
o
CM
O
-»
1
<->
CO
CM
a"
.VJ
O
->
1
.T
1C
T
*»"
sr
V)
O
OO
Jt
o
o
o
t-l
0
r-
o
to
o
o
1
•0
r»»
CM
**•
CM
to
0
O
1
<>
CM
•O
in
ro
ro
O
f
0
tn
«-i
•"•
"
CM
r-j
.3
1
t>
,>»
CM
;M
CM
0
^
1
^y
O
^
o
ro
CM
O
0
1
r-
m
0-
CM
3
l/»
0
0
1
o
o
o
o
1
CM
-O
in
O
o
|
o
in
to
•"•
CM
to
0
O
1
to
CO
if\
•^
to
ro
O
?
«-«
r*-
ro
ar
«-
v\l
•3
-3
1
£
«~|
-O
-
CM
••3
3
1
•O
,_,
fO
co
-
CM
;3
0
1
r-
-o
73
-O
«
3
4/1
a*
O
1
in
«-4
O
lit
0
0
CM
CO
oo
*°.
fO
a
o
l
in
0
•0
0
*
to
0
a
i
CM
0
ar
ar
-
to
f3
O
ar
O
a"
O
ro
.0
;••*
.-}
1
3
CM
O
33
CM
O
^
1
-3
CM
CM
CM
-
iO
*75
'O
i
r-
oo
O
CM
0-
3
3
O
1
ar
•0
ar
CO
.•o
O
o
ar
ar
"
0
o
1
in
00
•o
fO
rH
to
O
o
1
r-
-o
r*
*"•
CM
rO
O
Cl
ro
r*
ro
CO
ro
CM
.-"I
"3
1
to
IO
ra-
ft
!M
O
*•}
I
3Q
»-«
>n
iii
•-4
01
o
tD
|
0
co
CM
rv:
^
3
ar
O
1
O
•o
o
1
co
-o
»-l
1
o
0
1
ar
CM
•*
0
10
0
o
1
tH
*"O
o
00
CM
to
t5
3
CO
•-(
l>-
3s
ro
•M
»-•,
•^i
1
u]
«H
IO
iM
iM
C'3
15
1
•n
r—
Ovl
t>
-
-M-
«*
~3
1
.—
•>
to
-1
^
3
)
O
•O
to
o
o
1
o-
in.
to
to
to
o
o
1
a*
ar
*^
oo
to
rO
"3
?
a*
to
CO
ar
^-
N
-^
5
i
r-
u^
'-^
sr
CM
~
t3
1
o
ro
O
«-4
ar
CM
CJ
't^
1
^D
fO,
7)
-O
CM
3
Z
(O
"3
O
1
ar
b
o
i
-o
C\i
ro
O
O
1
CO
oo
0
•o
~
to
0
o
1
o
r««
to
o»
ro
ro
O
••3
r»
«H
ar
ar
0
N
•3
D
1
(^
r^
'*•
CM
CM
O
o
1
•o
•o
•o
r~
*
CM
O
75
1
ro
fi
r~
-'
ro
3
(0
C
1
ar
p-
-a
CM
ro
c
0
1
co
T-t
CM
CO
ro
to
o
o
I
in
o-
in
o
m
to
o
o
i
in
to
flO
o
ar
CM
n
O
1
3»
O
O
ar
rl
:M
•"~'
^5
1
cr
^
-"
•O
Al
O
""^
1
*o
-o
ar
•*•
»
Cvj
T*
*7)
i
o
p-
•-1
sr
(0
E
j[
CM
U1
O
•r-l
O
O
1
O
to
o
o
1
>H
c
o
CM
to
o
o
1
CO
in
o
•*•
ar
to
^
a
l
ro
in
ro
•23
*
••j
*~*
••71
i
t>
•^
•^
CM
T/1
~>
1
3D
i»»
'-^
rj
fO
.•M
/-\
*^
1
-.3
f
0
0-
O-t
UJ
2
i
O
r-
c\.
ro
1
to
OC
CM
^-
CM
O
O
1
sf
CM
UP.
to
^
0
o
(1
»H
ro
CM
•O
u-v
CM
^
P
i>j
u>
ar
•^
«
•4
•
S
1
sr
-'"
.M
";M
••".
^3
1
J"
O
~3
if
m
,M
» ^
""?
1
Tw
-
ex
,M
IM
7>
^
1
iT
p^
C>J
,0
rj
~ t
b
i
fi
•a"
-o
ro
UJ
5?
UJ
T
iO
o
tJ
CM
O
0-
to
o
o
1
,.4
CM
r~
CM
to
o
o
1
ro
•o
-0
sr
ar
ro
n
ut
CM
—4
ro
*
•M
~
1
1
-0
•n
.1
,o
•M
"?.'
*^
1
O
t>
o
l^1
u-v
CM
*™^
*^
1
•O
sr
o
'"'
'"•
lu
.•o
.M
iO
c.
ii
l*~
•4J
to
o
o
1
^,
f>.
ar
ec
CM
to
O
O
1
-o
r*
o
CO
to
ro
«*3
"i
-o
o
•H
9*
30
iM
•^^
~3
1
7?
f\r
*
^
•~
*^
i
j*
•o
-J
r*
•o
CM
O
•~^1
1
n
r-t
0
£*„*
20
UJ
UJ
"1
.1
r-
O
1
^-4
CM
Ul
to
o
1
o*
o-
r"
c
ro
ro'
O
O
I
>-Q
CM
r-.
"^
ar
ro
P
U)
3
iM
*
M
•••^
•j
1
a
cr
M
., »
*^j
1
j
_
•O
>o
4*
w
"%
TV
1
ro
0-
-c
UJ
I/I
1
o-
o
IJ
•o
CM
ro
o
0
1
to
CO
, rH
»o
o
•o
1
ah
in
o
*-"
to
ro
O
'l
tn
o
p-
-0
CM
-;
—)
1
in
in
r-.
04
iM
T5
-I
1
^
«-*
"M
*-•
if
^
• ^
"5
1
O
Cv.
*^
O
a-
UJ
tn
to
D-38
-------
S E
E O O (/)
O
OJ 4-> cn x •r-
•4—} fO »r~* .•"• fl3
r"" "X^J f^)L r" •
E O- i — £= 3
•i- -t- 0 0
CD =J C_3 r—
t/> E co rd
c T- oj c_j
0 S- E 0
•r- 3 O S- i —
4-> 4-> 3 OJ
re O T3 O T3
s- re oj oo o
4-» 4- 21
LO C 3 re r-
i—i oj E co re >>
• o re T- oj
Q c s: " s- •—
o =c -i-> »—
oj c_> t/> ^» i/> re
re 'r- r— t/) E "O
f— E CD OJ t-H E
oj ^ re
i/> re T3
s_ « s_^~.
«=t M- en re l —
o s- -a _i
T3 =J E O
OJ >» .0 re co
4-3 4-* (/) -f-^ *— 1
re -i— E co — •
EC" »p—
L_ f^
•r-" "i— i * f*- f~
-4-> O &. I/) i-
i/) -r— re re oj
UJ > S 3 1 —
'.J
U
z
^f
•^
r-
t/l
•-*
O
ct
o
T° -)
C r-
O CM
• in
U\ CM
c:
i
O fl
C5 • ro
3 0
• CM
O
o
1
o o»
{^ ^^
O *Q
• sr
f> M
o
i
0 C
0 «-l
?"i ?1
• o
CM m
.M
1
T> CM
Ti CM
T-< «H
•"M
^2
5
1
T Is-
'"* '!>
D in
. v>
CM
O
^3
1
*"^ O
-> o
-^ -o
• sr
O-
T-1
O
o
1
r f~i
^ f*.
m ^j
« M
•"
IA.
••o
o
1
•X)
CM
fO
o
-3
1
ar
O
n
o
o
CM
O
O
o
o
1
KQ
0
-o
o
'fl
o
-3
O
CM
O
in
•M
•*^
->
1
O
~~-
o*
•-D
^
l\i
*m
."•>
i
"^j
c?*
o
fO
^
rvj
O
*3
1
•M
CO
••M
"•
O
3
VI
l/J
ar
1
Cf)
03
1
in
a
•o
iii
o
0
1
in
*H
o
o
i
Q
T-I
sr
o
a
i
sr
•0
33
-M
"7^
O
1
r»»
T^
^
s-n
-N
nj
*^
-^
i
•H
-o
^J
o
fO
:xj
i*5
.73
i
cr
Tl
r—
r*.
Ui
3
V)
-3T
O
1
U)
«•
a
0
i
r-
0
o
0
1
T-I
CM
in
o
a
i
31
"•
b
o
«r
T-I
CO
*
r>j
o
l
•U
»-1
iO
J1
-
CM
*3
-3
1
j^
ar
n
ar
<>,
,M
*"*
b
i
CO
r-
CM
r-
CM
3
1/1
2
sr
O
1
CM
ar
CO
ro
O
O
1
r-
o
o
o
o
1
•o
r-
r-
O
O
ar
CM
CO
CM
O
O
o
CO
in
»
.M
•3
1
r-
o*
i sj
ar
CM
CM
O
O
1
U"\
ij-
CxJ
C\i
fO
IM
0
1
oc
Is-
0
r,
3
o
1
CO
oc
o
o
1
co
0
0
sr
00
o
o
o
1
flO
CM
O
•ts
ar
oo
=r
M
^
->
1
X**
cr,
-1!?*
j-l
••n
'.M
rr
-s
i
M
Is-
T^
" "'
»
CM
T!)
-^
1
i'O
•T^
Is-
I'l
sr
3
3
O
1
r-
o
o
1 ;
-O
P-
co
o
1
O
ar
in
o
o1
1
in
T5
0
f,
fl
n
o
CM
«o
0
0
•M
.••^j
-3
1
«•»
CD
~^O
J5
•°'
r-d
^>
-^
I ;
~^
^
•o
>"*
eft
^
•*"3
^3 •
1
-o
c
sr
CM
a '
3
2 ;'
p
1
O-
b
3
1
-O
O-
oo
o
o
1
•o
If
CO
fl
o
o
•Q
»
CM
O
sr
O
-
•j
o
1
U
T-I
sr
-0
*
CM
0
1
^*
T-i
T-4
at
o«
Csl
^
O
1
TM
!!••.
X>
O-
Oi
3
g
O
•5
1
-O
CM
O
o
1
-o
fl
0
0
o
CO
CM
in
o
o
1
CM
o
in
CM
o
o
r-
Sf
-o
~<
^*M
~ •,
.3
1
-3
CV»
^\J
r*
o.
r-1
r^
•3
1
^4
CO
O
c;
~
T-1
.^
"3
1
r-
o
CM
^
~
Z
O
1
c
T^
ar
03
CM
CM
0
o
1
•o
o
o
CM
0
0
o»
0
ar
m
CM
o
n
CM
CM
-
'."J
.--;
T
1
•^
sr
OJ
rvi
U1
.-J
••7
— ^
1
3-
151
,»»
-0
•-
TH
-V
-s
1
f\ '
*-+
r-t
""^
-
UJ
^
o
J
CO
co
r?
i
CO
cv
b
t
o
ft.
0
fl
o
0
1
•~o
sr
•o
iM
CM
-
•M
-•
'3
i
-O
ft
O*
OJ
n
i^
^j-
"•
I
•*"l
r^»
•J»
sr
0.
_
r.
i
o
T-I
-1
-
UJ
K
t
r-
O
1
d>
sr
t>
o
0
1
in
-o
o
•0
CM
O
0
1
o
sr
b
i
CM
T-I
UI
O
"•J
. -j
1
^r»
^i
JO
•i
^
N
•^
^
l
o
|V.
ivj
T-I
r>-
^
•'3
1
fl
•AJ
|0.
..,'
T-I
UJ
UJ
1
T!
=r
I
ar
0
CM
fl
O
O
1
in
0
CM
o
o
01
o
u>
CM
O
O
ar
o
CM
•H
•\J
••*
~
1
t^
<-l
t^
--r
O
_,
~*
—l
1
^*
-r
-O
«-'
-
»-l
;*^
*">
i
•^
~
fi
o
r*
u,
fi .^ .-•>
3 3 °
/ / ^
O -r> "J
•0 Cf- C
t^J cr C
fl CO l*»,
O C: O
000
1 t 1
o CM m
CM a- o
fl CM O
0 O 0
O O 0
1 1 1
"• o a>
CM O* sf"
in ~o o
000
ft fl CM
fO f*O f\
o o o
O o O
1 1 1
in r* in
ft fi .0
ar -0 "n
fi ^3 in
ar ar f»
CM ^4 fl
f^ O **^
? ? ?
»H « 2
^ »-4 nj
«-« t-t oo
••M .M M
" ^ 13 'TJ
i i i
\i J ,->
Is- co . -•
r*- ^r Ti
-il i— tr\
o o sr
.-> -t .j
—; - «; .' >
— ^ ^j -^
1 1 1
*; o ••*»
•o ^ -o
o ~r CM
Cu „ fn
^< r-i m
TH ,-1 T-1
-5 -2 ~
"3 '"S "^
i i F
iM O 'M
•*•' u\ *^
C'J vj- i".
C\J T-I *-•
UJ uj UJ
UJ ' l/">
en
CO
i
O
D-39
-------
D.5 ANALYTICAL UNCERTAINTIES APPLICABLE TO THE CALCULATIONS OF PUBLIC
HEALTH RISKS CONTAINED IN THIS APPENDIX
D.5.1 The Unit Risk Estimate
The procedure used to develop the unit risk estimate is described in
reference 2. The model used and its application to epidemiological data
have been the subjects of substantial comment by health scientists. The
uncertainties are too complex to be summarized sensibly in this appendix.
Readers who wish to go beyond the information presented in the reference
should see the following Federal Register notices: (1) OSHA's "Supplemental
Statement of Reasons for the Final Rule", 48 FR 1864 (January 14, 1983); and
(2) EPA's "Water Quality Documents Availability" 45 FR 79318 (November 28,
1980).
The unit risk estimate used in this analysis applies only to lung
cancer. Other health effects are possible; these include skin cancer,
hyperkeratosis, peripheral neuropathy, growth retardation and brain
dysfunction among children, and increase in adverse birth outcomes. No
numerical expressions of risks relevant to these health effects is included
in this analysis.
Although the estimates derived from the various studies are quite
consistent, there are a number of uncertainties associated with them. The
estimates were made from occupational studies that involved exposures only
after employment age was reached. In estimating risks from environmental
exposures throughout life, it was assumed, through either the relative-
risk model (1) or the absolute-risk model (2), that the increase in the age-
specific mortality rates of lung cancer was a function only of cumulative
exposures, irrespective of how the exposure was accumulated. Although this
assumption provides an adequate description of all of the data, it may be
in error when applied to exposures that begin very early in life. Similarly,
the linear models possibly are inaccurate at low exposures, even though they
provide excellent descriptions of the experimental data.
The risk assessment methods employed were severly constrained by the
fact that they were based only upon the analyses performed and reported by
the original authors—analyses that had been performed for purposes other
D-40
-------
than quantitative risk assessment. For example, although other measures of
exposure might be more appropriate, the analyses were necessarily based upon
cumulative dose, since that was the only usable measure reported. Given
greater access to-the data from these studies, other dose measures, as well
as models other than the simple relative-risk and absolute-risk models, could
be studied. It is possible that such wide analyses would indicate that other
approaches are more appropriate than the ones applied here.
D.5.2 Public Exposure
D.5. 2.1 General
The basic assumptions implicit in the methodology are that all exposure
occurs at people's residences, that people stay at the same location for 70
years, that the ambient air concentrations and the emissions which cause
these concentrations persist for 70 years, and that the concentrations are
the same inside and outside the residences. From this it can be seen that
public exposure is based on a hypothetical premise. It is not known whether
this results in an over-estimation or an underestimation of public exposure.
D.5.2.2 The Public
The following are relevant to the public as dealt with in this analysis:
1. Studies show that all people are not equally susceptible to cancer.
There is no numerical recognition of the "most susceptible" subset of the
population exposed.
2. Studies indicate that whether or not exposure to a particular
carcinogen results in cancer may be affected by the person's exposure to
other substances. The public's exposure to other substances is not
numerically considered.
3. Some members of the public included in this analysis are likely to
be exposed to inorganic arsenic in the air in the workplace, and workplace
air concentrations of a pollutant are customarily much higher than the
concentrations found in the ambient, or public air. Workplace exposures
are not numerically approximated.
D-41
-------
4. Studies show that there is normally a long latent period between
exposure and the onset of lung cancer. This has not been numerically
recognized.
5. The people dealt with in the analysis are not located by actual
residences. As explained previously, people are grouped by census districts
and these groups are located at single points called the population centroids.
The effect is that the actual locations of residences with respect to the
estimated ambient air concentrations are not known and that the relative
locations used in the exposure model may have changed since the 1980 census.
However, for the population sectors estimated to be at highest risk, U.S.
Geological Survey topographical maps were checked to verify that people did
live or could live in locations near the sources as modeled predictions
estimated. Maps in certain instances were old and the possibility could
not be excluded that additional areas near sources have been developed
since publication of the maps.
6. Many people dealt with in this analysis are subject to exposure to
ambient air concentrations of inorganic arsenic where they travel and shop
(as in downtown areas and suburban shopping centers), where they congregate
(as in public parks, sports stadiums, and schoolyards), and where they work
outside (as mailmen, milkmen, and construction workers). These types of
exposures are not numerically dealt with.
3.5.2. 3. The Ambi ent Ai r Concentrati ons
The following are relevant to the estimated ambient air concentrations
of inorganic arsenic used in this analysis:
1. Flat terrain was assumed in the dispersion model. Concentrations
much higher than those estimated would result if emissions impact on elevated
terrain or tall buildings near a plant.
2. The estimated concentrations do not account for the additive impact
of emissions from plants located close to one another.
3. The increase in concentrations that could result from re-entrainment
of arsenic-bearing dust from, e.g., city streets, dirt roads, and vacant lots,
is not considered.
-D-42
-------
4. Meteorological data specific to plant sites are not used in the
dispersion model. As explained, HEM uses the meteorological data from the
STAR station nearest the plant site. Site-specific meteorological data
could result in significantly different estimates, e.g., the estimates of
where the higher concentrations occur.
5. In some cases, the arsenic emission rates are based on assumptions
rather than on emission tests.
D-43
-------
D.6 References
1. National Academy of Sciences, "Arsenic," Committee on Medical and
Biological Effects of Environmental Pollutants, Washington, D.C. , 1977.
Docket Number (OAQPS 79-8) II-A-3.
2. Health Assessment Document for Inorganic Arsenic - Final Report EPA-600/
8-83-021F March 1984, OAQPS Docket Number OAQPS 79-8, II-A-13.
3. U.S. EPA, et.al., "Environmental Cancer and Heart and Lung Disease,"
Fifth Annual Report to Congress by the Task Force on Environmental Cancer
and Health and Lung Disease, August, 1982.
4. OAQPS Guideline Series, "Guidelines on Air Quality Models." Publication
Number EPA-450/2-78-027 (OAQPS Guideline No. 1.2-080).
5. Systems Application, Inc. , "Human Exposure to Atmospheric Concentrations
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-1/250-2.
D-44
-------
TECHNICAL REPORT DATA
(Please read Instruction! on the reverse before completing)
EPA-450/3-83-011b
RECIPIENT'S ACCESSION NO.
TITLE AND SUBTITLE
Inorganic Arsenic Emissions from Glass Manufacturing
Plants - Background Information for Promulgated Standarc
.REPORT DATE
May 1986
i. PERFORMING ORGANIZATION CODE
AUTMOR(S)
8. PERFORMING ORGANIZATION REPORT NO.
•ERFORMING ORGANIZATION NAME AND ADDRESS
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
68-02-38/6
2. SPONSORING AGENCY NAME AND ADDRESS
DAA for Air Quality Planning and Standards
Office of Air and Radiation
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final
14. SPONSORING AGENCY CODE
EPA/200/04
5. SUPPLEMENTARY NOTES
6. ABSTRACT
Standards to control emissions of inorganic arsenic from glass manufacturing
plants are being promulgated under the authority of Section 112 of the Clean Air Act
This standard applies to new and existing glass manufacturing plants.
This document contains a summary of public comments, EPA responses, and a
discussion of differences between the proposed and promulgated standard.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS C. COSATI Field/Group
Air Pollution
Glass Manufacturing Plants
Pollution Control
Emission Standards
Arsenic
Air Pollution Control
13-B
tie. DISTRIBUTION STATEMENT
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
148
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (Rev. <_77)
PREVIOUS EDITION IS OBSOLETE
-------
------- |